Bryn Mawr College 
Monographs7 
REPRINT SERIES, Vol. XII 
- 
Contributions from the Psychological Laboratory 
BRYN MAWR, PENNA. 
March, 1922 BRYN MAWR COLLEGE MONOGRAPHS 
The Bryn Mawr College Monographs are issued in two series: the 
Monograph Series, containing studies that appear here for the first time, 
and the Reprint Series, containing reprints of articles that have appeared 
in other journals. 
Of the Reprint Series, volumes I, II, III, V, VI, VII, IX, and X contain 
contributions from the biological laboratory; volumes IV and VIII, contri- 
butions from the mathematical and physical departments, and volumes XI 
and XII from the psychological laboratory. Bryn Mawr College 
Monographs 
REPRINT SERIES, Vol. XII 
Contributions from the Psychological Laboratory 
BRYN MAWR, PENNA. 
March, 1922 Table of Contents 
I. Miscellaneous Experiments on the 'Efficiency of the Eye Under 
Different Conditions of Lighting. By C. E. Ferree and 
G. Rand. 
(Reprinted from Ophthalmology, XII, July, 1916.) 
II. A Resume of Experiments on the Effect of Different Conditions 
of Lighting on the Eye. ' By C. E. Ferree and G. Rand. 
(Reprinted from the Annals of Ophthalmology, XXV, July, 1916.) 
III. A Simple Daylight Photometer. By C. E. Ferree and Gertrude 
Rand. 
(Reprinted from the American Journal of Psychology, XXVII, 
July, 1916.) 
IV. A Substitute for an Artificial Pupil. By C. E. Ferree and Ger- 
trude Rand. 
(Reprinted from the Psychological Review, XXIII, September, 1916.) 
V. A New Method of Heterochromatic Photometry-A Reply to 
Dr. Johnson. By C. E. Ferree and Gertrude Rand. 
(Reprinted from the Psychological Review, XXIV, March, 1917.) 
VI. A Note on the Needs and Uses of Energy Measurements for Work 
in Psychological Optics. By C. E. Ferree and Gertrude 
Rand. 
(Reprinted from the Journal of Philosophy, Psychology and Scien- 
tiftcfMethods, XIV, August, 1917.) 
VII. Radiometric Apparatus for Use in Psychological and Physio- 
logical Optics. By C. E. Ferree and Gertrude Rand. 
(Reprinted from the Psychological Monographs, XXIV, 1917, 
/ No. 103.) 
| NVfl. The Selectiveness of the Achromatic Response of the Eye to 
Wave-length and its Change with Change of Intensity of 
Light. By C. E. Ferree and Gertrude Rand. 
(Reprinted from Studies in Psychology, Titchener Commemorative 
Volume, Worcester, Mass., 1917.) 
IX. Some Areas of Color Blindness of an Unusual Type in the Per- 
ipheral Retina. By C. E. Ferree and Gertrude Rand. 
(Reprinted from, the Journal of Experimental Psychology, II, 
August, 1917.) X. The Power of the Eye to Sustain Clear and Comfortable Seeing 
with Different llluminants. By C. E. Ferree and G. Rand. 
{Reprinted from the American Journal of Ophthalmology, I, April, 
1918.) 
XI. The Inertia of Adjustment of the Eye for Clear Seeing at Different 
Distances. A Study of Ocular Functions with Special Refer- 
ence to Aviation. By C. E. Ferree and Gertrude Rand. 
{Reprinted from the American Journal of Ophthalmology, I, Novem- 
ber, 1918.') 
XII. A Note on Vision-General Phenomena. By C. E. Ferree and 
Gertrude Rand. 
{Reprinted from the Psychological Bulletin, XV, December, 1918.) 
XIII. Chromatic Thresholds of Sensation from Center to Periphery of 
the Retina and Their Bearing on Color Theory. Part I; 
By C. E. Ferree and Gertrude Rand. 
{Reprinted from the Psychological Review, XXVI, January, 1919.) 
XIV. Chromatic Thresholds of Sensation from Center to Periphery of 
the Retina and Their Bearing on Color Theory. Part II. 
By C. E. Ferree and Gertrude Rand. 
{Reprinted from the Psychological Review, XXVl, March, 1919.) 
XV. The Absolute Limits of Color Sensitivity and the Effect of 
Intensity of Light on the Apparent Limits. By C. E. Ferree 
and Gertrude Rand. 
{Reprinted from the Psychological Review, XXVII, January, 1920.^ 
XVI. The Limits of Color Sensitivity: Effect of Brightness of Preex- 
posure and Surrounding Field. By C. E. Ferree and Ger- 
trude Rand. 
K {Reprinted from the Psychological Review, September, 1920.) 
XVII. The Campperimeter-An Illuminated Perimeter with Cam- 
pimeter Features. By C. E. Ferree and G. Rand. 
{Reprinted from the Transactions of the American Ophthalmological 
Society, 1920.) 
XVIII. Factors which Influence the Color Sensitivity of the Peripheral 
Retina. By C. E. Ferree and G. Rand. 
{Reprinted from the Transactions of the American Ophthalmological 
Society, 1920.) 
XIX. A Note on the Selectiveness of the Achromatic Response of the 
Eye to Wave-length and its Change with Change of Intensity 
of Light. By C. E. Ferree and G. Rand. 
{Reprinted, from the Psychological Bulletin, XVII, April, 1920.) XX. A Method of Standardizing the Color Value of the Daylight 
Illumination of an Optics Room. By C. E. Ferree, G. Rand 
and I. A. Haupt. 
{Reprinted from the American Journal of Psychology, XXXI, 
January, 1920.') 
XXI. An Apparatus for Determining Acuity of Low Illuminations, for 
Testing the Light and Color Sense and for Detecting Small 
Errors in Refraction and in their Correction. By C. E. 
Ferree and Gertrude Rand. 
{Reprinted from the Journal of Experimental Psychology, III, 
February, 1920.) 
XXII. The Use of the Illumination Scale for the Detection of Small 
Errors in Refraction and in their Correction. By C. E. 
Ferree and Gertrude Rand. 
{Reprinted from the Journal of Experimental Psychology, III, 
August, 1920.) 
XXIII. A Study of Ocular Functions with Special Reference to the 
Lookout and Signal Service of the Navy. By C. E. Ferree, 
G. Rand and D. Buckley. 
{Reprinted from the Journal of Experimental Psychology, III, 
October, 1920.) MISCELLANEOUS EXPERIMENTS ON THE EFFICIENCY 
OF THE EYE UNDER DIFFERENT CONDITIONS 
OF LIGHTING. 
C. E. Ferree and G. Rand, 
BRYN MAWR COLLEGE. 
As one feature in the work of a preceding paper ("The Efficiency 
of the Eye Under Different Conditions of Lighting." (Ophthalmol- 
ogy, Vol. X, July, 1914, p. 622) we undertook to determine the most 
favorable intensities for the three types of lighting we had selected 
for investigation-direct, semi-indirect and indirect, and the effect 
of varying intensity with the particular grouping of distribution fac- 
tors* represented in each case. This work was completed for the 
direct and semi-indirect systems, but not for the indirect. In the 
present paper results will be given for a similar series of experi- 
ments pertaining somewhat more broadly to the hygienic employ- 
ment of the eye. 
The tests are made in the same room, with the same fixtures, 
and in general with the same conditions of installation and meth- 
ods of working as were employed in the work of the preceding 
paper. To secure the various degrees of intensity needed, tungsten 
lamps of different wattages were used. The series began with 25- 
watt lamps and included 25, 40, 60 and 100-watt lamps. The 
results of these experiments are given in Chart 1. In this chart 
are also included for the sake of comparison graphic representa- 
tions of the results obtained by a similar variation of intensity for 
the direct and semi-indirect systems. In drawing conclusions from 
these results the effects on the eye should of course be correlated 
with the illumination effects produced. For a full specification 
of these effects, also those treated in the next paragraph, see Trans- 
actions of the Illuminating Engineering Society, 1915, X, pp. 
♦The distribution factors are evenness of illumination, evenness of sur- 
face brightness, diffuseness of light and angle at which the light falls on 
the work. 
(REPRINTED FROM OPHTHALMOLOGY, JULY, 1916) 2 
Ferree and Rand. 
Showing a comparison of the effect on visual efficiency or power to 
sustain clear seeing of varying the intensity of light for the four in- 
stallations of lighting used: the indirect, semi-indirect and direct sys- 
tems, 8 lamps; and the direct system, 16 lamps.1 
CHART I.-INTENSITY SERIES. 
Lighting system: Semi-indirect 
Foot-candles 
f 
Hori- 
Verti- 
Watts 
Volts 
zontal 
cal 
45° 
A 
200 
107 
1.6 
0.45 
1.15 
B 
200 
110 
1.72 
0.484 
1.29 
C 
320 
107 
2.2 
0.58 
1.52 
D 
320 
110 
2.31 
0.62 
1.61 
E 
480 
107 
3.3 
0.94 
2.4 
F 
800 
107 
6.8 
1.82 
4.5 
X 
760 
107 
5.8 
1.45 
4.0 
Lighting system: Indirect 
Foot-candles 
Hori- 
Verti- 
A 
Watts 
Volts 
zontal 
cal 
45° 
A 
200 
107 
1.33 
0.39 
0.87 
B 
320 
107 
1.7 
0.49 
1.08 
C 
480 
107 
3.0 
0.765 
1.97 
D 
800 
107 
5.2 
1.36 
3.5 
Lighting system: Direct (8 lamps) 
Foot-candles 
r 
Hori- 
Verti- 
Watts 
Volts 
zontal 
cal 
45° 
A 120 
107 
0.64 
0.32 
0.49 
B 200 
107 
1.16 
0.45 
0.85 
C 320 
107 
1.97 
0.65 
1.39 
D 480 
107 
2.6 
1.02 
2.0 
Lighting system: Direct (16 lamps) 
Foot-candles 
Hori- 
Verti- 
Watts 
Volts 
zontal 
cal 
45° 
A 240 
107 
1.23 
0.54 
0.935 
B 365 
107 
1.6 
0.6 
1.33 
C 400 
107 
1.86 
0.8 
1.46 
X 880 
107 
4.2 
1.41 
2.6 Efficiency of Eye and Lighting. 
3 
434-4-12 and 469-473. Foi' the semi-indirect installation it will 
be seen that the eye fell off heavily in the power to sustain clear 
seeing for all intensities with the exception of a very narrow range 
on either side of 2.2 foot-candles measured at the point of work 
with the receiving test plate of the photometer in the horizontal 
plane. (In lighting practice 5 foot-candles is usually recommended 
as the value to be given to this component of illumination for 
ordinary work.) For the direct installation no intensity could be 
found for which the eye did not.lose a great deal in power to sustain 
clear seeing as the result of work. For the indirect installation, 
however, it was found to be possible to use a comparatively wide 
range of intensities without causing the eye to suffer any consid- 
erable depression of functional power as measured by the test. 
As was the case for the semi-indirect reflectors used in the work 
of the preceding paper, socket extenders had to be used with the 
25 and 40-watt lamps. That is, without the extenders these lamps 
owing to their smaller size came so low in the reflector as to change 
the distribution effects given by the reflectors. For example, with- 
out the extenders for these shorter lamps, the spot of light on the 
ceiling was made smaller and correspondingly more brilliant. It 
was considered to be a point of interest in relation to the general 
problem to determine whether this comparatively small change in 
illumination would cause any difference in the eye's ability to hold 
its power to sustain clear seeing. A comparison of the results for 
the indirect reflectors with and without socket extenders is shown 
in Chart II. 
Also in addition to the work on the distribution series reported 
in the previous paper it was decided to make a test of the effect 
on the eye of position in the room for the three systems of lighting 
for one of the intensities of light employed. Accordingly four rep- 
resentative positions in the median line of the room were selected: 
positions at which respectively six, four, two, and no lighting units 
were in the field of view. This variation of position at which the 
observation was made accomplishes two purposes. (1) It gives 
a more representative idea of the difference in the effect on the 
eye of the three types of lighting employed; and (2) it shows the 
effect of varying the number of surfaces in the field of view pre- 
senting brightness differences, more particularly the number of 
primary sources. As usual the intensity of light was as nearly as 
possible equal at the point of test for these installations, and a 
supplementary specification was given of the lighting effects in 
the remainder of the room (see Trans. I. E. S., 1915, X, pp. 414- 4 
Ferree and Rand. 
Showing the effect on less of visual efficiency or power to sustain 
clear seeing of changing the height of the light source in the reflector 
of the indirect lighting fixtures. The effect on surface brightness is 
primarily to change the area and surface brilliancy of the spot of light 
thrown on the ceiling. Chart A shows the results when height of 
source in the reflector is changed; Chart B, the results when the 
height is kept approximately constant. 
CHART IL-INTENSITY SERIES. 
CHART A 
CHARTB 
422 and 452-466). The lamps employed totaled 800 watts for the 
indirect system, 760 for the semi-indirect system, and 880 for the 
direct. An inspection of the tables of measurements referred to 
above shows in general a falling off in the magnitude of brightness 
differences for all systems from Positions I-IV. This falling off, 
however, is greatest for the direct system, next greatest for the 
semi-indirect system and least for the indirect. Thus there is not 
only a decrease in the number of surfaces in the field of view 
showing a high brilliancy from Positions I-IV, but also a decrease 
in the magnitude of brightness differences between the surfaces of 
high brilliancy and the test card, between these surfaces and the 
reading page, etc., especially for the direct and semi-indirect 
systems. An inspection of the chart for loss of efficiency shows, 
roughly speaking, a correspondingly marked decrease in loss of 
efficiency from Positions I-IV for the systems which show the 
marked decrease in brightness differences, that is, for the direct 
and semi-indirect systems. The decrease in loss of efficiency is, it 
will be noted, practically nothing for the indirect system. Thus 
not only much less loss of efficiency is sustained by the eye for the 
indirect units used, but the results are much more independent 
of the position of the observer in the room. 
The comparative effects on the eye for the four positions in the 
room are shown in Chart III. Efficiency of Eye and Lighting. 
5 
CHART III-DISTRIBUTION SERIES. 
Showing the effect on loss of visual efficiency of varying the ob- 
server's position in the room, or the number of bright sources, primary 
and secondary, in the field of vision. 
POSITION I 
POSITION II 
POSITION III 
POSITION IV 
In constructing the above charts the figure expressing the ability 
to sustain clear seeing during the three-minute record before and 
after work is plotted along the ordinate and the hours of work 
along the abscissa. These two points are connected by a straight 
line the slope of which gives a graphic representation of the change 
in the power of the eye to sustain clear seeing from the beginning 
to the end of the three-hour period. During the working period 
the observer read steadily from uniform type and paper. In the 
selection and use of the observers for the work the following are 
some of the precautions that were taken. Care was exercised in 
the first place to choose only those that had already shown a satis- 
factory degree of precision in other work in physiological optics 
and whose clinic record showed no uncorrected defects of conse- 6 
Ferree and Rand. 
quence. All were under 30 years of age. Before being allowed 
to take part in the actual work of testing, each observer was trained 
to a satisfactory degree of precision for the three-minute acuity 
record under a given lighting condition and in the three-hour test 
for several of the conditions which were to be tested. In the actual 
work of testing the results were compiled from several observations 
and the precision checked up by the size of the mean variation. No 
results were accepted as significant unless the variations produced 
by changing the conditions to be tested were largely in excess of 
the mean variation or mean error for each condition tested. This, 
the accepted conventional check on the influence of variable ex- 
traneous factors, was carefully applied at each step of the work. 
In our choice of the first set of conditions to be tested, it will 
be remembered from our previous work that our purpose was to 
make a selection that would give a wide variation in illumination 
effects. The direct reflectors chosen were not of the most modern 
make, although they may be said to give effects very similar to 
much of the lighting in actual use at the present time. They were 
of porcelain ware 16 inches in diameter and only slightly concaved. 
When placed above the lamps employed (clear tungsten) they 
served merely to distribute the light to the working plane. No 
protection from the brilliancy of the light source was afforded 
to the eye. For the semi-indirect system inverted alba reflectors, 
11 inches in diameter, were employed. These reflectors were of 
modern design and represent very well glassware of medium den- 
sity. In case of the indirect system corrugated mirror reflectors 
were used enclosed in brass bowls. These reflectors were also of 
modem design and give effects which may be taken to represent 
very well those obtained in good indirect lighting. In later papers 
results will be given for the smaller differences in illuminating 
effects that may be obtained by using semi-indirect and direct re- 
flectors differing in density and design. A large number of re- 
flectors will be used chosen with special reference to their repre- 
sentative character by designers of both classes of reflectors. A 
great deal of this work has already been completed. 
Eye Shade Series. 
This series of experiments has been conducted for the following 
reasons: (1) In general two methods are used to protect the eye 
from the source of light, eye shades and lamp shades. It is desir- 
able to know whether the eye is protected equally by both; and if 
the eye shade can be substituted for the lamp shade, what type of 
shade would best serve the purpose. (2) And the statement has Efficiency of Eye and Lighting. 
7 
been made to us many times that with an eye shade the three sys- 
tems of artificial lighting we have used should give equally good 
results; and results, moreover, as good as those given by the in- 
direct system without an eye shade. There are in general two 
classes of eye shades, the translucent and the opaque. Up to this 
time we have confined our work to the opaque shade. So far as 
we know, it is customary to make the opaque shade with a dark 
lining. This kind of lining is employed probably because of some 
notion that it is restful to the eye to darken as much of the field 
of vision as is possible.* 
The tests were begun with the opaque shade with the dark lin- 
ing. What we found as the result of these tests was somewhat in 
contradicition to the predictions that had been made. The shade 
did give pretty nearly the same results for the three systems, but 
it did this contrary to prediction by improving direct and semi- 
indirect systems and making worse, by an almost equal amount, the 
indirect system. That is, protected by the opaque shade the eye 
lost in efficiency for the three systems by an amount somewhere 
near the mean of the losses experienced by it for the three systems 
without a shade. Nor is this result surprising when one reflects 
upon the conditions imposed upon the eye by an opaque shade 
with a dark lining. While it protects the eye from the sources of 
light, such a shade does not by any means eliminate harmful bright- 
'ness differences in the field of vision. It in fact creates for the eye 
a very unnatural brightness relation, i. e., it renders the whole upper 
half of the field of vision dark in sharp contrast with the brightly 
lighted lower half. The direct effect of this is a strong brightness 
induction (physiological) over the lower half of the field of vision 
which manifests itself to the observer by causing glare in surfaces 
which have no glare and by increasing the glare in surfaces in which 
glare is already present. This it is scarcely necessary to point out, 
operates against the discrimination of detail and puts the eye under 
strain to see its objects clearly. Moreover, the unusual and strongly 
irregular character of the image formed on the retina probably 
also sets up a warfare in the incentives given to the muscles which 
adjust the eye. That is, the upper half of the field of vision is 
dark and presents no detail. The effect of this is probably to 
exert a tendency to cause the muscular relaxation characteristic of 
of the darkened field of vision. The lower half of the field is light 
*Another popular view might be, so far as protection to the eye is 
concerned, to regard the opaque shade as the analogue of the opaque or 
perhaps the indirect reflector and the translucent shade as the analogue 
of semi-indirect reflector. 8 
Ferree and Rand. 
and filled with detail. The incentive here is towards the best pos- 
sible adjustment of the eye for the discrimination of detail in the 
object, while the rim of the shade, the sharply marked boundary 
between the dark and light halves of the field of vision and much 
nearer to the eye than the objects viewed,* serves as a constant 
and consciously annoying distraction to fixation and accommoda- 
tion. These complex and somewhat contradictory impulses given 
to the muscles of the eye might very well and doubtless do cause 
an excessive and unnatural loss of energy and efficiency in case 
of the prolonged adjustment of the eye needed for a period of work. 
Early in the course of the tests it occurred to us that we might 
render the brightness distribution in the field of view presented to 
the eye wearing a shade, more natural and thereby improve the 
effect of the shade on the eye, by employing a white instead of a 
dark lining. By using a mat white paper* with a reflection coef- 
ficient of about 75 per cent, for this lining, the following effects 
were produced. The two halves of the field of vision were rend- 
ered much more nearly of equal brightness; the glare in the lower 
half of the field of vision was very noticeably lessened and the dis- 
crimination of detail was correspondingly improved; the upper half 
of the field of view no longer tended to give to the eye the reflexes 
of the darkened field of vision; and the rim of the shade did not 
stand out nearly so distinctly in the field of view to distract accom- 
modation and fixation. That is, the whole lining of the shade was 
darkened just enough by being shielded from the light of the 
room by the shade itself to make it nearly equal in brightness to 
the rest of the field of vision. The effect of this was to make the 
shade merge into the field of view rather than to stand out dis- 
tinctly from it. A shade to give the best effects should be seen 
as little as possible. It thus offers a minimum of distraction to 
the proper adjustment of the eye for its work. 
The results of the test for loss of efficiency show, moreover (See 
Chart IV), that our surmise with regard to the effect on the eye 
of this change in the lining of the shade was correct. The action 
of the white lining was greatly to improve the ability of the eye 
to maintain its efficiency for a period of work. As good results 
were not gotten, however, with the shade for any of the systems 
as were given by the indirect system without the shade. Since 
there was a still greater evenness of surface brightness in the field 
*This rim is about three inches in front of the observer's eye when 
the shade is in position. 
*Hering standard white paper was used for this lining. The reflection 
coefficient of the dark lining was about 6-8 per cent. Efficiency of Eye and Lighting. 
9 
of view in case of the indirect system with the eye shade than with- 
out, the question arises why at least as good results were not ob- 
tained with the shade as without. The answer, we believe, is to be 
found in terms of the distraction to fixation and accommodation 
caused by the eye shade even when a light lining was used. For 
the effect of a shade on the eye even when the most favorable lin- 
ing is employed is that of a constantly present distracting object 
with its lower margin not far removed from the center of the field 
of vision, and much nearer to the eye than are the objects which 
the observer is called upon to discriminate. Without doubt the 
best results cannot be obtained without changing also the shape or 
design of the shade. It will be noticed also in Chart IV that the 
results were never so good for either kind of shade for the direct 
and semi-indirect systems as for the indirect. Since the evenness 
of surface brightness in the field of view was not very different for 
the three systems in both cases, this again probably indicates that 
the evenness of surface brightness is not the only one of the dis- 
tribution factors that has to be taken into account in studying 
the effect of different conditions of lighting on the eye. 
As yet we have not determined the effect of translucent shades 
on the eye. In attempting to deal in a general way with this class 
of shades we have the same type of difficulty to face that we have 
in case of the semi-indirect reflector. That is, we may have shades 
varying from translucent to opaque and sharing in the merits and 
demerits of each extreme. Our judgment would be, however, that 
it would be very difficult to get a translucent shade that would 
give as good results as the opaque shade with a light lining; for 
the translucent shade when made sufficiently opaque to give the 
needed reduction to the image of the source will darken too much 
the upper half of the field of vision and thereby simulate too much 
the condition given by the opaque shade with the dark lining to 
give the best results for comfortable and efficient seeing. More- 
over, from the results that have already been obtained with the 
opaque shade and from the principles it seems fair to infer from 
these results it seems very probable to us that as good effects for 
seeing should not be expected from the use of any kind of eye 
shade as may be gotten from lamp shades. That is, if we are to 
secure the best results for seeing, the shade should be put on the 
lamp, not on the eye. However, the relative inexpensiveness of eye 
shades, their independence of the limitations which militate against 
the use of certain types of lamp shades, their ready availability 
to those who have the least chance to escape from the effects of 10 
Ferree and Rand. 
CHART IV.-EYE SHADE SERIES. 
Showing the effect on loss of visual efficiency or power to sustain 
clear seeing of opaque eye shades with dark and with white lining for 
the installations direct, semi-indirect, and indirect with the same in- 
tensity of light at the point of work. Chart A shows results without 
shade; Chart B, with shade having dark lining; Chart C, with shade 
having white lining. 
bad lighting, namely the subordinate and the employee, should 
constitute a strong incentive for the development of this type of 
protection to the eye as a provisional and immediate aid in solving 
the problem of bad lighting. 
For a fuller statement of results and a specification of the illum- 
ination and brightness measurements and brightness ratios that 
should be taken into account in considering the results of the tests, 
see Transactions of the Illuminating Engineering Society, 1915, 
X, pp. 475-483. 
The Angle at Which the Light Falls on the Work. 
The object of these experiments was to find out whether the 
difference in the angle at which the light falls on the work pro- 
duces an effect on the eye that can be detected by the test we have 
used for loss of efficiency. For the purpose of this preliminary in- 
vestigation it was decided to make the general illumination of the Efficiency of Eye and Lighting. 
11 
room such as to cause the eye little loss of efficiency as the result 
of a period of work; and to add to that at the point of work a 
component of light which was less diffuse in order that the amount 
of light entering the eye would be more dependent upon the angle 
at which the reading page was held. 
The general illumination was obtained from the indirect system 
used in the work of the preceding sections with lamps totalling 
800 watts. The less diffuse component at the point of work was 
obtained from a 60-watt lamp with a porcelain reflector of the 
desk lamp type. This lamp was turned into the horizontal posi- 
tion and was placed behind the observer and to the left so that the 
light came over the left shoulder. When in the position for which 
the test was taken, the tip of the lamp was slightly above the level 
of the observer's eye and at a distance of 1 meter from the left eye. 
For the specification of the illumination and brightness measure- 
ments and the brightness ratios for our test room illuminated by 
the indirect system 800 watts, see Trans. 1. E. S., 1915, X, pp. 
469-471. For a specification of the significant changes in these 
measurements and ratios produced by the addition of the 60-watt 
lamp behind the observer, see ibid. p. 484. The brightness of the 
reading page in the position that gave the least amount of specular 
reflection was 0.0059 cp. per sq. in.; and in the position that gave 
the greatest amount of specular reflection 0.0077 cp. per sq. in. 
A mirror surface was used as an aid in locating the position of 
least and greatest specular reflection. The comparative effects on 
the eye of these two positions of the reading page are shown in 
Chart V. 
CHART V.-THE ANGLE AT WHICH THE LIGHT FALLS ON THE 
WORK. 
Showing the effect on loss of visual efficiency oi' power to sustain 
clear seeing of the angle at which the light falls on the work. 12 
Ferree and Eand. 
The Effect of Different Conditions of Lighting on the Fixation 
Muscles of the Eye. 
The test we have employed thus far in the conduct of our work 
is one designed to show the effect of different conditions of light- 
ing on the ability of the eye to hold its efficiency for clear seeing 
for a period of three minutes. In itself this text is not analytical 
in principle. The results, as is stated above, are expressed in terms 
of an aggregate loss of function. The contributive factors may be 
inferred from the nature of the test but the test is not in itself 
designed to separate them out. And indeed it is a question whether 
any practical good can accrue to the practice of lighting from a 
knowledge of just what part of the visual apparatus it is that falls 
off in function as a result of an unfavorable condition of lighting. 
Obviously the chief need is to find out what are the conditions that 
cause the eye to lose its ability to see clearly and to avoid these 
conditions in planning and installing a lighting system. From 
the beginning we have had in mind, however, an analysis of effect. 
Our tests for the sensitivity and functional state of the retina 
(sensitivity to color and brightness, lag in coming to a full response, 
rate of exhaustion and rate of recovery) showed, for example, that 
very little, if any, of the difference in results we have gotten for 
the four types of lighting we have employed can be ascribed to a 
loss in the efficiency of the retina, or the light sensitive part of 
the visual apparatus. Three sets of factors are involved in clear 
seeing: (1) the sensitivity of the eye to colored and white light; 
(2) the ability to make fine space discriminations which is in part 
dependent upon our third factor; and (3) accurate fixation and 
accommodation. Both fixation and accommodation are the result 
of muscular action. When the muscles lose in tone because of ex- 
cessive use or by sharing in a general condition or state of the 
body, the eye loses correspondingly in its power to sustain clear 
seeing. If, for example, the muscles of accommodation have fallen 
off in efficiency, the lens is no longer held in the adjustment need- 
ed to bring the light to a sharp focus on the retina and loss of 
detail and blurring result; or if it be the fixation muscles that have 
suffered the loss, the eyes cannot be continuously held in such a 
position that the images of the object viewed fall symmetrically on 
the fovea of each. When this latter condition is present, loss of 
detail results from two causes. (1) The fovea and region immed- 
iately surrounding it are the most highly developed parts of the 
retina and the best fitted for the light and space discriminations 
needed for clear seeing. Moreover, the refracting media of the Efficiency of Eye and Lighting. 
13 
eye give the clearest images when the axis of the cone of rays from 
the object viewed deviates as little as possible, consistent with the 
mechanism of the eye, from the optic axis. And (2) if the images 
in the two eyes do not fall more or less symmetrically upon the 
fovea of each, they are not accurately combined into one, and 
blurring and loss of detail result from the doubling of the objects 
seen. It is our purpose as fast as possible to isolate the effect of 
the three systems of lighting we have used on each of the above 
named factors. In the work of the present section the effect of 
these systems on the fixation muscles has been studied only in a 
tentative and provisional way. 
The doubling of the image seen when the fixation muscles lose 
their power of coordinated action furnishes us with the clue for 
a test for loss of efficiency of these muscles. That is, just as blur- 
ring and loss of ability to discriminate detail is taken as the 
criterion of the loss of acuity of vision, so will the doubling of the 
image seen be taken as our index of the loss of the coordinated 
action of the fixation muscles. If one were to stare continuously 
for an interval of time with natural vision at a single test object, 
for example, a vertical line, doubling might be expected especially 
if there had been protracted strain or considerable loss of power to 
coordinate. For the purpose of our work, however, greater sensi- 
tivity than this would be needed. Obviouslv sensitivity can be add- 
ed by putting the eyes under strain to combine their images. When 
this is done, even when the muscles are fresh, if the object is looked 
at or fixated for an interval of time it will be seen alternately as 
one and as two. The proportion or the ratio of the time seen as 
one to the time seen as two can be regulated by the amount of 
initial strain under which the eyes are put to combine their images. 
The regulation of this ratio is empirical and of importance; for 
as in the case of the test for loss of efficiency for clear seeing, the 
sensitivity of the test depends to a considerable extent upon the 
initial value that is given to this ratio. The eyes may be put under 
strain to combine their images by interposing between them and 
the object viewed weak prisms and so adjusting them and regulat- 
ing the distance of the object from the eye that with the maximum 
of effort to see it as one it is seen alternately as one and as two in 14 
Ferree and Rand. 
the proportion desired.* In onr provisional experiments on this 
point we found that an adaptation of the Brewster stereoscope 
afforded a convenient method of putting the eyes under a strain 
to combine their images. In this case a stereograph consisting of 
two vertical lines exactly alike may be used as the test object. In 
the stereograph used in our test the vertical lines were 2.5 cm. long 
and were printed on the card 4.5 cm. apart or at 2:25 cm. from 
the center of the card. When this was put in a sliding carrier 
and was made to approach the eyes, a position was reached at which 
with the maximum of effort the observer was no longer able to 
see the two vertical lines as one.* They were seen alternately as 
one and two. In making the test the hood was removed from the 
stereoscope so that the eyes were fully exposed to the conditions of 
the illumination that were being tested. The stereoscope was 
mounted in front of the eyes of the observer in position at the point 
of work. The distance of the carrier containing the test object from 
the observer's eyes was adjusted until the proper ratio of time 
seen as one and time seen as two was obtained. Having determined 
*It is obvious that the greater part of the strain may be put at will 
upon the internal or external muscles by the proper rotation of the 
prisms. It will be understood that the work reported above is intended 
to be little more than suggestive of possibilities. 
It would seem also that the principle advanced here might be utilized 
to advantage by the ophthalmologist as a supplement to his tests of the 
extrinsic muscles of the eye. The abduction and abduction tests, for 
example, determine only what the muscles are able to do by momentary 
effort. Obviously, however, it is not what the muscles are able to do 
by a momentary effort or jerk that measures their ability to hold the 
eyes continuously adjusted for work. It is rather their endurance or 
what they are able to accomplish in an interval of time. An expression 
may be had for this either for the eyes conjointly or separately by the 
method described above. That is, prisms may be put in front of either 
one or both eyes and the ratio be determined of the time the object is 
seen as one or as two for whatever interval of time the operator may 
select. Similarly, it seems to the writers that the time element might be 
introduced to advantage into the visual acuity test used by the ophthal- 
mologist when the cycloplegic is not employed, for example, in cases of 
post-cycloplegic refraction. Is it enough to know that the eye in these 
cases has 20/20 acuity or can discriminate a certain standard visual angle 
by a momentary effort? Would it not give a more complete representa- 
tion of the functional condition of the eye to know what it can discrimi- 
nate clearly through an interval of time; or better still perhaps, for what 
proportion of an interval of time it can discriminate a certain detail or 
standard visual angle clearly? For example, just as a fatigued eye 
may for a moment under the spur of the test overcome the functional 
results of fatigue, so might small errors of refraction be overcome for 
the moment by muscular effort, especially in the cases in which the 
muscles of the eye are unusually strong. But just as the fatigued muscle 
cannot do this through an interval of time, so it would seem that a 
residual error of refraction might not be so easily masked through an 
interval of time by means of muscular effort. In short, this form of 
test is suggested as affording possibly a closer approximation to the con- 
ditions and demands imposed upon the eye during a period of work than 
is afforded by the acuity test based upon the momentary judgment. In 
making this suggestion, however, we recognize that in the work of the 
clinic the advantage of such a test may be more theoretical than feasible 
and practicable. 
*The observer, whose results are given in this paper, preferred to 
have the working position of the sliding carrier beyond the position at 
which the combination of the two lines was effected the most easily, 
instead of nearer, as is described above. It is obvious that either posi- 
tion may be used for testing the ability of the muscles to sustain coordi- 
nation of action. Efficiency of Eye and Lighting. 
15 
this position a record was made of the time seen as one and the 
time seen as two for three minutes at the beginning and close of 
work. The ratio of these intervals may in either case be taken as 
a measure at that time of the power of the fixation muscles to 
act in coordination for three minutes of continuous effort; and 
the decrease in this ratio from the beginning to the close of work 
may be taken as a measure of the loss in that power sustained as 
the result of work. In making this test the same recording appa- 
ratus was used as was employed in the test for loss of efficiency 
for clear seeing. That is, the record was traced on a kymograph 
by means of an electro-magnetic marker and a telegraph key, and 
a time line was run underneath the record by means of a Jacquet 
chronograph registering seconds. 
This test was made under the same installations, conditions of 
work, and with the same observers that were used in the distribution 
series of the former paper. For a more complete account of this 
series, see also Transactions I. E. S., 1915, X, pp. 484-490. The 
test was made at only one of the positions in the room that were 
used in that series, namely, the position at which the greatest loss 
of power to sustain clear seeing was obtained. At this point, it 
will be remembeerd, six of the lighting units were in the field of 
view. The specifications of the lighting effects produced by these 
installations are given Transactions I. E. S., 1915, pp. 452-459. 
Nothing need be added here to these specifications but the bright- 
ness of the stereograph, or the test object, in position for the three 
systems of lighting, and the illumination measurements at the test 
object. The brightness of the test object, corrected for the absorp- 
tion of the prisms of the stereoscope, was for the direct system 
0.00172 cp. per sq. in.; for the semi-indirect system 0.00163 cp. 
per sq. in.; and for the indirect system 0.00167 cp. per sq. in. 
New illumination measurements were needed at the test object 
because it had to be moved closer to the eyes than was the case 
in the tests for loss of power to sustain clear seeing. This brought 
it into a region of different illumination. These measurements are 
given in Chart VI. The results of this test for loss of coordinating 
efficiency of the fixation muscles are given also in this chart. These 
results show (a) that very little loss of coordination was suffered 
by the fixation muscles as the result of three hours of work under 
the systems selected; and (b) that there was very little difference 
in the effect for the three systems. Since there are no obvious 
reasons for thinking that this test has not somewhere nearly as 
great sensitivity as the test for loss of efficiency for clear seeing, 16 
Ferree and Rand. 
and since the same observers, conditions of lighting and working 
were used as in the former tests, it does not seem to us at this time 
that the large differences in the loss of efficiency for clear seeing 
that is sustained under these conditions, as shown by the former 
tests, can be ascribed to any great extent to an effect on the muscles 
of fixation. The point, however, cannot be considered as finally 
nettled because we have not made long enough study of the test 
itself and the limitations of its application to the study in question 
to make its results certainly comparative with those of the test 
for the ability of the eye to sustain clear seeing. 
CHART VI-FIXATION MUSCLE SERIES. 
Showing the loss of efficiency of the fixation muscles as the result 
of a three-hour test under the direct, semi-indirect, and indirect systems 
of lighting employed. 
Foot-candles 
Lighting system 
Watts 
Horizontal 
Vertical 
45° 
Indirect 
800 
4.2 
0.99 
2.5 
Semi-indirect 
760 
4.8 
0.98 
2.6 
Direct 
880 
3.9 
1.0 
1.99 
The Effect of Motion Pictures on the Efficiency of the Eye. 
The belief that motion pictures subject the eye to undue strain 
is too prevalent to need more than mention in passing. All are 
familiar with the conditions-the initially dark-adapted and high- 
ly sensitized eye, the comparatively brilliant screen with its dark 
surrounding field, the flickering light, the shifting and very often 
unsteady pictures. We have already seen that differences in sur- 
face brightness of considerable magnitude in the field of vision 
cause loss of efficiency and produce discomfort and we have dis- 
cussed the causes for these effects. We have nothing further to 
add to that discussion here. We are, however, facing for the first 
time in our work the question of the effect upon the eye of a flick- 
ering light and lack of steadiness in the object viewed. The fol- 
lowing reason is suggested why a flickering light or unsteady pic- 
ture may cause loss of efficiency. The eye is so constituted that 
when its images lose in clearness or distinctness it is incited to a Efficiency of Eye and Lighting. 
17 
muscular readjustment to bring about the clearness needed. Ordi- 
narily in seeing, the conditions for loss in clearness come about 
primarily through the difference in distance or direction from the 
eye of the objects which are successively viewed. In motion pic- 
tures, however, the changing clearness in the objects viewed is not 
due to any change in their distance or direction from the eye; nor 
to anything in fact which the readjustment of the- eye can remedy 
to any considerable degree. The effort expended, therefore, is of 
little avail for seeing, if indeed the new setting of the parts is 
not a detriment to clear seeing and a condition which in turn must 
be corrected. This should, and doubtless does, lead to muscular 
strain and loss of efficiency. It was decided therefore, to make an 
explorative investigation to determine whether there is an effect 
of motion pictures on the eye which can be detected by our test 
for loss of efficiency. The tests were taken in a local theatre, select- 
ed primarily because of the favorable conditions that prevailed. 
The definition at the screen was good and the pictures were un- 
usually steady and free from flicker. The conditions were, we 
think, fairly representative of what is found in the better class 
of moving picture houses. 
The tests were taken immediately before and after two hours 
of. observation of the pictures. During the exhibition the observer 
sat directly in front of the center of the screen. The observation 
was made at successive times at three distances from the screen- 
in the front, the middle and the back of the house. These positions 
were respectively 25, 48 and 71 ft. (7. 62, 14.6 and 21.6 m.) from 
the screen. The room in which the pictures were shown was 78 
ft. (23.7 m.) long and 48 ft. (14.6 m.) wide. The tests were 
taken in a room 14 ft. (4.2 m.) long, 9 ft. (2.74 m.) wide, 11 ft. 
(3.35 m.) high adjoining the stage. The walls and ceiling of 
this room were of rough plaster, painted a flat white. When tak- 
ing the tests the observer sat facing one of the side walls of the 
room, 1.5 m. distant. The room was lighted for the purpose of 
the test with one 100-watt and one 60-watt clear tungsten lamp 
suspended behind and slightly to the right of the observer when 
in position for the test at about 2 ft. (.06 m.) above the level of 
his eyes. The source of light was entirely out of the field of view 
and the light fell evenly and without shadow on the test card and 
the wall in front of the observer. At the point of the test card, 
the illumination measured with the receiving test plate of the photo- 
meter in the horizontal plane was 1.3 foot-candles; in the vertical 
plane 1.9 foot-candles; and in the horizontal plane 2.3 foot-candles. 18 
Ferree and Rand. 
The surface brightness of the test card was 0.003256 cp. per sq. 
in. The distribution of surface brightness on the wall which the 
observer faced was very even. At the point of maximum bright- 
ness to the right of the observer, as nearly as that point could be 
located, the brilliancy was 0.00308 cp. per sq. in.; and to the left 
of the observer, 0.002024 cp. per sq. in. 
In order that there might be no intermission between the pic- 
tures for changing the films, two projection machines were used. 
The following is the specification of the apparatus employed, as 
given by the operator: 
Type of machine, Powers 6--A Projector. 
Lens equipment, 1 pair pearl white condensers, 6% in. F. L., 1 Bausch 
and Lomb objective combination, 4% in. E. F. 
Lamp, 1 10,000-cp. adjustable arc. 
Carbons, % in. cored bio's. 
Current, 22 volt. a. c. through Halberg transformer. 
Line current, 28-30 amperes. 
Arc voltage, 45-50 volts. 
Length of throw or distance from objective to screen, 72 ft. (21.9 m,). 
Screen, sheet muslin sized and coated with flat white alabastine. 
Speed of film through machine, 66 ft. 8 in. (20.3 m.) per min. 
Number of pictures per 1 ft. (0.3 m.) of film, 16. 
Size of picture on film, % in. (1.9 cm.) high by 15/16 in. (2.38 cm.) 
wide. 
Size of picture on screen, 11 ft. (3.35 m.) high by 14 ft. (4.26 m.) wide. 
Approximate brightness of screen with film removed from projector, 
3.47 cp. per sq. in. 
Exceptional steadiness, it may be said, is given to the movement 
of the film, and therefore to the picture in this type of projector 
by the special type of intermittent movement that is employed. 
Details of this movement need not be given here. As has already 
been stated, our reason for making the test in this particular the- 
atre was the comparative steadiness of the pictures and the com- 
parative freedom from flicker that was obtained. 
The results of the test are shown in Chart VII. Quite a great 
deal of loss of efficiency is shown as the result of two hours of 
observation. The nearer the observer was to the screen, the greater 
was this loss found to be. The loss, however, so far as we can 
tell, is no greater than is caused by steady work under the direct 
and semi-indirect installation of lighting used in our distribution 
series. Unfortunately we have not for the purpose of comparison 
results for the same observer for the same length of time of ex- 
posure for the two sets of condition. The loss for Observer R for 
the two hours observation of the motion pictures was not nearly 
so great as for the three hours of reading from good print and 
paper under the direct and semi-indirect systems of lighting. But 
comparing results for Observer G for two hours of reading from 
the same type and paper with those for Observer R for two hours' 
observation of the pictures, the loss seems to be about the same. Efficiency of Eye and Lighting. 
19 
CHART VIL-MOTION PICTURE SERIES. 
Showing the loss of visual efficiency or power to sustain clear see- 
ing of the eye caused by two hours observation of motion pictures. 
Position A, 25 ft. from projection screen 
Position B 48 ft. from projection screen 
Position C 71 ft. from projection screen 
That is, our results indicate that while the eyes are strained a great 
deal by the observation of moving pictures, even in the better mov- 
ing picture houses, they are damaged little more by that in all 
probability than they are by reading steadily the same length of 
time under the greater part of the lighting that is now in actual 
use. For the sake of comparing the effect of motion pictures on 
the eyes with the effect of reading steadily under the direct, semi- 
indirect and indirect systems of lighting we have employed, Chart 
VIII has been prepared. 
The Tendency of Different Lighting Conditions to Produce Discom- 
fort and a Comparison of the Tendency of These Conditions 
to Cause Loss of Efficiency and to Produce Discomfort. 
In the former papers we have held that the general level or scale 
of efficiency of the fresh eye, loss of efficiency as the result of work, 
and the tendency to produce discomfort are all separate aspects of 
the problem bf lighting in its relation to the eye, and that our 
knowledge of each must be obtained by different methods of inves- 
tigation. A correlation between these three moments is doubtless 
possible, but that correlation should be founded upon the results 
of careful investigation; it should not be assumed. It is our pur- 
pose in this section of the paper to show the relative tendency of 20 
Ferree and Rand. 
Distribution Series (Observer G) 
Showing the loss of visual efficiency or 
power to sustain clear seeing as the re- 
sult of a test two hour under the sys- 
tems of direct, semi-indirect, and indi- 
rect lighting used, and daylight. 
Lighting 
system 
Watts Volts 
Foot-candles 
Hor. Ver. 45° 
A Daylight ... 
- - 
5.5 1.32 4.2 
B Indirect ..... 
800 107 
5.2 1.36 3.5 
C Semi-indir't 
.. 760 107 
5.8 1.45 4.0 
D Direct 
880 107 
4.2 1.41 2.6 
Motion Pictures Series 
(Observer R) 
Showing the loss of visual ef- 
ficiency or power to sustain clear 
seeing caused by two hours obser- 
vation of motion pictures. 
Position A, 25 ft. from projection 
screen. 
Position B, 48 ft. from projection 
screen. 
Position C, 71 ft. from projection 
screen. 
CHART VIII. 
Distribution Series (Observer R) 
Showing the loss of visual efficiency or 
power to sustain clear seeing as the re- 
sult of a two hour test under the sys- 
tems of direct, semi-indirect, and indirect 
lighting used, and daylight. 
Lighting r 
Foot-candles A 
system 
Watts 
Hor. 
Ver. 
45° 
Daylight 
5.5 
1.32 
4.2 
Indirect 
800 
5.2 
1.36 
3.5 
Semi-indirect 
.. 760 
5.8 
1.45 
4.0 
Direct 
..... 880 
4.2 
1.41 
2.6 Efficiency of Eye and Lighting. 
21 
the different conditions of lighting we have used to produce dis- 
comfort, and to make a rough comparison of the tendency of each 
condition to cause loss of efficiency and to produce discomfort. 
Any comparative study of the conditions producing discomfort 
necessitates a means of estimating discomfort. It is obvious that 
the core of the experience of discomfort is either a sensation or 
a complex of sensations. As such it should have a limen or thres- 
hold just as other sensations have; and just as we are able in gen- 
eral to estimate sensitivity in terms of the threshold value, so 
should we in this- case be able to use the threshold value in estimat- 
ing the eye's sensitivity or liability to discomfort under a given 
lighting condition. Threshold values are usually determined by 
finding out how much energy or intensity of a given stimulus, 
applied for a short interval of time, is required to arouse a just 
noticeable sensation. This form of procedure, however, is not 
adapted to the needs of our problem. It is much better to reverse 
the process and find how long the eye has to be exposed to a stimu- 
lus of given intensity to arouse just noticeable discomfort. Our 
threshold thus becomes a time threshold and is measured in units 
of time instead of units of intensity. In order to determine whether 
the judgment of the threshold of discomfort can be made with cer- 
tainty, and to perfect the method and to test in general its feasibil- 
ity, an abstract investigation was undertaken first, running through 
an entire year, in which a better and more convenient control of 
conditions could be secured than is possible in the investigation of 
a concrete lighting situation. That is, we undertook to determine 
the comparative sensitivity of the eye to discomfort when a single 
source of light was exposed in different parts of the field of vision. 
In order to carry out that investigation a lamp house with a cir- 
cular opening in one side 3 cm. in diameter was attached to the 
arm of a perimeter in such a way that the opening was always 
directed towards the observer's eye. In the lamp house could be 
placed a lamp of whatever candlepower was desired. The arm of 
the perimeter could be shifted to any meridian in which it was 
desired to work and the' lamp house could be moved at will along 
this arm. It was thus possible to expose the light for any length 
of time in any part of the field of vision that was desired. Work- 
ing in this way we have not only investigated the effect of many 
types of variation of the position of the light in the field of view, 
the effect of intensity of light, etc.; but we have studied and stand- 
ardized the factors that influence the sensitivity and reproducibility 
of the judgment and have given our observers the training that 22 
Ferree and Rand. 
was needed for the concrete investigation. In making the con- 
crete investigation we have used every variation of the conditions 
of lighting described in this and the preceding paper. That is, 
the tendency to produce discomfort, measured in terms of the time 
threshold, has been determined for all the conditions of lighting 
we have used in the tests for loss of efficiency. Two cases of the 
investigation may be made-a determination of the tendency to 
cause discomfort when the eye is at rest, and a determination of 
this tendency when the eye is at work. Both of these cases were 
included in our investigation. The following determinations were 
made, (a) The time threshold of discomfort was gotten when 
the observer was sitting with the accommoadtion muscles relaxed 
and with the fixation muscles as nearly relaxed as was practicable 
under the conditions; that is, the observer sat in the positions used 
in the distribution series (one with six, one with four, one with 
two and one with no fixtures in the field of view) and took an easy 
fixation of an area at the level of the eye on the opposite wall of 
the room. The fixation distance, for example, for the first of these 
positions was 22 ft. Since blinking was found to be one of the 
variable factors which influence the tendency to produce discom- 
fort, the amount of blinking was made constant from test to test. 
This was accomplished by having the observer blink at equal inter- 
vals during the test, timing himself by means of the stroke of 
a metronome. The interval most natural and suitable for this 
purpose was determined for each observer separately. In the re- 
sults given in the following table a three-minute interval was used. 
And (b) the time threshold of discomfort was determined when 
the observer was reading from print and paper similar to that used 
in the loss of efficiency tests. In these tests all the conditions were 
kept as nearly the same as they were in the work on loss of ef- 
ficiency as was possible. The results of both of these sets of experi- 
ments on the tendency to produce discomfort are shown in Tables 
I-IV. The tendency to produce discomfort should be estimated, 
roughly speaking, probably as inversely proportional to the time 
it was required for discomfort to be set 'up. The time required 
for discomfort to be set up is given in the tables. In order to make 
convenient a comparison of the tendency of the various conditions 
of lighting to cause loss of efficiency and to produce discomfort, 
the percentage loss of efficiency caused by the given lighting condi- 
tion is given in a parallel column in each table. The percentage 
loss of efficiency was computed by dividing the loss in the ratio 
of time seen clear to time seen blurred sustained as a result of Efficiency of Eye and Lighting. 
23 
TABLE I.-DISTRIBUTION SERIES. 
Showing a comparison of the tendency of the direct, semi-indirect, and indirect installations of lighting used in the 
distribution series to cause loss of visual efficiency or power to sustain clear seeing, and to produce discomfort. The loss 
of efficiency is the result of a three hour test. The tendency to produce discomfort is estimated by the time required for 
just noticeable discomfort to be set up. 
Foot-candles 
Per cent. 
Time limen 
of discomfort 
Time limen 
of discomfort 
Position 
- 
A 
loss of 
in seconds 
in seconds 
of observer 
Lighting system 
Watts 
Horizontal 
Vertical 
45° 
efficiency 
(not reading) 
(reading) 
1. 
Indirect 
.... 800 
5.2 
1.36 
3.5 
8.6 
263 
100 
Semi-indirect 
.... 760 
5.8 
1.45 
4.0 
72.0 
15 
8 
Direct 
.... 880 
4.2 
1.41 
2.6 
81.0 
10 
9 
II. 
Indirect 
.... 800 
5.1 
1.98 
4.2 
6.3 
259 
103 
Semi-indirect 
.... 760 
6.1 
2.5 
4.7 
37.0 
26 
14 
Direct 
.... 880 
4.65 
2.75 
4.4 
58.3 
20 
13 
III. 
Indirect 
.... 800 
3.9 
2.1 
4.0 
7.7 
255 
99 
Semi-indirect 
.... 760 
5.0 
2.6 
5.4 
22.0 
120 
35 
Direct 
.... 880 
4.0 
2.9 
4.6 
31.0 
55 
24 
IV. 
Indirect 
.... 800 
2.9 
2.1 
3.6 
6.6 
265 
101 
Semi-indirect 
... 760 
3.4 
3.0 
4.4 
19.0 
240 
87 
Direct 
... 880 
3.0 
3.4 
4.5 
23.0 
235 
57 24 
Ferree and Rand. 
TABLE II.-INTENSITY SERIES. 
Showing a comparison of the tendency of the direct, semi-indirect and indirect installations of lighting for the different 
intensities used in the intensity series to cause loss of visual efficiency or power to sustain clear seeing, and to produce dis- 
comfort. The loss of efficiency is the result of a three hour test. The tendency to produce discomfort is estimated by the 
time required for just noticeable discomfort to be set up. 
Lighting system Watts 
Foot-candles 
Per cent, 
loss in 
efficiency 
Time limen 
of discomfort 
in seconds 
(not reading) 
Time limen 
of discomfort 
in seconds 
(reading) 
Horizontal 
Vertical 
45° 
Indirect 800 
5.2 
1.36 
3.5 
8.6 
263.0 
100 
480 
3.0 
0.765 
1.97 
8.0 
265.0 
103 
320 (with socket extenders) 
1.7 
0.49 
1.08 
9.1 
256.0 
98 
200 (with socket extenders) 
1.48 
0.407 
0.95 
5.7 
251.0 
104 
320 (without socket extenders) 
1.33 
0.39 
0.87 
23.0 
50.0 
33 
200 (without socket extenders) 
1.16 
0.37 
0.76 
40.0 
20.0 
14 
Semi-indirect 320 
2.2 
0.58 
1.52 
11.4 
102.0 
35 
200 
1.6 
0.45 
1.15 
40.9 
62.0 
16 
480 
3.3 
0.94 
2.4 
50.0 
50.0 
15 
760 
5.8 
1.45 
4.0 
72.0 
15.0 
8 
800 
6.8 
1.82 
4.5 
78.0 
14.0 
3 
Direct (16 lamps).. 240 
1.23 
0.54 
0.935 
57.4 
23.5 
17 
365 
1.6 
0.6 
1.33 
62.0 
14.0 
11 
400 
1.86 
0.8 
1.46 
65.0 
12.0 
11 
880 
4.2 
1.41 
2.6 
81.0 
10.0 
9 
Direct (8 lamps).... 200 
1.16 
0.45 
0.85 
34.3 
56.0 
27 
120 
0.64 
0.32 
0.49 
45.5 
52.0 
15 
320 
1.97 
0.65 
1.39 
55.5 
23.0 
13 
480 
2.6 
1.02 
2.00 
67.0 
20.0 
12 Efficiency of Eye and Lighting. 
25 
work by 3.5, the standard ratio to which all the ratios at the be- 
ginning of work were reduced. A rough correspondence of the 
tendency to produce discomfort and to cause loss of efficiency will 
be noted in every case. This correspondence by no means amounts 
to a 1:1 correlation, however. In Table I is given the compari- 
son of the tendency to cause loss of efficiency and to produce dis- 
comfort for the distribution series; in Table II for the intensity 
series; in Table III for the eye shade series; and in Table IV for 
the series showing the effect of the angle at which the light falls 
on the work. 
TABLE III.-EYE SHADE SERIES. 
Showing a comparison of the tendency of the direct, semi-indirect, 
and indirect installations of lighting used in the distribution series to 
cause loss of visual efficiency or power to sustain clear seeing, and 
to produce discomfort when the eye was protected by an opaque eye 
shade with a dark lining and by an opaque eye shade with a white 
lining. The loss of efficiency is the result of a three hour test. The 
tendency to produce discomfort is estimated by the time required for 
just noticeable discomfort to be set up. 
Lining of eye shade 
Lighting system 
Watts 
Foot-candles 
Per cent, loss of efficiency- 
Timelimen of discomfort in 
seconds (not reading) 
Tfme limen of discomfort in 
seconds (reading-) 
r 
Horizontal 
Vertical 
45° 
White 
Indirect 
. 800 
5.2 
1.36 
3.5 
9.1 
85 
50 
Semi-indirect 
. 760 
5.8 
1.45 
4.0 
10.6 
81 
48 
Direct 
. 880 
4.2 
1.41 
2.6 
12.0 
75 
45 
Dark 
Indirect 
800 
5.2 
1.36 
3.5 
33.0 
23 
19 
Semi-indirect 
. 760 
5.8 
1.45 
4.0 
33.4 
19 
15 
Direct 
. 880 
4.2 
1.41 
2.6 
35.0 
16 
13 
TABLE IV.-THE ANGLE AT WHICH THE LIGHT FALLS ON 
THE WORK. 
Showing a comparison of the tendency to cause loss of visual ef- 
ficiency or power to sustain clear seeing, and to produce discomfort 
of the angle at which the light falls on the work. The loss of ef- 
ficiency is the result of a three hour test. The tendency to produce 
discomfort is estimated by the time required for just noticeable dis- 
comfort to be set up. 
Foot-candles 
Time 
limen of 
discomfort 
c 
Reflection from 
Hori- 
Verti- 
Per cent, loss 
in seconds 
reading page 
zontal 
cal 
45° 
of efficiency 
(reading) 
Diffuse , 
5.3 
1.84 
3.9 
6.6 
95 
Specular ... 
..... 5.3 
1.84 
3.9 
14.3 
30 Reprinted from the Annals of Ophthalmology, 
July, 1916. 
A RESUME OF EXPERIMENTS ON THE EFFECT OF 
DIFFERENT CONDITIONS OF LIGHTING 
ON THE EYE. 
C. E. Ferree, Ph. D., and G. Rand, Ph. D., 
Bryn Mawr College. 
The work of which this paper is a brief outline was done 
under the auspices of the American Medical Association's sub- 
committee on the Hygiene of the Eye, of which Dr. William 
Campbell Posey of Philadelphia is chairman, and has been in 
progress five years. The object of the work has been to com- 
pare the effect of different lighting conditions on the eye and 
to find the factors in a lighting situation which cause the eye 
to lose in efficiency and to experience discomfort. 
Confronting the problem of the effect of different lighting 
conditions on the eye, it is obvious that the first step towards 
systematic work is to obtain some means of estimating effect. 
The prominent effects of bad lighting systems are loss of 
efficiency, temporary and progressive, and eye discomfort. 
Three classes of effect, however, may be investigated: (1) 
The effect on the general level or scale of efficiency for the 
fresh eye; (2) loss of efficiency as the result of a period of 
work; and (3) the tendency to produce discomfort. A de- 
scription of tests designed especially for the investigation of 
these effects has already appeared in print.1 Some of these 
tests have been designed to determine the eye's aggregate 
loss in functional power, others to aid in the analysis of this 
effect. Space can be taken here for the mention only of the 
one with which the greater part of the work was done- 
namely, a test for determining the power of the eye to sustain 
clear seeing. Just two principles are involved in this test. 
One is that visual acuity or clearness of seeing may be meas- 
ured by the smallest visual angle the eye is able to discriminate ; 2 
EFFECT OF CONDITIONS OF LIGHTING ON THE EYE. 
the other, a principle equally old, is that a loss of efficiency in a 
machine, apparatus, or a living organ or organism will show 
out more plainly when a prolonged rather than a momentary 
performance is required. These principles have been combined 
in their simplest terms into a test of the comparative ability of 
the eye to maintain its power of clear seeing or aggregate 
functional activity under different conditions of lighting and 
under different kinds and conditions of use. In operation the 
test method may be described briefly as follows : The power 
of the eye to sustain a certain standard of acuity for three 
minutes is measured before and after a period of reading from 
uniform type and paper under the lighting conditions to be 
tested. That is, by means of a visual acuity test object with 
the proper auxiliary apparatus and a kymograph and a chrono- 
graph, records are made of the time the eye can be held up 
to this standard of performance and the time it drops below. 
The ratio of these quantities to each other or to the total time 
for which the record was made is taken as the measure of the 
ability of the eye to sustain its power of clear seeing before 
and after work under the lighting conditions to be tested. 
The following aspects of lighting sustain an important rela- 
tion to the eye: The evenness of illumination, the diffuseness 
of light, the angle at which the light falls on the object viewed, 
the evenness of surface brightness, intensity, and quality, or 
color value of the light. The first four of these factors, which 
may be grouped together as distribution factors, will be dis- 
cussed briefly with reference to types of lighting now in com- 
mon use. 
The.ideal condition with regard to the distribution factors, 
so far as the functional welfare of the eye is concerned, is 
to have the field of vision uniformly illuminated with light well 
diffused and no extremes o/ surface brightness. When this 
condition is attained the illumination of the retina will shade 
off more or less gradually from center to periphery, which 
gradation is necessary for accurate and comfortable fixation 
and accommodation. In the proper illumination of a room by 
daylight we have been able thus far to get the best control of 
the distribution factors. Before it reaches our windows or 
skylights, daylight has been rendered widely diffuse by in- 
numerable reflections; and the windows and skylights them- EFFECT OF CONDITIONS OF LIGHTING ON THE EYE. 
3 
selves, acting as sources, have a broad area and low intrinsic 
brilliancy, all of which features contribute towards giving the 
ideal conditions of distribution stated above. Of the systems 
of artificial lighting, the best control of the distribution fac- 
tors, speaking in general terms, is given by the indirect systems, 
and the semiindirect systems with a small direct component 
of light. In the indirect systems the source is concealed from 
the eye and the light is thrown against the ceiling or some other 
diffusely reflecting surface in such a way that it suffers one or 
more reflections before it reaches the eye. When properly 
installed the use of these reflectors introduces no extremes of 
surface brightness in the field of view greater than that which 
the eye is prepared to stand without a significant depression 
of functional power. Moreover, the brightest spots are on the 
ceiling and are, therefore, in rooms of ordinary height, pretty 
well removed from the zone of most harmful influence on the 
eye. The direct lighting systems are designed to send the 
light directly to the plane of work. There is in general in 
the use of these systems a tendency to concentrate the light on 
the working plane or object viewed rather than to scatter it in 
all directions, and therefore a tendency, especially with some 
types and kinds of installation, to create brightness differences 
in the field of view rather than to level them down. Much can 
be done to ameliorate this tendency, however, in constructing 
the reflector and grading its density and in choosing the height 
of installation above the working plane. Too often, too, the 
eye is not shielded properly from the light source, and fre- 
quently no effort at all is made to do this, although such prac- 
tice is now strongly condemned by the lighting engineer. How 
to retain as much as possible of the superior physical efficiency 
of direct lighting and at the same time to protect the eye from 
the harmful effects of badly controlled distribution factors, 
more especially from the glare of poorly concealed sources, 
the excessive brilliancy presented by the surfaces of reflectors 
of low density and by the openings of reflectors of high density, 
etc., is one of the most interesting and difficult problems pre- 
sented to the workers in this field at the present time. The 
semiindirect reflectors are intended to represent a compromise 
between the direct and indirect reflectors. A part of the light 
is transmitted to the plane of work through the translucent 4 
EFFECT OF CONDITIONS OF LIGHTING ON THE EYE. 
reflector placed directly beneath the source of light, and a part 
is reflected to the ceiling. Thus, depending on the density 
of the reflector, this type of lighting may vary between the 
totally direct and the totally indirect, and share in the respect- 
ive merits and demerits of each in proportion to its place in 
the,scale. By giving a better control of what we have called 
the distribution factors, this type of lighting is supposed also 
to be a concession to the welfare and comfort of the eye, and 
so it is in reflectors of high density. Our tests, however, show 
that the concession is not nearly so great as it was supposed 
to be in case of reflectors of low and medium density. In fact, 
installed at the intensity of illumination ordinarily used or at 
an intensity great enough for all kinds of work, little advan- 
tage seems to be gained for the eye for reflectors of low and 
medium density ; for with these intensities of light and den- 
sities of reflector the brightness of the source has not been 
sufficiently reduced to give much relief to the suffering eye. 
Until this is done in home, office and public lighting, we can- 
not hope to get rid of eye strain with its complex train of 
physical and mental disturbances. Moreover, the principles 
in accord with which the installation is made require that the 
reflector be brought further into the field of vision than is the 
case, for example, when a direct reflector is used. On this 
account a worse result is apt to be obtained with semiindirect 
reflectors of low and medium density than even with equally 
well designed direct reflectors of the same density. 
In the experimental work the following points have been cov- 
ered: The effect of varying the distribution factors on the abil- 
ity of the eye to maintain its maximal efficiency for a period of 
work; the effect of varying the intensity of light with various 
groupings of distribution factors ; and certain miscellaneous ex- 
periments relating to the hygienic employment of the eye. These 
latter experiments include the effect of varying the area, and 
conversely the intrinsic brightness of the ceiling spots above 
the reflectors in an indirect system of lighting; the effect of 
varying the angle at which the light falls on the work in a given 
lighting situation; the effect of using an opaque eye shade 
with light and dark linings with each of the lighting installa- 
tions used in the distribution and intensity series; the effect 
on the efficiency of the fixation muscles of a period of work EFFECT OF CONDITIONS OF LIGHTING ON THE EYE. 
5 
under each of these installations; the effect of motion pictures 
on the eye at different distances from the projection screen; 
and a determination of the tendency of each of the conditions 
of lighting employed to produce discomfort and a comparison 
of the tendency to produce discomfort and to cause loss of 
efficiency. 
The investigations were not abstract in character. All the 
variations obtained were gotten in actual concrete lighting sit- 
uations by employing lighting installations in common use. 
In order that a correlation might be made between lighting 
conditions and the effect on the eye, the following specification 
of illumination effects was made in each case: 
(1) A determination was made of the average illumination 
of the room under each of the installations of lighting used. 
The room was laid out into three-foot squares and illumina- 
tion measurements were made at sixty-six of the intersections 
of these squares and at the point of work. Readings were 
taken in a plane one hundred and twenty-two centimeters above 
the floor with the receiving test plate of the illuminometer in 
the horizontal, the forty-five degree and the ninety degree 
positions, measuring respectively the vertical, the horizontal, 
and the forty-five degree components of illumination. The 
one hundred and twenty-two centimeter plane was chosen be- 
cause that was the height of the test object. In the work on 
the distribution series the illumination was made as nearly as 
possible equal to the point of work. 
(2) A determination was made in candle power per square 
inch of the brightness of prominent objects in the room, such 
as the test surface, the reflectors for the semiindirect installa- 
tion, the reflectors and filament for the direct installation, the 
reading page, the specular reflection from surfaces, etc. The 
brightness measurements were made by means of a Sharp- 
Millar illuminometer with the test plate removed. The instru- 
ment was calibrated against a magnesium oxid surface ob- 
tained by depositing the oxid from the burning metal. By this 
method the reflecting surfaces were used as detached test 
plates. The readings were converted into candle power per 
square inch by the following formula: Brightness = 
Foot-candles. 
II x 144. 6 
EFFECT OF CONDITIONS OF LIGHTING ON THE EYE. 
(3) Photographs were macle of the room from three posi- 
tions under each system of illumination. 
In the selection and use of observers for the work the fol- 
lowing are some of the precautions that were taken: Care was 
exercised, in the first place, to choose only those who had 
shown already a satisfactory degree of precision in other work 
in physiologic optics and whose clinic record showed no un- 
corrected defects of consequence. All were under thirty years 
of age. Before being allowed to take part in the actual work 
of testing, each observer was trained to a satisfactory degree 
of precision in the three minute record under a given lighting 
condition, and in the three hour test under several of the con- 
ditions to be tested. In the actual work of testing the results 
were compiled from several observations, and the precision 
was checked up by the size of the mean variation. No results 
were accepted as significant unless the variation produced by 
changing the condition to be tested was largely in excess of 
the mean variation or mean error for each condition tested. 
This, the accepted conventional check on the influence of 
variable extraneous factors, was carefully applied at each 
step in the work. The following results were obtained: 
(1) Of the lighting factors that influence the welfare of 
the eye, those we have grouped under the heading of distribu- 
tion are apparently fundamental. They seem to be the most 
important we have yet to deal with in our search for the con- 
ditions that give us the minimum loss of efficiency and the 
maximum comfort in seeing. If, for example, the light is well 
distributed in the field of vision and diffuse, and there are 
no extremes of surface brightness, our tests indicate that the 
eye, so far as the problem of lighting is concerned, is prac- 
tically independent of intensity. That is, when the proper 
distribution effects are obtained, intensities high enough to 
give the maximum discrimination of detail may be employed 
without causing appreciable damage or discomfort to the eye. 
(2) For the control of distribution effects given by the semi- 
indirect reflectors of low and medium density and the direct 
reflectors presenting, as most of them do, excessive brilliancies 
due to opening, surface of reflector, or a wholly or partially 
exposed source, our results show unquestionably that too much 
light is being used in ordinary work for the comfort and wel- 
fare of the eye. That is, with these reflectors means have not EFFECT OF CONDITIONS OF LIGHTING ON THE EYE. 
7 
yet been found of producing this amount of light without intro- 
ducing harmful brilliancies into the field of vision. 
(3) The angle at which the light falls on the object viewed 
is an important factor, but not nearly so important, for ex- 
ample, as evenness of surface brightness in the field of vision. 
Extremes of surface brightness in the field of vision seem, 
in fact, to be the most important cause of the eye's discomfort 
and loss of efficiency in lighting systems as we have them at 
the present time. In lighting from exposed sources it is not 
infrequent to find the brightest surface from one million to 
two and one-half million times as brilliant as the darkest; and 
from three hundred thousand to six hundred thousand times 
a£ brilliant as the reading page. These extremes of bright- 
ness in the field of vision are, our tests show, very fatiguing 
to the eye. 
(4) Of the systems of artificial lighting testedthus far, the 
best results have been obtained for the indirect systems, and the 
semiindirect systems with reflectors having a high density. 
By means of these reflectors the light is well distributed in the 
field of vision, and extremes of surface brilliancy are kept 
within the limits which the eyes are prepared to stand. A 
great deal of loss of efficiency has been found to result from 
the use of semiindirect reflectors of low and medium density, 
and from the use of direct reflectors, especially those of shallow 
and medium depth. With regard to the degree of density that 
is most favorable for the eye, the direct reflectors seem, how- 
ever, to present a special case. With reflectors of medium 
depth our best results have been gotten so far with reflectors 
of medium density. This, however, is not in contradiction to 
our principle that extremes of brightness are fatiguing to the 
eye. For in case of the denser reflectors the ceiling and the 
reflectors are dark, while standing out in sharp contrast to 
them is the bright opening of the reflector. Moreover, if the 
physical efficiency of the reflector is not to be lowered by in- 
creasing its density, its opening must become lighter in some 
proportion to the increase of density, for in a totally opaque 
reflector all, and in the denser reflectors nearly all, of the light 
sent to the working plane must come from this opening. In 
the reflectors of medium density, however, the opening need 
not have such a high brilliancy, and there is little contrast 8 
EFFECT OF CONDITIONS OF LIGHTING ON THE EYE. 
between it and its surroundings. When installed on or near 
the ceiling in rooms of moderate height, the best results seem 
to be obtained when the opening, the surface of the reflector, 
and the ceiling have as nearly as possible an equal brilliancy. 
It seems probable that the effect on the eye of the denser 
reflectors can be very much improved by increasing the depth 
of the reflector and by other devices that will lower the 
brilliancy of the opening. 
(5) The problem of installing is not the same for the semi- 
indirect as for the totally indirect reflector. In the latter case 
the height should be so adjusted as to give as nearly as possible 
an even distribution of surface brightness on the ceiling and 
evenness of illumination on the working plane. In the case 
of semiindirect reflectors, especially those of medium densities, 
and in rooms of the height ordinarily found in dwelling houses, 
if the distance from the ceiling is made great enough to pro- 
duce these effects, the bright reflectors are dropped too low 
in the field of vision for the highest comfort and efficiency of 
the eye. Apparently the denser they are the more nearly they 
can afford to be installed as indirect reflectors, and the less 
dense they are the more nearly they should be installed as 
direct reflectors, so far as eye effects of the kind revealed by 
our tests are concerned. In this connection it may be pointed 
out that in current practice direct reflectors for general illu- 
mination are usually installed on the ceiling or as near to it 
as is possible, especially in rooms of low or medium height. 
(6) In the work of providing general illumination the most 
difficult feature presented in the problem of protecting the eye 
is encountered in the lighting of rooms of low and medium 
height. The difficulty decreases with increase of the height 
of the ceiling. In rooms whose ceilings are very high in pro- 
portion to the other dimensions of the room, it seems safe to 
say that comparatively good results should be gotten with 
almost any reflector of modern design; for it is much easier 
in such rooms to get the bright sources of light, primary and 
secondary, out of the zone of most harmful influence on the 
eye. 
(7) The loss of efficiency sustained by the eye in an un- 
favorable lighting situation seems to be muscular, not retinal. 
The retina has been found to lose little, if any, more in effi- EFFECT OF CONDITIONS OF LIGHTING ON THE EYE. 
9 
ciency under one than under another of the lighting systems 
employed (tested by power to discriminate color and bright- 
ness, lag of sensation, rate of exhaustion and rate of re- 
covery ). 
(8) Eye shades are apparently not an adequate substitute 
for lamp shades for the protection of the eye from the source 
of light. The best results were gotten by means of an opaque 
eye shade with a light lining. The usual opaque eye shades 
with a dark lining, while they shield the eye from the source 
of light, do not by any means eliminate harmful brightness 
differences in the field of vision. They in fact create for the 
eye a very unnatural brightness relation-L e., they make the 
whole upper half of the field of vision dark, in sharp contrast 
with the brightly lighted lower half. The direct effect of this 
is a strong brightness induction (physiologic) over the lower 
half of the field of vision which causes glare in surfaces which 
have no glare and increases the glare in surfaces in which glare 
is already present. Moreover, the unusual and strongly irreg- 
ular character of the image formed on the retina probably 
also sets up warfare in the incentives given to the muscles 
which adjust the eye-that is, the upper half of the field of 
vision is dark and presents no detail. The effect of this is 
probably to exert a tendency to cause the muscular relaxation 
characteristic of the darkened field of vision. The lower half 
of the field of vision is light and filled with detail. The in- 
centive here is for the best possible adjustment of the eye for 
the discrimination of detail in the object viewed, while the 
rim of the shade, the sharply marked boundary between the 
light and dark halves of the field of vision and much nearer 
to the eye than the objects viewed, serves as a constant and 
consciously annoying distraction to fixation and accommoda- 
tion. These complex and somewhat contradictory impulses 
given to the muscles of the eye might very well, and doubtless 
do, cause an excessive and unnatural loss of energy and ef- 
ficiency in case of the prolonged adjustment needed for a 
period of work. Translucent shades, when made sufficiently 
opaque to give the necessary reduction to the image of the 
source, darken too much the upper half of the field of vision 
and simulate thereby too much the effect given by the opaque 
shade with the dark lining to give the best results for efficient 
and comfortable seeing. 10 
EFFECT OF CONDITIONS OF LIGHTING ON THE EYE. 
(9) The observation of motion pictures for two or more 
hours causes the eye to lose heavily in efficiency. The loss de- 
creases rather regularly with the increase of distance from the 
projection screen. It seems little if any greater, however, than 
the loss caused by an equal period of working under much of 
the artificial lighting now in actual use. In making these tests 
care was taken to choose a projection apparatus which gave a 
picture comparatively steady and free from flicker. 
(10) In all the conditions tested a rather close correlation 
is found to obtain between the tendency of a given lighting 
condition to cause loss of visual efficiency and to produce ocu- 
lar discomfort. The tendency to produce ocular discomfort 
was estimated by the time required for just noticeable dis- 
comfort tccbet set up with the eye both working and at rest 
under the conditions to be tested. The results of this work 
were also carefully checked up by the determination of the 
mean variation. 
BIBLIOGRAPHY. 
1. Ferree, C. E.: Tests for the Efficiency of the Eye Under Dif- 
ferent Systems of Illumination and a Preliminary Study of the 
Causes of Discomfort. Trans. Ilium. Eng. Soc., 1913, 8, pp. 40-60. 
Untersuchungsmethoden fiir die Leistungsfahigkeit des Auges bei 
verschiedenen Beleuchtungssystemen, und eine vorlaufige Unter- 
suchung iiber die Ursachen unangenehmer optischer Empfindungen. 
Zeit. f. Sinnesphysiol., 1915, 49, pp. 59-78. The Efficiency of the Eye 
Under Different Systems of Lighting. Fourth Intern. Congress on 
School Hygiene, Buffalo, 1913, 5, pp. 351-364. Ophthalmology, July, 
1914, pp. 1-16. Mind and Body, 1913, 20, pp. 280-286, 345-353. The 
Problem of Lighting in Its Relation to the Efficiency of the Eye, 
Science, July 17, 1914, N. S., 15, pp. 84-91. 
Ferree, C. E., and Rand, G.: The Efficiency of the Eye Under 
Different Conditions of Lighting: The Effect of Varying Distribu- 
tion and Intensity. Trans. Ilium. Eng. Soc., July, 1915, 10, pp. 407- 
447. Further Experiments on the Efficiency of the Eye Under Dif- 
ferent Conditions of Lighting. Trans. Ilium. Eng. Soc., July, 1915, 
15, pp. 448-501 
Some Experiments on the Eye with Inverted Reflectors of Dif- 
ferent Densities. Trans. Ilium. Eng. Soc., 1915, 10, pp. 1097-1170. 
A Resume of Experiments on the Problem of Lighting in Its Rela- 
tion to the Eye. Jour, of Philos., Psychol, and Scientific Methods, 
1915, 12, pp. 657-663. 
See also J. R. Cravath: Some Experiments With the Ferree Test 
for Eye Fatigue. Trans. Ilium. Eng. Soc., 1914, 9, pp. 1033-1047; 
and C. E. Ferree: Discussion of Mr. Cravath's paper "Some Exper- 
iments," etc., ibid., pp. 1050-1059. A SIMPLE DAYLIGHT PHOTOMETER 
By C. E. Ferree and Gertrude Rand, Bryn Mawr College 
A need for a simple daylight photometer has long been felt, 
especially in the work of the undergraduate laboratory. The 
impossibility of making determinations of color sensitivity 
even with a degree of precision that is acceptable in under- 
graduate work without constancy of illumination especially 
when pigment papers are used as stimuli is too well known 
to need more than mention here. To make such an instru- 
ment broadly serviceable the following are some of the re- 
quirements which should be met. (a) The instrument should 
be compact and easily portable, (b) It should be so simple 
and inexpensive in construction as to be readily within the 
mechanical resources of the average laboratory. And (c) 
the standard and comparison fields should present little if any 
color difference. 
An instrument which we have constructed especially to meet 
the above needs is shown in Fig. I. It has been in use in 
our laboratory for more than a year and has proven so ser- 
viceable and convenient that we have thought it worth de- 
scribing for the possible benefit of others. It was designed 
and has been used by us primarily for the reproduction of a 
given intensity of illumination rather than for its measurement 
in photometric units, although it can be calibrated and be 
used for photometric measurements. The instrument consists 
of a photometer head, a short bar, a standard tungsten lamp 
with carriage which is moved back and forth along the bar 
by means of a rack and pinion, a millimeter scale which may 
be read outside of the photometer box, a finely graduated 
ammeter to regulate the supply of current to the lamp, and a 
tripod support. When operated as a daylight photometer one 
opening of the photometer head, the bar, and the standard 
lamp with its sliding carriage must be boxed in; and the other 
opening of the photometer head be suitably exposed to the 
illumination that is to be balanced against the light of the 
standard lamp. This boxing can be made as elaborate as one 
chooses or it can be made very simple. In this connection 
the different needs that may arise for a portable photometer 336 
C. E. FERREE AND GERTRUDE RAND 
should be kept in mind. One may want, for example, to 
determine the average illumination, or the distribution of light 
in a room which may or may not be evenly illuminated. To do 
this the room should be laid out in small squares and measure- 
ments be taken in several directions of horizontal, vertical and 
45 degree components of illumination at the corners of these 
squares. For such work it is obvious that a somewhat elabo- 
rate photometer is required, comprising, for example, a test 
plate that can be turned in different directions and a type of 
boxing that will permit of a quick adjustment of the lamp 
and reading of the scale from the outside. Such a photo- 
meter we are required to employ for the specification of the 
lighting effects in our work on the effect of different condi- 
tions of artificial lighting on the eye. An instrument of this 
kind, however, may cost from one hundred and fifty to three 
hundred dollars, which is of course more than is justified for 
the work of the general laboratory. If, however, an instru- 
ment is wanted primarily to reproduce the horizontal com- 
ponent of illumination or the light falling on a vertical sur- 
face such as a campimeter screen, rotating disk, etc., at a 
given point in a room, a very simple boxing is all that is 
required; for all that is needed here is to set the standard 
light at such a position on the bar as will balance the light 
in the room at that point and keep it there as long as that 
intensity of light is wanted. We have found it quite sufficient 
in one instrument we are using to make this boxing of light- 
proof cloth sliding on a suitably constructed' frame. This 
cloth may be folded back to the far end of the frame for the 
adjustment of the position of the lamp or it may be brought 
forward and hooked to the frame of the photometer head 
while the photometric balance is being made. In case of the 
instrument described in this paper a somewhat more elaborate 
but still simple boxing is used. This boxing is made of heavy 
sheet tin painted black outside and inside and carefully light- 
proofed. It is 18 inches long, 4.5 inches wide, and 10 inches 
deep.1 The photometer head forms one end of this box; the 
other end is of fiber fitted with binding posts which connect 
with the line and with cords running to the standard lamp, 
and with a knife switch to make and break the circuit. The 
1 It may seem that the boxing of this instrument is unnecessarily 
deep. It was made deep in order that lamps of ordinary sizes might 
be used as standards. The boxing shown in Fig. II is designed to 
take 25, 40, and 60-watt Mazda lamps and to allow for the adjustments 
of height needed to bring the centers of their filaments in line with 
the center of the opening of the photometer head. If smaller special 
lamps were used so much depth would not be needed and the instru- 
ment could be given a neater appearance. A SIMPLE DAYLIGHT PHOTOMETER 
337 
top of the box is covered with tightly fitting hinged lid which 
permits of a convenient and easy entrance to the box. Pro- 
jecting through the side of the box is a milled head which 
operates the rack and pinion adjustment of the position of 
the standard lamp on the bar. The instrument with the box 
ing is shown in Fig. II. 
The photometer bar is 24 inches long. At one end of this 
bar is a right-angled holder for the photometer head. The 
Figure I 
photometer head is supported on a brass rod 5 inches long 
which passes vertically through an opening in the right- 
angled holder. When adjusted to the height that is wanted 
it is held in position by means of a set screw. The carriage 
for the standard lamp is shown in Fig. I. This carriage is 
fitted also with a right-angled holder and set screw to hold 
the standard lamp and to provide for adjusting its height so 
that the center of the lamp may always be in line with the 
center of the adjacent opening in the photometer head. On 
the bottom of this carriage is fastened a rack 12 inches in 
length which is engaged by the pinion operated by the milled 
head already mentioned. To this carriage is also fastened a 
brass scale graduated in millimeters which extends through 338 
C. E. FERREE AND GERTRUDE RAND 
an opening in the fiber plate forming the encl of the photo- 
meter box opposite to the head. Thus as the lamp is run back 
and forth along the bar its position can be read, outside the 
box, from the divisions on the scale. To facilitate the read- 
ing of these divisions the scale runs immediately back of a 
short pointer fastened to the end of the photometer box. 
The photometer head employed is of the Bunsen type. 
This type of head is especially suitable for our purpose be- 
Figure II 
cause it combines to a favorable degree the features of accu- 
racy and simplicity of construction. The photometer screen 
may be very simply made. In the present case it consists 
merely of two pieces of Hering mat white paper 12.5 cm. long 
and 8 cm. wide smoothly pasted together with the mat side out. 
The screen so formed can be overlaid if desired with magne- 
sium oxide deposited from the burning metal. In the median 
line (horizontal) of this screen. 1.5 cm. from one end a circular 
opening, 1.5 cm. in diameter, with serrated margins, is cut. 
This opening may be filled with a layer of an extra good 
grade of tissue paper or other translucent material, the edges 
of which are held between the two layers of Hering paper. 
It is desirable to have a material to fill this opening whose A SIMPLE DAYLIGHT PHOTOMETER 
339 
coefficient of transmission is as nearly as possible equal to 
its coefficient of - reflection. This screen fits into a groove 
which runs from front to back in the median plane of the 
photometer head. Set into the back of the photometer head 
on either side of the screen and making an angle of about 
65 degrees with it are two mirrors of suitable size in which 
the images of the two sides of the screen are viewed by the 
eye in making the photometric comparison. On either side 
of the photometer head are two openings, 3.25 x 2.5 inches, 
for the illumination of the photometer screen. One of these 
admits the light from the standard lamp, the other the light 
from the room. Both of these openings are filled with a plate 
of single-thick milk glass (Belgian make) ground on one side. 
This glass diffuses the light and gives a more uniform illumina- 
tion of the two sides of the photometer screen. In order that 
the two sides may be illuminated by light of the same color 
quality, color filters must be employed. That is, either the 
standard lamp must be robbed of its excess of yellow and red 
light or the daylight must be colored to match the light from 
the standard lamp. Either of these effects can be readily ac- 
complished by means of thin sheets of colored gelatines, placed 
in the grooves in front of the sheets of milk glass. With gela- 
tines of a low coefficient of selective absorption it is not at all 
difficult to make a good match of the two lights as to color 
quality and thus to eliminate the difficulty that attends the at- 
tempt to make a judgment of equality of brightness between 
two surfaces which differ as to color quality. In making this 
match by means of filters it must be remembered that if the 
match is made by filtering the daylight, a slight physical error 
will be introduced because of the variable composition of day- 
light on different days and at different times of the same day. 
That is, a filter that transmits heavily in the yellow will let a 
greater total of light through when the daylight contains an 
excess of yellow than when it does not. This objection, how- 
ever, is considered by some photometrists to be of more theo- 
retical than practical consequence. To offset this objection 
the greater photometric sensitivity to yellow may perhaps be 
mentioned with some justification. The variable composition 
of daylight also causes some difficulty in maintaining an exact 
color match between the standard light and daylight. A filter 
that produces an exact match at one time may not at another 
time. For this reason it is of advantage to make the filters 
of thin layers of gelatine which can be added to or subtracted 
from with the proper corrections for absorption as the need 
arises. 
The bar carrying the standard lamp and the photometer 340 
C. E. FERREE AND GERTRUDE RAND 
head is supported by a tripod base and stem. The stem con- 
sists of a hollow tube split at the upper end and fitted with a 
collar and set screw. The stem telescopes over a rod 8 inches 
long which is screwed into the photometer bar 8.5 inches 
from the end supporting the photometer head. By means 
of the collar and set screw the apparatus may be adjusted and 
clamped at different heights. 
In order that the standard lamp may be operated directly 
from the line a rheostat and finely graduated ammeter are 
used to regulate and keep constant the supply of current. 
For the sake of portability the ammeter is fastened to a wood 
base which is screwed to two of the feet of the tripod. The 
ammeter is of Weston make, triple range, 0.5, 1, and 1.5 
amperes, combined in one case. The scale of the first of 
these ranges is graduated to 0.01 amperes. On account of 
its size, its graduations, and its comparative inexpensiveness, 
this ammeter is very well suited for the purpose. A speci- 
fication of the rheostat need not be given here. Any good 
rheostat of suitable carrying capacity and range of adjustment 
which permits of fine changes of resistance may be used. 
The use of the apparatus for the reproduction of any given 
illumination is as follows: The rheostat is adjusted to give 
the amperage at which the standard lamp is to be operated. 
A balance is then made at the point in the room in question 
between the light falling on the photometer head and the 
standard light, and a reading is taken of the photometer scale. 
When it is wished to reproduce this illumination the resistance 
is again adjusted to give the reading of the ammeter chosen 
as standard and the light of the room is varied until a photo- 
metric match is obtained. If it is wished to calibrate the 
instrument so that the reading of the scale can be translated 
into foot-candles, for example, a standard lamp is set up at 
such distance from the milk glass test plate on the photometer 
head as will give a balance with the photometer lamp adjusted 
for the different points on the scale. The amount of light 
falling on the test plate can be computed directly from the 
known flux of the standard lamp and the distance of the 
lamp from the test plate. This is correlated with the divi- 
sion on the scale for which the photometric balance is made. 
The different points on the scale are thus gone over one by 
one and the correlative foot-candle values are obtained. Dur- 
ing the calibration the photometer lamp must of course be 
operated at a constant amperage, and in the use of the cali- 
brated instrument this amperage must be reproduced else the 
calibrated values will not be valid. [Reprinted from The Psychological Review, Vol. XXIII, No. 5, Sept., 1916.] 
A SUBSTITUTE FOR AN ARTIFICIAL PUPIL 
BY C. E. FERREE AND GERTRUDE RAND 
Bryn Mawr College 
It is a well-known principle of physiological optics that 
the amount of light condensed into the retinal image of a 
given test object or source of light varies with the size of 
the pupil. It is obvious, therefore, that a complete plan or 
provision for the standardization of the factors which influence 
the precision of a color determination must take into account 
the possible sources of error due to a variable pupillary 
aperture.1 The remedy usually employed when any account 
is taken of the factor is an artificial pupil. Whether, how- 
ever, the use of an artificial pupil is to be recommended is 
seriously open to question. In our own work we have found 
its use to be attended by so many difficulties and to be 
open to so many objections as to be quite inadvisable, unless 
perhaps in case it is wanted to compare the response to 
stimuli differing rather widely in intensity. Space can not 
be taken here for a detailed discussion of these difficulties 
and objections. It will be sufficient perhaps to say that 
with the best possible adaptation of the size of the artificial 
to the natural pupil, its distance from the eye, etc., four 
difficulties remain which seem inherent and very difficult if 
not impossible to overcome, (i) The influence of the 
brightness of the surrounding field on the sensitivity of the 
retina to the test field can not be satisfactorily controlled. 
1 We will say, however, that we believe this variable error to be very small when 
a constant light flux is used for stimulus, and a light-adapted eye is employed, working 
at a moderately high and constant intensity of illumination. We believe this because 
of the very small mean variations we have obtained in determinations of color sensi- 
tivity when all the other factors influencing the response of the eye, which we have 
discussed in previous papers, are controlled and the pupil is left to regulate itself. 
In fact we have not been able so far to reduce appreciably the size of this mean error 
by any artificial regulation of the constancy of the light flux entering the eye from a 
given constant source. Regulation is desirable, however, when the effect on the 
eye of different intensities of light is to be compared. 
380 381 
C. E. FERREE AND GERTRUDE RAND 
Theoretically considered it would not be possible to attain 
this control unless the artificial diaphragm could be brought 
approximately into the plane of the iris. (2) The response 
of the retina can not be investigated out to the peripheral 
limits of the field of vision. With the intensity of light 
attainable with the apparatus described in a preceding 
paper,1 red, blue, and yellow can be sensed out to 920 and 
green to 700 in the nasal meridian. (3) The relation of size 
of pupil to the cross section of the beam of light which the 
artificial pupil admits to the eye can not be under observation 
while the color determination is being made to see whether 
the regulation needed is actually accomplished in any given 
case. The adaptation of the size of the artificial to the 
natural pupil must rest upon a probability established by a 
number of measurements made on the reaction of the pupil 
under a set of conditions as nearly as possible identical with 
those used in the case in question. And (4) the device used 
to give the pupillary aperture can not be gotten so close to 
the natural pupil as not to disturb the adjustment of the 
eye and otherwise to serve as an annoying distraction in the 
field of vision. 
In the apparatus described in the preceding paper the 
method of presenting the stimulus to the eye is such as to 
permit of a substitute for the artificial pupil which so far as 
we are able to determine is entirely free from the objections 
and difficulties attending the use of a diaphragm in front of 
the iris. In this substitute plan instead of cutting down the 
cross section of the beam of light at the eye by means of an 
apparatus which interferes with the natural functioning of 
that organ, the regulation needed is accomplished further 
back in the optical system, out of range of the anterior reac- 
tions of the eye and out of the road of the manipulation 
needed to control the factors which directly influence the 
response of the eye to its stimulus. That is, the stimulus 
light is focused by means of the lens (Z2)2 upon the pupil of 
1 Ferree and Rand, 'A Spectroscopic Apparatus for the Investigation of the 
Color Sensitivity of the Retina, Central and Peripheral,' J. of Exp. Psychol., 1916, 1, 
247-284. 
2 Op. cit., pp. 254-255. 382 
A SUBSTITUTE FOR AN ARTIFICIAL PUPIL 
the eye, forming an image of the analyzing slit of the spectro- 
scope. Obviously the size of this image can be regulated 
by controlling the height of the slit, for example, or by means 
of the lens system so as always to fall within the aperture of 
the pupil reacting to the intensity of the light used in any 
given case. It adds very much to the precision of the regu- 
lation also that its correctness can be checked up if desired 
for every color determination while the determination itself 
is being made.1 That is, not only can the size of image 
needed for a given intensity of light and set of conditions be 
determined empirically in a number of trials but its relation 
to the size of the pupil may be under observation all of the 
time in a series of determinations of sensitivity. With the 
lens system and breadth of slit we are now using, we have 
found that it is not necessary to alter the breadth of the 
image. This would be done, if it were necessary, by means 
of an alteration in the lens system. The height is reduced 
the desired amount by cutting down the height of the ana- 
lyzing slit which was made variable over a wide range pri- 
marily for this purpose, as was stated on pp. 253 and 261 of the 
preceding article: 'A Spectroscopic Apparatus for the Investi- 
gation of the Color Sensitivity of the Retina, Central and 
Peripheral.' This control of the constancy of the amount 
of light entering the eye we have found to be entirely feasible 
and practicable. With it in fact the observation is attended 
with no more difficulty than if the apparatus were used with 
no attempt to exercise this control. 
1 Owing to the brightness and sharpness of the image on the cornea with its 
dark background of iris and pupil, this comparison of the size of the image with the 
size of the pupil is, it is obvious, not difficult to make. The possibility of having this 
feature constantly under observation gives this method of regulating the amount of 
light entering the eye no small advantage over the artificial pupil. [Reprinted from The Psychological Review, Vol. XXIV, No. 2, March, 1917.] 
DISCUSSION 
A NEW METHOD OF HETEROCHROMATIC PHOTOM- 
ETRY-A REPLY TO DR. JOHNSON 
In the September number of this journal appears a discussion 
entitled £A Note on Ferree and Rand's Method of Photometry,' by 
Dr. H. M. Johnson, of the Nela Park Laboratory. This discussion, 
we may perhaps be pardoned for noting, is remarkable chiefly for 
its numerous mistakes and incorrect or misleading representations, 
a few of which we take opportunity here to rectify. The net service 
of the discussion is to call the authors' attention to the omission of a 
decimal point in the original article, for which they duly acknowledge 
their debt. 
I. In his opening paragraph Dr. Johnson says: "The authors 
claim for their method that with respect both to sensitivity and 
reproducibility it surpasses the equality of brightness method, even 
when the photometer head used is of the best Lummer-Brodhun 
type." In regard to this statement we beg to point out that Dr. 
Johnson has omitted from what was actually said all that makes a 
difference between a reasonable and an absurd claim. We had 
claimed in our paper greater reproducibility of setting for the method 
in question as compared with the equality of brightness method 
only in case of heterochromatic photometry, in which respect as is 
well known the equality of brightness method is notably deficient. 
The possibility of a service to heterochromatic photometry alone 
is the reason given in the paper for applying to the rating of artificial 
lights a principle formerly used by us for an entirely different pur- 
pose. Also the special reference to heterochromatic photometry 
was featured in the title. 
2. Dr. Johnson next says: "The authors assumed that the two 
elements making up the photometer screen 'received equal amounts 
of light from the source to be measured.' Even if the elements 
were equidistant from the lamp . . . the truth of this assumption 
does not follow from the data given. In some of the work the 
results of which are presented in the authors' table, the angular 
separation of the compared elements was 140 to 150 at the source. 
Now the radiation from a carbon or tungsten lamp is not equal in 
159 160 
C. E. FERREE AND GERTRUDE RAND 
all directions as is that from an ideal point source. In fact, for 
lamps of such types, differences of several per cent, in different 
directions normal to the long axis of the lamp are the rule, and a 
considerable difference might occur in a range of 150." 
With reference to the above statements we wish to note in the 
first place that it was never assumed by us that there were only two 
elements in the photometer screen. This erroneous interpretation 
of the principle on which the method is based can be attributed to 
Dr. Johnson only. Secondly, that when the angular separation of 
the elements referred to (the stimulus patch and the measuring 
disc), is correctly computed from the data contained in the original 
article it is found to vary between 4.50 and 110,1 and not to have a 
range of 150. And thirdly, that when the question of the influence 
of the distribution curve on the general applicability of the method 
to working practice was raised by us in a paper presented to the 
Philadelphia Section of the Illuminating Engineering Society in 
February, 1914,2 it was the consensus of opinion in the discussion 
that followed that the possibility of error from this source is of neg- 
ligible consequence in a field presenting so many difficulties as 
heterochromatic photometry, and that the effective check on these 
and many other points which were raised by us at that time-in 
addition to those now raised by Dr. Johnson-must come in a 
comparison of the results with those obtained by the equality of 
brightness method. Because of this confirmatory opinion of a group 
of specialists fully familiar with all the technical and working details 
of photometry and because of the check experiments we had run on 
the point to convince ourselves of the negligible influence of the 
factor for the conditions under which we worked (see this paper, 
p. 165), we had not considered it necessary to raise the discussion 
in the preliminary exposition of the principles on which the pro- 
posed method is based, contained in the article in question. How- 
ever, since the point has been raised by Dr. Johnson, the following 
comments may not be out of place. 
1 It is assumed here of course that Dr. Johnson referred to the angle for the colorless 
light. There can have been no reasonable doubt in his mind that the colored light was 
not obtained from the naked carbon or tungsten lamps to which his comments on 
distribution refer. (See footnote, original article p. 9). 
2 With reference to the foregoing point and to others taken up in this discussion it 
is scarcely needful to state that principles and descriptions of conditions of a technical 
nature were taken up in a fuller and more detailed way when a statement of the method 
was presented to auditors technically interested in photometry than was done in the 
article criticized by Dr. Johnson. HETEROCHROMATIC PHOTOMETRY 
161 
(a) A general statement of the type which Dr. Johnson has 
made about the inequality of distribution of carbon and tungsten 
lamps is incomplete to the point of being somewhat misleading. 
As is well known, the distribution curve of an incandescent fila- 
ment lamp depends upon the shape of the filament. While, for 
example, the single oval filament of the ordinary carbon lamp 
gives considerable unevenness of distribution, if wide enough angles 
are considered, the single loop tungsten filament of the Mazda 
lamp, series type, gives a curve which deviates so little from 
Fig. 1. Showing the distribution curve in the horizontal plane of a 50-watt 
carbon lamp, single oval filament-readings taken at 5 ft. radius; lamp operated at 
6.8 horizontal cp.; watts per horizontal cp., 2.97. 
a circle as to be scarcely detectable with the exception of a very 
small region in the plane of the filament. The curves for these 
lamps are appended in Figs. I and 2. In Fig. 3 is given also the 
curve for the ordinary type B Mazda lamp.1 This curve shows 
more variation than the series lamp but it is so nearly uniform as 
to be considered circular for practical purposes. However, neither 
this nor the single oval filament carbon lamp have ever been used 
1 The determinations represented in these curves were made by the photometric 
laboratory of the General Electric Co., Schenectady, N. Y. 162 
C. E. FERREE AND GERTRUDE RAND 
by us in connection with the method of photometry in question 
without some device to secure greater uniformity of distribution of 
light. In case a naked lamp were used at all it has always been of 
the series type, single-loop tip-anchored filament, and care has been 
taken to have the lamp set on the bar so that the light was taken at 
right angles to the plane of the filament or from the most uniform 
part of the curve. But even were a carbon lamp used and the arrow 
Fig. 2. Showing the distribution curve of a 60 cp. series Mazda lamp (clear), 
single loop tip anchored filament, 6.6 amps.-readings taken at 5 ft. radius; lamp 
operated at 60 horizontal cp.; watts per horizontal cp., 1.18. 
or 'fiducial' mark scratched in a plane at right angles to the plane 
of the filament, the distribution would fall off so evenly on either 
side (see Fig. i)1 that the difference in the illumination of the 
stimulus patch and measuring disc, not exceeding 5.50 on either 
side, should be negligible. 
1 It should be noted that in making the cuts for the curves in Figs. I and 3 the 
true deviations from uniformity have been exaggerated by small but considerable 
amounts. HETEROCHROMATIC PHOTOMETRY 
163 
(If) So far as the question of uniformity of angular distribution of 
light is concerned, stress seems to be laid in the criticism on the 
equality of illumination of the stimulus patch and the measuring 
disc alone from the lights to be photometered. This is not at all 
in keeping with a correct interpretation of the method, for the 
photometric balance does not consist in the judgments of the actual 
amounts of light falling on the stimulus patch and measuring disc. 
Fig. 3. Showing the distribution curve in the horizontal plane of a 40-watt G. E. 
Mazda lamp (clear), regular type small bulb, no volts-reading taken at 5 ft. radius; 
lamp operated at 32.5 horizontal cp.; watts per horizontal cp., 1.23. 
The apparent brightness of the stimulus patch is, for example, the 
result of three factors: the actual amount of light falling on the 
stimulus patch, the amount falling on the surrounding screen (rather 
in both cases the amount reflected to the eye), and the physiological 
induction caused by the difference in the brightnesses of these two 
surfaces. Dr. Johnson, however, as indicated above, in considering 
the question of the distribution of the illumination and its probable 
effect on the results of the method, seems throughout his discussion 164 
C. E. FERREE AND GERTRUDE RAND 
to take into account only the relative amounts of light received by 
the stimulus patch and the measuring disc, and in so doing shows a 
fundamental misunderstanding of the principle on which the method 
is based. The illumination of the field surrounding the stimulus 
patch is just as important as the illumination of the stimulus patch 
itself, for it is an equal factor in producing the induction and is, so far 
as any one knows, effective for induction up to the measuring disc; 
and there is, it is scarcely needful to point out, not an angular 
separation of 15° between this screen and the measuring disc. The 
important point is rather that there shall be no effective difference 
in the collective situation influencing the induction and its measure- 
ment for the standard and the comparison lamp. That is, although 
the two surfaces are compared in each judgment, the comparison 
of the two light sources is based on the results of two judgments, 
and if there is no difference in the collective situation influencing the 
two judgments, no injustice is done to the lights compared. If, 
therefore, we were considering with Dr. Johnson the relative illu- 
mination of stimulus patch and measuring disc to the exclusion 
of other factors, and to what degree this relative illumination is 
influenced by the distribution curve of the light source, the impor- 
tant item is not that there is an angular separation between them 
of a given number of degrees and a possible difference of illumination 
in consequence, but how much this varies for the position of the 
standard and comparison lamps on the photometer bar. For the 
nearest position of the standard lamp, the difference in the angular 
separation for the two lamps was n°; for the farthest position for 
the distances as given in the table, the difference would have been 
4.5°. However, for the greater distances that would have been 
required for the standard lamp from the screen to establish a balance 
with the less intense colored lights, a part of the reduction was 
produced by sectored discs, because in the form and set-up of ap- 
paratus employed for that work, distances of light from screen of 
134-160 cm. (Table I., original article, p. 9) could not conveniently 
be attained owing to the angle of the shadow cast by the observer's 
head. This reduction was converted into terms of the law of 
squares to make the results comparable in the table with those 
obtained by the equality of brightness method. The actual setting 
of the lamp on the photometer bar for the greatest of these distances 
was 104 instead of 160 cm. The difference between the angular 
separation of stimulus patch and measuring disc was in this case, 
therefore, 70. The actual range of variation of angular separation HETEROCHROMATIC PHOTOMETRY 
165 
of stimulus patch and measuring disc was thus only from 70 to u°. 
There is, it is obvious, considerable difference between these values 
and the 150 with which Dr. Johnson confronts us. Furthermore, in 
the course of the original work we ran a series of check experiments 
to determine whether this difference in angular separation in case 
of the standard and comparison lights produced any significant 
error. That is, in these check experiments both lights were kept 
in the same position and the light for the more intense, the standard, 
was reduced by means of sectored discs very accurately cut from 
sheet aluminum, the open sectors of which were measured with a 
protractor provided with a Vernier scale reading to minutes. The 
results of these experiments are given in Table I. 
Table I. 
Showing a Comparison of the Results Obtained for the Lights Represented 
in the Original Table when the Photometric Balance was made (a) by 
Changing the Setting of the Lights on the Photometer Bar; and (Z>) by the 
Use of the Sectored Disc 
Source of Colored Light 
Color 
Dis- 
tance of 
White 
Light 
Giving 
Equal- 
ity of 
Illumi- 
nation, 
Cm. 
Ratio of 
Candle- 
power. 
Color; 
White 
Value of 
Open Sec- 
tor Giving 
Equality 
of Illumi- 
nation with 
Distance 
of White 
and Col- 
ored 
Lights 
Equal 
Ratio of 
Candle- 
power. 
Color: 
White 
Differ- 
ence in 
Ratio 
Differ- 
ence in 
Per 
Cent. 
Candle- 
power 
87 cp. 41 cm. distant 
from photometric 
Red 
66.6 
o-379 
137-5° 
O.382 
O.OO3O 
O.785 
screen 
Blue-green 
59-5 
0.4748 
172.0 
0.4778 
0.0030 
O.628 
52 cp. 38 cm. distant 
from photometric 
Red 
82.2 
0-2137 
77-75 
O.2l6o 
O.OO23 
1.06 
screen 
Blue-green 
70. s 
0.2905 
105-5 
O.293I 
O.OO26 
O.887 
13 cp. 38 cm. distant 
from photometric 
Red 
160.0 
0.0564 
20.5 
O.O5694 
O.OOO54 
O.948 
screen 
Blue-green 
134-9 
0.0793 
28.85 
O.O8OI4 
O.OOO79 
0.985 
Moreover, so far as inequalities of illumination of stimulus patch 
and measuring disc are concerned, we may point out that a naked 
lamp was not even used in the experiments the results of which are 
given in the original table. Partly because the colored light was 
secured by means of colored filters, and partly as a precaution 
against unevenness of illumination of stimulus patch, measuring 
disc, and surrounding field for a height and breadth sufficient for 
the purpose of the experiment, the light was placed in a lamp-house 166 
C. E. FERREE AND GERTRUDE RAND 
(see original article, footnote p. 9).1 This lamp-house was 24 cm. 
high, 14 cm. wide and 14 cm. deep. At the lower end of the lamp- 
house was an opening 5 cm. square through which the light passed 
to the screen. The lamp-house was lined with mat white paper so 
shaped as to round off the edges and corners and to give as much as 
possible in the lower part of the enclosure the effect of the segment 
of a sphere. No light passed directly from the lamp to the screen 
as the tip of the lamp was for the different lamps used from 2 to 9 
cm. above the opening for the emission of the light. Owing to the 
high absorption of the Wrattan and Wainwright filters the light 
from the lamp used to establish a balance with that transmitted 
from the filters had to be greatly reduced at the opening of the lamp- 
house by means of colorless absorbing screens, which served further 
to diffuse the light. To determine whether or not any serious dif- 
ference in the distribution of light to measuring disc and stimulus 
patch was present in case of this device, the light was photometered 
at stimulus patch and at measuring disc for each position of the 
lights on the bar. No difference could be detected for these two 
positions by the equality of brightness method. Also the distribu- 
tion curve for the light coming through this opening was found to be 
circular through an angle greater than the n° in question. Our 
statement in the original article then was correct that the equality 
of distance of the measuring disc and stimulus patch on either side 
of the photometer bar guaranteed that they receive equal illumina- 
tion from the light source employed. It would also be true within 
any reasonable margin of error for the single loop filament series 
lamp set as described above (without a lamp house), and even for 
the single oval carbon filament within a margin of error quite ac- 
ceptable for work in heterochromatic photometry. 
3. In a later paragraph (p. 394) Dr. Johnson conveys the im- 
pression that we claim an agreement between the results of the 
new method and those of the equality of brightness method within 
1 The lamp-house is not shown in the photograph of apparatus given in the original 
article. The photograph was a part of the general description of the method and the 
apparatus that might be used with it. In making this photograph the apparatus 
was regrouped, the object being merely to show the type of bar used, the screen and 
the measuring disc. In this photograph it will also be noted that the apparatus was 
not even shown in the position in which it is used in making the determinations. The 
use of a lamp-house is mentioned in another part of the article, namely the part treating 
of the results that were given as a sample of what might be obtained with the method. 
In the first photographs that were made the lamp-house was included, but its size and 
position in the foreground made it appear so disproportionately large that it was decided 
to omit it and to give the photograph the general character mentioned above. HETEROCHROMATIC PHOTOMETRY 
167 
a fraction of one per cent.1 Of this we have to say that no numerical 
value whatever was assigned to the agreement in the original article 
nor was any general statement made that would warrant the infer- 
ence that we claimed an agreement within so small a margin. All 
that appears in the article in this connection is a very brief table of 
results containing no reference whatever to the point in question 
accompanied by an 8-line paragraph stating that the table is ap- 
pended as a sample of the results obtained, that the results are 
averages from 25 determinations, etc. It is the custom in photom- 
etry when a numerical expression is made of agreements, mean 
deviations, etc., to give these in per cent, illumination or per cent, 
candlepower. When this is done for the table in question, the 
agreement shown by the data given falls within 1.5 per cent, instead 
of 'within a fraction of 1 per cent.' as is stated by Dr. Johnson. 
And this it will be remembered, is an agreement in the average. 
When the individual determinations are compared, the deviations 
reach values of the order of + 10 and- 12 per cent. Some idea of 
this may be had from an inspection of the per cent, mean variations 
appearing in the table for the results obtained by the equality 
of brightness method. Thus it will be seen that the actual 
closeness of agreement of results is not surprising. It has been 
made to appear so only by our critic's method of presentation. 
1 On p. 393 Dr. Johnson says: "The authors do not describe their mode of pro- 
cedure in making their measurements by the method of direct comparison. I assume, 
therefore,.... Under these conditions and working with the lamps beyond certain mini- 
mal distances from the photometer head, the luminous intensities of the compared 
sources would be inversely [italics ours] as the squares of their distances from the photom- 
eter screen at valid settings for equality of brightness on the two halves of the photometer 
field." We did not suppose that in an article on photometry it was necessary to give 
a description of the equality of brightness method over 100 years after its principles 
were laid down for all time (Pierre Bouguer, 1760, and Sir Benjamin Thompson, 
Count of Rumford, 1793). However, we do wish to say now that Dr. Johnson has 
raised the question that we conformed to all that is essential in his very elementary 
directions with the exception that we chose rather to follow the custom to which we 
know of no exception either in practice or recommendation, of calculating the luminous 
intensities of the light sources on the basis of the direct squares of the distances of these 
light sources from the photometer head, instead of the inverse squares. In replying 
to an advanced criticism on photometric method, one should not have to point out 
that the law of inverse squares applies to the intensity of illumination at different 
distances from a given source; while the converse of this relation, namely, the direct 
squares, applies to the comparative intensities of two sources which produce equal 
illumination on a given screen or photometer head. That is, the former is used in the 
computations of intensity of illumination: foot-candles, meter-candles, etc.; and the 
latter in the computation of the relative intensities of light sources: candlepower, lam- 
berts, millilamberts, etc. 168 
C. E. FERREE AND GERTRUDE RAND 
4. Also on p. 394 Dr. Johnson presents a table in which it is 
represented that the measuring disc in the work for which our table 
of results was submitted was 3 cm. nearer to the observer than the 
plane of the screen containing the stimulus patch. Applying the 
law of inverse squares he demonstrates that the illumination of the 
stimulus patch and measuring disc was in case of each light source 
unequal. Since the colored lights were all nearer the screen and 
measuring disc than the standard white light in proportions varying 
from 41 /59 to 38/160 (actually 41/59 to 38/ 106 because, as stated 
earlier, a sectored disc was used for the lights requiring the greater 
distance of setting from the screen), the 3 cm. caused a greater 
difference between the illumination of the measuring disc than of 
the stimulus patch for the colored lights than for the white light by 
percentages ranging from 5.4 to 15.5. From the showing of this 
table without further inquiry into causes, it was concluded that 
'the authors' procedure in making the settings was faulty,' the 
'method is insensitive' and that the evidence of agreement of the 
two methods is 'spurious,' for the explanation of which latter point 
there seems to have been no hypothesis worthy of mention but that 
the settings of one method were biased by a knowledge of the settings 
of the other-a smashing and uncompromising arraignment truly! 
However, we beg in passing to say a word of this table ourselves. 
In the first place, as a matter of only minor consequence to the 
present discussion, we wish to point out that in all of the compu- 
tations given by Dr. Johnson of the deviations in per cent, from 
proportionality of illumination of stimulus patch and measuring disc, 
errors have been made, and that in 5 out of a total of 6 cases ap- 
pearing in his table these errors have ranged from 1.8 to 11 per cent, 
of the correct value, with a leaning in some of the most important 
cases towards the advantage of the critic. This, we may be par- 
doned for noting, is under the circumstances somewhat surprising, 
and is of value perhaps chiefly in demonstrating that it is possible 
for mistakes to occur even in a critique levelled at the accuracy of 
the work of others without furnishing a justification for the im- 
pugning of motives and integrities. And secondly we wish to state 
that, as might have been suspected by our critic himself,1 the 3 cm. 
1 The above statement is made for the following reasons, (a) It is obvious on 
a -priori grounds to one having even the least rudimentary knowledge of the principles 
on which photometry is based, that a just balance could not be established between the 
colored and white lights involving so wide a difference in setting on the bar if the 
measuring disc was 3 cm. in front of the photometer screen. And (b) even an approxi- 
mate set-up of the apparatus with the lights in position demonstrates at a glance that HETEROCHROMATIC PHOTOMETRY 
169 
was a typographical error. In the original data still in our posses- 
sion, the distance of the measuring disc from the screen is given as 
.3 cm.1 When the law of inverse squares is applied to this, the dis- 
crepancy of illumination of stimulus patch and measuring disc for 
the distances used by Dr. Johnson in his computations ranges from 
.464 to 1.22 per cent., and for the actual distances used, from .464 
to 1.03 per cent.-an amount which the experienced photometrist 
will, we think, grant is relatively negligible among the much greater 
sources of error present in heterochromatic photometry. 
We have, however, been sufficiently curious to know what results 
would be obtained with the measuring disc placed 3 cm. in front of 
the screen to repeat the work represented in the original table for 
the four highest intensities with this change in the set-up. Dif- 
ferences from the results quoted in the original tables-also, as it 
happens for the cases tested, the amount of deviation from agree- 
ment with the equality of brightness results-ranged from 13.5 to 
25 per cent, when the determination was begun with the weaker 
light, and from 18.6 to 29 per cent, when the determination was 
begun with the stronger light.2 These figures indicate that rather 
than being remarkable for its insensitivity, as is charged by Dr. 
Johnson on the basis of too narrow a consideration of possibilities 
and apparently no first-hand knowledge whatever of the facts in 
question, the method shows by still another test a very high degree 
of sensitivity. 
5. The error in our critic's final conclusion (pp. 395-6) should by 
this time be so obvious as to need no comment. We will, therefore, 
rest our case so far as we recognize that a case has existed, until 
space can be had for a further presentation of results. In this 
regard it is hardly necessary to mention that we do not consider, 
the conditions produced are not compatible with the principles on which the method 
of making the balance is based. For example, when illuminated directly from the 
lamp on the bar a sharp shadow is cast by the disc on the screen, which is plainly 
in the view of the observer at the angle at which the observation is made. This is the 
equivalent of surrounding the disc with a black band which varies in width as the 
position of the lamp on the bar is changed. This is obviously not permissible. In 
fact the error is of a kind which is usually handled in a note of inquiry to the authors. 
1 Also there are, we might mention, a number of witnesses to the set-up of the 
apparatus used by us in the work on heterochromatic photometry. 
2 On account of the limited space allowed, an explanation of why such excessive 
deviations are obtained with this incorrect set-up will have to be deferred until later 
work; also the very obvious explanation of why a greater distance of measuring disc 
from screen was permissible, in fact of advantage, in the work in which the method 
was used to detect changes in the diffuse illumination of an optics-room (Psychol. 
Bull., 1913, 10, p. 371) than when it was applied to the rating of lights on a bar. 170 
C. E. FERREE AND GERTRUDE RAND 
as our critic seems to have thought, that a place has been won for 
our method among those hoary and worn with service on the basis 
of a single sample table appended to a preliminary description of 
method and apparatus and representing the results of only one 
observer for two colors and only six of the possible settings on the 
photometer bar. 
Note.-Dr. Johnson mentioned the use of a rotator toequalize the light radiation in 
different directions; also the deviations found by Wright from Lambert's law of re- 
flection for mat surfaces. Since neither of these points was raised in the original article, 
it might be inferred that they were not known and taken into account by the authors. 
It will probably not be prejudicial to either side of the case to mention here that one of 
the writers supervised the construction of his first lamp rotator for work in photometry 
in 1901 while a teacher of physics, and is well acquainted with the uses and need of a 
rotator. Also in 1903 while a graduate student of physics he was assigned a study of 
the reflection from mat surfaces as a problem for investigation, the object being to 
continue along the lines mapped out by Wright. Both from his reading and in- 
struction with regard to the work of Wright and others, however, he is totally unable 
to concur in a single comment that Dr. Johnson has made on the subject of diffuse re- 
flection in the footnote on p. 392. Dr. Johnson says: "Another source of error which 
the authors appear not to have taken into account may be worthy of mention. The 
angles at which the light was diffusely reflected into the eye from the stimulus patch 
and the disc at the fixation point were not the same. The ■percentage of incident light 
reflected into the eye would have been different, therefore, even if the two surfaces had 
been of the same material. Furthermore, the difference in percentage of incident light 
reflected in the direction of the eye is not constant for any two positions of the source. 
Cf. Wright, H. R., 'Photometry of the Diffuse Reflection of Light on Matt Surfaces,' 
Philos. Trans., 1900, 49, Ser. 5, pp. 199-216." Of the sentences quoted the second is 
the only one that can be said to be true. The angle of emission e from the stimulus 
patch in relation to the eye was approximately o°; while for the measuring disc it was 
250. The reflection, therefore, in the direction of the eye from a given point or unit 
surface in the area fixated of the measuring disc was less than that from the stimulus 
patch by an amount equal to the cosine of 250. Dr. Johnson, however, neglects to take 
into account in considering the case presented by our method that the observation is 
not confined to a single point or unit of area and that the area of surface viewed increases 
as the secant (the reciprocal of the cosine) of the angle at which the surface is 
viewed measured from the normal. That is, the increase of the area viewed just 
compensates for the lessened amount of reflection from unit area. Nutting, for ex- 
ample, says: "A red hot metal plate is of the same brightness viewed at any angle since 
the foreshortening of the area just compensates for the variation in the radiation from 
a given area. Lambert's law holds for mat surfaces for both emitted and reflected 
radiation." Even the author referred to by our critic, in discussing the two possible 
methods of making the photometric determination in his investigation of the reflection 
from mat surfaces, says in effect the same thing (cf. Wright, p. 205), so without exception 
does every other author after whom we have read. Therefore when two mat surfaces 
are observed whose areas are not limited, the apparent brightness of these surfaces is 
the same for different angles of observation provided that the angle of incidence and 
amount of incident light are the same for both-surfaces as was the case for the stimulus 
patch and measuring disc in our work for any one setting of the light on the bar; for HETEROCHROMATIC PHOTOMETRY 
171 
although the reflection from unit area decreases as the cosine of the angle of reflection, 
the area from which the eye receives its light increases as the secant of the same angle; 
from which it follows that the amount of light entering or reflected in the direction of 
the eye is independent of the angle at which the surface is viewed. 
It is obvious, then, that Dr. Johnson's statement that the percentage of incident 
light reflected in the direction of the eye would have been different, even if the two 
surfaces had been of the same material, is not true. From this it is equally obvious 
that his next statement also is not true, namely, that the difference in the percentage 
of incident light reflected in the direction of the eye is not constant for any two posi- 
tions of the source, for as shown above there is no difference in the percentage of 
incident light reflected to the eye from the two surfaces for any given setting of the 
light on the bar. In other words, the possible bearing of Lambert's law and Wright's 
results with regard to this law, is not what Dr. Johnson has stated it to be. Just 
what this bearing is will be discussed further on in this note. What we wish to do 
at this point is to show that even if it were true that the percentage of incident light 
reflected to the eye were different for any one setting of the light on the photometer 
bar, this would make no difference whatever in the results obtained by our method. 
That is, if less light were reflected to the eye from the measuring disc than from the 
stimulus patch for the first light set upon the bar, it would mean merely that the coef- 
ficient of reflection of the measuring disc would have to be reduced by a corresponding 
amount to obtain the match. Then when the comparison light was placed on the 
bar and its distance adjusted until as much light was given to the screen as was re- 
ceived from the first light, the stimulus patch and measuring disc would again match, 
for neither the difference in angle of reflection to the eye nor the reflection coefficients 
would have been changed. Dr. Johnson's point, granting its verity, would have appli- 
cation only if the stimulus patch were illuminated alone by one of the lights and the 
measuring disc by the other and the method of balancing consisted in bringing these 
two surfaces to equality-then it would be necessary that each reflect to the eye the 
same percentage of the light received by it; but the point is clearly quite irrelevant to the 
method described by us in which the two surfaces are illuminated for each judgment by 
only one of the lights, and the balance consists in so adjusting the distance of the two 
lights in the successive judgments that the match for the one based on the amount of 
induction produced at the stimulus patch holds also for the other. In this case it is 
important only that the physical situation and other factors be kept the same for both 
judgments-not that they be equal each to each for the single judgment-for the balance 
is based on the principle that if all the factors are kept constant the amount of induction 
at the stimulus patch will always be the same when the same amounts of light are re- 
ceived on the screen. It is obvious also that the same considerations are true with 
regard to the materials forming the stimulus patch and measuring disc. Moreover, 
with reference to this point, it may also be said that there was, as a matter of fact, very 
little difference in the materials forming the two surfaces; for one sector of the measuring 
disc was identical with the stimulus patch and the other sector was a darker gray of the 
same series of papers (Hering's series of standard grays). 
In concluding our comments on this footnote which has revealed so much of our 
critic's point of view, we will indicate briefly and only in a general way the relation of 
Lambert's law of reflection from mat surfaces and Wright's findings with regard to this 
law to the practical working of our method. As already shown, Dr. Johnson's criticism 
was based both on an erroneous understanding of this law as applied to the making of 
the photometric judgment by any method whatsoever and on a wrong conception of 172 
C. E. FERREE AND GERTRUDE RAND 
the principles of the method criticized. Our actual chance of error in terms of Lam- 
bert's law is that the angle of incidence (Johnson's 'difference in angle of reflection1 
has nothing whatever to do with photometry from mat surfaces) on the stimulus patch 
and its surrounding field is different for the light from the standard and comparison 
lamps when they are of different intensities and a different setting on the bar is required 
to establish the photometric balance. That is, according to Lambert's law the intensity 
of the illumination of the stimulus patch and its surrounding field is proportional to 
the cosine of the angle of incidence (the cosine i). 
Now considering for the sake of simplicity the stimulus patch alone, the variation 
in the cosine of the angle of the incident light for the entire range covered in the work 
criticized from the least to the greatest distance of the source of light from the screen, 
falls within I per cent. While this would mean only a comparatively slight difference 
in the induction situation from the lights compared, we have from the beginning in 
our own thinking frankly faced it as a small source of error in case the reductions of the 
light on the screen are produced by changing the position of the lamps on the bar. 
However, it would not enter in at all, as will be readily seen, if the reduction of light is 
produced by means of a sectored disc or any device: absorbing screen, Nicol's prism, 
grating, etc., which does not change the distance of the source of light from the screen 
and, therefore, the angle of incidence of the light on the stimulus patch. In this 
regard it should be remembered too that our photometer is no more at fault in physical 
principle than the equality of brightness photometer after Rumford as ordinarily con- 
structed, in which also the angle of incidence is changed with a change of the position of 
the light on the bar-not so much at fault perhaps, for compensating factors operate 
in our method of getting the balance which are not present in the Rumford method. 
The relation of Wright's results to the situation described here is that he found that 
there are certain small deviations from the law of the cosine i as the angle of incidence 
is changed. Now just how great the chance of error is in our method from the law of 
the cosine i considered in relation with the results of Wright it is utterly impossible to 
estimate with any acceptable degree of precision from the principles involved for the 
following reasons: (a) The surrounding field as well as the stimulus patch must be 
taken into account in applying the law of cosines. The difference in the angle of in- 
cidence for the different points in this field vary for any two positions of the light on the 
bar-towards zero as a limit, for example, for the points between the stimulus patch 
and the end of the bar, and differently in other directions, lb} The effect is not direct 
but operates through induction, the quantitative relations of which are not definitely 
known. And (c) Wright apparently considered it worth while to make no change of 
angle of incidence smaller than 20°, while the entire range of variation of this angle in 
our work from greatest to least distance of lamp from screen was for the colorless 
light 2° and for both the colored and colorless lights 50. 
Rather, therefore, than indulge in bootless speculation in regard to the possibilities 
of error from these sources, it is obviously much more to the point to get some empirical 
measure of their effective importance. The effective importance of this factor along 
with others not mentioned by Dr. Johnson may be checked up (a) by a comparison of 
results in the average with those obtained by the equality of brightness method (see 
table in original article, p. 9); and lb} still more definitely and directly by comparing 
the results obtained by the method when the reductions of the light on the screen are 
produced by changing the distances of the sources from the screen and when the dis- 
tance of the source and, therefore, the angle of incidence of the light is kept constant 
and the reductions are made by means of a sectored disc (see Table I. of this discussion). HETEROCHROMATIC PHOTOMETRY 
173 
Even had these comparisons not been made, the probable relative unimportance of 
these sources of error as compared with the high variable error obtained for one or any 
small number of determinations by the equality of brightness method, should, we 
think, be obvious to all who have a working familiarity with the latter method in hetero- 
chromatic photometry. On the point of sureness of principle, moreover, it is instructive 
to compare the agreements of the induction and equality of brightness methods shown 
in the tables referred to above with those obtained for the equality of brightness and 
flicker methods, for example, for lights presenting the same amount of color difference. 
Bryn Mawr College, 
C. E. Ferree, 
Gertrude Rand. 
[The above discussion, which exceeds our usual limits, has been 
accepted by the Editors in order that the authors might have ample 
opportunity to clear up the points raised in Dr. Johnson's Note. 
The questions at issue are so specialized and technical that we be- 
lieve it unprofitable to continue the discussion in the pages of the 
Review. A committee of experts acceptable to both parties may 
be suggested as the best means of settling any differences which re- 
main between the writers and their critic.-The Editors.] Reprinted from The Joubnal of Philosophy, Psychology and Scientific Methods, 
Vol. XIV., No. 17, August 16,1917. 
A NOTE ON THE NEEDS AND USES OF ENERGY 
MEASUREMENTS FOR WORK IN PSYCHO- 
LOGICAL OPTICS 
A BRIEF discussion of this subject was given by us five years ago 
in an article, entitled "A Note on the Determination of the 
Retina's Sensitivity to Colored Light in Terms of Radiometric 
Units. ' 'x Since that time some dispute has arisen with regard to the 
comparative merits of the subjective and objective types of measure- 
ment of the stimulus light for work in psychological optics. Time 
alone can, of course, reveal the full range of needs and uses of the ob- 
jective type of measurement. A few words in the way of general 
perspective, however, may not be out of place at this time. 
Considered in its relation to the eye, two points of view may 
be recognized in the rating of lights. One of these is involved in 
their rating for the use of the eye as an organ of seeing. In such a 
rating it is obvious that the method should take into account all of 
the eye's peculiarities of response to the different wave-lengths of 
light. In the production of illuminating effects this is the work of 
photometry, which should be done by the eye or some instrument 
calibrated to give results in terms of the responses of the eye. 
Another and quite a different point of view, however, is involved in 
their rating for the purpose'of investigating the eye's peculiarities 
or characteristics of response in every way in which it is capable of 
giving response. In such work it is obvious that the ultimate method 
of making the rating should be free from the peculiarities to be in- 
vestigated, that is, should not be made by the eye itself. In general, 
in work of this kind, two needs arise. (1) A method of specifica- 
tion is required which will make possible an accurate and convenient 
reproduction of intensities from time to time and from laboratory to 
laboratory. The difficulty of doing this by photometry with lights 
differing widely as to wave-length, as do in most cases the stimuli 
employed in psychological optics, is too well known to need emphasiz- 458 
THE JOURNAL OF PHILOSOPHY 
ing here. Obviously what is needed for certitude in this work, is a 
measuring instrument which can be calibrated directly against the 
standard of radiation, or black body, and which is non-selective in 
its response to wave-length-not an instrument like the eye, the 
selenium cell, the photo-electric cell, or the photographic plate, the 
responses of which are not only selective to wave-length, but vary 
in their amounts of selectiveness with change of intensity of light, 
differ greatly in both of these regards (especially the eye) from 
instrument to instrument or from sense organ to sense organ, and 
can not be calibrated against the total of radiation of a black body. 
The insistence on a subjective method of rating for standardiz- 
ing purposes, when the objective method is available, is not only 
difficult to understand, but is entirely contradictory to current prac- 
tise in other sense fields. No one would think, for example, of 
specifying, for the purpose of securing reproducibility, the weights 
used in an investigation of skin sensitivities in terms of the skin's 
own responses when the means of making the physical measurement 
is at hand; yet there should be more chance of successfully estab- 
lishing from laboratory to laboratory a system of calibration of skin 
measurements in terms of some common standard than there is of 
accomplishing the analogous task in case of light. It is scarcely 
conceivable that the most ardent advocate of subjective ratings in 
case of light would recommend the substitution of the subjective for 
the objective method for work on the skin for the simple reason that 
it would be so undesirable as not to be tolerated unless, for want 
of an objective method, it was rendered absolutely necessary. With 
the objective method available from the beginning, the possibility of 
using the subjective method has not even been raised in work on the 
skin. And indeed the subjective method has been used in rating 
light intensities only because (a) for more than a hundred years no 
other method was available, and (6) it was desirable to rate lights 
for use in seeing by a method which gave results corresponding to 
the eye's powers of response. The former of these reasons for its 
use has now disappeared. Only the latter, with a few laboratory 
exceptions, remains and marks off for the subjective method of 
rating a separate and special field which is clearly recognized as 
such by physicists and the engineers dealing with the problem of 
lighting. 
As a brief, however, for the continuation of the use of the eye 
for the measurement of its own stimuli, although such measurements 
would not be subjective, it may be claimed that in time it will be 
possible to calibrate the eye by means of the non-selective radiom- 
eters so that it can be used to measure the visible energies directly. 
For example, just as it is possible to measure a linear dimension with PSYCHOLOGY AND SCIENTIFIC METHODS 
459 
a meter rod and to convert the results into terms of the English 
system, and vice versa; so it may be possible to measure the differ- 
ent wave-lengths of light by the eye and convert the results obtained 
into energy values. The difficulties in the way of this, as we have 
already pointed out, consist in differences in the sensitivity of dif- 
ferent eyes for a given wave-length; the selectiveness of the eye's 
response to wave-length and to intensity and its variations in both 
of these regards from observer to observer; the lack of a fixed scale 
from observation to observation, even in the case of a single ob- 
server, etc. In short, to complete the analogy suggested above, it 
would be an exceedingly difficult task to convert measurements from 
the metric into the English system and vice versa if very few of the 
measuring rods employed represented the same amounts of linear 
space; if in case of a given rod the dimensions of some objects were 
underestimated and others overestimated and the magnitude of this 
underestimation and overestimation varied with the dimensions of 
the object measured by amounts as yet undetermined, etc., as 
happens in case of the eye's evaluations of the wave-lengths of the 
visible spectrum. Obviously if the eye's ratings are to be converted 
into energy values, this conversion can come only after a very great 
deal of investigation and calibration against radiation standards by 
means, for example, of the non-selective radiometers, which but con- 
stitutes one of the subdivisions of what we have included under the 
second of the needs we are giving for energy measurements in the 
study of the responses of the eye. However, to represent the calibra- 
tion as now completed and available for use instead of scarcely begun, 
as is being done in some quarters, is chimerical and visionary to a 
degree which we can consider compatible only with an inadequate 
knowledge and understanding of all that is involved in the problem. 
(2) The second and perhaps more fundamental need for energy 
measurements for work on the eye is, as stated in the general head- 
ing, for a method of rating the stimulus which will make possible a 
quantitative comparison of the eye's power of response to its stimuli 
in every way in which it is capable of giving a response; for we can 
know the kind and amount of its selectiveness of reaction to the 
different wave-lengths of light only when they are compared with 
those of an instrument as a standard which shows equal power or 
capacity of response to all wave-lengths. Only with such an instru- 
ment, or rather with such an evaluation of the stimuli as a common 
or invariable standard to which to refer the eye's evaluations or 
responses, can the work of comparing its powers or peculiarities of 
response to its stimuli be put on a basis that can be called quanti- 
tative for a single eye or from eye to eye. To this it may be de- 
murred, however, that in some problems it is required as one of the 460 
THE JOURNAL OF PHILOSOPHY 
features of the investigation that the stimuli have equal power to 
arouse the eye's response or sustain some subjective relation to each 
other. This need we have always freely recognized both in our 
work and in our recommendations.2 It in no way conflicts with, 
however, or supplants the more fundamental one already given, but 
is rather supplementary to it in certain types of investigation; for 
even in the cases where the subjective relation is demanded to 
fulfil the requirements of the investigation there is still great need 
for the ultimate purposes of the science that the physical amounts 
of light required to produce this subjective relation for the given 
observer be determined and specified. Again to use the analogy of 
work on skin sensation, it would be a careless investigator indeed 
who would fail to specify, if it were possible to do so, the physical 
measure of the weights he used to give equal pressure responses, 
for example. In short, it seems a paradox that one should even feel 
the need to make a special pleading for the introduction of ob- 
jective measurements into the work of psychological optics when it 
is the current practise to use objective ratings of the stimulus in 
every other field of psychological investigation in which it is possible 
to do so, the intensity ratings in vision and audition alone being the 
conspicuous outstanding exceptions and these being so only be- 
cause adequate methods for making such ratings have been slow in 
coming. 
As examples of needs for regulating the stimuli to give certain 
subjective relations we may quote here the following cases that we 
have already formally recognized. In a recent investigation of 
the comparative lags of the achromatic response to wave-length 
made in our laboratory, the stimuli employed were made photo- 
metrically equal and the amounts of light used to give these equal 
responses were measured radiometrically. The photometric equaliza- 
tions were made because the data were wanted in an interpretation 
of the characteristic overestimations and underestimations found in 
the results of certain observers iin photometry by the method of 
flicker as compared with their results by the equality of brightness 
method. In another case in a determination of whether stimuli 
which have the same power to arouse the achromatic response have 
also the same power to make the eye lose in its capacity to give this 
response as a result of prolonged stimulation, the stimuli were as a 
matter of course made subjectively equal as one of the essential 
conditions of the investigation; but again the amounts of light re- 
quired to produce this subjective relation were determined radio- 
metrically for the purpose of ultimate specification. Also in our 
2 See, for example, American Journal of Psychology, Vol. XXIII., pp. 
329-331. PSYCHOLOGY AND SCIENTIFIC METHODS 
461 
original note on energy measurements we recognized quite broadly 
the possible need of establishing subjective relations between the 
stimuli used. For example, in discussing methods of determining 
after-image and contrast sensitivity, we state: "It is conceivable 
that two points of view may be held with regard to what is meant 
by after-image and contrast sensitivity. (1) After-image and con- 
trast sensitivity may express a relation between the amount of light 
required to arouse after-image and contrast sensations and the unit 
of light used. (2) It may express a relation between the amount of 
light required to arouse the after-image and contrast sensations and 
the amount required to arouse the positive sensation."3 In the 
former case the after-image or contrast sensations are treated as 
one of the eye responses the selectiveness of which to wave-length 
is to be determined; in the latter a figure is sought which expresses 
the relation between the after-image and contrast and the positive 
sensitivities. On the same page and the one following we say: 4 ' Simi- 
larly, two views may be held with regard to the determination of 
the comparative rates of fatigue, and of the development-time of 
sensation. (1) Lights equalized in energy may be used. (2) The 
energy of the lights may be made proportional to the sensitivity of 
the eye to the different colors." Also in discussing the investiga- 
tion of the peripheral limits of sensitivity, we state: " (a) The limits 
may be considered in relation to the comparative sensitivity of the 
retina to the different colors. (6) They may be considered in rela- 
tion to existing color theories. In the first of these problems the 
limits should be obtained with stimuli equalized in energy. So ob- 
tained, the results will constitute merely another expression of the 
comparative sensitivity of the retina to the different colors." 
"The second problem is more complicated and will be made the 
subject of a separate paper." Indeed, as these citations abundantly 
show, we have never failed to recognize that the stimuli in certain 
types of investigation must be made to conform to some type of 
subjective relation, but these investigations constitute in immediate 
importance only a minor part of the work that is to be done in 
getting a thorough knowledge of the eye's characteristics of response; 
and even in these investigations there is as great need of an invari- 
able standard of reference as there is in any field, psychological 
or otherwise, where the value of quantitative work or measurement 
is recognized. 
Perhaps the general character of the discussion will not be 
deviated from too widely if we add in conclusion a word on the 
determination of retinal sensitivities which will indicate in a con- 
crete case the type of treatment that should in our opinion be given 
3 Op. cit., p. 329. 462 
THE JOURNAL OF PHILOSOPHY 
both to the response and to the stimulus, when possible, in quantita- 
tive work in psychological optics. If the sensitivity of the retina is 
to be measured in a way that is comparable with the measurement 
of the sensitivity of the physical recording instruments, two condi- 
tions must be fulfilled: (a) the amounts of response in terms of 
which the comparison is to be made must be numerically comparable; 
and (b) the amounts of stimulus used in arousing the response must 
also be numerically comparable or commensurable. The sensitivity 
of two galvanometers could not be compared, for example, were it 
not known that the divisions on the scale of each were either equal 
or commensurable; likewise the amounts of current used to produce 
the given deflections must be known in terms of the same or com- 
parable units. With the introduction of the radiometric treatment 
of the stimulus the second of the above conditions is fulfilled, and 
for the first time in a way that can be considered quantitative to a 
degree that would be acceptable in rating the sensitivity of a phys- 
ical instrument. With reference to the first condition we are con- 
fronted with a situation somewhat similar to that which obtains in 
heterochromatic photometry. That is, in general, five different 
quantities have been used or suggested in the work of measuring 
sensitivities (the liminal threshold, the just noticeable difference, 
the average error, equal amounts of response and equal sense dif- 
ferences), but only the last two of these conform to the requirement 
that is considered absolutely necessary in determining the sensitivity 
of a physical instrument, namely, that the amounts of response as 
well as the amounts of stimulus must be numerically comparable. 
Moreover, in the absence of sureness of principle in case of the other 
three, the empirical check of agreement in result with those that 
have the needed sureness of principle has never been offered; yet 
sensitivities are determined just as if this condition did not exist, 
comparisons are made and conclusions are drawn. In short it may 
not be out of place to call attention here to the looseness of thinking 
and practise that prevails more or less generally with regard to the 
work of determining physiological sensitivities as compared with the 
analogous physical determinations. For the sake of consistency it 
might well be urged either that this work be revised on the basis of 
the standards set for the physical instruments with all of the inter- 
checking of methods that is needed, or that the term sensitivity 
with its definite quantitative connotation be abandoned in all cases 
in which this standard can not be lived up to. 
C. E. Ferree, 
Gertrude Rand. 
Bbyn Mate College. Vol.XXIV PSYCHOLOGICAL REVIEW PUBLICATIONS Whole No. 103 
No. 2 1917 
THE 
Psychological Monographs 
EDITED BY 
JAMES ROWLAND ANGELL, University of Chicago 
HOWARD C. WARREN, Princeton University (Review') 
JOHN B. WATSON, Johns Hopkins University (J. of Exp. Psych.) 
SHEPHERD I. FRANZ, Govt. Hosp, for Insane (Bulletin) and 
MADISON BENTLEY, University of Illinois (Index) 
Radiometric Apparatus for Use in 
Psychological and Physiological 
Optics 
Including a Discussion of the Various Types of Instru- 
ments that have been used for Measuring 
Light Intensities 
BY 
C. E. FERREE and GERTRUDE RAND 
Bryn Mawr College 
PSYCHOLOGICAL REVIEW COMPANY 
PRINCETON, N. J. 
and LANCASTER, PA. 
Agents: G. E. STECHERT & CO., London (2 Star Yard, Carey St., W. C.); 
Leipzig (Koenigstr., 37); Paris (16 rue de Cond6) PREFACE 
Six years ago, realizing the fundamental relation of energy 
measurements to quantitative work in psychological optics, 
we undertook to procure a non-selective radiometer which not 
only would be sufficiently sensitive for work in the visible spec- 
trum, but the operation of which would be within the technical 
possibilities of the laboratories in which research is done in 
psychological optics. At that time we decided upon an instru- 
ment of the surface type because (a) we wished to measure all 
of the light falling on the opening of our campimeter screen 
rather than compute it from several measurements with the 
linear type of instrument; and (b) we believed that sensitivity 
would be added to the instrument in some proportion to the in- 
crease in area of the receiving surface (see this paper p. 12). 
A quick acting surface thermopile, because of its superior steadi- 
ness and ease of operation, seemed to be best adapted to our 
purpose. Since such an instrument could not be obtained in 
the market, Dr. W. W. Coblentz of the radiometric division of 
the Bureau of Standards who had at that time just finished 
a comparative study of the radiometric instruments showing the 
above-mentioned advantages of the thermopile, undertook the 
design and construction of the thermopiles, surface and linear, 
and the auxiliary radiometric apparatus which we have used in 
our work. These thermopiles are the first of their type made 
by Dr. Coblentz (see Bulletin of the Bureau of Standards, 1913, 
p, pp. 15-29) and have been used by us for five years. Thermo- 
piles of this type are now in use also in many physical labora- 
tories and in other laboratories in which a sensitive and con- 
venient means is needed for measuring spectrum energies.* We 
*For example, the recent work of Nutting (Phil. Mag., 1915, 29, (6), p. 
301), Ives, Coblentz and Kingsbury (Phys. Rev., 1915, 5, (2), p. 269), and 
Coblentz and Emerson (Bull. Bur. of Standards, 1917, 14, p. 167), on the 
visibility of radiation and the mechanical equivalent of light, of interest to 
psychologists, has been done and was made conveniently possible by these 
improved thermopiles designed and constructed by Dr. Coblentz. IV 
PREFACE 
are not thus recommending in the following pages an apparatus 
the feasibility and convenience of which for quantitative work 
of the kind needed in psychological optics is untried. 
Of late some dispute seems to have arisen with regard to the 
need and uses of energy measurements for work in psychological 
optics. A brief discussion and statement of opinion on this 
point was given by us in an article published in the American 
Journal of Psychology in 1912. (A Note on the Determination 
of the Retina's Sensitivity to Colored Light in Radiometric 
Units, 23, pp. 328-332.) Time alone can of course reveal 
the full range of needs and uses of this type of measurement 
and render a just verdict on disputed points. A few words in 
the way of general perspective, however, may not be out of 
place here. 
Considered in its relation to the eye two points of view may 
be recognized in the rating of lights. One of these is involved 
in their rating for the use of the eye as an organ of seeing. In 
such a rating it is obvious that the method used should take 
into account all of the eye's deviations from equality of response 
to the different wave-lengths of light. In the production of 
illuminating effects this is the work of photometry which should 
be done by the eye or some instrument calibrated to give results 
in terms of the responses of the eye. Another and quite dif- 
ferent point of view, however, is involved in their rating for 
the purpose of investigating the eye's peculiarities or character- 
istics of response in every way in which it is capable of giving 
response. In such work it is obvious that the ultimate method 
of making the rating should be free from the peculiarities to 
be investigated, that is, should not be made by the eye itself. In 
general in work of this kind two needs arise. (1) A method of 
specification is required that will make possible an accurate and 
convenient reproduction of intensities from time to time and 
from laboratory to laboratory. The difficulty of doing this by 
photometry with lights differing widely as to wave-length as 
do in most cases the stimuli employed in psychological optics, 
is too well known to need emphasizing here. Obviously what 
is needed for certitude in this work is a measuring instrument PREFACE 
V 
which can be calibrated directly against the standard of radia- 
tion, or black body, and which is non-selective in its response 
to wave-length,-not an instrument like the eye, the selenium 
cell, the photo-electric cell, or the photographic plate, the re- 
sponses of which are not only selective to wave-length but vary 
in their amounts of selectiveness with change of intensity of 
light, differ greatly in both of these regards (especially the eye) 
from instrument to instrument or from sense organ to sense 
organ, and can be calibrated against the radiation standard or 
black body, if at all, only with a great deal of difficulty and 
with many chances of cumulative error. 
The insistence on a subjective method of rating for stand- 
ardizing purposes when the objective method is available, is 
not only difficult to understand but is entirely contradictory to 
current practice in other sense fields. No one would think, for 
example, of specifying for the purpose of securing reproduci- 
bility, the weights used in an investigation of skin sensitivities 
in terms of the skin's own responses when the means of making 
the physical measurement is at hand; yet there should be more 
chance of successfully establishing from laboratory to laboratory 
a system of calibration of skin measurements in terms of some 
common standard than there is of accomplishing the analogous 
task in case of light. It is scarcely conceivable that the most 
ardent advocate of subjective ratings in case of light would 
recommend the substitution of the subjective for the objective 
method for work on the skin for the simple reason that it would 
be so undesirable as not to be tolerated, unless for want of an 
objective method it was rendered absolutely necessary. With 
the objective method available from the beginning, the possibility 
of using the subjective method has not even been raised in work 
on the skin. And indeed the subjective method has been used 
in rating light intensities only because (a) for more than a 
hundred years no other method was available, and (b) it was 
desirable to rate lights for use in seeing by a method which gave 
results corresponding to the eye's powers of response. The 
former of these reasons for its use has now disappeared. Only 
the latter, with a few laboratory exceptions, remains and marks VI 
PREFACE 
off for the subjective method of rating, a separate and special 
field which is clearly recognized as such by physicists and the 
engineers dealing with the problem of lighting. 
As a brief, however, for the continuation of the use of the 
eye for the measurement of its own stimuli, although such meas- 
urements would not be subjective, it may be claimed that in time 
it will be possible to calibrate the eye by means of the non- 
selective radiometers so that it can be used to measure the visible 
energies directly. For example, just as it is possible to measure 
a linear dimension with a meter rod and to convert the results 
into terms of the English system, and vice versa; so it may be 
possible to measure the different wave-lengths of light by the 
eye and convert the results obtained into energy values. The 
difficulties in the way of this, as we have already pointed out, 
consist in differences in the sensitivity of different eyes for a 
given wave-length; the selectiveness of the eye's response to 
wave-length and to intensity and its variations in both of these 
regards from observer to observer; the lack of a fixed scale from 
observation to observation, even in the case of a single observer; 
etc. In short to complete the analogy suggested above, it would 
be an exceedingly difficult task to convert measurements from 
the metric into the English system and vice versa if very few 
of the measuring rods employed represented the same amounts 
of linear space; if in case of a given rod the dimensions of some 
objects were over-estimated and others under-estimated and the 
magnitude of this over-estimation and under-estimation varied 
with the dimensions of the object measured by amounts as yet 
undetermined; etc.,-as happens in case of the eye's evaluations 
of the wave-lengths of the visible spectrum. Obviously if the 
eye's ratings are to be converted into energy values, it can come 
only after a very great deal of investigation and calibration 
against radiation standards by means, for example, of the non- 
selective radiometers, which but constitutes one of the sub- 
divisions of what we have included under the second of the 
needs we are giving for energy measurements in the study of the 
responses of the eye. However, to represent the calibration as 
now completed and available for use instead of scarcely begun, PREFACE 
VII 
would be chimerical and visionary to a degree which we can 
consider compatible only with an insufficient knowledge and 
understanding of all that is involved in the problem. 
Since the foregoing was written, Troland (Journal of Experimental 
Psychology, 1917, 2, pp. 7-13) has advised that, instead of the thermopile 
or other non-selective radiometer, the psychologist may, with sufficient ac- 
curacy for his purpose, use the eye as a selective radiometer and convert 
the results into units of energy by means of a value for the mechanical equiv- 
alent of light that has recently been determined by Nutting (Philos. Mag., 
1915, 2g, (6), p. 301). This point can not be discussed here in detail. We 
would, however, recommend that the reader consult this work on the mechan- 
ical equivalent, which has been done by means of the flicker photometer and 
the thermopile, and judge for himself how unreliable it would be to attempt 
to follow Dr. Troland's advice and use a result obtained with a given limited 
group of observers for only one intensity of light, to convert the photo- 
metric results of individual observers in other laboratories and for other 
intensities of light into anything at all closely approximating the correct 
energy values. It is obvious that in order to make the conversion in any 
given case with the same order of accuracy with which the direct energy 
measurements may be made, the same observers would have to be used, the 
same state of adaptation and sensitivity of the eye, the same intensity of 
light or approximately so (at least so far as adequate proof to the contrary 
for a large part of the intensity scale is concerned), the exact same range of 
wave-lengths and distribution of energy within the group of wave-lengths, and 
the same degree of purity of light as were used in making the original de- 
termination of the visibility curve which is meant to serve as the basis for 
making the conversion. Considering the first of these points alone, it will be 
remembered that Ives, working through the spectrum with the flicker photo- 
meter, found in a group of eighteen observers disagreements as great as 159 
per cent for .487^; 114 per cent for .498/1; 26 per cent for .518^; 18 per cent 
for .537/z,; 13 per cent for .556/1; 10 per cent for .576/1; 28 per cent for -595/x; 
65 per cent for .615/^; 86 per cent for .635//,; and 122 per cent for .655/t. 
(Philos. Mag., 1912, 24, Ser. 6, pp. 856-863.) From this showing of low agree- 
ment from observer to observer with the flicker photometer, it is clear that the 
results for individual observers could not be used for the purpose of making 
the conversions recommended unless some means were had of correcting these 
results to those of the group for which the original visibility curve and the 
mechanical equivalent were determined. Space can not be taken here to 
discuss the complications and approximations that would be involved in 
making such a correction. It will be sufficient to say that if it were made 
in the most approved manner-an adequate method of doing it has not by 
any means as yet been devised-and the mechanical equivalent were applied, 
the results would scarcely be accepted as correct even by an ordinarily care- 
ful worker unless they could be checked up by a direct energy measurement. 
In this connection it is interesting and important to note that the visibility 
curve as determined by Nutting does not agree with that determined by 
Ives, also that the curves of Nutting and Coblentz agree only when certain VIII 
PREFACE 
corrections are made in the energy measurements of Nutting. In short the 
attempt to get a set of figures that will express for the different wave-lengths 
the relation of the lumen as evaluated by a number of eyes to the watt is an 
interesting bit of work and may present, perhaps, when the proper computa- 
tions are made, a rough analogy to the determination of the mechanical 
equivalent of heat; but the attempt to use these figures to convert the photo- 
metric results of the individual observers in the different laboratories into 
the correct energy values is quite a different matter, and can scarcely be con- 
sidered as the intent of those who have made the determination. This ques- 
tion will be discussed in greater detail in a later paper. 
However, the idea of using the eye indirectly to determine the energy 
values of light is by no means new. Before the direct type of measurement 
had been made as feasible as it now is, several attempts were made to use 
the eye for this purpose. (See, for example, Lummer and Pringsheim. 
Jahresber. d. Schles. Ges. f. vaterl. Kultur, 1906, pp. 95-97; Beibl., 1907, p. 466; 
Thiirmel, Das Lummer-Pringsheimsche Spektral-Flickerphotometer als op- 
tisches Pyrometer, Ann. der Phys., 1910, 33, (4), pp. 1139, 1160; etc.) 
(2) The second and perhaps more fundamental need for en- 
ergy measurements for work on the eye is, as stated in the gen- 
eral heading, for a method of rating the stimulus which will 
make possible a quantitative comparison of the eye's power of 
response to its stimuli in every way in which it is capable of 
giving a response; for we can know the kind and amount of 
its selectiveness of reaction to the different wave-lengths of light 
only when they are compared with those of an instrument as 
standard which shows equal power or capacity of response 
to all wave-lengths. Only with such an instrument, or rather 
with such an evaluation of the stimuli as a common or invariable 
standard to which to refer the eye's evaluations or responses, 
can the work of comparing its powers or peculiarities of response 
to its stimuli be put on a basis that can be called quantitative 
for a single eye or from eye to eye. To this it may be demurred, 
however, that in some problems it is required as one of the fea- 
tures of the investigation that the stimuli have equal power to 
arouse the eye's response or sustain some subjective relation to 
each other. This need we have always freely recognized both 
in our work and in our recommendations. (See Amer. 
Jour, of Psychol., 1912, 25, pp. 328-332.) It in no way con- 
flicts with, however, or supplants the more fundamental one 
already given, but is rather supplementary to it in certain types PREFACE 
IX 
of investigation; for even in the cases where the subjective rela- 
tion is demanded to fulfill the requirements of the investigation 
there is still great need for the ultimate purposes of the science 
that the physical amounts of light required to produce this sub- 
jective relation for the given observer be determined and speci- 
fied. Again to use the analogy of work on skin sensation, it 
would be a careless investigator indeed who would fail to specify, 
if it were possible to do so, the physical measure of the weights 
he used to give, for example, equal pressure responses. In 
short, it seems a paradox that one should even feel the need 
to make a special pleading for the introduction of objective 
measurements into the work of psychological optics when it is 
the current practice to use objective ratings of the stimulus in 
every other field of psychological investigation in which it is 
possible to do so, the intensity ratings in vision and audition 
alone being the conspicuous outstanding exceptions and these 
being so only because adequate methods for making such ratings 
have been slow in coming. 
As examples of needs for regulating the stimuli to give cer- 
tain subjective relations we may quote here the following cases 
that we have already formally recognized. In a recent investiga- 
tion of the comparative lags of the achromatic response to wave- 
length made in our laboratory, the stimuli employed were made 
photometrically equal and the amounts of light used to give 
these equal responses were measured radiometrically. The photo- 
metric equalizations were made because the data were wanted 
in an interpretation of the characteristic overestimations and 
underestimations found in the results of certain observers in 
photometry by the method of flicker as compared with their 
results by the equality of brightness method. In another investi- 
gation now in progress, namely, a determination of whether 
stimuli which have the same power to arouse the achromatic re- 
sponse have also the same power to make the eye lose in its ca- 
pacity to give this response as a result of prolonged stimulation, 
the stimuli are as a matter of course being made subjectively 
equal as one of the essential conditions of the investigation; but 
again the amounts of light required to produce this subjective re- X 
PREFACE 
lation will be determined radiometrically for the purpose of ulti- 
mate specification. Also in our original note on energy measure- 
ments we recognized quite broadly the possible need of establish- 
ing subjective relations between the stimuli used. For example, 
on p. 329 in discussing methods of determining after-image and 
contrast sensitivity, we state: "It is conceivable that two points 
of view may be held with regard to what is meant by after-image 
and contrast sensitivity. (1) After-image and contrast sensitivity 
may express a relation between the amounts of light required 
to arouse after-image and contrast sensations and the unit of 
light used. (2) It may express a relation between the 
amount of light required to arouse the after-image and contrast 
sensations and the amount required to arouse the positive sensa- 
tion." In the former case the after-image or contrast sensations 
are treated as one of the eye's responses the selectiveness of 
which to wave-length is to be determined; in the latter a figure 
is sought which expresses the relation between the after-image 
and contrast and the positive sensitivities. On the same and 
the succeeding page we say: "Similarly two views may be held 
with regard to the determination of the comparative rates of 
fatigue, and of the development-time of sensation. (1) Lights 
equalized in energy may be used. (2) The energy of the lights 
may be made proportional to the sensitivity of the eye to the 
different colors." Also in discussing the investigation of the 
peripheral limits of sensitivity, we state: "(a) The limits may 
be considered in relation to the comparative sensitivity of the 
retina to the different colors, (b) They may be considered in 
relation to existing color theories. In the first of these problems 
the limits should be obtained with stimuli equalized in energy. 
So obtained the results will constitute merely another expression 
of the comparative sensitivity of the retina to the different 
colors. The second problem is more complicated and will be 
made the subject of a separate paper." Indeed as these citations 
abundantly show, we have never failed to recognize that the 
stimuli in certain types of investigation must be made to con- 
form to some type of subjective relation, but these investigations 
constitute in immediate importance only a minor part of the PREFACE 
XI 
work that is to be done in getting a thorough knowledge of the 
eye's characteristics of response; and even in these investigations 
there is as great need of an invariable standard of reference 
as there is in any field, psychological or otherwise, where the 
value of quantitative work or measurement is recognized. 
Perhaps the general character of the discussion will not be 
deviated from too widely if we add in conclusion a word on the 
determination of retinal sensitivities which will indicate in a 
concrete case the type of treatment that should in our opinion 
be given both to the response and to the stimulus, when possible, 
in quantitative work in psychological optics. If the sensitivity 
of the retina is to be measured in a way that is comparable with 
the measurement of the sensitivity of the physical recording 
instruments, two conditions must be fulfilled: (a) the amounts 
of response in terms of which the comparison is to be made 
must be numerically comparable; and (b) the amounts of stim- 
ulus used in arousing the response must also be numerically com- 
parable, or commensurable. The sensitivity of two galvanome- 
ters could not be compared, for example, were it not known 
that the divisions on the scale of each were either equal or com- 
mensurable; likewise the amounts of current used to produce 
the given deflections must be known in terms of the same, or 
comparable units. With the introduction of the radiometric 
treatment of the stimulus the second of the above conditions 
is fulfilled, and for the first time in a way that can be considered 
as quantitative to a degree that would be acceptable in the rating 
of the sensitivity of a physical instrument. With reference to 
the first condition we are confronted with a situation somewhat 
similar to that which obtains in heterochromatic photometry. 
That is, in general five different quantities have been used or 
suggested in the work of measuring sensitivities (the liminal 
threshold, the just noticeable difference, the average error, equal 
amounts of response and equal sense differences), but only the 
last two of these, so far as has yet been demonstrated, conform 
with certainty to the requirement that is considered absolutely 
necessary in determining the sensitivity of a physical instrument, 
namely, that the amounts of response as well as the amounts of XII 
PREFACE 
stimulus must be numerically comparable. Moreover, in the 
absence of sureness of principle in case of the other three, the 
empirical check of agreement in result with those that have the 
needed sureness of principle has never been offered; yet sensitiv- 
ities are determined just as if this condition did not exist, com- 
parisons are made and conclusions are drawn. In short it may 
not be out of place to call attention here to the looseness of 
thinking and practice that prevails more or less generally with 
regard to the work of determining physiological sensitivities as 
compared with the analogous physical determinations. For the 
sake of consistency it might well be urged either that this work 
be revised on the basis of the standards set for the physical in- 
struments with all of the interchecking of methods that is needed, 
or that the term sensitivity with its definite quantitative connota- 
tion be abandoned in all cases in which this standard can not be 
lived up to. TABLE OF CONTENTS 
PAGE 
I. Introduction I 
II. Methods and apparatus that have been used for the 
measurement of light intensities and their applica- 
bility to the investigation of retinal sensitivities.... 2 
A. The thermopile 9 
1. Important points in the construction of sen- 
sitive thermopiles 10 
a. The metals used to form the thermo- 
electric junctions 10 
b. The size of the wire used in forming 
the couples 11 
c. The dimensions of the pile and the 
number and arrangement of the ther- 
mo-couples 12 
d. The type of connection of the couples 13 
e. The relation of internal to external re- 
sistance 13 
f. Nicety of construction 14 
g. Provisions to secure steadiness of re- 
sponse 15 
2. Advantages of the thermopile 15 
B. The Nichols radiometer 16 
1. Significant points with regard to the radio- 
meter 17 
2. Comparative advantages and disadvantages 
of the radiometer 17 
C. The radio-micrometer 18 
D. The bolometer 20 
1. Important points in the construction of sen- 
sitive bolometers 21 
a. The kind of material used for the re- 
ceiving surface 21 
b. The area of the receiving surface. ... 21 
c. The thickness of the strip used as a re- 
ceiver 21 
d. The most favorable resistance of bolo- 
meter and balancing coils 22 
XIII XIV 
CONTENTS 
PAGE 
e. The slide-wire for balancing the resist- 
ance 22 
f. The protection of the bolometer from 
air currents 22 
g. The strength of current 23 
2. Points in the construction of the auxiliary 
galvanometer 23 
3. Possible sources of difficulty in the use of the 
bolometer 25 
4. The comparative advantages and disadvan- 
tages of the bolometer 25 
E. The selenium cell 26 
1. Points to be considered in the construction 
and use of selenium cells 27 
a. Method of preparing the selenium. ... 27 
b. Purity of the selenium 28 
c. Material and size of the electrodes. . . 28 
d. Strength of the battery current 29 
e. Direction of the battery current 30 
f. Duration of the battery current 30 
g. Temperature at which cell is operated 30 
h. Pressure 31 
i. Moisture 31 
j. Age of cell 31 
k. The amount of polarization gradually 
set up in the cell 32 
1. Photo-electric currents 32 
2. Factors which render it difficult to use the 
selenium cell for quantitative work either 
as an ohmic resistance or as a light-measur- 
ing instrument 32 
3. Factors which apply especially to its use as a 
light-measuring instrument 33 
a. The preexposure of the cell 33 
b. The time of exposure to the light to 
be measured 34 
c. Wave-length of spectrum light 34 
(1) Factors which have been found 
to influence the selectiveness of 
response to wave-length for a 
given cell 35 
(a) Intensity 35 CONTENTS 
XV 
PAGE 
(b) Length of time of ex- 
posure . 35 
(c) Temperature, humidity 
and voltage 36 
(d) Photo-electric currents.. 36 
(2) Factors which have been found 
to influence the selectiveness of 
response to wave-length in dif- 
ferent cells 36 
d. Intensity of mixed or white light. ... 37 
4. Theories of the action of light on selenium. . 39 
F. The photo-electric cell 41 
1. Factors that have been taken into account in 
the construction and use of photo-electric 
cells 42 
a. The metal used to hold the negative 
charge 42 
b. The residual gas 42 
c. The pressure of the residual gas 43 
d. The potential difference between anode 
and cathode 43 
e. The galvanometer or electrometer em- 
ployed . 43 
f. The angle of the incident light 44 
g. Dark effects and after-effects 44 
h. Fatigue effects 44 
2. Comparative advantages and disadvantages 
of the photo-electric cell 45 
G. The photographic plate 47 
1. The blackening of the plate and the factors 
which influence this action 47 
2. The possibilities of using the photographic 
plate in quantitative work 52 
H. The eye 52 
1. The two possibilities of using the eye in 
light measurements 52 
2. The comparative advantages and disadvan- 
tages of the use of the eye for balancing 
energies of light of the same spectro-radio- 
metric composition 53 
III. A convenient and sensitive radiometric apparatus for 
work in psychological and physiological optics 54 XVI 
CONTENTS 
PAGE 
A. The linear thermopile 55 
B. The surface thermopile 57 
C. The radiation standard 59 
D. The galvanometer 61 
E. The auxiliary apparatus 63 
1. Sensitivity tester for the galvanometer .... 63 
2. Special resistance coils 64 
3. The telescope and scale 64 I. INTRODUCTION 
In a previous paper the purpose has been expressed of describ- 
ing apparatus for work on the color sensitivity of the retina con- 
sisting of spectroscopic and radiometric features. In partial 
fulfillment of this purpose apparatus was described in a recent 
number of the Journal of Experimental Psychology1 designed to 
meet the following needs: (1) to stimulate any part of the retina 
with the light of the spectrum and to control as desired the condi- 
tions of preexposure and surrounding field; and (2) to regulate 
the amounts of light used within the small gradations needed for 
threshold and just noticeable difference determinations. It is the 
purpose of the present paper to describe apparatus with which it is 
possible to specify the amount of light used in energy units. 
This completes the description of a group of apparatus by means 
of which it is possible to determine the sensitivity of the eye 
to wave-length in terms that are commensurable, and thus to 
place the investigation of the responses of the eye on a methodo- 
logical plane comparable with the study of the responses of the 
physical recording instruments. This could not be done until 
a means was had of estimating light intensities which is not 
only independent of the achromatic and chromatic functioning of 
the eye itself, but which gives results directly proportional to the 
physical value or energy of the light waves. An instrument 
which gives responses directly proportional to the intensity of the 
light-waves is, we scarcely need to point out, equally sensitive to 
all wave-lengths. With the responses of such an apparatus as 
standard, the deviations of the eye from equal sensitivity to the 
different wave-lengths can readily be determined and compared. 
1 Ferree and Rand. A Spectroscopic Apparatus for the Investigation of 
the Color Sensitivity of the Retina, Central and Peripheral. J. of Experi- 
mental Psychology, 1916, 1, pp. 247-283. 
1 II. METHODS AND APPARATUS THAT HAVE BEEN 
USED FOR THE MEASUREMENT OF LIGHT IN- 
TENSITIES AND THEIR APPLICABILITY TO 
THE INVESTIGATION OF RETINAL 
SENSITIVITIES 
No problem in optics probably has presented more difficulty to 
the investigator and the various committees which have been 
appointed for the purpose by scientific and engineering societies, 
bureaus, etc., than that of standardizing the intensity of lights 
differing in color value. In the investigation of retinal sensitivi- 
ties the problem of standardizing presents two aspects. (i) As 
the prime requisite of scientific work a method of specification is 
needed that will make possible an accurate and convenient repro- 
duction of light intensities. Without this no investigation can 
have the certitude that comes from repetition and verification. 
And (2) an important item in the determination of retinal sensi- 
tivities has been a comparison of the sensitivity to different wave- 
lengths. This has been made a feature of the general problem 
for the sake both of knowing the characteristics of the eye as a 
sense-organ and measuring instrument, and of being able to meet 
the many practical needs that have arisen in the attempt most 
effectively to adapt light to the service of the eye in the work of 
lighting, etc. As we have already pointed out, if the sensitivity 
or responsiveness of the eye to lights of different wave-lengths is 
to be compared, it is obvious that a common unit must be had, 
independent of the functioning of the eye itself, in terms of which 
to measure the quantity or intensity of the light employed; or to 
'express the need in another form, the light to be used in the in- 
vestigation should be rated by instruments whose responses are 
directly proportional to the energy value of the light waves.1 
Such instruments being non-selective in their response to wave- 
1 For a further statement of the conditions that must be fulfilled if the 
sensitivity of the retina is to be measured in a way that is comparable with 
the measurement of the sensitivity of the physical recording instruments, see 
Preface, p. xi. 
2 RADIOMETRY IN PSYCHOLOGICAL OPTICS 
3 
length and giving the true physical value of the stimulus are the 
logical standard of reference in the comparative study of instru- 
ments or organisms whose responses are selective. Fortunately 
instruments for measuring light intensities which fulfill the above 
requirements have within the last few years reached such a stage 
of advancement as to make this kind of treatment of the prob- 
lem not only possible but feasible and even convenient. On this 
account a brief history may not be out of place here of the at- 
tempts that have been made to attain measurements of light 
intensities which are purely physical. 
Light of any wave-length is universally conceded to be a form 
of motion in a transmitting medium. By common agreement 
among physicists quantitative estimates of motion are made in 
terms of what is called energy of motion; or of mass and rate of 
motion. Owing to the small quantities of energy involved in the 
waves of the visible spectrum, it is obvious that light energies can 
not be estimated directly in terms of mass and rate of motion. 
Some instrument or apparatus which responds to light must be 
used and the response of this instrument be calibrated against a 
source of energy the radiation from which per unit of surface 
per unit of time is known. Once calibrated, such an instrument 
with proper checks on its sensitivity may be used for the measure- 
ment of the visible radiations from any source. The requisites 
of a satisfactory instrument for the physical measurement of 
light are obviously as follows. (i) It must give a response which 
is directly proportional to the energy of the light wave or must 
be capable of calibration against an instrument which does give 
such responses. (2) It should be non-selective in its response to 
wave-length and to intensity, i.e., it should be no more sensitive 
to one wave-length of the spectrum than to another and its sensi- 
tivity should not vary with the intensity of the light used. This 
requirement is an obvious corollary to the preceding. If an in- 
strument is used which is selective in its response to wave-length, 
the amount of its selectiveness must be a constant else correction 
factors cannot be determined which will be valid for all intensi- 
ties. (3) It should be sufficiently sensitive to respond to the small 
amounts of light present in the visible spectrum. And (4) it 4 
C. E. FERREE AND GERTRUDE RAND 
should give results which have a satisfactory degree of reproduci- 
bility, or if erratic within known limits or conditions it must be 
calibrated against some instrument which does give reproducible 
results, and correction factors be determined. 
As comparators of light intensities the following instruments 
and the human eye have at various times been employed or in- 
vestigated,-the Nichol's radiometer, the radio-micrometer, the 
micro-radiometer, the bolometer, the thermopile, the selenium 
cell, the various types of photo-electric cell, and the photographic 
plate.2 The following comparisons may be made of these.instru- 
ments with regard to the above mentioned requirements, (a) 
The radiometer, the radio-micrometer, the micro-radiometer, the 
bolometer, and the thermopile, depending initially for their action 
on heating effects, give responses which are directly proportional 
to the energy of the incident light. They are, therefore, non- 
selective in their reaction both to wave-length and to intensity.3 
The selenium cell, the photo-electric cell, the photographic plate 
and the human eye, however, do not give responses which are 
proportional to the energy of the incident light. They are all 
known to be selective in their reaction to wave-length; and the 
amount of this selectiveness in case of the selenium cell, the 
photographic plate and the human eye has been found to change 
with the intensity of light, (b) All the instruments which are 
selective in their responses to wave-length, namely, the human 
eye, the selenium cell, the photo-electric cell and the photographic 
plate, have a high degree of sensitivity to light. The photographic 
plate possesses the additional advantage that the action may be 
integrated over an interval of time. The instruments which are 
non-selective to wave-length are as a class less sensitive to light. 
Recent improvements in the construction of such instruments, 
2 The use of these instruments for the measurement of light is based on 
the following effects produced by incident light: (i) heating effects; (2) a 
change in the resistance of certain metals to the flow of a current; (3) a 
decrease in the power of certain metals to hold a negative charge in a 
partial vacuum; (4) chemical action; and (5) visual sensation,-used chiefly 
in connection with the various types of photometer. 
3 It should be mentioned, however, that their windows absorb selectively 
some of the wave-lengths of the invisible spectrum. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
5 
however, have increased their sensitivity greatly. Of this class 
of instruments, the thermopile because of its greater ease of con- 
trol and greater reliability is probably best adapted for use in 
laboratories of physiological and psychological optics. Moreover, 
as was stated in our introduction, it has been developed to a high 
degree of sensitivity. In fact a comparative study of the non- 
selective instruments has shown that in the present stage of de- 
velopment of such instruments, the thermopile possesses as high 
a sensitivity as the others when operated in air and probably also 
when operated in a vacuum, (c) The factors which influence 
the response and use of the instruments which are selective in 
their action to wave-length have proven to be so hard to control 
that the results obtained have shown a comparatively low degree 
of reproducibility. Of the non-selective instruments the bolo- 
meter is perhaps the hardest to control. The factors which in- 
fluence the action of the thermopile, the radiometer, and the 
radio-micrometer are on the other hand comparatively easy to 
control. A comparative statement of the advantages and disad- 
vantages of these instruments will be given later in the paper. 
Before considering these instruments in greater detail, it may 
be of service perhaps to give a brief statement of the type of 
action that is produced in each by the radiant energy falling upon 
its receiving surface. As was stated earlier in the discussion, the 
measurement of energy by the types of instrument that are men- 
tioned here is not direct. The instrument is available because it 
gives to a greater or lesser degree some regular and constant type 
of response to radiant energy, the value of which in energy units 
is determined by calibration against the known radiations from a 
black body, or from some other source whose radiations have been 
determined by comparison with that from a black body. In 
instruments of the type of the radiometer, micro-radiometer, 
radio-micrometer, bolometer and thermopile, all radiations are 
transformed into heat at the receiving surface of the instrument. 
In case of the radiometer, for example, the absorbed energy pro- 
duces thermodynamic effects in the rarified gas contained in the 
housing of the instrument, which in turn causes regular deflec- 
tions of a delicately suspended vane; in case of the radio-microm- 6 
C. E. FERREE AND GERTRUDE RAND 
eter and the thermopile, the absorbed energy acting upon a 
thermo-electric couple causes a flow of current which deflects the 
needle of a sensitive galvanometer in circuit with it; and in case 
of the micro-radiometer and bolometer, the absorbed energy 
changes the resistance to the flow of current in a delicately bal- 
anced electric circuit which is also detected by means of a sensi- 
tive galvanometer. The action of the remaining instruments is 
not due to heating effects; also these instruments are not respon- 
sive to all radiations. The selenium cell and the eye, for example, 
are sensitive only to the visible spectrum (Brown and Sieg, how- 
ever, Phys. Rev., 1914, 4, (2), pp. 48-61, report one cell that has 
considerable sensitivity as far out as .85^.) the photographic 
plate, when properly sensitized to red, and the photo-electric cell 
are sensitive both to the visible and the ultra-violet radiations. 
In case of the selenium cell the visible radiations falling on a 
strip of metallic selenium placed in one arm of a delicately bal- 
anced electric circuit so change the resistance of the selenium 
to the flow of current that the electromotive balance between the 
two arms of the circuit is disturbed, and a flow of current takes 
place between two given points which were before at equal poten- 
tials. This current deflects a galvanometer. The use of the 
selenium cell is attended with a great deal of difficulty and there 
are many opportunities for cumulative error. The following is a 
brief statement of some of these difficulties. A detailed statement 
will be given later in the paper. (1) As an instrument to be 
used in the process of measuring, it can be employed without cali- 
bration (e.g., the determination of sensitivity curves for differ- 
ent intensities of light expressing a relation between response 
and energy) only to identify equal amounts of energy; and since 
it is as a general case responsive only to light waves it can be 
used to equalize only light energies. Its employment in this way 
as a measuring instrument for the visible spectrum energies pre- 
supposes, therefore, a standard light source, the energy values of 
the visible spectrum from which are known, against which to 
balance the unknown lights. But as stated earlier in the paper, 
the light energy emitted from sources ordinarily available can 
not be determined directly. It must be determined by compari- RADIOMETRY IN PSYCHOLOGICAL OPTICS 
7 
son with the radiations from some body the amount of which 
can be directly estimated. This comparison may be most con- 
veniently made by means of some measuring instrument such as 
the thermopile which is responsive to the total of radiation and 
which is non-selective in its response to wave-length, and a black 
body radiating known amounts of energy to furnish the standard 
for the comparison. In short, without the possibility of ultimate 
recourse to such instruments as the thermopile, the radio-micro- 
meter, etc., which are non-selective in their response to wave- 
length, instruments of the class of the selenium cell would be 
practically useless for radiometric purposes. Moreover, the two- 
fold nature of the measuring operation, the difficulty of main- 
taining constancy of conditions in the employment of the sec- 
ondary standard, and more especially the many factors extrane- 
ous to light which influence its response, make its use very liable 
to error. And (2) since the selenium cell is selective in its re- 
sponsiveness to the different wave-lengths of light, the standard 
light source must in every case be of the same spectro-radiometric 
composition as the light against which it is to be balanced, other- 
wise the cell can not be relied upon to give to the unknown light 
a fair radiometric evaluation. That is, if the light to be meas- 
ured is not of the same wave-length or composition as the stan- 
dard light, correction factors have to be used which represent 
the amount of the selectiveness of action. Furthermore, the 
amount of selectiveness of the action changes with the intensity 
of light, therefore correction factors established for one intensity 
will not serve for all intensities. 
The action of the photo-electric cell depends on the power of 
light to cause certain metals to lose a negative charge of electricity 
in a partial vacuum. Much that has just been said of the selenium 
cell applies also to the photo-electric cell. (1) It is not sensitive 
to the infra-red spectrum, hence can not be calibrated directly 
against the total of radiation of a black body. (2) It is selective 
in its response to the different wave-lengths of the visible spec- 
trum. Griffith and Dember claim that it is also selective to in- 
tensity. (3) Its use in measuring the energy of the visible spec- 
trum presupposes, for example, either some calibration similar 8 
C. E. FERREE AND GERTRUDE RAND 
to that noted above for the selenium cell, or the availability of a 
light source the values of the visible radiations of which are 
known to serve as a standard against which to balance the un- 
known wave-lengths. And (4) its sensitivity is influenced by so 
many factors difficult of control as to give it a comparatively low 
reproducibility of response. 
The photographic plate responds to light by a chemical change 
in its sensitive film, known as the 'blackening" of the plate. Its 
convenient use as an energy measuring instrument depends upon 
whether or not this blackening sustains any constant relation to 
the amount of incident light. If not, its use would necessitate 
such an elaborate calibration as to render it impracticable as a 
radiometer. Like the selenium and photo-electric cells, it too is 
selective both to wave-length and to intensity; its employment 
as an energy measuring instrument presupposes a standard light 
source, the energy value of the radiations from which is 
known; and its responses are subject to the influence of many 
variable factors which tend to give them a low degree of repro- 
ducibility. 
The eye gives two responses to light waves, the chromatic and 
the achromatic. As yet the achromatic response alone has been 
used in the measurement of light intensities. Two possibilities 
are presented for the use of the eye as a measuring instrument: 
photometric or the rating of lights in terms of their power to 
arouse the achromatic sensation; and radiometric in the sense of 
balancing or equalizing the energy values of lights of the same 
spectro-radiometric composition. As an energy comparator the 
eye is like the selenium cell in the following regards. (1) It is 
responsive only to the visible spectrum. Its employment, there- 
fore, presupposes the provision of a light source, the energy value 
of the visible radiations from which are known. And (2) since 
it is selective in its response to wave-length, it can without cor- 
rection factors be used to establish an energy balance only be- 
tween lights having the same spectro-radiometric composition. 
While not generally used or classed as radiometric, the eye can 
like the selenium and photo-electric cells be used very sensitively 
to balance energies of lights of the same spectro-radiometric RADIOMETRY IN PSYCHOLOGICAL OPTICS 
9 
composition and has in this respect a similar claim to be con- 
sidered as one of the radiometric possibilities. In fact our con- 
trol of the factors which influence the response of the eye is per- 
haps enough greater than that of the selenium cell, the photo- 
electric cell, etc., to render its use for this purpose preferable 
from the standpoint of precision. 
A. The Thermopile.4 
The thermopile is probably the most celebrated of the radio- 
metric instruments. To it we are indebted for the researches 
of Melloni and Tyndall as well as for the most notable 
advances that have been made in the study of radiation. 
The instrument was invented by Nobili and is based on a discov- 
ery made by Seebeck about 1820 that when two wires of different 
metals are joined end to end so as to form a closed circuit, an 
electric current passes around the circuit when one of the junctions 
is heated or cooled, and this current continues to flow as long as 
any difference of temperature exists between the two junctions.5 
4 With regard to the non-selective radiometers we are indebted heavily to 
the publications of Dr. Coblentz for data and for guidance in the compila- 
tion of data. 
5 There are three thermo-electric effects in metals: the Seebeck effect, the 
Peltier effect, and the Thomson effect. The Seebeck effect is described above 
and is the one on which the action of the thermopile is based. The Peltier 
effect discovered in 1834 is the converse of the Seebeck effect, i.e., when a 
current is passed through a junction of dissimilar metals, the junction is 
either heated or cooled depending upon the direction of the current with 
reference to the thermo-electric relation of the metals. For example, if the 
current passes from the electro-negative to the electro-positive, work is done 
and the temperature of the junction is raised; but if it passes from the 
electro-positive to the electro-negative, the temperature of the junction is 
lowered. The result of the Peltier effect in a thermo-couple, therefore, is 
to lower the temperature of the exposed junction. This effect, however, is 
not considered to be sufficient to make an appreciable change in the results 
gotten with a thermopile-galvanometer combination of the sensitivity 
ordinarily obtained. The Thomson effect is a heat effect manifested when a 
current flows between points at different temperatures in the same metal. 
This effect differs in different metals. For example, when a current flows 
from a hot to a cold point in copper, it evolves heat; but when it flows from 
a cold to a hot point, heat is absorbed. In iron, however, the reverse is true. 
When the current flows from a hot to a cold point, heat is absorbed. This 
effect for the small temperature differences involved is also considered negli- 
gible by Altenkirch (Phys. Zeit., 1909, 10, p. 560) in his discussion of the 
efficiency of thermopiles. 10 
C. E. FERREE AND GERTRUDE RAND 
Like the bolometer the thermopile owes its effective sensitivity in 
part to its own construction and in part to the auxiliary galva- 
nometer. 
1. Important points in the construction of sensitive thermopiles. 
The problem in thermopile construction appears to be to secure a 
low resistance, a low heat capacity and heat conductivity, and a 
high thermo-electric power. The following have been found to be 
important points in the construction of thermopiles, a. The 
metals used to form the thermo-electric junctions. This point is 
of importance because metals are found to differ in their thermo- 
electric power, i.e., in their electromotive force per degree centi- 
grade when compared with the standard metal, lead. The fol- 
lowing are some of the thermo-electric metals: bismuth, silver, 
German silver, lead, platinum, copper, zinc, iron, antimony, con- 
stantan, tellurium, and selenium. A very small amount of im- 
purity may make a great difference in the thermo-electric power 
of a metal, and some of the alloys and metallic sulphides show a 
very high thermo-electric power. Some of the combinations 
most commonly used in making thermo-couples are bismuth and 
antimony, iron and constantan, and bismuth and silver. The 
bismuth and silver couple has been chosen by Coblentz because of 
its high thermo-electric power and low resistance. Silver 
was selected to complete the element with bismuth more 
especially because of its low resistance, its pliability and the ease 
with which it can be cleaned6 and annealed. The latter two points 
are of great importance in the construction of the pile. Nicety 
of construction is of fact of greater importance to a high radia- 
tion sensitivity, Coblentz declares, than a high thermal E. M. F.7 
provided the material has a correspondingly high resistance. 
6 It is important that the metal chosen be easily cleaned for completeness 
of contact in soldering. A preliminary heating can be given the silver wire 
which serves the double purpose of cleaning and annealing. This preliminary 
heating could not, for example, be given to copper and iron wire. 
7 Coblentz (Bulletin of Bureau of Standards, 1914, II, pp. 148-150) found, 
for example, in eight samples of bismuth wire with diameters of 0.06, 0.08, 
0.1 and 0.15 mm. that the thermo-electric power when coupled with silver 
varied from 75 to 82 microvolts per degree, depending upon the purity of 
the material. Haken (Verh. Phys. Gesell., 1910, 12, p. 229) and Gelhoff and RADIOMETRY IN PSYCHOLOGICAL OPTICS 
11 
b. The size of the wire used in forming the couples. The chief 
defects in the older types of thermopiles were their great heat 
capacity and their consequent lag in reaching a temperature 
equilibrium. The larger the wire used in making the couple, the 
greater, of course, will be the heat capacity. In the recent at- 
tempts that have been made to improve the linear thermopile a 
prominent item of change has been the use of finer wires, which 
not only decreases the heat capacity and lag and increases the 
radiation sensitivity, but permits more elements to be placed in a 
given area. The decrease in the size of the wire, however, in- 
creases the internal resistance which must of course be taken into 
account in planning for sensitivity. For example, Coblentz8 
found in experiments with surface thermopiles that a bismuth 
wire 0.15 mm. in diameter had sufficient heat capacity to require 
a half minute to attain thermal equilibrium, while a wire o. 1 mm. 
in diameter gave satisfactory results. Using this wire in con- 
junction with one of silver 0.0513 mm. in diameter as a standard 
of sensitivity, a silver wire of 0.041 mm. in diameter gave a sensi- 
tivity of 1.13; one of 0.03 mm. in diameter, a sensitivity of 1.20; 
and one of 0.021 mm. diameter, a sensitivity of only 1.12. That 
is, when the wire has reached an optimum fineness, any further 
decrease in size so increases the internal resistance as to more than 
Neumeier (ibid., 1913, 15, p. 876) found that an alloy of bismuth with 9 to 
10 per cent antimony gives a thermo-electric power which varies from 77 to 
87 microvolts. Coblentz (Op. cit., p. 149) found that an alloy of 5 to 6 per 
cent of tin gives a thermal E. M. F. of -44 to -45 microvolts per degree; 
and a thermo-element made of high grade bismuth and this alloy gives a 
thermo-electric power of 125 to 127 microvolts per degree. 
While having 50 to 60 per cent higher thermo-electric power than a bis- 
muth-silver pile, piles made of the bismuth alloy showed only about 10 
per cent higher radiation sensitivity. The alloy is so much harder to handle 
that the same nicety of construction is not possible, also as high a durability 
is not attained. Since in making the silver-bismuth couple a bead of tin is 
used in soldering the two wires together, an alloy of bismuth and tin is made 
at the junction. 
A bismuth-iron thermo-element (op. cit., pp. 151-154) was found to give 
a thermal E. M. F. which was 18 per cent higher than was obtained from 
bismuth and silver. No increase of radiation sensitivity was obtained, how- 
ever, because the initial resistance was almost doubled by the use of the iron. 
8 Coblentz, W. W. Bulletin Bureau of Standards, 1913, 9, pp. 21-22. 12 
C. E. FERREE AND GERTRUDE RAND 
compensate for the advantage gained by the lessened heat capa- 
city. Johansen further says9 that the radii of the two wires of the 
thermo-element should be so chosen that the ratio between the 
heat conductivity and the electrical resistance is the same in both. 
c. The dimensions of the pile and the number and arrange- 
ment of the receiving thermo-couples. In a recent theoretical con- 
tribution to the construction of thermopiles for the measurement 
of radiant energy, more especially the construction of vacuum 
thermopiles, Johansen10 arrives at the conclusion that the radia- 
tion sensitivity is proportional to the square root of the exposed 
surface in case of the thermopile as it is in case of the bolometer. 
In extensive experimental determinations of the point, however, 
Coblentz11 finds (a) that in single thermo-couples the sensitivity 
is not proportional to the square root of the area exposed to 
radiation, but that the area has an optimum value which gives a 
considerably higher sensitivity than is required compatible with 
the square root law; and (b) that the highest sensitivity is at- 
tained by building up a composite receiver of elements having in- 
dividual receivers of a size giving the maximum sensitivity.12 It 
is obvious, therefore, that sensitivity can be added to the instru- 
ment by increasing the total area of the receiving surface and con- 
sequently the number of thermo-couples, the individual receivers 
of which make up the total area; and that the maximum increase 
can be attained by having each individual receiver of the optimum 
size. In one of his more recent models of linear thermopiles 
9 Johansen, E. S. Ann. der Phys., 1910, 33, (4), p. 517. 
10 Johansen, E. S. Loc. cit. 
11 Coblentz, W. W. Bulletin of Bureau of Standards, 1914, 11, p. 142. 
12 From the data obtained in constructing the receiving surface in this 
way, he concludes that the gain in sensitivity over what is indicated by the 
square root law amounts probably to as much as 50 per cent. 
According to Coblentz the requisite of the optimum size is that it shall 
absorb radiant energy at a rate which will just compensate for the loss of 
heat from conduction along the wires. If this size is exceeded, the loss 
from emission becomes even greater than the loss by conduction along the 
wires and the two together operate to give less than the maximum difference 
of temperature attainable between the "hot" and "cold" junctions of the 
couple. A lag in reaching a thermal equilibrium also results because the 
heat is drained off from the center of the receiver faster than from the 
edges by conduction along the wire. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
13 
Coblentz13 uses, for example, 22 junctions of bismuth and silver 
mounted in a space 10.5 mm. long. The width of this pile was 5 
mm. and its resistance was 10.8 ohms.14 In the surface thermopile 
greater sensitivity may of course be attained than in the linear. 
The surface pile is in effect built up of contiguous linear piles. 
d. The type of connection of the couples. In the older types 
of thermopile it was the custom to connect the couples in series. 
Coblentz15 has found, however, that it is of advantage to substi- 
tute a series-parallel connection. In the series connection one 
thermo-couple is attached to each of the overlapping receivers on 
the front of the pile, while in the series-parallel arrangement two 
couples are soldered to each receiver. The effect of this type of 
connection is in the first place to reduce the number of overlapping 
receivers by one-half. This reduces the superfluous metal at the 
lap and the amount of insulation required, and gives the apparatus 
a quicker response. And secondly the internal resistance is re- 
duced to one-fourth what it would be if the elements were all con- 
nected in series; so that although their E. M. F. is reduced by 
one-half by the series-parallel arrangement, there is a gain of 
from 10 to 12 percent, in radiation sensitivity. 
e. The relation of internal to external resistance. It has been 
a commonly accepted principle in the construction of thermopiles 
that the highest sensitivity is attained when the resistance in the 
thermopile is equal to the resistance of galvanometer and con- 
necting wires. Rayleigh,16 for example, in his computation of the 
thermodynamic efficiency of the thermopile has shown that the 
useful work done externally attains a maximum when the external 
resistance is equal to the internal resistance. In these computa- 
tions only the specific resistances and the thermal conductivities 
were considered. Altenkirch,17 1909, however, contends that the 
13 Coblentz, W. W. Bulletin of Bureau of Standards, 1913, 9, p. 292. 
14 The linear pile that has been used in the work that has been done in this 
laboratory is of this type with the exception that the receiving surface, de- 
signed for spectroscopic work, has a breadth of only 2 mm. 
15 Coblentz, W. W. Bulletin of the Bureau of Standards, 1914, 11, pp. 
138-142. 
16 Rayleigh. Phil. Mag., 1885, 20 (5), p. 361. 
17 Altenkirch, E. Phys. Zeit, 1909, 10, p. 560. 14 
C. E. FERREE AND GERTRUDE RAND 
external resistance may be two or three times the internal resist- 
ance without seriously affecting the maximum efficiency of the 
thermopile, and Coblentz,18 1914, finds that the external resistance 
may be two or three times the internal resistance without decreas- 
ing the sensitivity of the instrument more than 5 to 10 percent. 
f. Nicety of construction. Coblentz makes the statement that 
the attainment of a high radiation sensitivity in a thermopile is 
at the present stage of development of thermopile making mainly 
a question of nicety of construction, for upon this more than 
any other point depends the low heat capacity, conductivity and 
emissivity needed for a sensitive instrument. The following are 
some of the points that should be taken into account in attaining 
the most effective relation between capacity, conductivity and 
emissivity,-the kind of materials, the size and form of the wires 
used for the couples, the length of wire, the size of the receiving 
surface, the type of connection of the couples, the relation of size 
of slit to size of receiving surface, the amount of insulation 
material, etc.19 The object to be attained by a low heat capacity, 
conductivity and emissivity is of course that the energy falling on 
the receiving surface shall cause a maximum rise of temperature 
and that there shall be as little lag as is possible in the rise to 
18 Coblentz, W. W., op. cit., p. 175. 
19 Coblentz attributes a great deal of his success in the construction of 
thermopiles to the use of his electrically heated welding device. (See Bull. 
Bureau of Standards, 1913, 9, p. 16) ; to the choice of silver wire which is 
easily cleaned and annealed; and to his use of pure tin in the process of 
welding which produces an alloy which is not brittle. 
He cites cases to show the effect on sensitivity of deviations from the 
general method of construction. For example, the central line of receivers 
of one thermopile was given an additional coat of shellac to cause the indi- 
vidual receivers to adhere, instead of causing the adhesion by merely 
moistening the insulating layer with alcohol. The instrument was slow in 
responding to a radiation stimulus and was besides insensitive. This extra 
shellac was then removed by means of blotting paper wet with alcohol, and 
the surfaces resmoked. The radiation sensitivity was increased 40 to 50 
per cent. In another case a thermopile was made of 0.1 mm. wire pressed 
flat. This thermopile had a radiation sensitivity 25 to 30 per cent less than 
the average sensitivity of a number of similar thermopiles made of round 
wire. The flat wire which presented a greater surface for radiation in- 
creased the emissivity and thus lowered the sensitivity of the instrument. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
15 
thermal equilibrium. When this is attained the instrument will 
respond quickly and give its maximum response. 
g. Provisions to secure steadiness of response. The main 
source of unsteadiness of response is exposure to air currents. 
This of course can be completely eliminated by isolating the in- 
strument from the air. The best success of isolation is evacua- 
tion which doubles the sensitivity. Water jackets and combina- 
tions of water and air jackets have been used also. Unlike the 
bolometer, however, the thermopile is noted for its steadiness of 
response in air. This is one of the strongest recommendations 
for its general use. 
Older forms of thermopiles used by Melloni, Tyndall and 
others were subject to a "drift"; i.e., there was a permanent 
E. M.F. which caused a permanent deflection of the galvanometer. 
This permanent E. M. F. seems at least in part to have been due 
to lack of symmetry in the construction of the "hot" and "cold" 
junctions. In our own linear thermopile this tendency to drift 
was overcome by soldering on the "cold" junctions receiving 
surfaces of tin of the same dimensions as were carried by the 
"hot" junctions. 
2. Advantages of the thermopile. (1) It is non-selective in 
its response to wave-length. (2) It is readily portable and is 
easily adapted to the many needs for which a sensitive radio- 
meter is needed. (3) In its most improved forms it is very 
steady in its action. Even when used in air there is compara- 
tively little drift. (4) A high degree of sensitivity has been 
obtained. Coblentz20 with a single thermo-couple in a vacuum 
and a 3-foot telescope has recently made quantitative measure- 
ments of the radiations of stars of the fifth magnitude and de- 
tectable responses were obtained from stars of the seventh 
magnitude. The instruments used by us are abundantly sensi- 
tive to measure the visible spectrum. ( 5) It is already attainable 
in forms adapted to special purposes. Coblentz21 for example, 
describes thermopiles for the following purposes: for stellar 
20 Coblentz, W. W. Publications of the Astronomical Society of the Paci- 
fic, 1914, 26, pp. 169-178. 
21 Coblentz, W. W. Bulletin of Bureau of Standards, 1914, II, p. 163. 16 
C. E. FERREE AND GERTRUDE RAND 
measurements, and the measurements of other nocturnal radia- 
tions; for the measurements needed in physical photometry; for 
the determination of whether or not heat is generated in the 
tetanization of a nerve (an ingenious device in which the 
thermo-couple is made into a U-shaped trough for the reception 
of the nerve) ; and for various miscellaneous purposes which 
need not be gone into here. Its feasibility and wide range of 
utility are attested by the fact that it is now being used with suc- 
cess and a fair degree of convenience in radiation work in phys- 
ical, chemical, biological and psychological laboratories. Owing 
to the recent improvements that have been made in its sensitivity, 
its quickness and steadiness of response, and the ease and con- 
venience with which it can be operated, it seems to be the most 
promising of the radiation instruments now available, especially 
for the use of the experimenter who is not a radiometric special- 
ist. These improvements mark, it is to be hoped, an epoch in the 
quantitative study of phenomena in the production of which radia- 
tion plays a part. 
B. The Nichol's Radiometer 
The radiometer was first described by Crookes22 in 1874 
as an interesting scientific toy. Some years later it was used 
in a somewhat modified form to investigate the infra-red spec- 
trum to about 1.5/U3 The first really useful radiometer was 
developed by Nichols in 1896.24 It was further developed and 
improved by Coblentz in 1905.25 In its modern form the 
radiometer consists of two similar thin vanes of mica or 
platinum blackened on one side which are held together by glass 
fibres and are suspended in a vacuum by means of a fine quartz 
fiber. The vanes are about 3 mm. from an opening or window in 
the housing of the apparatus. The radiations to be measured fall 
upon one of the vanes which becomes slightly warmed. This 
22 Crookes, W. Philos Trans. 1874, 164, p. 501; 1875, 165, p. 519; 1876, 
166, p. 325. 
23 Pringsheim, E. Ann. der Phys. 1883, 18, p. 32. 
24 Nichols, E. F. Berichte der Berliner Akad., 1896, p. 1186; Phys. Rev., 
1897, 4, P- 297. 
25 Coblentz, W. W. Investigations of Infra-red Spectra. Carnegie Pub- 
lication, No. 35, Washington, 1905, p. 21. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
17 
causes the residual gas molecules to rebound with increased ve- 
locity from the blackened surface and the reaction pushes this 
vane from the window causing a rotation about the axis of sus- 
pension. A small mirror is attached to the glass fiber which forms 
the axis of rotation, and the deflection is observed by means of a 
telescope and scale. 
1. Significant points with regard to the radiometer. The 
behavior of the radiometer has been worked out theoretically by 
Maxwell26 in his paper on "Stresses in Rarefied Gases Arising 
from Inequalities of Temperature." Crookes, Nichols, and others 
have shown that the sensitiveness of the radiometer is a function 
of the pressure of the residual gas surrounding the vanes, of the 
kind of gas, and of the distance of the exposed vanes from the 
window. Investigation by Coblentz27 has also brought out the 
following points. (1) For vanes of small dimensions such as 
must be used in practical work, the deflections are found to be 
proportional to the area of the exposed surface of the vane. 
(2) The sensitiveness varies with the diameter of the suspension 
fiber. (3) The instrument is not selective in its response. 
2. Comparative advantages and disadvantages of the radio- 
meter. As a working instrument the radiometer is said to have 
the following advantages, (a) Its sensitiveness is comparatively 
easy to control since it can be made to depend almost entirely 
upon the pressure of the residual gas. (b) It is not 
influenced by magnetic and thermo-electric disturbances which 
render work with a very sensitive galvanometer tedious and un- 
satisfactory. (c) It is not so sensitive to temperature changes 
as is, for example, a bolometer, and it can be more easily shielded 
from changes in temperature than can a bolometer with its galva- 
nometer, battery, etc. It has the following disadvantages, (a) 
It is not portable, which may cause inconvenience in certain types 
of work, (b) For maximum sensitiveness its period is very long 
as compared with that of a bolometer or thermopile and galva- 
nometer combination. (Nichols used a period of 8-12 sec., single 
26 Maxwell, J. C. The Scientific Papers of, 2, p. 681; Philos. Trans., 1879, 
170, p. 231. 
27 Coblentz, W. W. Bull. Bur. Standards, 1907, 4, p. 405. 18 
C. E. FERREE AND GERTRUDE RAND 
swing; Coblentz 30-45 sec.) This makes the instrument slow 
to operate. As a compensating feature, however, as Coblentz 
points out, the readings are always trustworthy so that there is no 
need to repeat them, (c) Its window or preferably double win- 
dow is selective in its transmission in the invisible parts of the 
spectrum. A correction has, therefore, to be applied to the re- 
sults for this inequality when working in this region. 
C. The Radio-micrometer. 
This instrument was invented independently by d'Arsonval28 
and by Boys.29 It combines in one instrument the thermo- 
couple which in response to the radiant energy generates the 
electric current, and the galvanometer which indicates by its 
deflections the comparative amounts of current. That is, the 
radio-micrometer is essentially a moving coil galvanometer, 
the moving coil of which contains one or more thermo- 
junctions. In the instrument devised by d'Arsonval a single loop 
of wire was used, one part of which was silver and the other 
palladium. In the instrument devised by Boys the moving coil 
consisted of a loop of copper wire to which was soldered a 
thermo-junction of bismuth and antimony. These instruments 
not having been found to possess the sensitivity attributed to them 
by their inventors, various attempts have been made to improve 
them but with little success. Paschen,30 for example, tried to in- 
crease the sensitivity by increasing the number of thermo- 
junctions.31 Different thermo-couples have been employed. 
28 d'Arsonval. Soc. Franc, de Phys., 1886, pp. 30, 77. 
29 Boys, C. V. Proc. Roy. Soc. 1887, 42, p. 189; 1888, 44, p. 96; 1890, 47, 
p. 480; Philos. Trans., 1889, 180A, p. 159. 
30 Paschen, F. Ann. der Phys., 1893, (3) 48, p. 272. 
31 An advantage is gained in the thermopile by increasing the number of 
thermo-couples, but not in the radio-micrometer. In a thermopile the 
highest efficiency is attained when the resistance of the thermo-couples is 
equal to the combined resistance of the connecting wires and the auxiliary 
galvanometer. Since the resistance of a single thermo-couple is much less 
than this combined resistance, it is of advantage to use several pairs of 
junctions. In the case of the radio-micrometer, however, the connecting loop 
of wire has a negligible resistance, hence there is nothing to gain by using 
more than a single pair of junctions; for as the electromotive force is in- 
creased by the addition of junctions, there is a proportionate increase of 
resistance and the throw of current remains constant. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
19 
Fery,32 for example, used silver and constantan; Schmidt33 bis- 
muth and antimony; and Coblentz,34 also bismuth and antimony, 
and later bismuth and silver.35 Hollnagel36 added greatly to the 
sensitivity and constancy of the instrument described by Schmidt 
by operating it in a vacuum; Coblentz37 increased the sensitivity of 
his bismuth-silver radio-micrometer by enclosing it in a vacuum; 
and Rubens and Hollnagel,38 and Rubens and Wood39 succeeded 
in obtaining an increase of sensitivity for the instrument de- 
scribed by Schmidt by using a concentrating or conical receiver. 
Coblentz also found that the sensitivity of his instrument was 
lowered by para- and dia-magnetic effects produced by the field 
magnets. He was able to lessen these effects and add thereby to 
the delicacy of response by using weak field magnets; or if strong, 
by placing them as far above the thermo-junctions as was possible. 
The elimination of these effects he considers one of the chief 
obstacles to be overcome in the future construction of the 
instrument.40 
As a working instrument the radio-micrometer may be said to 
have the following advantages: (a) It is self-contained; (b) it is 
non-selective in its response to wave-length; (c) it is little subject 
to magnetic perturbations; and (d) it has a high constancy of 
the zero-reading. 
32 Fery, C. Comptes Rendus, 1909, 148, p. 915. 
33 Schmidt, H. Inaug. Diss., Berlin, 1909; Ann. der Phys., 1909, 29 (5), 
p. 1004. See also U. Meyer. Ann. der Phys., 1909, 30 (5), p. 612. 
34 Coblentz, W. W. Bulletin Bureau of Standards, 1906, 2, p. 479. 
35 On p. 10 (Bulletin Bureau of Standards, 1913, 9), Coblentz says: "From 
later experience it seems desirable to try constantan instead of bismuth." 
36 Hollnagel, H. Inaug. Diss., Berlin, 1910. 
37 Coblentz, W. W. Bull. Bureau of Standards, 1906, 2, p. 479; 1908, 4, 
p. 396. 
38 Rubens, H. and Hollnagel, H. Sitz Ber. d. konig. Preuss. Akad. Wiss., 
Berlin, 1910, No. 2, p. 26. 
39 Rubens, H. and Wood, R. W. Ibid., 1910, No. 52, p. 1122. 
40 For further reports of work with the radio-micrometer see Lewis, 
E. P. Astrophysical Journal, 1895, 2, p. 1; Wilson, W. E. Proc. Rov. Soc., 
1894, 55, P- 246; 1895, 5$, P- 174; 1896, 60, p. 377; and Julius, W. T. Hand- 
lungen, 5, de Nederlandisch Natuur en Geneeskundig Congress, 1895. C. E. FERREE AND GERTRUDE RAND 
20 
D. The Bolometer. 
As has already been stated the bolometer is an instrument 
depending for its response to radiant energy on the change 
in resistance with change in temperature offered by a metal to 
the flow of an electric current. The instrument is 
essentially a Wheatstone bridge, two arms of which are 
made of very thin blackened metal strips (of high electrical re- 
sistance and high temperature coefficient), one or both of which 
are exposed to radiation. When thus exposed there is a change 
of temperature which unbalances the bridge, and the resulting 
deflection of the needle of the galvanometer in circuit with the 
bridge gives a measure of the energy absorbed. In order that the 
instrument shall be sensitive to small radiation quantities, it is 
obvious that the metal used should have a high temperature 
coefficient of resistance, a small specific heat, and a low heat con- 
ductivity. Such metals are nickel, platinum, tin, and iron. For 
various reasons relating to mechanical construction, however, 
platinum is much more frequently used than the others. 
The earliest account of an instrument depending on change of 
electrical resistance for measuring or detecting radiant energy 
appears to be that of Svanberg, 1851,41 who for this purpose in- 
troduced a flat spiral of blackened copper wire into one of the 
arms of a Wheatstone bridge. Langley, 1881,42 was the first 
however, to invent a practical instrument and demonstrate its 
superiority to all radiation meters existing at that time for ac- 
curacy, quickness of action, and adaptability. As is shown in the 
reference appended below, his improvements of the instrument 
extended over a long period of time. The value of these improve- 
ments may be shown by comparing the sensitivity of his earlier 
and later instruments. The first had a sensitivity of 0.000020 per 
mm. deflection of the galvanometer, and the latest recorded a 
temperature change of 0.0000010 per mm. deflection when used 
with a galvanometer having a figure of merit of 2-5 x io-10 
ampere. 
41 Svanberg, A. F. Pogg. Ann. der Phys., 1851, 84, p. 416. 
42 Langley, S. P. Proc. Amer. Acad., 1881, 16, p. 342; Chemical News, 
1881, 43, p. 6; Brit. Assoc. Report, 1894, p. 465; Annals Astrophys., Obs. I. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
21 
1. Important points in the construction of sensitive bolom- 
eters. Obviously effective sensitivity can be added in two 
ways in the use of the bolometer: (1) by improving the bolom- 
eter itself; and (2) by making more delicate the auxiliary 
galvanometer. In the attempt to construct sensitive bolometers 
with as great constancy of the zero as is possible, the following 
are some of the points that have received attention. 
a. The kind of material to be ttsed for the receiving surface. 
As has been stated the problem is to get a metal having a high 
temperature coefficient of resistance, low specific heat; and low 
conductivity of heat.43 The following metals have at various 
times been used: platinum,44 tin,45 nickel,48 and iron.47 
b. The area of the receiving surface. The sensitiveness has 
been found to be closely proportional to the square root of this 
surface. In spectral energy work, therefore, where the bolometer 
strip is narrow the sensitiveness attainable for the bolometer is 
limited. 
c. The thickness of the strip used as receiver. There is a 
mechanical limit to the thinness of the strip that can be used when 
exposed to the air. Langley48 found that platinum strips, for 
43 According to Lummer and Kurlbaum (Wied. Ann. der Phys., 1892, 46, 
p. 208) the following equation expresses the relation between the sensitive- 
ness of the bolometer, S; the bolometer current, I; the temperature coefficient 
of the area exposed to radiation, e; the area of the strip exposed to radia- 
tion, a; the resistance of the bolometer strips, r; the absorption coefficient 
of the surface exposed to radiation, A; the emissivity of the whole surface, 
E; the area of the whole surface, F; the heat capacity, W; and the galva- 
nometer constant, k. 
s_ I eak Ir ~ 
From this equation it will be seen that the sensitiveness is increased by 
decreasing the heat capacity and the emissivity; and by increasing the 
bolometer current, the temperature coefficient, the resistance, the absorp- 
tion coefficient, and the surface. 
44 Langley, S. P. Loc. cit. 
43 Angstrom, K. Wied. Ann. der Phys., 1885, 26, p. 253; 1889, 36, p. 715; 
1893, 68, p. 493, and others. 
46 Julius, W. T. Licht und Warmestrahlung, 1890, p. 31. 
47 Rubens, H., (used tin, iron, and platinum). Wied. Ann. der Phys., 
1889, 37, p. 255; and 1892, 43, p. 238. 
48 Langley, S. P. Op. cit. 22 
C. E. FERREE AND GERTRUDE RAND 
example, less than 0.002 mm. thick are inadvisable, thinner ones 
being disturbed mechanically by air currents. In recent work 
Coblentz49 finds it permissible to use platinum strips 0.001 mm. in 
thickness. In vacuum bolometers, however, much thinner strips 
(0.0005 mm.) are used to advantage. 
d. The most favorable resistance of bolometer and balancing 
coils. Lummer and Kurlbaum50 considered the bolometer a 
simple Wheatstone bridge which has its maximum sensitiveness 
when the four arms and the galvanometer are all of equal re- 
sistance. In fact in the construction of their instrument instead 
of using one or two bolometer surfaces, they used four just alike, 
each forming one of the arms of the bridge.51 Child and Stew- 
art,52 however, have shown experimentally that the sensitiveness 
is increased by having the resistance of the balancing coils several 
times that of the bolometer strips. Abbot53 has also shown that 
the maximum sensibility is very nearly attained when the resist- 
ance of the balancing coils is four times or more that of the 
bolometer strips, and the galvanometer resistance is not less than 
0.6 or more than four times the resistance of the bolometer 
strips.54 
e. The slide wire for balancing the resistance. The question of 
a satisfactory slide wire has required considerable attention. Cob- 
lentz has found that slide wires of platinum 0.5 and 1 mm. in 
diameter used in connection with a mercury contact give the best 
satisfaction. 
f. The protection of the bolometer from air currents. This 
can be done adequately only by putting the bolometer in a 
vacuum. Contact with the air renders the bolometer both in- 
sensitive and inconstant in its response. On the former point, 
49 Coblentz, W. W. Bulletin of the Bureau of Standards, 1912, 9, p. 37. 
50 Lummer, O. and Kurlbaum, F. Wied. Ann. der Phys., 1892, 46, p. 204. 
51 The tendency among foreign investigators has been to make the four 
arms of equal resistance. On the other hand just as strong a tendency has 
been shown among American investigators to make the resistance of the 
balancing coils greater than that of the bolometer strips. 
52 Child, C., and Stewart, O. Phys. Rev., 1897, 4, p. 502. 
53 Abbot, C. G. Annals of the Astrophysics Obs., I. 
54 See also Reid, H. F. Amer. Jour. Sci., 1888, 33, (3), p. 160. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
23 
Warburg, Leithauser, and Johansen35 have shown that the heat 
lost by air conduction for a bolometer 1 mm. wide is 4.5, and for 
a bolometer 0.2 mm. wide, 14.8 times as great as it is from 
radiation. And in a vacuum a bolometer 0.2 mm. wide when 
operated with a small current was found to be ten times as sensi- 
tive as it was in air.56 
g. The strength of current. The radiation sensitivity of a 
vacuum bolometer is found to be proportional to the current for 
small values but for a large current the radiation sensitivity of a 
narrow bolometer passes through a maximum. This maximum is 
obtained for a current density at which the radiation sensitivity of 
the air bolometer does not depart appreciably from proportionality 
with the current. The manner in which the radiation sensitivity 
varies with the gas pressure and with the bolometer current is 
shown, for example, by Buchwald.57 
2. Points in the construction of the auxiliary galvanometer. 
It is scarcely necessary to mention that the effective sensitivity of 
the bolometer depends in a large measure upon the auxiliary 
galvanometer. The first great step in improving the moving 
magnet galvanometer is due to Kelvin who decreased the weight 
of the moving parts to a few milligrams, and introduced the static 
system of magnets. Snow58 was among the first to give much at- 
tention to the possibility of adding sensitivity to the bolometer by 
improving the galvanometer. Paschen59 continued the work in 
this direction and constructed the most sensitive galvanometer 
used up to that time. DuBois and Rubens,60 Mendenhall and 
55 Warburg, E., Leithauser, G. and Johansen, E. Ann. der Phys., 1907, 
24, (5), P- 25. 
56 A noteworthy vacuum spectro-bolometer is described by A. Trowbridge 
(Phys. Rev., 1908, 27, p. 282; Philos. Mag., 1910, (6), 20, p. 768) in which 
the bolometer and the optical parts of the spectroscope are in a vacuum. 
Coblentz also describes a very sensitive vacuum bolometer and gives results 
with it at different pressures (Bull. Bureau of Standards, 1913, 9, pp. 39-43). 
Other less adequate methods of shielding have been to surround the bolom- 
eter by a double wall with an air space between; to enclose it in a water 
jacket (Langley, op. cit. and Abbot, op. cit.) ; etc. 
57 Buchwald, E. Ann. der Phys., 1910, (4), 35, p. 928. 
58 Snow, B. W. Phys. Rev., 1895, I, p. 31. 
59 Paschen, F. Wied. Ann. der Phys., 1893, 48, p. 272. 
60 DuBois and Rubens. Ann. d. Phys., 1900, (4), 2, p. 84. 24 
C. E. FERREE AND GERTRUDE RAND 
Waidner,61 Abbot,62 Ingersoll,63 and Coblentz64 have all described 
sensitive galvanometers. Some of the points to be considered in 
the construction of a sensitive galvanometer are form and size of 
coil, size of wire, the kind of magnet and the dimensions and con- 
struction of the needle system, the astaticizing of the magnet sys- 
tem, the shielding of the system from influences due to the earth's 
field and neighboring objects, etc. The proper form and method 
of winding galvanometer coils to secure a maximum effect from a 
given weight or resistance of copper has been thoroughly dis- 
cussed by Maxwell.65 He shows that the greatest effect is ob- 
tained by winding the coils with different sizes of wire, beginning 
with the smallest size and winding each layer so that it lies within 
the surface the polar equation of which is r2=d2 sin 6, where r is 
the length of the radius making an angle 0 with the axis of the 
coil, and d the value of r when 6=go°. Abbot66 has computed the 
most efficient coils for meeting these conditions, and gives results 
for coils wound with a single wire and for coils wound with 
three sections of wire of different diameters. He found that the 
total force exerted at the center is closely proportional to the 0.45 
power of the total resistance and that coils composed of three 
sections of the best sizes of wire give 1.4 times the force of a coil 
of the best single size wire of the same total resistance. In his 
best 25 ohm coil, wound in three sections the diameters of the 
wires were 0.08, 0.16, and 0.32 mm.; the lengths were 256, 1031, 
and 4144 cm.; and the external diameter of the completed coil 
was 3.3 cm. 
In the construction of the needle system the greatest sensitivity 
is attained when the ratio of the magnetic moment to the moment 
of inertia of the system is a maximum. The best dimensions and 
construction of needle systems have been extensively investigated 
61 Mendenhall, C. E. and Waidner, C. W. Amer. Journ. Sci., 1901, 12, (4), 
P- 249. 
62 Abbot, C. G. Astrophys. Journ., 1903, 18, p. 1. 
63 Ingersoll, L. R. Philos. Mag., 1906, (6), 11, p. 41. 
64 Coblentz, W. W. Bulletin Bureau of Standards, 1908, (3), 4, pp. 424- 
435 J 1916, 13, P- 423. 
65 Maxwell, J. C. Electricity and Magnetism, 2, p. 322. 
GG Abbot, C. G. Loc. cit. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
25 
by Paschen, Mendenhall and Waidner, and by Abbot. The 
shielding of the galvanometer from magnetic perturbations, etc., 
is done by means of housings of soft iron. For an inexpensive 
and convenient method of shielding, for a simplification of the 
moving coil galvanometer for convenience of shielding, and for 
the astaticization of the needle system, see Coblentz (loc. cit.). 
3. Possible sources of difficulty in the use of the bolometer. 
The preceding discussion though brief may be enough to indicate 
that the bolometer is a difficult instrument to operate. The fol- 
lowing are the possible sources of trouble. (1) The auxiliary 
galvanometer is subject to magnetic perturbations and if exposed 
to great temperature changes its sentiveness is changed, due to a 
variation in the resistance of the coils. The sensitiveness and zero 
reading are also subject to frequent changes due to variations in 
the magnetic field. (2) The bolometer strip is affected by air 
drafts, and, if very thin, by mechanical vibration. (3) The elec- 
tric circuits are subject to temperature (resistance) changes which 
cause variations in the bolometer current. (4) The storage bat- 
tery current is irregular due to changes in temperature and to 
polarization. (5) The gases surrounding the bolometer may 
affect the readings. Lummer and Pringsheim67 found, for ex- 
ample, that variations in the amount of moisture in the air 
change the sensitiveness of the bolometer. A part or all of these 
causes tend to make the readings in work with the bolometer ex- 
tremely variable. These variations are of two kinds, (a) A 
slow drift of the zero scale reading due to changes in the resis- 
tance of the bridge; and (b) fluctuations of the reading due to 
air currents and magnetic perturbations. 
4. The comparative advantages and disadvantages of the 
bolometer. The bolometer has, however, the following advan- 
tages. It has a high degree of sensitivity. It is portable. It is 
non-selective in its response to wave-length. It can be calibrated 
directly against a black body. It is quick in its action and is, 
therefore, well adapted to work in which a quick registration 
of the galvanometer deflections is desired. Its chief disadvan- 
fi" Lummer, O. and Pringsheim, E. Ann. der. Phys., 1897, (3), 63, p. 398. 26 
C. E. FERREE AND GERTRUDE RAND 
tage, as has been stated, is its unsteadiness and the difficulty of 
operating. It is not nearly so easy to operate as the thermopile, 
for example; and the unsteadiness of its zero point renders it 
untrustworthy for small readings in spite of its high intrinsic 
sensitivity. 
E. The Selenium Cell 
To Willoughby Smith belongs the discovery that has led to 
the use of selenium as a light-measuring instrument. In 1873 
while using a resistance made of selenium in connection with ex- 
periments in telegraphy, he discovered that its electrical conduc- 
tivity is raised by exposure to light. The immediate result of 
this discovery was twofold: light-measuring instruments were 
constructed of selenium, and a long series of investigations was 
begun to determine (a) the factors extraneous to light that in- 
fluence the resistance of selenium and thus affect its applicability 
to the measurement of light; (b) the factors that influence the 
action of light on the conductivity of a selenium cell and the pos- 
sibility of the use of the cell either as a photometer or a radio- 
meter; and (c) the nature of the action of light on the specific 
resistance of selenium. 
A selenium cell is a device consisting essentially of a mass of 
crystalline selenium furnished with two metallic electrodes. Crys- 
talline selenium is obtained by keeping molten vitreous selenium 
at a temperature of from 150 to 2100 C. for several hours. It 
then takes on a metallic appearance and becomes opaque even in 
case of very thin films. Selenium has such a very small conduc- 
tivity that in making resistances of it, one feature of the construc- 
tion is to offer several paths or rather one continuous broad path 
for the flow of current. One way in which this has been accom- 
plished in the construction of light-measuring cells has been to 
wind a strip of mica or slate with two parallel wires less than 1 
mm. apart. This is covered with powdered selenium. The selen- 
ium is melted, worked into a smooth surface and cooled quickly. 
It is then heated again and cooled very slowly. One end of each 
wire is connected to the battery terminals, the others end in the 
selenium. The wires thus in reality form the electrodes, and the 
circuit is completed through the intervening selenium. This type RADIOMETRY IN PSYCHOLOGICAL OPTICS 
27 
of cell has been used among others by Bidwell, Siemans, Sabine, 
and Pfund. (For construction of selenium cells, see Bidwell,68 
Fritts,69 Berndt,70 Townsend71 and Dieterich.72) 
1. Points to be considered in the construction and use of 
selenium cells. The following are some of the points to be consid- 
ered in the construction and use of sensitive selenium cells. 
a. The method of preparation. The sensitiveness of the cell 
to light depends largely upon its initial specific resistance. This 
has been pointed out by Pochettino73, Giltay,74 and especially by 
Brown.75 According to Brown, the higher the resistance of 
crystalline selenium the greater is its sensitivity to light; and con- 
versely the lower its resistance, the less is the sensitivity. Brown 
gives results showing the resistance of the cell and its sensitivity 
in terms of the ratio of conductivity in light to conductivity in the 
dark. For example, a cell with a resistance of io9 ohms had a 
sensitivity of 200:1 in an arbitrary scale; a cell with a resistance 
of 400,000 ohms, a sensitivity of 30:1; 3,000 ohms, 2:1; 1,700 
ohms, 1:1. In its vitreous state selenium is practically a non-con- 
ductor. To become a conductor it must be brought to the crys- 
talline form. For example, when it is heated to a temperature of 
100 to 1500 C. its conductivity is slight and variable; but when 
heated repeatedly to temperatures of 190 to 21 o° C. and cooled, it 
passes into a coarsely granular crystalline state and acquires and 
retains a greater conductivity. That the temperature to which 
selenium has been heated is the chief factor in determining its 
conductivity and sensitivity to light was pointed out by Siemans76 
as early as 1875, w^° stated that when heated to 2100 C. cells of 
greater conductivity, constancy, and light sensitivity were pro- 
duced than when heated to 1500 C. A systematic study of the 
68 Bidwell, S. Phil. Mag., 1891, Ser. 5, 31, pp. 250-256; 1895, 40, pp. 233-256. 
69 Fritts. Electrical Review, 1885, p. 208. 
70 Berndt, G. Phys. Zeit., 1904, 5, pp. 121-124. 
71 Townsend, F. Electrician, Oct. 7, 1904, 53, pp. 987-990. 
72Dieterich, E. O. Phys. Rev., 1914, 4, Ser. 2, pp. 467-476. 
73 Pochettino, A. N. Cimento, 1911, 1, Ser. 6, pp. 147-210. 
74 Giltay, J. W. Phys. Zeit., 1910, II, p. 419. 
75 Brown, F. C. Phys. Rev., 1911, 33, pp. 1-26. 
76 Siemans, W. Phil. Mag., 1875, 50, p. 416. 28 
C. E. FERREE AND GERTRUDE RAND 
effect of temperature and the duration of annealing on the resist- 
ance of selenium has been made by Dieterich.77 His results show 
that when maintained at a temperature of 200 to 2100 C. for six 
hours, a cell was produced with a resistance of 233,000 ohms; 
at 2100 for four hours, a resistance of 358,000 ohms; at 2100 for 
five hours, 490,000 ohms; at 1800 for three and one-half hours, 
1,400,000 ohms; and at 1900 for two hours, 3,690,000 ohms. In- 
asmuch as his cells of highest resistance were not permanent, he 
was not able unfortunately to work out the correlation between 
resistance and sensitivity. Pochettino,78 Aichi and Tanakadate,79 
Brown,80 and Dieterich81 all think that a change of structure takes 
place when selenium is annealed at a temperature of 210 to 2200 
which causes the increase of conductivity. 
b. Purity of the selenium. Bidwell82 found that insensitive 
selenium cells were increased in sensitivity by the addition of a 
small quantity of cuprous selenide. Marc83 recommends the addi- 
tion of o. 1-0.5% of silver to increase its sensitivity. Townsend8'1 
claims that 1 or 2% of copper or nickel selenide may be present 
without affecting to a marked degree the sensitivity of the cell. 
Pfund85 found that the sensitivity could be increased by the pres- 
ence of 3% of a selenide. He believes it to be of advantage, 
however, to start with a chemically pure selenium and add im- 
purity of a definite kind and amount. 
c. Material and size of electrodes. Sale,86 and Adams and 
Day87 used platinum electrodes; Bell88 used brass; Bidwell,89 
77 Dieterich, E. O. Loc. cit. 
78 Pochettino, A. N. Cimento, 1911, I, Ser. 6, pp. 147-210. 
79 Aichi, K. and Tanakadate, T. Math, and Phys. Soc., Tokyo, 1904, (2), 
16, pp. 217-221. 
80 Brown, F. C. Loc. cit. 
81 Dieterich, E. Loc. cit. 
82 Bidwell, S. Phil. Mag., 1895, 40, Ser. 5, pp. 233-256. 
83 Marc, R. Z. Anorg. Chem., 1906, 48, pp. 393-426. 
84 Townsend, F. Loc. cit. 
85 Pfund, A. H. Phil. Mag., 1904, 7, Ser. 6, pp. 26-39. 
88 Sale. Proc. Roy. Soc. of London, 1873, 2/, PP- 283-285. 
87 Adams and Day. Philos. Trans., 1877, 167, pp. 313-349. 
88 Bell, G. Nature, 1878, 22, p. 500. 
89 Bidwell, S. Loc. cit. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
29 
copper; Pfund00 and Berndt,91 carbon. Dieterich92 tried copper, 
nickel, platinum, German silver and Advance wire. He found 
that copper, German silver and Advance wire have the disad- 
vantage that at the temperature of annealing a film of oxide is 
formed. This so materially increases the resistance of the cell 
as to make it practically useless except with very sensitive auxili- 
ary apparatus. Nickel wire is much less easily oxidized and 
proved as satisfactory as platinum wire besides being less ex- 
pensive. Pfund and Berndt consider carbon electrodes prefer- 
able in that selenium forms no conducting compound with car- 
bon. On the question of size of electrodes, there seems to be 
general agreement that large surface contact between the junc- 
tions and the selenium is necessary to avoid high junction resist- 
ance and consequent diminished sensitivity of the cell. 
d. Strength of the battery current. The sensitivity of the 
selenium cell has been found to vary with the battery current. It 
was first noted by Adams and Day93 that the resistance of selenium 
diminished as the battery current is increased. Sabine94 held this 
to be true only after a certain intensity of current had been 
reached. For lower intensities of current, increase of current 
caused increase of resistance. Minchin95 increased the conduc- 
tivity of his cell fourfold by increasing the voltage from 2 to 12 
volts. Brown96 found that with a Ruhmer cell the conductivity 
varied by an amount almost directly proportional to the voltage; 
with a Giltay cell, however, the variation decreased in amount as 
the voltage was increased. Ries97 claims that conductivity in- 
creases with increase of voltage for a range of from ©.4 to 4 
volts. Luterbacher98 states that this change is greater for direct 
than for alternating current. The necessity for an accurately 
90 Pfund, A. H. Loc. cit. 
91 Berndt, G. Loc. cit. 
92 Dieterich, E. O. Phys. Rev., 1914, 4, Ser. 2, p. 468. 
93 Adams and Day. Op. cit., p. 319. 
94 Sabine, R. Phil. Mag., 1878, 5, ser. 4, pp. 401-416. 
95 Minchin, G. M. Phil. Mag., 1891, 31, ser. 5, pp. 207-238. 
90 Brown, F. C. Loc. cit. 
97 Ries, C. Phys. Zeit., 1911, 12, p. 529. 
98 Luterbacher, J. Ann. der Physik, 1910, 35, p. 1392. 30 
C. E. FERREE AND GERTRUDE RAND 
constant battery current when using a selenium cell as a measur- 
ing instrument is obvious. 
e. The direction of the battery current. Adams and Day" 
found that the passage of a current in any direction at any period 
in a series of observations produces a condition which tends to 
facilitate the subsequent passage of a current in the opposite 
direction but obstructs one passing in the same direction. He in- 
terprets this condition as a slight "set" of the molecules. The 
effect is particularly marked in case of the first current sent 
through the selenium and is more or less permanent. This result 
was confirmed by Sabine100 who thinks the changes are in the 
resistance of both the selenium and the junctions. This fact 
combined with the changes in resistance caused by changes in the 
strength and duration of current led Sabine to state that selenium 
is very unsuitable for the production of a constant resistance for 
measuring purposes. 
f. Duration of battery current. Adams and Day101 found that 
the resistance of selenium increases continuously during the time 
of the passage of the battery current. They point out, for ex- 
ample, that on this account the precaution should always be 
taken to shut off the current between observations. This pre- 
caution, however, does not eliminate; it only lessens the effect 
of the variable factor. 
g. Temperature. Bidwell102 claims that there is an optimum 
temperature for each cell above and below which the resistance 
decreases. For six cells this temperature was 24, 23, 14, 30, 25 
and 220 C. Brown and Stabbins103 tested the effect of tempera- 
tures ranging from 400 to 2000 C. and found that the resistance 
of selenium decreases with increase of temperature. Tempera- 
tures above or below these were not used, so their results contain 
nothing that bears directly on the claim made by Bidwell. For a 
change of temperature ranging from 13.20 to 73.4° C. they found 
that a given amount of light incident on the cell caused changes of 
99 Adams and Day. Op. cit., p. 323. 
100 Sabine, R. Loc. cit. 
101 Adams and Day. Op. cit., p. 314. 
102 Bidwell, S. Phil. Mag., 1881, 11, Ser. 5, p. 302; 1895, 40, ser. 5, p. 242. 
103 Brown and Stebbins. Phys. Rev., 1908, 26, pp. 273-298. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
31 
resistance varying in percentage from 24.9 to 3.7. The effect of 
temperature on the sensitivity of the cell is so marked that Pfund, 
for example, worked in a room in which the temperature was kept 
constant to 1/100. 
h. Pressure. According to Brown,104 Brown and Stebbins,105 
and Monten,106 increase of pressure decreases the resistance of 
selenium and lowers its sensitivity to light. Brown found that 
these effects were present up to a pressure of 1,000 atmospheres. 
In case of a single crystal of selenium, he increased the conduc- 
tivity about 120 times by an increase of pressure of 180 atmo- 
spheres. Brown and Stebbins found the percentage change of 
resistance for one atmosphere to vary between 0.05 and 0.30 for 
different cells. 
i. Moisture. Ries,107 Bidwell108 and others have shown that 
humidity affects the electrical properties of selenium. Ries thinks 
this effect is sufficient to explain the discrepancies existing in the 
results of different observers. On this account cells of the Giltay 
type, which are constructed so that there is free communication 
between the outer air and the selenium surface, show wide varia- 
tions in conductivity. Nicholson109 improved the constancy of a 
cell of this type by enclosing it in an air-tight box with a glass 
window. 
j. Age of cell. Adams and Day110 found the sensitivity of the 
selenium cell to be greatly reduced after one year. Bidwell111 
found no material loss at the end of one year, but the cells were 
practically useless after four years. Dieterich,112 however, con- 
structed two cells of remarkably high sensitivity which lost of 
their sensitivity within a month. 
104 Brown, F. C. Phys. Rev., 1905, 20, pp. 185-186; Phys. Rev., 1914, 4, 
ser. 2, pp. 85-98. 
105 Brown and Stebbins. Loc. cit. 
106 Monten, F. Ark. for. Mat. Astron, och. Fysik, Stockholm, 1908, 4, 
pp. 1-6. 
107 Ries, C. Phys. Z., 1908, p, pp. 569-582. 
108 Bidwell, S. Phil. Mag., 1895, 40, ser. 5, p. 245. 
109 Nicholson, P. J. Phys. Rev., 1914, 3, Ser. 2, p. 8. 
110 Adams and Day. Op. cit., p. 348. 
111 Bidwell, S. Phil. Mag., 1891, 31, Ser. 5, pp. 250-256. 
112 Dieterich, E. O. Phys. Rev., 1914, 4, Ser. 2, p. 471. 32 
C. E. FERREE AND GERTRUDE RAND 
k. The amount of polarization gradually set up in the cell. 
The presence of polarization currents produced by the passage of 
a battery current though selenium was found by Adams and 
Day.113 This effect was increased by the exposure of the selenium 
to light. Bidwell114 says the polarization current is very trouble- 
some in making accurate resistance tests by the bridge method. 
The intensity of this current is increased by humidity. While 
this factor and the next to be considered, the presence of photo- 
electric currents, can hardly be said to influence the sensitivity of 
the cell in a way similar to the preceding factors, they undoubtedly 
affect its use as a light-measuring instrument; for with the 
presence of polarization and photo-electric currents of unknown 
intensity, an exact determination of the conductivity of the cell 
under a given set of conditions can not be made. 
1. ' Photo-electric .currents. The presence of photo-electric cur- 
rents in selenium due to an exposure to light was noted first by 
Adams and Day,115 later by Bidwell116 and by Minchin.117 Adams 
and Day found that this current was often more intense than the 
polarization current and was sufficient to overbalance a weak 
battery current. 
2. Factors which render it difficult to use the selenium cell 
for quantitative work either as an ohmic resistance or as a light- 
measuring instrument. 
A part of the foregoing factors are of importance chiefly in 
making it almost impossible to construct two selenium cells of 
similar properties. They do not affect the use of a given cell once 
constructed. The remainder, however, apply to the responses of 
a single cell and are so difficult if not impossible of control as to 
make it exceedingly doubtful whether the selenium cell can 
be used as an instrument of precision. In fact the consensus 
of opinion among the investigators has been that it can not be 
used with a degree of precision which is acceptable in quantitative 
113 Adams and Day. Op. cit., p. 328. 
114 Bidwell, S. Phil. Mag., 1895, 40, Ser. 5, p. 244. 
115 Adams and Day. Op. cit., p. 333. 
116 Bidwell, S. Op. cit., p. 251. 
117 Minchin, G. M. Phil. Mag., 1891, 31, Ser. 5, pp. 207-238. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
33 
work. These factors are : (1) The passage of the battery cur- 
rent through the cell in a given direction produces a condition 
which tends to facilitate the subsequent passage in the opposite 
direction, but obstructs one in the same direction. Since this 
effect can not be completely eliminated and the cell restored to its 
original condition by reversing the current, the cell continually 
changes its state of conductivity with use; hence two measure- 
ments can never be made with it in the same condition. This 
difficulty is further increased by the fact that the longer the cur- 
rent is allowed to flow, the greater is the change of conductivity. 
The greater number of times the cell is used, therefore, and the 
longer the current is allowed to flow, the greater will be the 
progressive change in the properties of the cell. (2) Over and 
above the effect of current is a loss of sensitivity with age. 
Measurements made by the cell at intervals at all .widely separated 
are, therefore, not comparable. (3) The polarization currents 
due to the passage of the battery current and increased by the 
exposure to light, and the photo-electric currents which are even 
stronger than the polarization currents and strong enough accord- 
ing to Adams and Day to overcome a weak battery current, pro- 
duce a variability in the action of the cell for which there seems 
to be no remedy. When to these apparently insuperable obstacles 
is added the fact that the strength of the battery current, the 
temperature of the cell and the humidity of the atmosphere must 
be kept constant within small limits, one gets some idea of the 
difficulties attendant on the use of the selenium cell as an instru- 
ment of precision. 
3. Factors which apply especially to its use as a light measur- 
ing instrument. The foregoing properties of selenium, it may be 
noted, apply to its use as material for the construction both of 
ohmic resistances and of instruments for the measurement of 
light. In addition to these the following points which apply 
specifically to its use in the measurement of light are to be 
considered. 
a. The preexposure of the cell. Adams and Day118 claim that 
118 Adams and Day. Op. cit., p. 315. 34 
C. E. FERREE AND GERTRUDE RAND 
selenium is more sensitive in its response to light after it has 
been kept in the dark for several hours than after it has been 
exposed to light several times; hence the result obtained from the 
first of a series of measurements is generally not comparable with 
those gotten later. Townsend119 says that after prolonged ex- 
posure to light there is a fatigue effect which takes place im- 
mediately and lasts at least four hours. Nicholson120 says that 
fatigue effects are present when the cell has not been allowed 
sufficient rest between readings. Marc121 finds that the sensitivity 
to red light is greatly modified by a previous strong illumination 
with white light or by a long continued blue illumination. Grant- 
ham122 investigating the recovery period of the cell, found that 
for a short time after the exposure to light was cut off, the re- 
sistance decreased still further; it then increased rapidly at first, 
then more slowly. After constant use the cells became at times 
temporarily almost insensitive. 
b. The time of exposure to light. For Pfund123 a maximum 
response was reached in 2 to 3 sec.; then a slight "creeping effect" 
took place. In later work he124 used an exposure time of 12.5 
seconds. Brown125 claims that the change of conductivity is a 
function of the time of illumination. For effect of exposure 
time on response to monochromatic light, see pp. 35-36. 
c. The wave-length of the spectrum light and the factors 
which influence the selectiveness of response to wave-length. 
Sale128 was the first to report that selenium is selective in its re- 
sponse to wave-length. He found the greatest change in re- 
sistance was caused by red light near the end of the solar spec- 
trum; next by red of shorter wave-length; then in order by 
orange, green, blue and violet. Adams and Day127 found the 
119 Townsend, F. Loc. cit. 
120 Nicholson, P. J. Op. cit., p. 9. 
121 Marc, R. Z. Anorg. Chern., 1903, 37, pp. 459-475. 
122 Grantham, G. E. Phys. Rev., 1914, 4, Ser. 2, pp. 259-266. 
123 Pfund, A. H. Phil. Mag., 1904, 7, Ser. 6, pp. 26-39. 
124 Pfund, A. H. Phys. Rev., 1912, 34, p. 370. 
125 Brown, F. C. Phys. Rev., 1911, 33, pp. 14-15. 
120 Sale. Proc. Roy. Soc. of London, 1873, 21, pp. 283-285. 
127 Adams and Day. Op. cit., p. 317. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
35 
greatest light effect in the greenish yellow, next in the red and 
least in the violet. Pfund123 used lights equalized in energy by a 
thermopile and later by a radiomicrometer. He got the maximum 
response near .7/*. This maximum was not changed when selen- 
ides of lead, mercury, copper, and silver were introduced. Brown 
and Sieg129 found the curve of response to wave-length to vary 
for different types of cell. With reference to selectiveness of re- 
sponse to wave-length two sorts of investigation have been made, 
-one to determine the factors that influence this selectiveness in 
a given cell; the other to determine the factors which influence 
selectiveness in different cells. 
(1) Factors which have been found to influence the selective- 
ness of response for a given cell. The following factors have 
been found to influence the selectiveness of response in a given 
cell. 
(a) Intensity. Pfund130 found the sensitivity curve for wave- 
length to vary with the intensity of the incident light. This was 
confirmed by Brown and Sieg131 and by Nicholson.132 With ref- 
erence to the changes in the sensitivity curve Pfund contributes 
the following formula: d=DP3 where d= change of conductiv- 
ity; I- energy of illumination; and D and are constants depen- 
dent on the wave-length of the incident light. For exposures of 
12.5 sec., he found D to be constant for any particular wave- 
length; /? was very nearly % for regions of the spectrum from 
the violet to the yellow; but its value increased as red was ap- 
proached, equalling 1 for deep red and infra-red. Nicholson 
verified both the formula and the constants for an exposure time 
of 12.5 sec. 
(b) Length of exposure time. Nicholson,133 however, found 
the formula contributed by Pfund to hold only for an exposure 
time of 12.5 sec. For longer and shorter exposures, the value of 
/3 changed. For 10 sec. exposure, it increased and the region in 
128 Pfund, A. H. Phil. Mag., 1904, 7, Ser. 6, pp. 26-39; Phys. Rev., 1909, 
28, pp. 324-336. 
129 Brown, F. C. and Sieg, L. P. Phys. Rev., 1914, 4, Ser. 2, pp. 48-61. 
130 Pfund, A. H. Phys. Rev., 1909, 28, pp. 324-336. 
131 Brown, F. C. and Sieg, L. P. Phys. Rev., 1913, 2, Ser. 2, pp. 487-494. 
132 Nicholson, P. J. Phys. Rev., 1914, 3, Ser. 2, pp. 1-24. 
133 Nicholson, P. J. Loc. cit. 36 
C. E. FERREE AND GERTRUDE RAND 
which fl becomes 1 shifted towards the shorter wave-lengths. 
With longer exposures (15 to 20 sec.), the value of fl decreased. 
With unlimited exposures or until a steady state of resistance of 
the selenium was attained, fl equalled 0.5 throughout the spec- 
trum, except at about 600^ where it equalled only 0.4. This 
change in the value of fl for different wave-lengths with exposure 
time is probably in accord with Nicholson's further demonstra- 
tion that selenium has a different inertia of response for different 
wave-lengths. This is particularly marked for the red and infra- 
red of the spectrum. Brown and Sieg134 also note a change in 
the shape of the sensitivity curve for exposures of 30 and 0.4 sec. 
(c) Temperature, humidity and voltage. That the selective- 
ness of response of selenium to wave-length varies with the tem- 
perature of the cell is mentioned by Marc135 and Nicholson;136 
Marc finds it to vary also with the intensity of the current used; 
and Nicholson with the humidity. 
(d) Photo-electric currents. Minchin137 using seleno-alumi- 
nium cells, found that the intensity of electromotive force pro- 
duced by the action of light on the cell varies with the wave-length 
of the incident light. It is greatest in order for yellow, orange, 
green, red and blue. 
(2) Factors which have been foitnd to influence the selective- 
ness of response to wave-length in different cells. The main 
cause of difference in the selectiveness of response to wave-length 
from cell to cell is according to Brown and Sieg138 and to Dieter- 
ich139 the temperature at which the cell was made and annealed. 
In general there are two groups of cells,-those with the maxi- 
mum response at wave-lengths greater than 640^; and those 
with the maximum at a wave-length less than this. Cells of the 
former group are produced by annealing at lower temperatures, 
e.g., annealing at 1700 C. gives a pronounced red maximum; 
134 Brown and Sieg. Phys. Rev., 1913, 2, Ser. 2, pp. 487-494. 
135 Marc, R. Z. Anorg. Chern., 1903, 37, pp. 459-475. 
136 Nicholson, P. J. Loc. cit. 
137 Minchin, G. M. Phil. Mag., 1891, 31, Ser. 5, pp. 207-238. 
138 Brown, F. C. and Sieg, L. P. Phys. Rev., 1914, 24, Ser. 2, pp. 48-61. 
139 Dieterich, E. O. Phys. Rev., 1914, 4, Ser. 2, pp. 467-476. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
37 
those of the latter group by annealing at high temperature, e.g., at 
2100 C. By partial annealing at 2100 and completing the process 
at a lower temperature, the maximum response is given in the 
blue and a secondary maximum in the red. Brown140 confirms 
this result with his selenium "crystal forms" produced by the 
sublimation of the vapor either in a high vacuum or at atmos- 
pheric pressure. Among these forms he finds types which 
give the different wave-length sensitivity curves found by 
Dieterich in the different cells annealed at the various tempera- 
tures. Brown believes that there are at least three forms of 
metallic selenium of widely different electrical resistivity. These 
forms are produced at different temperatures. That is, at high 
temperatures, for example, crystals of maximum sensitivity to 
red light are not allowed to form. 
d. The intensity of white light. Attempts have been made to 
use the selenium cell both as a radiometer and a photometer. In 
the latter case, the following laws of change of resistance with 
change of intensity have at different times been formulated. 
When m= conductivity, i= light intensity, R= resistance, and 
the other quantities are constants, Rosse,141 Adams and Day,142 
and Berndt143 give the formula i- cm2; Hopius,144 i= cm3; 
Athanasiadis,145 i= m(m-a)b; Hesehus,146 i=bm-1; Ruhmer,147 
R0/Rb=: (b/a)a; Stebbins, i=cm. (See Brown, Phys. Rev., 
1911, 33, pp. 1-26.) Brown states that although the illumination 
used, time of exposure, and construction of cell varied in the 
work of the above men, it appears obviously futile from these 
results to look for a universal law of conductivity for the selenium 
cell as a function of intensity of illumination. 
In summarizing the difficulties that apply to the use of the 
140 Brown, F. C. Phys. Rev., 1914, 4, Ser. 2, pp. 85-98; see also Dietrich, 
ibid, p. 474. 
141 Rosse. Phil. Mag., 1874, 47, Ser. 4, pp. 161-164. 
142 Adams and Day. Op. cit., p. 318. 
143 Berndt, G. Phys. Z., 1904, 5, pp. 121-124. 
144 Hopius. Jurn. Russk. Fisik Chimicesk. Obscestva, 1903, 35, pp. 581-585. 
145 Athanasiadis, G. Ann. der Physik, 1908, 25, pp. 92-98. 
146 Hesehus, N. A. Jurn. Russk. Fisik. Chimicesk. Obscestva, 1905, 37, 
pp. 221-231; Phys. Zeit., 1906, 7, pp. 163-168. 
147 Ruhmer, E. Phys. Zeit., 1902, 3, pp. 468-474. 38 
C. E. FERREE AND GERTRUDE RAND 
selenium cell in the measurement of light, the following points 
then may be noted. (i) The fatigue effects and the effects of 
previous exposure to light are so great that it is exceedingly 
difficult to keep the cell in a state of constant sensitivity (2) 
The amount of response is not only a function of the time of 
exposure to the light, but apparently rather complexly so. (3) 
There is not only selectiveness of response to wave-length but 
the amount of this selectiveness varies with the intensity of light, 
with the strength of the battery current, with the temperature of 
the cell and with humidity.148 While there is a possibility of con- 
trolling the last three of this latter group of factors, there seems 
no way to deal satisfactorily for any wide use of the cell with the 
first, or what may be termed roughly a "Purkinje phenomenon." 
Because of this factor a calibration of the cell for wave-length 
for one intensity of light would not hold for all intensities, which 
would limit the use of the cell to the intensity of light for which 
it was calibrated or for ranges for which there is no change in 
relative sensitivity to wave-length. That is, any wide use of the 
cell would require both a wave-length and an intensity calibra- 
tion in terms, for example, of the responses of the non-selective 
instruments. And (4) there seems to be no regular relation of 
the amount of response to the amount or intensity of light used 
even when the lights are of the same composition. At least ac- 
cording to Brown this is the conclusion that must be drawn from 
the work that has been done with white light. If this be true, 
the possibilities of use of the selenium cell as a radiometric in- 
strument seem in general practice to be limited to the equaliza- 
tion of light intensities and this, unless correction factors are 
used, only in case the lights are of the same composition. In this 
regard its case is similar to that of the eye when considered as a 
possible radiometric instrument.149 
148 It will be remembered also that the intensity of the photo-electric cur- 
rents that are set up by the action of light on selenium which are an im- 
portant factor in the variability of action of the cell, are different for the 
different wave-lengths. This factor obviously can not be controlled and 
there seems no satisfactory calibration for it. 
149 In this lack of a simple relation between the amount of intensity of 
light and amount of response, the selenium cell again presents an analogy to RADIOMETRY IN PSYCHOLOGICAL OPTICS 
39 
4. Theories of the action of light on selenium. It may be of 
interest to append here a brief account of the theories that have 
been advanced to explain the action of light on selenium. The 
change of resistance of selenium under the action of light has at 
various times been thought to be a heating effect, to be electro- 
lytic in nature, to be electronic, or to be of chemical origin. The 
first view was held by Sale, Sabine and Moser. Sabine, for ex- 
ample, thought that the action is similar in character to that of a 
dielectric "more or less charged with conducting crystals." 
In such a case the light by its heating effect would modify the 
surface tension of the selenium, which modification would prob- 
ably cause an expansion of its crystalline surface and this in 
turn would result in a closer contact among the superficial crys- 
tals. This view was disproved by the demonstration that the 
light effect of the different wave-lengths on selenium does not 
correlate with their heating effect. 
The theory that the action is electrolytic was first proposed by 
Adams and Day. They did not claim that actual electrolysis 
takes place, however, but that the molecular structure or crystal- 
line condition of the selenium is altered or modified by the action 
of a current of electricity in such a manner as to produce effects 
analogous to those which would occur if the selenium were an 
electrolyte and were actually decomposed by the current. Further- 
more, they thought that the action of light falling on selenium 
is to promote crystallization and thus to diminish its resistance to 
an electric current, inasmuch as in changing to the crystalline 
state selenium becomes a better conductor of electricity. And as 
this crystallization is greatest in the exterior layers of the 
the eye. And as in case of the eye this relation has as yet proven incapable 
of mathematical formulation. Fechner in his attempt to give a mathematical 
expression of the relation between stimulus and response for sensation in 
general was only trying to do what a number have tried to do with regard 
to this reaction both of the selenium cell and the photographic plate. When 
one knows how signally the attempt to find an expression separately for 
either of these reactions has failed, one realizes still more clearly the 
a priori improbability of finding a single expression that will apply to the 
reactions of five sensory mechanisms so differently constituted as they seem 
to be. 40 
C. E. FERREE AND GERTRUDE RAND 
selenium, a flow of energy from within outwards is produced 
which under certain circumstances appears to produce an electric 
current (the photo-electric current). Bidwell thought the action 
is really electrolytic, impure selenides having been used or selen- 
ides having been formed between the selenium and the metallic 
electrodes. 
Bidwell's view was disproved by Pfund and later by Berndt, 
both of whom used purified selenium and carbon electrodes and 
got greater sensitivity to light with the purified than with the 
impure selenium. Pfund developed an electronic theory of the 
action, an explanation that had previously been suggested by 
Nagaoka. He considered the effect due to a resonance of the 
electrons in the atom under the action of light, causing explo- 
sions which lead to an increase in the number of conducting elec- 
trons. There is, moreover, a "critical depth" of penetration above 
and below which the action on selenium is less pronounced. This 
fact accounts for the selective response to wave-length of light, 
the maximum response being to that light which penetrates to 
the "critical depth"; also for the change in this selectiveness of 
response to wave-length with change of intensity of the incident 
light. This view is held also by Ries and Nicholson. 
The chemical theory has been followed among others by Marc, 
Monten, Kruyt, Pochettino and Berndt. These men have ob- 
tained evidence that leads them to believe that there exist at least 
two forms of metallic selenium of widely different electrical re- 
sistivity, and they assume that illumination brings about a trans- 
formation from the less to the more conductive of the two. 
Brown claims to have found and isolated three forms of se- 
lenium crystals. They are produced at different temperatures in 
the annealing process and possess different conductivity. These 
"crystal forms" have also a different selectivity of response to 
wave-length. In his opinion the character of the conductivity 
curves for the four known varieties of light-sensitive selenium can 
be explained by assuming the existence of three components in dy- 
namic equilibrium under a given illumination, temperature, pres- 
sure and electrical potential difference. Any agency that changes RADIOMETRY IN PSYCHOLOGICAL OPTICS 
41 
the conductivity of selenium is of such a nature that it alters the 
rate of interchange between these components. 
F. The photo-electric cell. 
The action of the photo-electric cell depends upon the effect 
of light on the capacity of certain metals to hold a negative 
charge of electricity. Knowledge of the action of light on 
the conduction of electricity goes back to the discovery by 
Hertz in 1887 that the incidence of ultra-violet radiations 
on a spark gap facilitates the sparking. This led to a gen- 
eral investigation of the effect of light on the conduction of 
electricity.150 The discoveries which paved the way directly for 
the invention of the photo-electric cell were those pertaining to the 
effect of light on the electrical condition of certain metals. It 
was found, for example, that a zinc plate exposed to light becomes 
slightly positively charged; that a negatively charged plate be- 
comes less negatively charged; and that a positively charged plate 
is not affected. Later studies showed that the electrical condition 
of all metals is' changed to some extent by the action of light. 
Those affected most are, according to Elster and Geitel,151 
rubidium, potassium, alloy of potassium and sodium, sodium, 
lithium, magnesium, thallium, and zinc. 
The essential parts of the photo-electric cell are as follows. 
There must be juxtaposed in a glass tube or vessel a negatively 
charged surface of the metal in question (the cathode) and a 
conductor or anode to receive the charge as it is lost by the 
cathode under the action of light. Connected in series with the 
cell is either an electrometer or a galvanometer. In some cases 
the inside of the cell is coated with the metal forming the cathode 
and a receiving wire is suspended in the cell. In others, the 
negatively charged metal is suspended in the cell and the body 
of the tube silvered on the inside serves as anode. The photo- 
150 See Hallwachs, W. Ann. der Phys., 1888, 5?, p. 301; Hoor, M. Reper- 
tonniere des Physik, 1889, 25, p. 91; Righi, A. Comptes Rendus, 1888, 106, p. 
1349 and 107, p. 559; Stoletow, A. ibid., 1888, 106, p. 1149, 1593 and 107, p. 91; 
1889, 108, p. 1241; Physikalische Revue, Stuttgart, 1892, 1. 
151 Elster, J. and Geitel, H. Nature, 1894, 50, p. 451. 42 
C. E. FERREE AND GERTRUDE RAND 
electric current may be measured in five ways: (1) by the rate 
of drift of an electrometer needle; (2) by the ballistic method or 
the measurement of the charge acquired in a definite exposure 
time by an electrometer connected with the cell; (3) by 
measuring the potential across the terminals of a high resistance 
in series with the cell; (4) by balancing the photo-electric current 
with a current variable in a known manner, by means of either an 
electrometer or a sensitive galvanometer; and (5) by the deflec- 
tions of a sensitive galvanometer. Ives152 commenting on these 
recommends the third method. He finds the first inadvisable 
because the rate of drift is not uniform; the second, because the 
deflection varies with the exposure time; and the fifth, because it 
is insensitive. 
1. Factors that have been taken into account in the construc- 
tion and rise of photo-electric cells. The following are some of 
the factors that have been taken into account in the construction 
and use of sensitive cells. 
a. The metal used. The metal used should have a high emis- 
sive power and should permit of a certain ease in handling. 
Different metals have been used by different experimenters. 
Compton and Richardson,153 for example, used aluminium, plati- 
num, sodium and caesium. Potassium and sodium have been 
most frequently employed. For a summary of the metals used 
by different investigators, see Allen, Photo-electricity, 1913, p. 68. 
b. The residual gas. It is desirable of course to have for the 
residual gas one which ionizes easily but not to such an extent that 
the recombination of the ions is measurable, also one which is 
easy to handle. Elster and Geitel154 tested the rate of loss of 
charge from an illuminated surface through air, carbon dioxide, 
oxygen and hydrogen; and found that the rate of leak through 
carbon dioxide was much faster than for any of the other gases. 
Hydrogen, helium or argon have been most frequently used in 
the more recent work with photo-electric cells. 
152 Ives, H. E. Astrophys. Jour., 1914, 39, p. 428. 
153 Compton, K. T. and Richardson, O. W. Philos. Mag., 1913, 26, Ser. 6, 
p. 561. 
154 Elster, J. and Geitel, H. Ann. der Phys., 1890, 41, p. 166. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
43 
c. The pressure of the residual gas. There are two factors 
here which work against each other. If we consider the current 
from the cathode to the suspended loop of wire as the discharge 
of negative electrons from the cathode, a vacuum would offer the 
least impedance. An advantage is gained, however, by adding to 
the electrons sent off by the metal, electrons freed by ionizing the 
gas in the tube. The effect of pressure on the intensity of the 
photo-electric discharge has been investigated by Stoletow,155 
Schweidler156 and Kemp157 by comparing the intensity of the 
current for different pressures under otherwise constant condi- 
tions. For Stoletow the most favorable range of pressures is 
from 0.275-2.48 mm.; for Schweidler, from 1-2 mm.; for Kemp, 
from 2-3 mm. In a recent work Ives, Dushman and Karrer158 
found that the pressure giving the greatest sensitivity varies with 
the voltage; further that the "photo-electric sensitiveness does not 
disappear when the metal is made as gas-free as possible and the 
degree of vacuum is made as high as possible." 
d. The potential difference between anode and cathode. There 
must be a sufficient difference of potential between anode and 
cathode to guarantee that all of the electrons are drawn 
to the anode, i.e., there must be a saturation difference of 
potential. The difference must not, however, be so great as to 
cause sparking, and it must be kept fairly constant. From about 
20 to 180 volts are generally used. For an investigation of this 
effect, see Stoletow159 and Schweidler.160 
e. The galvanometer or electrometer. The electrometer is 
more sensitive than the galvanometer, but it is so much harder to 
manage that it seems to be the general opinion that it should be 
used only for measuring very low intensities.161 
155 Stoletow, A. Comptes Rendus, 1888, Z07, p. 91; J. de Phys., 1890, 9, 
p. 468. 
156 Schweidler, E. R. Sitzungsber. der Wien. Akad., 1898, 107, 2a, p. 881. 
157 Kemp, J. G. Phys. Rev., 1913, 1, Ser. 2, p. 274. 
158 Ives, Dushman and Karrer. Astrophys. Jour., 1916, 43, p. 9. 
159 Stoletow, A. Comptes Rendus, 1889, 108, p. 1241. 
100 Schweidler, E. R. Loc. cit. 
161 See Richtmyer, F. K. Transactions Illuminating Engineering Society, 
1913, 8, p. 461. 44 
C. E. FERREE AND GERTRUDE RAND 
f. The angle of the incident light. Elster and Geitel162 first 
reported that the angle at which the light strikes the cathode plate 
causes a difference in the effect on the plate. This effect has been 
investigated among others by Pohl163 and Kunz.164 Elster and 
Geitel claim to have overcome the effect due to angle of incidence 
by using a diffusing screen so that the light falls equally at all 
angles on the plate.165 Pohl and Pringsheim166 have found in 
addition that the curve of selective response to wave-length of 
the photo-electric cell varies with the angle of incidence. 
g. Dark effects and after-effects. When a cell is charged and 
left in the dark, a slight leakage or discharge is found to take 
place which is increased following an exposure to light. This 
leakage Elster and Geitel167 believe to be due to a conduction of 
current over the surface of the glass from the cathode to the 
anode circuit. In any event guard rings of metal connected to 
earth, placed on the inside and outside of the tube between the 
cathode and anode circuit, completely eliminated the leakage. 
h. Fatigue effects. Some metals when freshly prepared throw 
off many electrons when acted upon by light, then the number 
becomes less. This decrease in the responsiveness of the metal is 
apparently rapid at first, then becomes slower, ceasing perhaps in 
a few days. Allen,168 for example, measured the photo-electric 
activity at different intervals from 2 to 100 minutes after polish- 
ing the surface of a zinc plate. He found a decrease in activity 
which was rapid for the first few minutes, then more gradual 
after 20 to 30 minutes. Sadzewicz169 reports a similar result. The 
effect has also been investigated by Holman,170 Hallwachs,171 
162 Elster, J. and Geitel, H. Wied. Ann. der Phys., 1894, 52, p. 433; 1893, 
55, p, 684; 1897, 61, p. 445- 
103 Pohl, R. Phys. Z., 1909, IO, p. 542. 
104 Kunz, J. Phys. Rev., 1909, 29, p. 174. 
165 Elster, J. and Geitel, H. Phys. Z., 1912, 13, p. 740. 
168 Pohl, R. and Pringsheim, P. Deutsch. Phys. Gesell., 1910, 12, p. 215. 
167 Elster, J. and Geitel, H. Phys. Z., 1913, 14, p. 741. 
168 Allen, H. S. Proc. Roy. Soc., 1907, 78, Ser. A., p. 483. 
109 Sadzewicz, M. Acad. Sci. Cracovie, Bull., 1907, 5, p. 497. 
170 Holman, W. F. Phys. Rev., 1907, 25, p. 81. 
171 Hallwachs, W. Ann. der Phys., 1907, 23, (4), p. 459. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
45 
Compton and Richardson,172 Buisson,173 Ladenburg174 and Bergo- 
witz.175 Bergowitz claims, however, that there is no fatigue in 
case of cells whose negative poles are formed of alkali metals. 
This statement is confirmed by Elster and Geitel176 who say: 
"sogenannte 'Ermudungserscheinungen' an Alkalimetallzellen 
nicht auftreten." Ives, Dushman and Karrer, however, apparently 
found fatigue effects in potassium cells. 
2. Comparative advantages and disadvantages of the photo- 
electric cell. The photo-electric cell has the advantage of com- 
paratively high sensitivity. To offset its sensitivity, however, it 
has the following serious disadvantages, (a) It is selective in its 
response to wave-length. The shorter wave-lengths are 
overestimated. This selectiveness moreover, varies with the metal 
used in the cell. Pohl, and Pohl and Pringsheim have plotted 
the curves of response to wave-length for the following metals: 
mercury,177 platinum and copper,178 potassium-sodium alloy,179 
rubidium, potassium and sodium,180 barium, 181 lithium and 
sodium,182 magnesium and aluminium,183 and calcium.184 The 
selectiveness for calcium is found to be very similar to the 
selectiveness of the eye to wave-length. Richtmyer has 
plotted the curve for sodium ;185 Hallwachs for potassium ;186 
Kunz for sodium-potassium alloy;187 and Elster and Geitel for 
172 Compton, K. and Richardson, O. Philos. Mag., 1913, 26, Ser. 6, p. 561. 
173 Buisson, A. Comptes Rendus, 1900, 130, p. 1298; Ann. Chim. Phys., 
1901, 24, p. 320. 
174 Ladenburg, E. Ann. der Physik, 1903, 12, p. 558. 
175 Bergowitz, K. Phys. Z., 1907, 8, p. 373. 
176 Elster and Geitel, H. Phys. Z., 1913, 14, footnote p. 742. 
177 Pohl, R. Phys. Gesell. Verh. 1909, 11, p. 609. 
178 Pohl, R. Ibid., 1909, II, p. 339. 
179 Pohl, R. Ibid., 1909, 11, p. 715; Pohl and Pringsheim; Ibid., 1910, 12, p. 
215, 349, 682, 697. 
180 Pohl and Pringsheim. Ibid., 1910, 12, p. 1039; 1911, 13, p. 219. 
181 Pohl and Pringsheim. Ibid., 1911, 13, p. 474. 
182 Pohl and Pringsheim. Ibid., 1912, 14, p. 46. 
183 Pohl and Pringsheim. Ibid., 1912, 14, p. 546. 
184 Pohl and Pringsheim. Ibid., 1913, 15, p. ill. 
185 Richtmyer, F. K. Phys. Rev., 1910, 30, p. 385. 
186 Hallwachs, W. Ann. der Phys., 1909, 30, (4), p. 593. 
187 Kunz, J. Phys. Rev., 1909, 29, p. 212. 46 
C. E. FERREE AND GERTRUDE RAND 
rubidium, potassium and sodium.188 (b) Griffith189 working with 
ultra-violet light, and Dember190 working with the visible spec- 
trum, both claim that it is also selective in its response to inten- 
sity. That is, they do not find a constant relation between in- 
tensity of radiation and photo-electric current. Elster and 
Geitel,191 however, found on the other hand a constant relation 
between intensity of light and the response of the cell except in 
case of very intense light. As source they used the light of the 
sun, a mercury arc, a Nernst lamp of 32 cp. and a 2-volt carbon 
lamp. A variable resistance was used with the 2-volt lamp; also 
in a part of the work its light was passed through a blue filter. 
Richtmyer192 found the photo-electric current from a sodium sur- 
face under the action of light from an incandescent lamp to be 
proportional to the intensity of the incident light for very low 
intensities (0.007 candle-foot) up to 620 candle-feet. Ives in 
1914 found that the relation between illumination and photo- 
electric effect is not'linear and differs from cell to cell. In 1916, 
however, working with Dushman and Karrer he reports that the 
cause of these varied relationships lies in "focusing effects." "By 
this term is meant a change of direction of the electron stream as 
the number emitted changes, whereby a different proportion of 
the whole number of electrons reaches the receiving electrode" 
(p. 25). For the elimination of these effects he recommends a 
cell of special construction having absolutely no free surfaces on 
which electric charges can collect (see p. 30). This cell, he 
finds, gives a rectilinear relationship between intensity of illumi- 
nation and photo-electric effect.193 (c) The cell is not sensitive to 
188 Elster and Geitel. Ann. der Phys., 1894, 52, (3), p. 438. 
189 Griffith, I. iPhil. Mag., 1907, 14, (6), p. 297. 
190 Dember, H. Ber. d. kgl. sachs. Akad. d. Wiss., 1912, 64, p. 266. 
191 Elster and Geitel. Phys. Z., 1913, 14, p. 741; 1914, 15, p. 610. 
192 Richtmyer, F. K. Phys. Rev., 1909, 29, p. 71, 404. 
193 Kunz recently reports an investigation of cells of special construction 
the responses of which to white light for a wide range of intensities deviates 
so little from a linear function of intensity of light as to permit of the use 
of the cell for many photometric purposes. He verifies Ives's claim that in 
case of the older spherical form of cell, the photo-electric current is not 
proportional to the intensity of light. He found also that the Talbot-Plateau 
law holds for the responses of the cells described. (Astrophysical Jour., 
1917, 45, PP- 69-88.) RADIOMETRY IN PSYCHOLOGICAL OPTICS 
47 
heat radiations, hence can not be calibrated against the total of 
radiation of a black body. The value of its responses in terms 
of energy units can be determined feasibly and conveniently only 
by the aid of some other radiometric instrument which responds 
to the total of radiation, e.g., the thermopile, the bolometer, the 
Nichol's radiometer, etc. (d) With the present knowledge and 
control of the factors which influence its sensitivity, the cell can 
scarcely be recommended as giving results with a degree of 
reproducibility which is entirely satisfactory. Thus from the 
standpoint both of reproducibility and selectiveness of response 
the use of the cell in its present stage of development even as an 
energy comparator of the visible radiations can scarcely be con- 
sidered as advised by the radiometric specialist. However, the 
cell is still being developed and perfected and may yet be of 
service in measuring light intensities. 
As a light measuring instrument the photo-electric cell has the 
following advantages, (a) It has a comparatively high sensi- 
tivity especially to the shorter wave-lengths.194 And (b) it re- 
sponds very quickly to the light stimulus. Richtmyer claims, for 
example, that one of the special fields in which the cell promises 
to be serviceable is for exposures too short for the eye to be 
used with accuracy and convenience.193 
G. The photographic plate. 
1. The blackening of the plate and the factors which influence 
this action. When light acts on a photographic plate a chemical 
change takes place in the sensitive film which renders it opaque to 
light when the plate has been developed. This is called the black- 
ening of the plate, and is the response that must be calibrated if 
the plate is to be used as a light-measuring instrument. Unlike 
194 The shape of the curve representing the photo-electric response to wave- 
length is for calcium very similar to that of the eye. The maximum for 
most of the other metals that have been investigated, occurs much nearer 
to the violet end of the spectrum. The position of this maximum, as has 
been stated above, depends on the kind of metal used to give the photo- 
electric effect. 
195 Richtmyer claims advantage for the cell only in certain special fields 
of work. (See. Trans. Illuminating Eng. Soc., 1913, 8, p. 459.) 48 
C. E. FERREE AND GERTRUDE RAND 
the thermopile, for example, which gives its maximum response 
when once a thermal equilibrium has been attained and is from 
then on practically independent of the time of exposure to the 
light, the blackening of the photographic plate is a function of the 
time of exposure to the light as well as of its intensity and wave- 
length. An important problem in calibrating has been, therefore, 
to find out whether the amount of blackening sustains any regu- 
lar relation to these two factors. If so, the relation can be ex- 
pressed in terms of a formula; if not, the calibration must at 
every point be empirical. With regard to the degree of regular- 
ity of the blackening there has been a great deal of disagreement. 
In 1862, for example, Bunsen and Roscoe announced that the 
blackening may be expressed by the formula S=i t, in which S 
represents the blackening, i the intensity of light and t the time of 
its action on the plate.196 
196 That equal amounts of blackening are always produced by equal inten- 
sities of light and equal times of exposure was first accepted as a theoretical 
principle by Malaguti (Ann. de Chemie et de Phys., 26, p. 5; Pogg. Ann. der 
Phys., 1840, 49, p. 567). 
It was first experimentally demonstrated within a narrow range of light 
intensities (1 to 2*4 in an arbitrary scale) by Hankel (Abhandl. der k. Sachs. 
Gesell. d. Wiss. z. Leipzig, 1864, 9, p. 55). Its first claim to establishment as 
a general law came from Bunsen and Roscoe (Annal. der Phys, und Chemie, 
1862, 117, pp. 529-562). They say: "So wird man den Satz als feststehend 
betrachten diirfen, dass innerhalb sehr weiter Granzen gleichen Producten 
aus Lichtintensitat und Insolationsdauer gleiche Schwarzungen auf Chlorsil- 
berpapier von gleicher Sensibilitat entsprechen." The law may be applied 
under the following conditions. "(1) Wenn die bei Messungen des gesamm- 
ten Himmelslichts in Betracht kommenden Lichtstarken nur noch von so 
kurzen Inductionsphanomenon begleitet sind, dass die dadurch erzeugten 
Storungen innerhalb der erlaubten unvermeidlichen Beobachtungsfehler fal- 
len ; (2) wenn es moglich ist, eine photographisch sensibele Schicht von 
vollig constant Empfindlichkeit darzustellen; (3) wenn sich eine unverand- 
liche, zu jeder Zeit und an jedem Orte leicht wieder hervorzubringende 
Schwarzung finden lasst, die eine sichere Vergleichung mit einer photo- 
graphisch geschwarzten Flache zulasst." 
The first suggestion that photographic action may be used as a means of 
measuring light intensities seems to have come from Sir J. F. W. Herschel 
who describes an "actinograph or self-registering photometer for meteor- 
ological purposes" in Section VIII of paper "On the Chemical Action of the 
Rays of the Solar Spectrum on Preparations of Silver and other Substances, 
both metallic and non-metallic, and on some Photographic Processes (Philos. 
Trans., 1840, 130, pp. 1-61) ; and suggests on pp. 46-47 that the photographic RADIOMETRY IN PSYCHOLOGICAL OPTICS 
49 
This statement that the effect of the light on the plate may be 
expressed by the product of the intensity and time of exposure or 
is directly proportional to the energy of the light incident on the 
plate, has become widely known in the subjects of Chemistry and 
Physiology as the Bunsen-Roscoe law. If it were true that the 
blackening is proportional to the energy falling upon the sensitive 
surface, the photographic plate could be used directly as an energy 
measurer and would serve as an exceedingly useful means of 
measuring light intensities, for it has the additional advantages 
of great sensitivity and of integration of action through an inter- 
val of time. More recent investigations of the blackening, how- 
ever, beginning with Abney in 1874,197 give little support to the 
law formulated by Bunsen and Roscoe. The blackening does not 
vary regularly with the intensity and time of exposure for any 
considerable range of intensities and times of exposure. It is 
also, as is well-known, selective in its response to wave-length and 
this selectiveness varies with the intensity of the light. That is, 
like the eye the photographic plate is selective in its response to 
intensity (a crude analogy to the Purkinje phenomenon) and is 
irregular in its action through an interval of time.198 Moreover, 
plate may be used to measure light. Later A. Claudet describes the "photo- 
graphometer, an instrument for measuring the intensity of the chemical ac- 
tion of the rays of light on all the photographic preparations, and for com- 
paring with each other the sensitiveness of these different preparations" 
(Philos. Mag., 1848, Ser. 3, 33, pp. 329-335). Claudet also refers to T. B. 
Jordan who invented an instrument which he called a heliograph consisting 
of a cylinder covered with sensitized paper placed parallel to the axis of the 
ecliptic, which turned to follow the sun. The object of this apparatus was 
to get the actinic value of the sun's rays at different times in the day. The 
instrument was improved by R. Hunt in 1845 and called by him an actino- 
graph. In the work of these men we find the somewhat obscure beginnings 
of the subject of actinometry which is later to compete with photometry and 
radiometry as methods of measuring light intensities. 
197 Abney, W. deW. Philos. Mag., 1874, 48, Ser. 4, pp. 161-165. 
198 In addition to the above three characteristics crudely analogous to the 
eye, the action on the photographic plate is said by Bunsen and Roscoe to 
show an inertia or lag in coming to its full value. (See Pogg. Annal., 1855, 
100, p. 481-516; also Photochem. Untersuch., Ostwald's Klassiker, 34, p. 363.) 
The analogy in the response of the photographic plate to the optical 
Purkinje phenomenon was first mentioned by Miethe, Zur Aktinometrie 
astronomisch-photographischer Fixsternaufnahmen, Gottingen, 1889. The 50 
C. E. FERREE AND GERTRUDE RAND 
the response varies with other factors which are in practice 
difficult to control and exceedingly troublesome if not impossible 
to take into account in the derivation of formulae.199 
The tendency in fact among recent investigators has been to 
question whether a mathematical expression can be given to the 
action for any considerable range of intensities and times of ex- 
posure and to disagree widely with regard to the formula that 
should be used for the range of intensities and times of exposure 
regarded as most favorable to regularity of action. At differ- 
ent times the following formulae have been derived to express the 
action: S=i tp in which p represents a constant with a value of 
0.86 (Schwarzchild) ;200 S=k. i tp in which k and p are both con- 
stants (Leimbach) ;201 S=k. log. i tp (Parkhurst) ;202 and S-log. 
(k. im tn) in which k, m and n are all constants (Stark).203 
The validity of the above formulae has been the subject of con- 
siderable experimental investigation. Renwick,204 for example, 
phenomenon is discussed and tested experimentally within certain limits of 
difference in wave-length by Schwarzchild, (Sitzungsber. d. Wien. Akad., 
Math.-Naturwiss. Classe, 1900, 109, 2a, pp. 1127-1135). Discussing the in- 
fluence of the optical Purkinje phenomenon on the systematic differences in 
the different brightness catalogues of stars, he says: "Aber es besteht auch 
fiir die Photographic ein dem Purkinje-Phanomen ganz analoger Ubelstand. 
Zwei verschiedenfarbige Lichtquellen, die bei einer gewissen, fiir beide 
gleichen Expositionszeit gleiche Schwarzung ergeben, erfiillen diese Bedin- 
gung nicht mehr, wenn mann die Expositionszeit andert oder die Intensitaten 
der Lichtquellen im selben Verhaltnisse verstarkt" (p. 1128). 
199 Blackening is said to vary with wave-length of light, with kind of plate, 
with process of developing, with intensity of light, with time of exposure, 
with temperature of plate, and with type of exposure-continuous or inter- 
mittent. With regard to type of exposure Abney (Journ. of the Phot. Soc., 
1893-4, p. 63), Eder (Sitzber. d. Wiener Akad., Math-Nat. Classe, 1899, 108, 
2a, p. 1433), Englisch (Archiv. f. Wiss. Photog., 1899, 1, p. 117; 1900, 2, 
p. 131), and Schwarzchild (Astrophys. Journ., 1900, II, pp. 92-100) all find 
a less effect with an intermittent than with a continuous exposure. Abney, 
for example, finds the retardation to be more pronounced the greater is the 
ratio of closed to open sector, the greater the speed of rotation employed, 
and the less the light intensity. 
200 Schwarzchild, K. Astrophys. Journ., 1900, 11, pp. 89-92. 
201 Leimbach, G. Zeit. f. wiss. Photog., 1909, 7, p. 174. 
202 Parkhurst, J. A. Astrophys. Journ., 1909, 30, p. 33. 
203 Stark, J. Annal. der Phys., 1911, 35, (4) pp. 461-486. 
204 Renwick, F. F. Photog. Journ. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
51 
Fi.9. I- 
contends that the Bunsen-Roscoe law falls short about 1.17 per- 
cent. Schwarzchild205 finds for p a value of 0.86; while Tik- 
hoff206 claims that for the photographic rays p varies from 0.67 
to 0.79; and for the green-yellow rays, from 0.91 to 0.96. Park- 
hurst207 states that p is not a constant but a variable depending 
for its value (a) upon the density of the image, (b) the kind of 
plate used, and (c) the light filter employed. Geiger208 finds that 
the law formulated by Schwarzchild is approximately correct 
within certain limits of time of action on the plate. Keeping the 
intensity the same and plotting the log. of the time against the 
blackening, Geiger obtains the curve given in Fig. 1. So plotted 
the curve should be a straight line according to the formula 
S=i tp if p is a constant. Between the points a and b the line is 
almost straight, he finds. Between these limits alone then the 
action is capable of approximate formulation, and the plate 
should not be used for light measurements for lengths of ex- 
205 Schwarzchild, K. Loc cit. 
206 Tikhoff, G. Mittheilungen d. Nikolai-Hauptsternwarte zu Pulkow, 
1909, 31, (3), p. 31; Comptes Rendus, 1909, 148, p. 268. 
207 Parkhurst, J. A. Astrophys. Journ., 1909, 30, p. 34. 
208 Geiger, L. Annal. der. Phys., 1911, 37, (4) pp. 68-78. 52 
C. E. FERREE AND GERTRUDE RAND 
posure time that do not fall within these limits. Abney209 conducts 
an investigation for the special purpose of showing the depend- 
ence of the blackening upon wave-length. Stark210 finds that 
k, m and n in the formula S=log. (k. imtn) depend upon the 
wave-length; Eder211 gets results showing that the time exponent 
depends upon the wave-length; while Leimbach212 contends that 
bdth the intensity exponent m and the time exponent n are inde- 
pendent of wave-length. Stark213 claims that the time exponent 
n is a constant for a range of light intensities from 1 to 1600 in an 
arbitrary scale. The intensity exponent m within this range 
varies 5 percent, for some emulsions and for others it varies 
widely, k also varies quite widely, k, m and n may be considered 
as constants over a range of intensities varying from 1 to 100 
for the "normalbelichtung." 
2. The possibilities of using the photographic plate in quanti- 
tative work. While the above may be considered only as the brief- 
est mention of the quantitative work that has been done on the 
blackening of the photographic plate, still it is enough to show 
that the plate can scarcely be considered as a feasible light- 
measuring instrument. Its quickness of response, its sensitivity, 
and especially its integration of action through an interval of 
time make it very valuable, however, for many kinds of scientific 
work in which quantitative comparisons are not important. 
H. The eye. 
1. The two possibilities of rising the eye in light measure- 
ments. As an instrument for the measuring or comparing of 
light intensities, the eye may be regarded in two ways. (1) It 
may be used to rate and compare lights designed for its own 
service. In the production of illumination effects this is the 
work of photometry and should be done by the eye or some in- 
strument calibrated to give results in terms of the response of 
209 Abney, W. deW. Proc. Roy. Sec., 1901, 68, pp. 300-321. 
210 Stark, J. Annal. der Phys., 1911, 35, (4), pp. 461-486. 
211 Eder, J. M. Op. cit., p. 1473. 
212 Leimbach, G. Diss. Gottingen, 1909; Zeit. f. wiss. Photog., 1909, 7, 
p. 257. 
213 Stark, J. Loc. cit. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
53 
the eye. And (2) it may be used in balancing the energies of 
lights of the same spectro-radiometric composition. Used as 
such it is one of the most sensitive of the energy-comparing in- 
struments. It can not be used, however, to balance radiometric- 
ally lights differing in composition without elaborate calibration, 
because of the degree of selectiveness of its response to wave- 
length.214 
2. The comparative advantages and disadvantages of the use 
of the eye for balancing energies of light of the same spectro- 
radiometric composition. When used to balance lights of the 
same spectro-radiometric composition, the eye has the following 
advantages, (a) It is highly sensitive, among the most sensitive 
of the light measuring instruments, (b) It is quick in its 
action, reaching its maximum of response in times variously esti- 
214 It is in fact our interest in making a quantitative determination of the 
selectiveness of this response both to wave-length and to intensity in all the 
ways in which the eye responds to its stimulus, that has led us to attempt 
to help bring about means of rendering energy measurements feasible for 
the work in psychological optics. Under this heading would come, for ex- 
ample, the investigation of the selectiveness of the eye's achromatic response 
both to wave-length and to intensity with its wide application to photometry 
and light specification; the selectiveness of the chromatic response; the 
selectiveness to wave-length and intensity shown in the rise and decay of 
both types of response; the selectiveness found in after-image and contrast 
response; etc. In fact neither the characteristics and possibilities of the eye 
as a measuring instrument nor its peculiarities as a sense organ can be defi- 
nitely known without a common unit in terms of which to evaluate the dif- 
ferent wave-lengths to which it gives response. It is, moreover, obvious 
that neither the unit nor method of measurement must in any way involve 
the peculiarities of the responses of the eye itself. 
In this work our point of view is to investigate the responses of the eye 
just as the physicist investigates the responses of his instruments. Too fre- 
quently this investigation has received its direction from theories and doc- 
trinal conceptions. Such investigations can not help but be narrow in their 
scope and are moreover apt to lead to wrong conclusions. Much more will 
be accomplished, we believe, by holding theoretical and doctrinal interests in 
abeyance for a time and to approach the study of the eye's responses with 
the broader purpose of finding out what they are from the purely descrip- 
tive point of view, using methods and technique designed for such a purpose 
and not for the confirmation or destruction of a theory. In any event the 
two points of view should be kept separate in the work of investigation, and 
in the evaluation of results it should be clearly recognized to what degree a 
result is the product of the method of working employed. 54 
C. E. FERREE AND GERTRUDE RAND 
mated from 0.014 to 0.541 sec. And (c) it possesses great ad- 
vantages in the ease and convenience with which it may be used. 
Its disadvantages are very similar to those of the photo-electric 
and selenium cells, (a) Like both of these instruments it is 
very selective in its response to wave-length and can not be used, 
therefore, as an energy comparator of lights differing in spectro- 
radiometric composition without elaborate calibration, (b) Like 
the selenium cell and photographic plate it is selective also in its 
response to intensity. It is perhaps more selective in this regard 
than the selenium cell. And (c) it responds only to the visible 
spectrum and can, therefore, be calibrated against a black body 
only with exceeding difficulty and with many chances of error 
both in the calibration and its subsequent use. (See preface, 
pp. vi-vii. If calibrated, the thermopile or some other instrument 
which is sensitive to the total of radiation would ordinarily be 
employed and the calibration be made in terms of its responses. 
The work of calibration presents, moreover, great difficulties in 
case of the eye because of the lack of a fixed or closely repro- 
ducible scale of responses capable of numerical rating which can 
be correlated with the responses of the calibrating instrument. 
Two possibilities for calibrating suggest themselves: (a) the 
determination of sensitivity curves, valid only for the particular 
eye, the exact physical and physiological conditions, etc. for 
which the calibration was made, and for ranges of intensity in 
which no changes in relative sensitivity occur; and (b) the cor- 
relation of a just noticeable difference series with the correspond- 
ing energy values for such parts of the spectrum as are most 
frequently used. Neither of these possibilities, it is obvious, 
would be of much service for any very wide use of the eye as a 
measuring instrument. It will be the work of later papers to 
show in detail the selectiveness of the eye's response to wave- 
length and to intensity. 
III. A Convenient and Sensitive Radiometric Apparatus 
for Work in Psychological and Physiological 
Optics. 
The radiometric apparatus which we have used in our work 
for the past four years consists of two thermopiles, a very Fig. II RADIOMETRY IN PSYCHOLOGICAL OPTICS 
55 
sensitive Thomson galvanometer and auxiliary apparatus for both 
thermopile and galvanometer. The two types of thermopile, the 
galvanometer, and the auxiliary apparatus for the thermopiles 
and galvanometer were constructed by Dr. W. W. Coblentz of the 
Radiometric Division of the Bureau of Standards. This appar- 
atus is shown in Fig. II. LT is the linear thermopile in its brass 
mounting; ST is the surface thermopile; G is the galvanometer 
with its magnetic shields; X is the auxiliary apparatus for 
thermopile and galvanometer; and Y is the telescope and scale. 
A. THE LINEAR THERMOPILE 
We are at present using two types of thermopile, a surface 
and a linear pile. By using the two the energy of the light em- 
ployed for the colored stimulus can be measured at three places: 
at the opening of the campimeter screen with the surface pile 
which has a receiving area just large enough to cover this open- 
ing; at the analyzing slit of the spectroscope with the linear pile; 
and at the eye also with the linear pile. The linear pile with a re- 
ceiving surface of 2 x 12 mm. is broad enough to cover either 
the analyzing slit or the colored image of this slit in the plane 
tangent to the anterior surface of the eye. The linear pile mea- 
suring at the slit and at the eye would thus be adequate alone for 
our purpose. Some additional advantage is gained perhaps by 
having two instruments to serve as a check on each other and by 
measuring at three places instead of two. 
The thermo-elements in both piles are of bismuth and silver. 
The linear pile consists of 20 elements joined in series. These 
elements are in the form of wire, the bismuth 0.1 mm. in 
diameter and the silver 0.051 mm. in diameter. The total resist- 
ance of the pile is 8.4 ohms; the area of the receiving surface is 
2 x 12 mm. In Fig. HI this thermopile is shown drawn to scale. 
In the lower right hand corner is shown a front view of the pile. 
The row of junctions in the center are the "hot" junctions or 
those exposed to the radiations to be measured. The rows on 
either side of this are the "cold" or unexposed junctions. Di- 
rectly above is shown in detail the formation of a pair of "hot" 
and "cold" junctions. A bead of silver is used in welding the Fig. Ill RADIOMETRY IN PSYCHOLOGICAL OPTICS 
57 
bismuth and silver wires together to form the junctions. Over 
each of the "hot" junctions is fastened a receiving surface of tin 
2 mm. broad and long enough for the successive pieces to overlap. 
That the heat conducted to the "cold" junctions may rapidly 
radiate and thus maintain a temperature difference between the 
two junctions as large and as constant as possible, a surface of 
tin of appropriate size is fastened also over each of the "cold" 
junctions. In the diagram are also shown the top and end views 
of the pile mounted for use. The pile is mounted on a flat metal 
base which slides up and down in grooves constructed on the 
edges of the frame containing the analyzing slit for the spectro- 
scope. When in use it is lowered so that the analyzing slit opens 
directly upon the face of the pile. During the color observation, 
when not in use, the pile is raised to the upper part of the frame 
clear of the slit, and is fastened by means of a small hook which 
engages the upper edge of the frame. 
B. THE SURFACE THERMOPILE 
This thermopile was designed especially for our work by Dr. 
Coblentz. The object was to get a thermopile that would mea- 
sure directly all of the light that fell on the opening of the 
campimeter screen. This opening is 15 mm. in diameter. The 
surface of the thermopile was made 17 x 17 mm. The sur- 
face exposed to radiation was reduced to coincide with the stimu- 
lus-opening by means of a circular diaphragm 15 mm. in diam- 
eter. In order to shield the exposed junctions from the influence 
of air currents this aperture was covered with a thin sheet of 
clear glass (cover glass). 
The pile consists of three units joined in parallel. Each unit 
consists of 20 elements, bismuth and silver wire, joined in series. 
The bismuth wire is 0.1 mm. and the silver wire is 0.051 mm. in 
diameter. The total internal resistance of the pile is 3.65 ohms. 
Each of the "hot" junctions is covered with a surface of tin, 
6 x 1 x 0.02 mm. The "cold" junctions are to the rear of the 
"hot" junctions instead of on either side as is the case in the 
linear pile. That is, on leaving the "hot" junction, instead of 
running to either side each of the elements is bent backward. To 58 
C. E. FERREE AND GERTRUDE RAND 
simplify the construction, the radiating surfaces of tin which 
cover the "cold" junctions in the linear pile were omitted from 
the surface pile. 
A diagram of the surface pile is shown in Fig. IV. A shows 
the mounting for the ivory supports to which the sensitive ele- 
ments are attached; B shows the way in which the wires are bent 
from the "hot" junctions to form the "cold" junctions; C gives 
Fig. IV 
a side view of one unit of the pile consisting of 20 thermo- 
elements mounted on the ivory support; and D shows the way in 
which the units are connected. In order to reduce the resistance 
and thus increase the sensitivity of the pile they are joined in 
parallel. In Fig. V are shown the front and side view of the 
holder for the thermopile. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
59 
FRONT 
SIDE 
HOLDER FOR THERMOPILE 
Fig. V 
In common with all surface radiometers of high sensitivity this 
surface thermopile causes some drift of the zero of the galva- 
nometer unless the instrument and the shutter enclosing it are 
carefully protected from changes in temperature due to contact 
with the air of the room. Winding the terminals and body of the 
thermopile with cotton batting has overcome this effect almost 
entirely. 
C. THE RADIATION STANDARD 
The thermopile might be used for equalizing energies or re- 
producing given light intensities without calibration. However, 
if its responses are to be converted into C. G. S. units, calibration 60 
C. E. FERREE AND GERTRUDE RAND 
is necessary. For this a radiation standard is required. For this 
standard we employ a thoroughly seasoned carbon lamp the radi- 
ations from which have been carefully evaluated in terms of the 
primary standard conserved at the Bureau of Standards.1 From 
the known radiations of this lamp the sensitivity of the pile per 
unit area is determined. This is done as follows. First the value 
of the radiations from the standard incident upon unit area must 
be computed. This computation is based ultimately upon the 
Stefan constant of total radiation from a black body: T -5.7 x 
io'12 watt per sq. cm. For our standard operated at 102.1 volts 
giving 0.4 ampere of current, the value of the radiation at a 
distance of 2 meters is 90.70 x io-8 watt per sq. mm. With this 
radiation value known, the calibration of the thermopile becomes 
easy. It is set up at a distance of 2 meters from the radiation 
standard, and the deflections of the galvanometer, the sensitivity 
of which must have been determined just previous to the calibra- 
tion, is obtained. From this value and the total amount of energy 
falling upon the surface of the pile, the amount of energy re- 
quired to give a galvanometer deflection of 1 mm. is determined. 
This may be taken as a measure of the sensitivity of the radio- 
metric apparatus. The total amount of energy falling on the 
pile is obtained by multiplying the radiation per sq. mm. at 2 m. 
by the area of the receiving surface of the pile and correcting 
this value for the absorption of the glass cover (12% in case of 
our instrument). 
Since the galvanometer sensitivity may vary from time to time, 
it is necessary to establish for it a standard of sensitivity and to 
reckon the radiation sensitivity of the apparatus in terms of the 
reading at this standard sensitivity. In order to use the 
value so established in future work with the apparatus, it is 
necessary to determine each time the current sensitivity of 
the galvanometer which is quickly done with an especially devised 
testing apparatus, and to compute from this and the standard 
sensitivity a correction factor which has to be applied to all 
readings taken at this time. 
1 Coblentz, W. W. Bull. Bur. Standards, 1914, 11, p. 87. RADIOMETRY IN PSYCHOLOGICAL OPTICS 
61 
D. THE GALVANOMETER 
The galvanometer used was constructed specially for the ther- 
mopile employed. It is of the Paschen small-coil type, shielded 
from magnetic influence by four cylindrical soft iron shields. 
Its parts are as follows. The magnetic field of the instrument 
is given by four coils each having a resistance of 6 ohms. Each 
coil is wound in three layers, 2 ohms per layer. The wire used 
was B & S gauge Nos. 38, 30, and 26, single covered silk insula- 
tion, the diameter of the bare wires being respectively 0.101, 
0.255, 0.405 mm., and the lengths 92, 595, and 1375 cm. Each 
coil has a diameter of 2.8 cm. and a thickness of 7 mm. The 
coils are joined in pairs, series parallel, giving a total resistance 
of 6.58 ohms. The needle system consists of two groups of six 
magnets placed above and below the mirror. The magnets are 
of tungsten steel and are from 1.5 to 2.5 mm. in length; from 
0.3-0.4 mm. in width; and 0.1 mm. in thickness. They are 
mounted so that each group of magnets has the form of an 
ellipse. The mirror is of thin cover glass 2 mm. x 3 mm., platin- 
ized by cathode discharge. The magnet groups and the mirror 
are mounted on a segment of a very small glass rod in such a 
way that the centers of the groups are 33.5 mm. apart and the 
mirror is midway between them. At the lower end of the rod is 
attached a damping vane of bolometer platinum 5 x 4 x 0.003 mm. 
The needle system weighing 12-15 m&- was made heavy to mini- 
mize the influence of earth tremors. It is suspended between the 
coils by means of an extremely fine quartz fiber. 
The assembled galvanometer consisting of base provided with 
leveling screws, the coils and their ebonite supports, the needle 
system and the containing tube for its suspension, is drawn to 
scale in Fig. VI and needs little explanation. The coils, coated 
with paraffine to give the insulation needed, are attached to the 
ebonite supports by means of soft wax (a mixture of Venice 
turpentine and beeswax which hardens to the desired consist- 
ency on standing). Soft wax is used because it permits of an 
easy and convenient adjustment of the position of the coils, the 
faces of which should be brought to exact parallelism. This can 62 
C. E. FERREE AND GERTRUDE RAND 
be done very conveniently by moving up and down between the 
faces of the coils a rectangular object of the proper width, e.g., 
a microscope section glass. Having been rendered parallel the 
faces of the coil are brought into the vertical by means of the 
leveling screws on the base of the instrument. The quartz fiber 
which suspends the needle system passes through the containing 
tube and is attached at the top to a short brass pin with a 
milled head, held in place by means of a set screw. The 
needle system is fastened to the quartz fiber with shellac. The 
attachment of the needle system to the quartz fiber and this in 
turn to the brass pin is made on a special mounting board de- 
signed for the purpose. The housing of the galvanometer is of 
microscope section glass. This not only prevents disturbances of 
the needle system due to air currents but permits of an excellent 
illumination of the interior of the galvanometer. 
The magnetic shielding215 of the galvanometer consists of an 
inner laminated cylinder made up of six layers of transformer 
iron, 3% inches in diameter, and three sections of soft iron pipe 
5, 6 and 10 inches in diameter respectively, and 10 inches high. 
Each of these shields contains a horizontal window 1 cm. high 
and 8 cm. long through which the mirror is viewed. They rest on 
a slab of slate and a glass plate is put on top. Care is taken to 
keep the shields well annealed, but in spite of this some magnet- 
ism is acquired in handling. The influence of this and of the 
earth's field on the needle system has to be overcome if the galva- 
nometer is to have the long period needed for a high current sensi- 
tivity. This might be done in two ways. (1) The unknown and 
very complex field due to the iron shields could be ascertained by 
means of control magnets; or (2) a stronger and simpler known 
field could be created and this field be weakened to the desired 
amount. The latter is found to be the more feasible procedure. 
The field is created by a short magnet placed under the base of the 
galvanometer in such a position that the mirror and its reflected 
215 For working in localities close to street cars and other magnetic dis- 
turbances, it is necessary to use a more thoroughly shielded galvanometer, 
the coils of which are imbedded in soft iron. See Bull. Bur. Standards, 
1916, 13, p. 423- Fig. VI RADIOMETRY IN PSYCHOLOGICAL OPTICS 
63 
scale are brought into the field of the observing telescope. This 
field is then weakened the desired amount by a larger magnet of 
greater strength placed on the glass plate covering the top of the 
shields. The large magnet is rotated about the vertical axis of 
the galvanometer until the needle which follows it passes quickly 
through a neutral point and rotates in the opposite direction. In 
this position of the magnets the effective field is not so great as 
the moment of torsion of the needle system, and the galvanometer 
should have a long period and a high current sensitivity. 
E. THE AUXILIARY APPARATUS 
This consists of a device for testing the sensitivity of the gal- 
vanometer, a set of resistance coils to cut down the throw of 
current from the thermopile, and a reading telescope and scale. 
i. The sensitivity tester for the galvanometer. The sensitivity 
tester consists of a dry battery giving an E.M.F. of 1.43 volts, and 
of three shunt coils with the necessary switches mounted on a 
suitable base. The shunt coils have a resistance of 1000, 100 and 
1000 ohms respectively and are designed to pass 1.43 x io-8 am- 
pere of current through the galvanometer. This divided by the 
number of scale divisions of the deflections produced gives the 
value of the current required to give a deflection of one scale di- 
vision, or the sensitivity of the instrument. That is, the formula 
expressing the sensitivity is i = 1.43 x io-8, in which d is the 
d 
number of scale divisions in the deflection; 1.43 x io~8 is the total 
strength of the current; and i is the amount of current required to 
produce a unit deflection. For example, the galvanometer we use 
when adjusted to a 3 second period single swing gives a deflection 
of 40 scale divisions. This gives a sensitivity of 1.43 x io~8, or 
40 
3-575 x io"10. 
The galvanometer may be connected either with the sensitivity 
tester or the thermopile by means of two-pole, double-throw knife 
switch, as is shown at x in Figure II. When connected with the 
shunt coils of the sensitivity tester, the circuit is closed with the 64 
C. E. FERREE AND GERTRUDE RAND 
dry cell by means of a key shown at k. Sometimes when the 
galvanometer is adjusted to a high degree of sensitivity a big 
shift of the zero occurs when the double-throw switch is changed 
from the pole connecting the galvanometer with the sensitivity 
tester to the pole connecting it with the thermopile. This makes 
it necessary to readjust the magnets above the galvanometer to 
bring the zero again to the center of the scale. To avoid this a 
further means is provided for testing the sensitivity of the gal- 
vanometer without opening the switch which connects the thermo- 
pile with the galvanometer. This consists of an auxiliary coil 
of wire which is connected directly with the dry battery and a 
key to close the circuit, shown at o. This coil is mounted in a 
fixed position on the casing about the tube containing the suspen- 
sion above the needle system. When a current is sent through 
the coil a deflection is given which is proportional to the current 
sensitivity of the galvanometer and in fixed ratio to the deflection 
that is produced when the same amount of current is sent through 
the shunt coils of the sensitivity tester to the galvanometer coils. 
This fixed ratio may be determined once and for all by closing 
each of these two circuits in turn with the dry cell and comparing 
the deflections produced. 
2. Special resistance coils. The resistance coils designed to cut 
down the throw of current from the thermopile to the galva- 
nometer are thrown into the thermopile circuit by means of a 
series of single-pole, single-throw knife switches shown at A, B, 
C, D, and E, Fig. I. When A is closed no resistance is added, 
the full current passes to the galvanometer, and the true deflection 
is produced. When, however, B, C, D, and E are closed, the 
current is made to pass through io, 40, 100, and 191 ohms re- 
spectively, and the observed deflections must be multiplied by the 
following factors to give the true deflections: 6=1.657J £=3.63 ; 
0=7.57; and E-13.55. 
3. The telescope and scale. The telescope and scale used are 
of the kind ordinarily employed when precision is wanted in the 
reading of a sensitive galvanometer. The scale is graduated to 
millimeters and illuminated by two 40-watt tungsten lamps which 
can be moved along in front of the scale to give the maximum RADIOMETRY IN PSYCHOLOGICAL OPTICS 
65 
illumination of the part of the scale reflected by the mirror. The 
telescope is fitted with a lens system that will permit the clear 
reading of the scale at a distance of 2 meters. At this distance 
the definition is such that the scale can be read to 0.5 mm. 
We recommend to workers in psychological optics the appar- 
atus described in the foregoing pages as feasible, adequately sen- 
sitive, and precise. We feel that especial acknowledgment is 
due to Dr. Coblentz for a notable contribution to the apparatus 
available for the more definitely quantitative work in the subject. THE SELECTIVENESS OF THE ACHROMATIC RE- 
SPONSE OF THE EYE TO WAVE-LENGTH 
AND ITS CHANGE WITH CHANGE 
OF INTENSITY OF LIGHT 
By C. E. Ferree and Gertrude Rand, 
Bryn Mawr College. 
In a previous paper, " Radiometric Apparatus for Use in 
Psychological and Physiological Optics-Including a Discus- 
sion of the Various Types of Instruments that have been used 
for Measuring Light Intensities," Psychological Reviezv Mon- 
ographs, XXIII (5), 1917, we have pointed out that the selen- 
ium cell, the photographic plate, the eye and the photo-electric 
cell are all selective in their response to wave-length and that in 
case of the first three the amount of this selectiveness varies 
with the intensity of the light used.1 It was also stated that 
one of the purposes to which the apparatus described in that 
paper is to be devoted, is a study of the selectiveness of the 
eye's responses to wave-length and to intensity. Bearing on 
this point we need scarcely repeat that the kind and amount 
of selectiveness shown by the eye in its responses can be 
determined only by comparing it with an instrument whose 
responses are non-selective or directly proportional to the 
energy or physical value of the light waves employed. The 
instrument we have selected and described as most feasible 
for this purpose in the present stage of development of radio- 
metric apparatus is the thermopile.2 
Using the responses of the thermopile, therefore, as a 
1With regard to whether or not the selectiveness of response of the 
photo-electric cell changes with the intensity of the incident light, in- 
vestigators disagree. In the article referred to above will be found a 
resume of the literature on this point. 
2 For a similar investigation by means of the thermopile of the selec- 
tiveness of action of the selenium cell, see A. H. Pfund, Philosophical 
Magazine, vii., (6) 1904, p. 26; see also Brown and Sieg, Physical Re- 
view, iv., 1914, p. 48. 
It is obvious that the deviations from proportionality to the energy 
280 281 
STUDIES IN PSYCHOLOGY 
standard of reference we have undertaken a somewhat ex- 
tensive investigation of the selectiveness of both the achro- 
matic and chromatic responses of the eye to wave-length. 
Preliminary to the detailed report of this investigation which 
cannot be made for some time, it may not be out of place to 
give here a few comparisons of the achromatic responses of 
the eye with those of the thermopile. These comparisons 
were made four years ago and have been withheld from print 
in the hope of a speedier completion of a larger part of the 
study. They have in general taken three forms: (a) a com- 
parison at different intensities of the photometric value of 
stimuli made equal in energy value; (b) a similar comparison 
of stimuli having the relative energy values of a prismatic 
spectrum of a given type; and (c) a comparison at different 
intensities of the energy values of stimuli made equal photo- 
metrically. This, it scarcely need be pointed out, is only part 
of the programme that should be followed in making a study 
of the peculiarities and characteristics of the achromatic re- 
sponse to wave-length and in carrying out the interchecking 
of methods that is needed for making such a study; but as yet 
time has not been available to cover this broader field even in 
a preliminary way. 
In this preliminary work time has been had for the use of 
only a limited number of stimuli and intensities. In later 
work results will be given for a part of the work, at least, for 
a greater number of points in the spectrum and for a greater 
number and range of intensities of light. The lights used as 
stimuli were a red, orange, yellow, yellow-green, green, green- 
of the light waves which are shown in the responses of the selenium 
cell, the photoelectric cell and the photographic plate are in need of 
investigation as well as those which are shown in the responses of the 
eye. Until these deviations are known quantitatively and corrected for 
in each individual apparatus, instruments of the type can not be used 
directly to measure energy. They may be used to advantage, perhaps, 
as balancing or equating instruments, but then only when the lights em- 
ployed are of the same composition; or again in cases where it is pos- 
sible and feasible to determine and use correction factors. If the lights 
are not of the same composition it is obvious that the selectiveness of 
response of the instrument will make the balance a false one or one not 
proportional to the actual amounts of energy involved. STUDIES IN PSYCHOLOGY 
282 
blue, blue, violet and a mixed or white light. The colored 
lights unless otherwise specified were narrow bands taken 
from the following regions of the spectrum: 655/^, 6x6^, 
580/z/a, 522/x/z, 488/z/x, 463/i/a, and 439/z/z. In order to 
get the high intensities needed at the photometer head, the 
analyzing slit was purposely made somewhat wide,-0.5575 
mm. This width of slit was kept constant for all parts of the 
spectrum. If our problem had been such that smallness of 
range of wave-length had been a more important condition 
than intensity, a narrower slit would have been used. We 
have not considered, however, that smallness of range of wave- 
length was as important a condition to secure for the purposes 
of this investigation as the range of intensities that was made 
possible by using the greater width of slit. The white light 
was obtained by synthesizing the spectrum between wave- 
lengths 730 and 432/z/z. The above lights were used in mak- 
ing this comparison in part because of the importance to cur- 
rent work in heterochromatic photometry of a comparative 
knowledge of the eye's peculiarities of response to these 
lights; and in part because they are taken from those parts 
of the spectrum to which the eye shows its most significant 
changes in selectiveness of response with changes of light 
intensity. The lights, both colored and white, were taken 
from the spectrum (a) in order that the former should be as 
homogeneous as to wave-length as was practicable; and (b) 
in order that both should be free from the infra-red and ultra- 
violet radiations which would affect the thermopile but not 
the eye. In synthesizing the white light from the spectrum 
care was taken, therefore, to use only wave-lengths safely 
within the red at one end and the violet at the other.2a The 
light was examined in every case at the analyzing slit for im- 
purities by means of a small Hilger direct vision spectroscope 
provided with an illuminated scale. When found, impurities 
were absorbed out by thin gelatines selected so as to cut out 
as little of the useful light as possible. These gelatines were 
placed over the analyzing slit and were held in position by 
short clips fastened to the front surface of the jaws, the edges 
of which formed the slit. The importance of securing a high 
2a The work with the white light will be given in a later paper. 283 
STUDIES IN PSYCHOLOGY 
degree of purity when correlations are to be made between 
the responses of the eye for a given region of the spectrum 
and energy values is obvious, i. e., in proportion to their energy 
the alien wave-lengths would affect the eye differently from 
the wave-lengths under investigation.3 
In making this comparison between the relative photometric 
and radiometric values for the different wave-lengths for the 
first type of investigation mentioned above, it was desirable 
that the same amount and as nearly as possible the same radia- 
tion density of light should fall on the receiving surface of 
both instruments. This was accomplished by making both 
determinations at the same place and having the same cross 
section of the collimated beam of light fall on the receiving 
surface of the two instruments. In detail the procedure was 
as follows: The colored light was obtained by means of the 
spectroscopic apparatus described in a previous paper.4 To 
prevent an undue reduction by spreading and to make possible 
the requirements mentioned above with regard to the incidence 
of .equal amounts and densities of light on the receiving sur- 
faces of the two instruments, the waves of light emerging 
from the analyzing slit of this apparatus were rendered ap- 
proximately parallel by being passed through a collimating 
lens placed at a distance from the slit equal to its focal length. 
At a given place in the beam were mounted interchangeably 
the thermopile and the photometer head.5 The thermopile was 
of the surface type5 with a receiving area of 15x15 mm. 
3 In their work on the determination of the visibility of radiation in 
the red end of the visible spectrum, Hyde and Forsythe (Astrophysical 
Journal, xliv, 1915, p. 289) found impurities to the value of about 20 per 
cent, at A = 0.76^. In our own work determinations made with and 
without precautions for absorbing the scattered light show differences 
in result which are great enough to be considered of significance. In 
the work of Nutting and of.Ives on the visibility of radiation, nothing 
is said about stray light or about precautions for eliminating it or cor- 
recting for it. 
4 C. E. Ferree and G. Rand. A Spectroscopic Apparatus for the In- 
vestigation of the Color Sensitivity of the Retina, Central and Peri- 
pheral. Journal of Experimental Psychology, I., 1916, pp. 247-283. 
5 Owing to the smaller amount of energies represented in the shorter 
G C. E. Ferree and G. Rand. Radiometric Apparatus for Use in 
Psychological and Physiological Optics. Psychological Review Mono- 
graphs, xxiv., 1917. STUDIES IN PSYCHOLOGY 
284 
The photometer head consisted of a 6o° brass prism (a 
modified form of the type of photometer head known as the 
Ritchie wedge) of such dimensions that the two surfaces 
adjacent to the 6o° angle were each equal in area to the re- 
ceiving surface of the thermopile. These two surfaces were 
covered with magnesium oxide deposited from the burning 
metal. They received the standard and comparison lights 
and served as fields for the photometric comparison. In 
order that the same cross section of light should fall on the 
photometric surface as fell on the receiving surface of the 
thermopile, the arrangement of the apparatus was such that 
the photometer surface was normal to the beam of light. This 
relation of axis of beam of light to photometer surface was 
made to apply alike both to the standard and comparison lights. 
To compensate for possible inequalities in the coefficients of 
reflection of the two surfaces the two faces could be inter- 
changed by rotating the prism 1800 about its horizontal axis. 
In making the photometric determination the colored light 
was balanced against the light from a standardized lamp. The 
standard used was a single loop street series tungsten seasoned 
and standardized by the New York Electrical Testing Labora- 
tories to operate at 10.7 candle-power. Further to secure con- 
stancy of light flux this lamp was operated by storage bat- 
teries. In making the photometric-radiometric comparison 
the colored lights were, as stated above, first made radiometri- 
cally equal to the violet at the point of work. The thermo- 
pile was then removed, the photometer head put in its place, 
wave-lengths, for example, in the blue and violet, it was a matter of some 
difficulty to make an energy measurement of these wave-lengths with 
a satisfactory degree of precision at the position of the photometer 
head. It was thought better, therefore, to determine a reduction factor 
representing the relation of the total amount of energy emerging from 
the slit and the amount that was contained in the cross section falling 
on the photometer head. This was done, for example, by measuring 
the red at the slit with the linear pile and again at the photometer head 
with the surface pile, the latter receiving precisely the same cross section 
as fell on the photometer head. This factor having been determined, the 
blue light was measured at the slit with the linear pile and the factor 
was applied to give the energy value of the cross section incident on 
the photometer, head. 285 
STUDIES IN PSYCHOLOGY 
and the lights incident on the two surfaces of the prism 
brought to a photometric balance by adjusting the position of 
the standard lamp. 
Since the procedure is somewhat unfamiliar to psycholo- 
gists, it may not be out of place to give here a brief descrip- 
tion of how energy measurements are made by means of a 
thermopile. A description of the procedure at one of the 
places at which measurements were made, namely, the analyz- 
ing slit will be sufficient to show in a general way the method 
we have used in making these measurements. The thermo- 
pile to be used is placed in position immediately behind the 
slit and a blackened aluminum shutter is interposed in the 
path of the beam of light between the slit and the end of the 
objective tube of the spectroscope. Preliminary to the ex- 
posure of the thermopile to the light to be measured, the cur- 
rent sensitivity of the galvanometer is tested by means of a 
special device provided for this purpose in the construction 
of the galvanometer.7 With regard to this procedure it need 
scarcely be pointed out that the current sensitivity of the 
galvanometer varies with the period or time of the single 
swing of its needle system. Since it is not possible to control 
the field so as to get this period always the same, it is neces- 
sary, if results are to be compared, to take some sensitivity 
as standard and to convert all readings into deflections for 
the standard sensitivity by means of a correction factor deter- 
mined at each setting. For a detailed description of the 
method of determining this factor, see Psychological Review 
Monographs, XXIV, 1917. 
The thermopile is next connected with the galvanometer and 
the light allowed to fall on its receiving surface until a tempera- 
ture equilibrium is reached (ca. three seconds for our thermo- 
pile). The deflections are read by means of the telescope and 
scale and the readings are corrected to standard sensitivity by 
means of the factor previously determined. The final step in 
the process of measuring is the calibration of the apparatus, 
7 This device consists of a special galvanometer coil, dry battery 
circuit and switch board with finely graduated resistance. For a 
description of this device, see Psychological Review Monographs, xxiv., 
1917. STUDIES IN PSYCHOLOGY 
286 
i. e., the value of I mm. of deflection in radiometric units is 
determined for the area of thermopile exposed. To do this a 
radiation standard, the value of the radiations from which is 
already known, has to be employed. The standard used by 
us is a carbon lamp especially seasoned and prepared for the 
purpose by W. W. Coblentz8 of the radiometric division of 
the Bureau of Standards. This lamp is placed on a photo- 
meter bar 2 meters from the thermopile and operated at one 
of the intensities for which the calibration was made, in our 
case 0.40 ampere. The thermopile is exposed to its radiations 
with the same area of receiving surface as was used in case 
of the lights measured, and the galvanometer deflection is 
recorded. From the deflections obtained the value of 1 mm. 
of deflection, or the radiation sensitivity of the apparatus 
under the conditions given, is computed from the known 
amount of energy falling on the surface of the thermopile. 
Having the factor expressing the radiation sensitivity of the 
apparatus, the deflections produced by the wave-lengths of 
light measured are readily converted into energy units. The 
radiation sensitivity of the linear thermopile used by us was 
computed in a given case, for example, from the following 
data. The energy value of the radiations per sq. mm. at a 
distance of 2 m. from the standard lamp operated by 0.40 
amperes is 90.70 xio'8 watts. The deflections of the gal- 
vanometer corrected to a sensitivity of f=i x io'10 ampere pro- 
duced by this intensity of radiation falling on the same area 
of receiving surface as was used in measuring the lights em- 
ployed as stimuli, and corrected for the absorption of the 
glass cover of the thermopile, was 625.625 mm. The area of 
surface exposed was 6.862 sq. mm., and the time of exposure 
was 3 sec. The sensitivity of the instrument per sq. mm. of 
receiving surface was, therefore, 145 x io*11 watts. By 
means of this factor the galvanometer readings produced by 
the different wave-lengths of light may be readily converted 
into the energy value of light falling on the receiving surface 
of the thermopile. 
However, as previously stated, the comparisons treated of 
8 W. W. Coblentz. Measurements on Standards of Radiation in Ab- 
solute Value. Bulletin Bureau of Standards, xi., 1914, pp. 87-100. 287 
STUDIES IN PSYCHOLOGY 
in this paper were made four years ago. At that time we 
were not provided with a radiation standard for the calibration 
of our apparatus. Without this calibration, the stimuli em- 
ployed could be equalized in energy or their energy values 
could be compared from the galvanometer deflections, pro- 
duced, but the amounts of energy used could not be expressed 
in the conventional units. If an expression of the intensity 
of the stimuli were to be made in energy units, all the work 
done at that time would have had to be repeated. There has 
not been time to do this in season for the present paper. All 
of the radiometric values needed for the comparisons made in 
this paper will be given, therefore, in terms of relative galvano- 
meter deflections converted for convenience of representation 
into an arbitrary scale in which the largest deflection is given 
the value of 100. This scale has been constructed separately 
for each table in which relative energy values are represented. 
. TABLE I 
Showing the change in the relative selectiveness of the achromatic 
response of the eye to wave-length produced by varying the intensity of 
the light. In this table is given the photometric value in meter-candles 
of wave-lengths selected from different parts of equal energy spectra 
sustaining to each other the following ratios of intensity: A, 1/2 A, 
1/4 A and 1/12 A. 
Stimulus 
Photometric value of stimulus in 
meter-candles 
Intensity 
A 
Intensity 
1/2 A 
Intensity 
1/4 A 
Intensity 
1/12 
Red 
(655 mm) 
.... 0.98 
0.47 
0.15 
0.015 
Orange 
(616 mm) 
.... 1.345 
0.66 
0.35 
0.103 
Yellow 
(580 mm) 
.... 3.4 
1.5 
0.825 
0.35 
Yellow-green 
(553 mm) 
.... 4.02 
1.92 
0.96 
0.777 
Green 
(522 mm) 
.... 2.44 
1.34 
0.82 
0.56 
Green-blue 
(488 mm) 
.... 1.42 
1.06 
0.79 
0.523 
Blue 
(463 mm) 
.... 0.817 
0.546 
0.28 
0.25 
Violet 
(439 mm) 
.... 0.56 
0.26 
0.164 
0.128 STUDIES IN PSYCHOLOGY 
288 
The results for the first type of comparison are given in 
Tables I and II and Chart I. In Table I, Column 2, are given 
the photometric values of stimuli used in making the investi- 
gation, all made equal in energy value to the violet (439/4/x), 
width of analyzing slit 0.5575 mm., corrected for impurities, 
in a prismatic (CS2) spectrum given by a Nernst filament 
operated by 0.6 ampere of current. This will be called In- 
tensity A. In Columns 3, 4 and 5 are given the results of a 
comparison of the photometric values of these same groups 
of wave-lengths made equal in energy at three lower intensi- 
ties: 1/2 A, 1/4 A and 1/12 A. 
Spectra 1/2 A, 1/4 A and 1/12 A were obtained from A 
by the use of the sectored disc with a proper ratio of open to 
TABLE II 
Showing the change in the selectiveness of the achromatic response 
of the eye to wave-length produced by varying the intensity of light. 
In this table are shown in per cent of the original value the photometric 
values of each of the colored lights when their energy values have been 
reduced respectively to 1/2, 1/4, and 1/12 of the values present in equal 
energy Spectrum A. The changes in the deviations of the relative photo- 
metric from the relative radiometric values produced by the changes of 
intensity may be noted by comparing the percentages in Column 2 with 
50 per cent; in Column 3 with 25 per cent; and in Column 4 with 8.33 
per cent. 
Stimulus 
Relation of 
photometric 
value of 
1/2 A to A 
expressed in 
per cent 
Relation of 
photometric 
value of 
1/4 A to A 
expressed in 
per cent 
Relation of 
photometric 
value of 
1/12 A to A 
expressed in 
per cent 
Red 
47.96 
15.31 
1.53 
(655 /z/z) 
Orange 
49.07 
26.02 
7.66 
(616 /z/z) 
Yellow 
44.12 
24.26 
10.29 
(580 /z/z) 
Yellow-green... 
47.76 
23.88 
19.33 
(533 /z/z) 
Green 
54.92 
33.61 
22.95 
(522 /z/z) 
Green-blue 
74.65 
55.63 
36.83 
(488 /z/z) 
Blue 
66.83 
34.27 
30.60 
(463 /z/z) 
Violet 
46.43 
29.29 
22.86 
(439 /z/z) 289 
STUDIES IN PSYCHOLOGY 
Chart I 
Showing the change in the relative selectiveness of the achromatic 
response of the eye to wave-length produced by varying the intensity 
of the light. In this chart is represented the photometric value in 
meter-candles of wave-lengths selected from eight different parts of 
equal energy spectra sustaining to each other the following ratios of 
intensity: A, A, % A, and 1/12 A. 
closed sector inserted between the collimator lens and the 
prism of the spectroscope. The photometric-radiometric com- 
parison was made with the apparatus and by the method al- 
ready described. A graphic representation of these results is 
given in Chart I. 
In order to show the changes in the selectiveness of the 
achromatic response to wave-length produced by changing the 
intensity of light, Table II has been prepared. In the several 
columns of this table are shown in per cent, of the original 
value the photometric values of each of the colored lights 
when their energy values have been reduced respectively to 
1/2, 1/4 and 1/12 of the values present in Spectrum A. Since 
A is an equal energy spectrum, the reduced spectra are also 
equal energy spectra. The changes in the deviations of the 
relative photometric from the relative radiometric values pro- 
duced by the changes of intensity will be noted by comparing 
the percentages in column 2 with 50 per cent.; in col- 
umn 3 with 25 per cent.; and in column 4 with 8.33 
per cent. In this table a high value of the percentage STUDIES IN PSYCHOLOGY 
290 
expressing the relation of the photometric values of A, 1/2 A, 
1/4 A and 1/12 A indicates a relatively slow change in photo- 
metric value for a given change of intensity; and a low per 
cent, value indicates a relatively rapid change. Column 2 
shows that when the equal energy spectrum A is reduced one- 
half in intensity, green (X.52244), green-blue (X.48844) and 
blue (X46344) darken relatively slowly; and yellow (X58044) 
the most rapidly. Columns 3 and 4 show that for reductions 
to one-fourth and one-twelfth, the slowest rate of darkening is 
still from 52244-46344; the most rapid rate, however, has 
shifted to the red (X.65544). 
The results of the second type of comparison are given in 
Tables III and IV and in Chart II. In Table III are given 
the results of a comparison of the photometric values of the 
wave-lengths used in the preceding determinations from spec- 
TABLE III 
Showing the change in the achromatic response of the eye to wave- 
length produced by varying the intensity of light. In this table is given 
the photometric value in meter-candles of wave-lengths selected from 
different parts of prismatic spectra not equalized in energy, sustaining 
to each other the following ratios of intensity: A, 1/2 A, 1/12 A, and 
1/45 A. The source of light was a Nernst filament operated by 0.6 
ampere of current. 
Stimulus - 
Photometric value of stimuli in 
meter-candles 
Relative 
energy value 
of stimulus for 
- Intensity A 
at photo- 
meter head 
Intensity 
A 
Intensity 
1/2 A 
Intensity 
1/12 A 
Intensity 
1/45 A 
Red 
(655 /J./J,) 
2.86 
1.18 
0.405 
0.128 
100.00 
Orange.... 
(616 
6.96 
2.85 
0.70 
0.34 
58.74 
Yellow. ... 
(580 ,mm) 
7.74 
3.34 
0.92 
0.458 
30.36 
Yellow-green 
(553 mm) 
7.12 
3.1 
0.78 
0.42 
19.78 
Green 
(522 mm) 
5.50 
2.76 
0.55 
0.30 
13.59 
Green-blue. 
(488 mm) 
3.60 
2.25 
0.49 
0.25 
8.39 
Blue 
(463 mm) 
0.80 
0.535 
0.242 
0.12 
6.34 
Violet 
(439 mm) 
0.56 
0.26 
0.128 
0.07 
6.16 291 
STUDIES IN PSYCHOLOGY 
TABLE IV 
Showing the change in the achromatic response of the eye to wave- 
length produced by varying the intensity of light. In this table are 
shown in per cent of the original value, the photometric values of each 
of the colored lights when their energy values have been reduced to 1/2, 
1/12, and 1/45 of the values present in prismatic Spectrum A. The 
changes in the deviations of the relative photometric from the relative 
radiometric values produced by the change of intensity may be noted 
by comparing the percentages in Column 2 with 50 per cent; in Column 3 
with 8.33 per cent; and in Column 4 with 2.22 per cent. 
Stimulus 
Relation of 
photometric 
value of 
1/2 A to A 
expressed in 
per cent 
Relation of 
photometric 
value of 
1/12 A to A 
expressed in 
per cent 
Relation of 
photometric 
value of 
1/45 Ato A 
expressed in 
per cent 
Red 
41.26 
14.16 
4.47 
(655 mai) 
Orange 
40.95 
10.06 
4.89 
(616 mm) 
Yellow 
43.15 
11.89 
5.92 
(580 mm) 
Yellow-green 
43.50 
10.95 
5.90 
(553 mm) 
Green 
50.20 
10.00 
5.50 
(522 mm) 
Green-blue 
62.50 
13.61 
6.94 
(488 mm) 
Blue 
66.88 
30.25 
15.00 
(463 mm) 
Violet 
46.43 
22.86 
12.50 
(439 mm) 
tra of four intensities not equalized in energy: A, 1/2 A, 
1/12 A, and 1/45 A. Intensity A is that given by the Nernst 
filament operated at 0.6 ampere of current. Intensities 1/2 
A, 1/12 A, and 1/45 A were gotten by reducing Intensity A 
by means of a sectored disc. In making these comparisons 
there was, as is stated above, no attempt at an equalization of 
the energies of the lights employed. It was desired to find 
their relative photometric values at the four intensities with 
the type of distribution that occurs in such a spectrum as was 
used by us (prismatic CS2 with gelatines interposed at the 
analyzing slit for the absorption of impurities) with the given 
width of analyzing slit. 
This distribution is shown in Column 6, Table III and in 
Chart III. STUDIES IN PSYCHOLOGY 
292 
Chart II 
Showing the change in the achromatic response of the eye to wave- 
length produced by varying the intensity of light. In this chart is repre- 
sented the photometric value in meter-candles of wave-lengths selected 
from nine different parts of prismatic spectra not equalized in energy, 
sustaining to each other the following ratios of intensity: A, % A, 
1/12 A, and 1/45 A. The source of light was a Nernst filament ope- 
rated by 0.6 ampere of current. 
A graphic representation of the results of Table HI is given 
in Chart II. Supplementary to Chart II, Chart III has been 
prepared. In this chart is given a graphic representation of the 
relative energy values of the colored lights used for Intensity A, 
that is, a graphic representation at eight points of the distribu- 
tion of energy in the spectrum used by us when the width of 
the analyzing slit is equal to 0.5575 mm. This chart is con- 
structed from the results in Column 6, Table III. In plotting 
the curve wave-lengths are represented along the abscissa and 
relative energy values along the ordinate. 
Table IV has been prepared to make the same kind of 293 
STUDIES IN PSYCHOLOGY 
Chart III 
Showing a graphic representation of the relative energy values of the 
colored lights used for Intensity A, Chart II and Tables III and IV. STUDIES IN PSYCHOLOGY 
294 
showing of the results in Table III as was made in Table II 
for the determinations for the equal energy spectra. In this 
table are shown in per cent, of the original value the photo- 
metric values of each of the colored lights when their energy 
values have been reduced respectively to 1/2, 1/12, and 1/45 
of their values at Intensity A. The changes in the deviations 
of the relative photometric from the relative radiometric 
values produced by the changes of intensity may be seen 
by comparing the percentages in Column 2 with 50 per 
cent.; in Column 3, with 8.33 per cent.; and in Column 
4 with 2.2.2 per cent. In this table it is again seen 
that in general a rapid decrease of photometric value for a 
given decrease in energy is characteristic of the long wave- 
lengths and a relatively slow decrease of the short wave- 
lengths. 
The foregoing results express relations between the photo- 
metric and radiometric values of the light waves employed at 
the given intensities. While this type of relation may be of 
interest to the physicist and more particularly to the lighting 
specialist, it does not permit of a rating either of amounts of 
response, or of sensitivity or power of giving response in a 
way which is, strictly speaking, quantitative. That is, in order 
that the rating or comparison of sensitivities may be made 
quantitative, it must be possible from the data at hand to 
compare numerically both the amounts of response and the 
amounts of stimulus used to give the response. In this con- 
nection it need scarcely be pointed out that the photometric 
values plotted in the preceding curves cannot be considered 
as amounts of response or sensation quantities, or as sustain- 
ing any simple relation to amounts of response. Two sur- 
faces, for example, illuminated respectively by four and one 
meter-candles of light do not arouse sensations which sustain 
to each other the ratio of four to one, nor have we any knowl- 
edge of what ratio they do sustain to each other. That is, the 
photometric units: the candle, the lumen, the meter-candle, the 
lambert, etc., are all physical (not sensation) magnitudes dif- 
fering in kind essentially from the erg, the watt, etc., only in 
that in photometric practice their comparison with the stand- 
ard is made by the eye, the responses of which are not pro- 295 
STUDIES IN PSYCHOLOGY 
portional to the kinetic energy of the light waves. While, 
therefore, a determination of the photometric and the relative 
radiometric values of the wave-lengths shows that the eye is 
selective in its response to wave-length, the results are not 
obtained in a form that will permit of a simple numerical com- 
parison. To have done this by a photometric method the 
responses should have been brought to equality and the stimuli 
used to give the equal responses should have been estimated 
radiometrically instead of the converse procedure which was 
used in the preceding work. For with equal responses em- 
pirically determined and ratings of stimuli that can be put in 
a numerical scale, an expression can be given to the relative 
sensitivities which is itself numerical. 
That is, in connection with the problem of the quantitative 
rating of sensitivities, it is scarcely necessary to point out that 
without a means of rating the stimuli which treats all wave- 
lengths alike or which in other words gives values directly in 
terms of kinetic energy, an estimate cannot be made of rela- 
tive sensitivities which can be considered as quantitative. In 
other words, with the introduction of radiometric treatment 
of the stimulus in the various laboratories the possibility of a 
comparison of retinal sensitivities that can be considered as 
quantitative to a degree that would be acceptable in the rating 
of a physical instrument, has been presented for the first time. 
In addition, however, it is obvious that an equally important 
point to be considered in connection with the problem of rating 
sensitivities quantitatively, is what amounts of response can 
be employed with sureness of principle for the purpose. The 
amounts of response that can be determined with an acceptable 
degree of precision are, it will be remembered, equal responses, 
the threshold liminal and differential, equal sense differences, 
and the mean or average deviation of a determination of any 
one of these. Of these quantities the last can obviously be 
used with the least a priori sureness of principle in a quantita- 
tive rating of sensitivities, if, as we have stated, it is necessary 
in making the rating quantitative that we be able to compare 
numerically the amounts of response employed. The thresh- 
olds,liminal and differential could conform to this requirement 
only on the assumption that they represent equal amounts of STUDIES IN PSYCHOLOGY 
296 
response of the sense-organ. There is, however, no way of 
proving such an assumption. If they are accepted as equal, 
it must be on the grounds of logical self-evidence. Many, 
however, are unwilling to grant their equality on these grounds. 
Equality of response and equal sense differences seem alone, 
therefore, to be surely capable of numerical comparison as 
sensation quantities, and of these two the determination of 
the former is much the more feasible experimentally and has 
the much wider range of applicability. Now in the use of a 
sense-organ as a measuring instrument where several methods 
are proposed differing in sureness of principle, precision, range 
of applicability, etc., as is the case here, it is customary to 
choose the one having the greatest a priori sureness of prin- 
ciple as a standard and to check up the others against it. If 
they give results which agree with it in the average their use 
is considered permissible. For example, in photometry the 
equality of brightness method is generally conceded to have 
the greatest a priori sureness of principle for the rating of 
lights for the use of the eye and is accepted, therefore, as a 
standard for this purpose in terms of which to evaluate other 
methods which may have advantages of precision or conveni- 
ence of application in certain situations. The problem of the 
quantitative comparison of sensitivities seems to present an 
analogous case. The use of equal amounts of response seems 
to have the greatest sureness of principle; but it is not appli- 
cable to all cases in which a quantitative rating of sensitivities 
is desired. The use of the limen or just noticeable difference 
has, for example, a much broader applicability and its deter- 
mination has perhaps an advantage in precision when a com- 
parison is wanted between monochromatic stimuli differing in 
color value. Obviously, therefore, a comparative study should 
be made of the different possibilities of rating sensitivities in 
situations where all are applicable for the purpose of deter- 
mining whether or not a reasonable degree of agreement 
obtains.9 In our further work on the determination of the 
9 Because of the lack of a simple relation between the response and 
stimulus for a given wave-length or, more properly speaking, a small 
range of wave-lengths throughout the intensity scale, and of a similar 
relation for different wave-lengths at corresponding points in this scale. 297 
STUDIES IN PSYCHOLOGY 
achromatic sensitivity of the eye to wave-length, this will be 
made a prominent feature of the study. Achromatic sensi- 
tivity is selected for this purpose because it is a case for which 
all the determinations mentioned above may be made. 
Obviously in the development of methods of working in 
new fields-and the quantitative rating of sensitivities is 
from the standpoint of its degree of development, at least in 
vision, a new field-counsel should be taken of work and 
methods already well established. A pattern for the rating 
of sensitivities of sense-organs may be had in the practice 
with regard to the physical recording instruments. In the rat- 
ing of the sensitivities of two galvanometers it may be pointed 
out, for example, that the sensitivity to each is expressed for 
comparative purposes by the amount of current that is re- 
quired to produce one unit of deflection. Such a treatment 
of the sense-organ as a recording instrument is not possible 
unless it be assumed that its responses can be laid off in equal 
divisions or units. However, the underlying quantitative 
principle that both the amounts of response and amounts of 
stimulus must be numerically comparable is satisfied by mak- 
ing the comparison on the basis of equal amounts of response. 
While convenient it is by no means necessary that the results 
be expressed in unit terms. In short, the best that can be 
done is probably to accept equal responses as having a priori 
the possibility of quantitative comparison and evaluate the 
possibilities of other means of rating being strictly quantita- 
tive in terms of their agreement with this method taken as a 
standard. 
The work of rating sensitivities based on a correlation of 
equal sensation responses and the energy values required to 
produce these responses brings us to our third type of com- 
parison, namely, a comparison of the energy values of stimuli 
made photometrically equal. As an example of this method of 
it is scarcely to be expected that a close agreement will be obtained in a 
rating of sensitivities made by the limen and just noticeable difference 
with that obtained when equal amounts of response are employed. 
However, a more definite knowledge is needed on the point before there 
can be any systematic treatment of the problem or a fair comparison 
and evaluation of results. STUDIES IN PSYCHOLOGY 
298 
rating sensitivities Tables V-VIII and Charts IV-V have been 
prepared. 
' In carrying out this work the lights were made photo- 
metrically equal and their corresponding energy values were 
measured.10 The stimuli were taken from a prismatic (CS2) 
TABLE V 
Showing the change in the selectiveness of the achromatic response of 
the eye produced by varying the intensity of light. In this table are 
given the relative energy values of wave-lengths selected from seven 
different parts of spectra made photometrically equal at 75, 50, 25 and 
12.5 meter-candles. The results of this and the following table represent 
also a determination of sensitivity by a method according to which both 
the amounts of response and the amounts of stimulus are numerically 
comparable. The comparative sensitivities for a given intensity should 
be proportional to the reciprocals of the relative energy values given in 
this table for that intensity. 
Stimulus 
Relative energy value of stimuli at pupil of eye 
75 meter- 
candles 
50 meter- 
candles 
25 meter- 
candles 
12.5 meter- 
candles 
Red 
(660 
. .. 88.23 
44.50 
15.34 
6.14 
Orange 
(619 ,u/z) 
. .. 15.37 
8.14 
3.39 
1.58 
Yellow 
(582 /z/z) 
6.44 
3.47 
1.55 
0.744 
Yellow-green... . 
(560 /z/z) 
5.18 
2.85 
1.29 
0.647 
Green 
(523 mi) 
9.65 
5.29 
2.41 
1.114 
Green-blue 
(502 
.. 25.89 
15.54 
6.04 
3.02 
Blue 
(469 ju/z) 
.. 100.00 
53.54 
24.04 
9.62 • 
10 The method of working here is similar to that used by Nutting 
(Transactions Illumin. Engineering Society, iv, 1914, pp. 633-643; also 
Philosophical Magazine, xxix, 1915, (6), pp. 301-309) in the determina- 
tion of what he has called the visibility curve for the eye, with the 
following exceptions: (1) he used the method of flicker instead of the 
equality of brightness in making his photometric equalizations, i.e., a 
flicker balance instead of an equal sensation balance was obtained; and 
(2) compatible with his problem, namely, the determination of the visi- 
bility constants for a group eye, more especially the principal one, the 
maximum ratio of the candle to the watt, he used a greater number of 
observers and a much greater number of points in the spectrum. As a 
third point, it may also be mentioned that apparently he has used no 
precautions to obtain greater purity of light than is given by a single 
prism spectroscope. (See footnote, 3, p. 283.) 
Visibility curves have been determined also by Thiirmel (Das Lum- 299 
STUDIES IN PSYCHOLOGY 
spectrum of a Nernst filament operated at 0.7 ampere. They 
were narrow bands in the red (660/4/4), orange (619/4/4), yel- 
low (582/4/4), yellow-green (560/4/4), green (523/4/4), blue- 
green (502/4/4), and blue (469/4/4). Four intensities of light 
were used, made equal respectively to 75, 50, 25 and 12.5 
meter-candles, normal pupil. These higher intensities were 
selected because one of the objects of the investigation was 
to determine whether the selectiveness of the achromatic re- 
sponse to wave-length ceases at the higher intensities. Ives,11 
for example, determined his visibility curve at an illumination 
mer-Pringsheimsche Spektral-Flickerphotometer als optische Pyro- 
meter, Annalen der Physik, 1910, xxxiii, (4), p. 1139) and H. E. Ives 
(The Spectral Luminosity Curve of the Average Eye, Philosophical 
Magazine, xxiv., 1912, p. 853). Both of these men used the method of 
flicker but neither measured the energies of his lights directly. An at- 
tempt was made by both to get the energies of the lights employed by 
the use of the eye as a selective radiometer. (On the use of the eye as 
a selective radiometer, see, for example, Lummer and Pringsheim, 
Jahresbericht d. Schles. Ges. f. vaterl. Kultur, 1906, pp. 95'97 5 Beibl., 
1907, P- 466.) Throughout his work on the determination of the visi- 
bility curve Ives seems to have followed very closely the method used 
by Thurmel two years earlier. 
While the methods that have been used for determining the visibility 
curve are similar in general principle, with the exceptions just noted, 
to that which we have outlined for a quantitative determination of sen- 
sitivities, it is not our understanding of that work that the lights were 
made photometrically equal for the reason that is given above, namely, 
that if sensitivities are to be determined in a way that permits of numer- 
ical comparison the amounts of response as well as the amounts of 
stimulus used in making the determinations must be numerically com- 
parable. The reason that is assigned by Ives (Philosophical Magazine, 
xxiv, 1912, (6), p. 163), for example, is a technical one,-the photo- 
metric comparisons should all be made with the eye under the same 
illumination or in the same state of adaptation. In addition to this 
technical reason which is admittedly pertinent to the use of the eye in 
making any photometric balance between lights differing in color value, 
we have considered it of significance to call attention to this other 
reason which is of much more fundamental importance, we believe, to 
the quantitative determination of sensitivities for any purpose whatever, 
and which apparently has been overlooked. 
11 H. E. Ives. The Spectral Luminosity Curve of the Average Eye. 
Philosophical Magazine, xxiv, 1912, (6), pp. 853-863. The use of arti- 
ficial pupil by Ives does not seem to be a matter of design, but a condi- 
tion imposed upon the work by his apparatus. He says, p. 856: " In STUDIES IN PSYCHOLOGY 
300 
Showing the change in the selectiveness of the achromatic response of 
the eye to wave-length produced by varying the intensity of light. In 
this table are given the reciprocals of the scale values in Table V. The 
comparative sensitivity of the eye to the wave-lengths selected should 
be for a given intensity, directly proportional to the reciprocals of the scale 
values for that intensity.* 
TABLE VI 
Stimulus 
Reciprocals of scale values in Table VII 
75-meter- 
candles 
50 meter- 
candles 
25-meter 
candles 
12.5 meter- 
candles 
Red 
(660 ju/i) 
... 0.011334 
0.02247 
0.06519 
0.16287 
Orange 
(619 h/j.) 
. .. 0.0651 
0.12285 
0.2950 
0.6329 
Yellow 
(582 nn) 
. .. 0.15528 
0.2882 
0.6452 
1.3441 
Yellow-green 
(560 ai/z) 
... 0.19305 
0.3509 
0.7752 
1.5456 
Green 
(523 /z/z) 
... 0.10363 
0.1890 
0.41494 
0.9009 
Green-blue 
(502 mm) 
... 0.0386 
0.06435 
0.01655 
0.33113 
Blue 
(469 mm) 
... 0.01 
0.01868 
0.04160 
0.10395 
* The reader is cautioned against attempting from these data to com- 
pare sensitivities at different intensities. That is, while the amounts 
of stimulus are numerically comparable at the different intensities, ob- 
viously the amounts of response are not. 
which he estimated to be about 25 meter-candles for his own 
eye, normal pupil (300 meter candles falling on a pupillary ap- 
erture of 1 sq. mm.), claiming that at this intensity the achro- 
all the previous work an artificial pupil was used and the results were 
given in terms of meter-candles illumination as viewed through this 
i sq. mm. aperture [objective slit of spectroscope 0.5x2 mm.]. In 
working with a spectrometer the use of a small eye-slit is practically 
imperative. But in practical photometry an artificial pupil of this size 
would necessitate working at illuminations too high to be practicable 
with present illuminants if one had to attain the retinal illumination 
called for by the investigation here described. Were the pupils of all 
observers of the same size under the same conditions, a reduction factor 
might be obtained so that the luminosity curve could be found with the 
artificial pupil and used for a corresponding illumination with the 
natural pupil. Such, however, is not the case. In view of these facts 
it was considered advisable in the present research to have all curves 
made for a normal pupil illumination." 301 
STUDIES IN PSYCHOLOGY 
Showing the change in the selectiveness of the achromatic response of 
the eye to wave-length produced by varying the intensity of light. In 
this table are shown in per cent of the original value, the energy value of 
the stimuli when their photometric values have been reduced from 75 
to 50 meter-candles; from 75 to 25 meter-candles; and from 75 to 12.5 
meter-candles. A high per cent energy value indicates that a relatively 
small decrease in energy is needed to produce the desired decrease in 
photometric value or that there is a relatively rapid darkening of the 
color with decrease of energy. 
TABLE VII 
Stimulus 
Per cent of energy value to original value 
when photometric value has been 
reduced from 
75 to 50 
meter-candles 
7t> to 25 
meter-candles 
75 to 12.5 
meter-candles 
Red 
(660 
50.4 
17.4 
7.0 
Orange 
(619 nv) 
52.9 
22.1 
10.3 
Yellow 
(582 mm) 
53.8 
24.0 
11.5 
Yellow-green 
(560 mm) 
55.0 
25.0 
12.5 
Green 
(523 mm) 
55.8 
25.0 
11.5 
Green-blue 
(502 mm) 
60.0 
23.3 
11.7 
Blue 
(469 mm) 
53.5 
24.2 
9.6 
matic response is practically, if not entirely, free from Pur- 
kinje effects. Nutting12 for a similar reason used 350 meter- 
candles of light falling on a pupillary aperture of 1.465 sq. 
mm., contending that this intensity of illumination is " safely 
outside the range of the Purkinje effect." 
These views, however, it will be remembered, are quite the 
opposite from those of Helmholtz and others of the earlier 
writers who believed that the eye changes its selectiveness of 
response to wave-length with change of intensity of light at 
the higher as well as at the lower intensities of light. This 
conclusion is drawn from a statement made by them that 
beginning with a spectrum of fully saturated colors and in- 
creasing the intensitv of licdit. all the colors are found to tend 
12 P. G. Nutting. The Visibility of Radiation. Philosophical Maga- 
zine, xxix, 1915, (6), p. 303. STUDIES IN PSYCHOLOGY 
302 
TABLE VIII 
Showing the change in the selectiveness of the achromatic response of 
the eye to wave-length produced by varying the intensity of light. In 
this table the comparative sensitivities of the eye to the different stimuli 
at a given intensity are shown in a scale in which the highest sensitivity 
for that intensity is represented by 100. If there were no relative changes 
in the eye's sensitivity to wave-length with change of intensity of light 
for these high intensities, the values in this scale would be the same for 
each stimulus for the four intensities. 
Stimulus 
Relative sensitivity in a scale in which the 
highest sensitivity is represented by 100 
75 meter- 
candles 
50 meter- 
candles 
25 meter- 
candles 
12.5 meter- 
candles 
Red 
(660 mm) 
5.87 
6.40 
8.41 
10.54 
Orange 
(619 mm) 
. .. 33.72 
35.01 
38.06 
40.95 
Yellow 
(582 mm) 
. .. 80.435 
82.14 
83.23 
86.32 
Yellow-green... . 
(560 mm) 
... 100.00 
100.00 
100.00 
100.00 
Green 
(523 mm) 
... 52.18 
53.87 
53.53 
58.29 
Green-blue 
(502 mm) 
... 19.99 
18.34 
21.38 
24.74 
Blue 
(469 mm) 
5.18 
5.32 
5.36 
6.73 
towards white and in so doing to change their luminosities at 
different rates.13 (For example, see Helmholtz, Poggendorff 
Ann. der Phys., 1852, 86, p. 520; also Handbuch der physio- 
logischen Optik, 1896, 2te Aufl., pp. 465-466; A. Chodin, 
Sammlung phys. Abhandl. v. Preyer, 1877, 1, p. 33, ff., 
E. Britcke, Sitzungsber. der Wiener Akad., Math.-Natur. 
Klasse, 1878, 77, (3), p. 63.) 
Since in these as in all of the preceding determinations, a 
photometric value was wanted in terms of the power to arouse 
the achromatic sensation as the eye normally sees its bright- 
ness and not in terms of a flicker evaluation, the equality of 
brightness method was used in making the photometric bal- 
ance. That is, the comparisons are based on an equality of 
13 In this connection it should be borne in mind that Nutting and 
Ives presumably refer to determinations made by the method of flicker, 
while the writers referred to above are considering the eye as it nor- 
mally sees its brightnesses. 303 
STUDIES IN PSYCHOLOGY 
Chart IV 
Showing the change in the selectiveness of the achromatic response 
of the eye produced by varying the intensity of light. In this chart are 
represented the relative energy values of wave-lengths selected from 
seven different parts of prismatic spectra made photometrically equal 
at 75, 50, 25 and 12.5 meter-candles. 
brightness not a flicker balance. The plan of the apparatus 
used in making the photometric balance (spectroscope, photo- 
metric apparatus, etc.,) is indicated in Fig. I. The colored 
light was presented to the eye in the following manner. The 
eye-piece was removed from the spectroscope and a lens sys- 
tem was substituted consisting of two lenses Lx and L2, one 
to render the light emerging from the objective slit parallel 
and the other to focus it on the eye 30 cm. distant. By means 
of an extra set of jaws operating in the vertical, the length 
of the analyzing slit was reduced to a value which gave an 
image 3 x 1.49 mm. on the pupil of the observer's eye. This 
adjustment was maintained throughout the work. The size STUDIES IN PSYCHOLOGY 
304 
Chart V 
Showing a curve of achromatic sensitivity when the stimuli are made 
photometrically equal at 75 meter-candles. 
of the photometric field was limited by a screen S, containing 
a stimulus opening 15 mm. in diameter. This screen was 
placed 20 cm. from the eye. Between this screen and the 
lens L2 was inserted a small strip of metal, D, the inner edge 
of which was carefully beveled. When adjusted to the posi- 
tion used in making the photometric observation, this edge 
just bisected the photometric field. The surface of this strip 
was kept freshly coated with magnesium oxide deposited from 
the burning metal. This surface received the light from the 
standard lamp; the other half of the field was filled with light 
from the spectroscope. The spectroscope and lens system 
were shielded from the standard lamp by suitable screens. 
The photometric balance was obtained as follows. The stand- 
ard lamp, a seasoned 32 cp. carbon lamp giving the color value 
of the carbon standard of 4.85 watts per spherical candle, was 
set at the position on the bar required to give the desired 
intensity of light in the photometric field, and the intensity of 
the colored light filling the other half of the field was varied 
until a brightness match was obtained. The changes in the 
intensity of the colored light required to give the match were 305 
STUDIES IN PSYCHOLOGY 
not made by varying the width of the collimator slit, as is 
often the case, because changes in the width of the collimator 
slit tend to give a variable purity of light,-a condition which 
would have given us more trouble in the selection of our filters 
to absorb stray light. Specially constructed sectored discs 
with a single open sector adjusted by a finely threaded micro- 
meter screw and provided with a Vernier reading to minutes, 
were used instead for this purpose. That is, the collimator 
slit was set at a width which made the comparison field slightly 
brighter than the standard field for the group of wave-lengths 
in question and the gelatines required to absorb the alien 
wave-lengths were placed in position over the analyzing slit. 
These gelatines, as stated earlier in the paper, were held in 
place by short clips fastened on either side of the slit to the 
Fig. I 
front surface of the jaws, the edges of which formed the slit. 
The width of the collimator slit and the gelatines were then 
kept constant, and the reductions needed to give the four 
intensities were made by means of the sectored discs, inserted 
between the analyzing slit and the lens Lx. 
In Table V are given the relative energy values of the stim- 
uli to give equal achromatic responses at the four photometric 
intensities used. These values are shown graphically in Chart 
IV. In constructing this chart as in case of all of the preced- 
ing charts, the wave-lengths are spaced to approximate the 
distribution in the prismatic spectrum. The comparative sen- 
sitivity of the eye to the groups of wave-lengths used should 
of course be as the reciprocals of the relative energy values 
required to give equal achromatic responses. The values of 
these reciprocals are given in Table VI. A graphic repre- 
sentation of the results for the highest intensity in this table 
is given in Chart V. In this chart for the sake of a closer STUDIES IN PSYCHOLOGY 
306 
comparison with more recent work, the wave-lengths are given 
equal spacing along the abscissa and the scale employed (re- 
ciprocal at point of highest sensitivity = I) is the same as 
was used by Nutting in his visibility curve. 
In Table VII are shown in per cents of the values for 75 
meter-candles the energy values of each of the group of wave- 
lengths when they have been made photometrically equal at 
50, 25 and 12.5 meter-candles. In this table a high per cent, 
energy value indicates that a relatively small decrease in en- 
ergy is needed to produce the desired decrease in photometric 
value; or expressed in other terms, indicates a relatively rapid 
darkening of the color with decrease of energy. The results 
show first of all, it will be noted, that change in selectiveness 
of response with change of intensity is still present in the 
region of the intensity scale included between 50 and 75 meter- 
candles, as well as in the regions included between 25 and 75 
meter-candles and 12.5 and 75 meter-candles. A closer scru- 
tiny shows further that in case of the reduction from 75 to 
50 meter-candles the most rapid darkening with a decrease of 
energy occurs in the region of the blue-greem This effect is, 
it will be remembered, quite the opposite of that which was 
obtained at the lower intensities treated of earlier in the paper 
(see Table II, Columns 3 and 4). At these intensities the 
slowest darkening was obtained in the region of the blue- 
green and the most rapid in the region of the red. Moreover, 
for the reduction 75 to 12.5 meter-candles, the region of most 
rapid darkening shifts to the middle of the spectrum. In 
short, in passing from high to low intensities the region of 
most rapid change in selectiveness of achromatic response 
seems to shift from a region in the short wave-lengths at the 
high intensities, through the middle of the spectrum at the 
intermediate intensities, to the long wave-lengths at low in- 
tensities. In Table VIII the change in selectiveness with 
change in intensity is shown in still another way. In this 
table the comparative sensitivities of the eye to the different 
stimuli at a given intensity of light are represented in a scale 
in which the highest sensitivity for that intensity of light is 
arbitrarily given a value of 100. If there were no relative 
change in the eye's sensitivity to wave-length with change of 307 
STUDIES IN PSYCHOLOGY 
intensity of light, the values in this scale would be the same 
for the four intensities of light. A more detailed investiga- 
tion of this point for a greater number and range of intensi- 
ties and for a greater number of points in the spectrum will 
be carried out later. 
In conclusion we would again point out that the foregoing 
results are presented as preliminary and illustrative of some 
of the ways in which the selectiveness of the achromatic re- 
sponse of the eye to wave-length and its change with change 
of intensity may be studied by the help of energy measure- 
ments, rather than as a finished investigation of any one point. 
The work is discursive rather than intensive and in this re- 
gard was actuated by an entirely different purpose from that, 
for example, which has prompted the determination of the 
visibility constants for a group eye for the purpose of obtain- 
ing the mechanical equivalent of light, in which case a number 
of observers and a much greater number of points in the 
spectrum have been used. Moreover, since a sensation bal- 
ance as the eye normally sees its brightnesses was wanted for 
the different intensities of light used, and not a flicker balance, 
all subjective equalizations of light intensities were made by 
the equality of brightness method. This choice of methods 
we consider alone compatible with the purpose of such studies 
as are here outlined, even were all other points of dispute 
waived with regard to the selection of a photometric method 
for other purposes and problems which may be encountered 
in the handling of light intensities. Reprinted from Journal of Experimental Psychoi ogy. Vol. II, No. 4. Aug., 1917.] 
SOME AREAS OF COLOR BLINDNESS OF AN 
UNUSUAL TYPE IN THE PERIPHERAL 
RETINA 
BY C. E. FERREE AND GERTRUDE RAND 
Bryn Mawr College 
At the psychophysical section of the first Congress for 
Experimental Psychology held at Giessen April 18-21, 1904, 
F. Schumann reported what he termed an unusual case of 
color blindness (his own).1 So far as the ability to get the 
positive sensation is concerned, Dr. Schumann is, according 
to the report, totally blind to green and partially so to red.2 
In addition his case presents the following features. (1) 
While green light does not arouse a sensation of green, it 
does give red after-image and contrast sensations. Red light 
on the other hand gives a positive sensation but does not give 
either after-image or contrast sensations.3 (2) A colorless 
mixed light can be matched by combining homogeneous red 
and green lights, but quite a little greater proportion of green 
is needed to give the neutral sensation than is required for 
1 Schumann, F., 'Ein ungewohnlicher Fall von Farbenblindheit,' Bericht fiber 
den I. Kongress fur experimentelle Psychologic in Giessen, 1904, pp. 10-13. See 
also G. E. Muller's discussion of the report, ibid., pp. 20-21. 
2 His diagnosis that he is partially sensitive to red was based on two facts, (a) Red 
in the region of 670 gave a sensation which was plainly different from yellow and 
could not be matched by a full spectrum gray. And (2) orange which to the normal 
eye appeared distinctly reddish, appeared to him a pure yellow. From this point 
in the orange towards the short wave-length end of the spectrum three qualities were 
sensed: a yellow, a blue and a band between them which could be matched with a full 
spectrum gray. On these facts was based the diagnosis of blindness to green. 
3 While a red light does not give green contrast sensation, it does produce an effect 
on a neighboring field which raises the threshold or diminishes the sensitivity to red. 
That is, a gray ring on a red ground appears gray but an amount of red can be added 
to it without being sensed which is supraliminal when red is not present in the sur- 
rounding field. In other words a physiological induction seems to be present which 
inhibits the complementary excitation although the induced excitation does not itself 
arouse sensation. We have here, therefore, another evidence that the complementary 
and induction relations between red and green are intact, the ability of the green 
excitation to arouse sensation alone being absent. 
295 296 
C. E. FERREE AND GERTRUDE RAND 
the normal eye. And (3) a yellow of the spectrum can be 
matched by the combination of a red and green if properly 
selected. In this case also a considerably greater proportion 
of green is needed than is required for the normal eye. 
One of the writers1 mentioned several years ago in a partial 
report of a somewhat extensive investigation of the color 
sensitivity of the peripheral retina that small areas could be 
found in the periphery of the normal retina showing charac- 
teristics for the colors red, green, yellow and blue similar to 
the Schumann case for red and green. That is, areas may be 
found which are totally blind or deficient to one of these colors 
so far as the positive response is concerned, but which seem 
not to be correspondingly deficient in the after-image and 
complementary or cancelling reactions. In fact, so far as 
can be told, the after-image and complementary reactions are 
no different in these areas from those in the immediately 
adjacent normal portions of the retina. We found it infeasible 
to test for a contrast reaction in these comparatively small 
and remote areas. Since that time an examination has been 
made of the eyes of a number of observers, more especially in 
connection with the work of the undergraduate laboratory, 
with the result that although observers may differ widely 
with regard to the number of the spots, their location, and 
the color responses affected, we are inclined to believe that 
the presence of such spots in the peripheral retina may be 
considered the rule rather than the exception. 
A successful search of the peripheral retina for spots of the 
kind described above requires some means of making a rather 
minute investigation from center to periphery in a number of 
meridians. In our own work the rotary campimeter has been 
employed as a means of presenting the light to the different parts 
of the retina and the investigation has been made both with 
pigment papers and with the light of the spectrum as stimuli. 
In case the light of the spectrum was used, a very intensive 
stimulation was given. The lights were narrow bands taken 
1 Rand, G., 'The Factors that Influence the Sensitivity of the Retina to Color,' 
Psychol. Rev. Monog., 1913, 15, footnote, pp. 108-109. See also Ferree and Rand, 
'A Note on the Determination of the Retina's Sensitivity to Colored Light in Terms 
of Radiometric Units,' Amer. Jour, of Psychol., 1912, 23, p. 331. AREAS OF COLOR BLINDNESS 
297 
from the spectrum of a Nernst filament (corrected for impu- 
rities) operated by 0.6 ampere of current, and was of the order 
of intensity at the eye of 11.14 X io-8 watt per sq. mm. for 
the red; 9.676 X io~8 watt per sq. mm. for the yellow; 1.285 
X io-8 watt per sq. mm. for the green; and 0.878 X io~8 
watt per sq. mm. for the blue. The total amount of 
light entering the eye was in each case respectively 36.76- 
X io-8 watt; 31.93 X io-8 watt; 4.24 X io-8 watt; and 
2.90 X io-8 watt. Lights of this intensity for the same 
observers were sufficiently strong, for example, to be sensed 
as color clear out to the limits of white light vision for 
all the colors but the green; and the green, it may be men- 
tioned, could not for the observers used be made to coincide 
with the limits of white light vision whatever intensity of 
light was employed. The only effect of these greater intensi- 
ties v as to narrow in some cases the area of the spots previously 
mapped by means of the pigment paper stimuli. Thus it 
seems that the totally blind area is frequently bounded by a 
zone of weakened sensitivity. The investigation was made 
with three conditions of surrounding field: a gray of the bright- 
ness of the color, white and black. The only effect of the 
white and black surrounding fields was to increase the size 
of a spot mapped with a surrounding field of the gray of the 
brightness of the color. A preexposure of a gray of the bright- 
ness of the color was used in each case. Care was taken to- 
keep the intensity of the illumination of the room accurately 
constant throughout the investigation. 
In testing for the after-image reaction, a few seconds of 
pre exposure was given to a gray of the brightness of the color 
previous to the color stimulation. The color stimulation was 
given for 3 sec. and the after-image projected on a gray of the 
brightness of the color, placed in the path of the colored light 
just behind the campimeter-opening. The intensity of after- 
image was compared as well as could be with that obtained 
with the same intensity and conditions of stimulation on the 
immediately adjacent portions of the retina. So far as could 
be told there was characteristically no difference between the 
strength of the after-image reaction in the spot itself and in 298 
C. E. FERREE AND GERTRUDE RAND 
the immediately surrounding retina. In the investigation of 
the complementary or cancelling reaction, a combination of the 
complementary colors to gray was made for the surrounding 
portions of the retina and this stimulus was presented to the 
color-blind area. In each case it was seen as gray, not as the 
color complementary to that for which the spot was blind. 
We need scarcely point out that in making this test it was 
necessary to determine whether the amount of colored light 
employed would have aroused color sensation had it not been 
combined with its complementary color. This was done in 
two ways: (a) The amount of color used to form the combina- 
tion to gray was presented to the color-blind area, combined 
with the proper value of a substitute sector of the gray of the 
brightness of the color to which the area was blind; and (b} 
the threshold was determined in the color-blind area for the 
color complementary to that to which the area was blind, and 
its value compared with the amount used in making the com- 
bination to gray. The results of both determinations showed 
that a true complementary action was present. 
The peripheral retina spots while similar in a general way 
to the case described by Schumann present the following 
points of difference, (i) There was no detectable weakening 
of the sensitivity to the complementary or antagonistic color 
in the areas in question. And (2) no more of the color to 
which the area was blind was required to combine to gray 
with the antagonistic or complementary color than was needed 
on the normal areas of the retina immediately adjacent. 
In the presentation of results space will be taken for only 
two observers, selected because of rather wide differences in 
the number, size and location of the spots. In Chart I. is 
shown results for Observer R. and in Chart II. for Observer C. 
The Hering pigment papers were used as stimuli and the 
investigation was made with surrounding field and preexposure 
of the brightness of the color in each case. The limits of 
sensitivity were determined for each color in 16 meridians and 
these points are connected to give outline maps of color sensi- 
tivity for stimuli of the intensity used. In working in any of the 
above-mentioned meridians the stimulation was given at points AREAS OF COLOR BLINDNESS 
299 
separated by no more than i°. When a gap or a significant 
depression in sensitivity was found, the campimeter was rotated 
and the investigation made in a sufficient number of near- 
lying meridians to give a careful outline of the deficient area. 
These areas are represented on the charts in black, with letters 
to indicate the colors to which there is a deficiency. In case 
there is total blindness to the stimulus used, the spot is repre- 
Chart I 
Showing for Observer R the Areas of the Peripheral Retina having the Schumann 
Type of Color Blindness. These areas are represented in black, with letters to indi- 
cate the colors to which there is a deficiency. In case there is total blindness to the 
stimulus used, the area is represented in solid black; in case there is only a marked 
depression of sensitivity, the area is shaded. In this latter case areas are represented 
only when the depression amounts nearly to blindness. The only effect of using 
spectrum lights of very high intensity: red (670 mm), 36.76 X io-8 watt at pupil of 
eye; yellow (581 mm), 31-93 X io~8 watt at pupil of eye; green (522 mm), 4-24 X io-8 
watt at pupil of eye; and blue (471 mm), 2.90 X io-8 watt at pupil of eye, was to 
narrow in some cases the area of the spots previously mapped by means of the 
pigment paper stimuli. 300 
C. E. FERREE AND GERTRUDE RAND 
sented in solid black; in case there is only a marked depression 
of sensitivity the area is shaded. In this latter case areas are 
represented only when the depression of sensitivity amounts 
nearly to blindness. They were, for example, so insensitive 
that the color response could not be aroused when an unfavor- 
able brightness of surrounding field or preexposure was used. 
Chart II 
Showing for Observer C the Areas of the Peripheral Retina having the Schumann 
Type of Color Blindness. The conditions of investigation and method of repre- 
sentation is the same in this chart as in Chart I. The results for Observers R and C 
are selected for presentation here because of the somewhat unusually wide difference 
that was found in the number, size and location of their spots. 
In discussing his own case, Schumann seems to think that 
the phenomenon indicates that there must be more than one 
functional level involved in the production of visual sensation: 
peripheral or sub-cortical, and cortical. One of these, the 
peripheral or the sub-cortical, is the locus of the complementary 
or cancelling action, and the after-image and contrast reac- AREAS OF COLOR BLINDNESS 
301 
tions. Green light in his case arouses these three reactions 
because the level concerned in producing them is functionally 
normal. Green does not arouse the positive sensation, how- 
ever, because there is functional deficiency in the remaining 
level or levels. G. E. Muller, who also made supplementary 
tests and experiments on Schumann and discussed Schumann's 
report at the Congress in Giessen, concurs strongly in the con- 
ception that more than one functional level is needed to explain 
the Schumann case. At this same Congress Muller1 discusses 
seven types of color-blindness which he further believes can 
be explained best on the conception of more than one specific 
functional level, the processes of which may be separately 
deficient. His theory of color vision is in fact here elaborated 
to include both peripheral and central visual processes. While 
we have no wish to engage in theoretical discussions at this 
stage in our own work, it may not be out of place, in addition 
to the results presented in this paper, to call to mind in this 
general connection our experiments of the effect of the achro- 
matic excitation on the chromatic which seemed to indicate 
very strongly that this effect both quantitative and qualitative 
takes place at some level posterior to that of the cancelling, 
after-image, and contrast reactions.2 We have also obtained 
other results, as yet unpublished, on contrast induction in the 
far periphery of the retina which seem to indicate that the 
deficiency which for these portions of the retina prevents the 
color stimulation from producing sensation is in part at least 
posterior to the level at which induction takes place. More- 
over, unless the complementary action were intact in case of 
1 Muller, G. E., 'Die Theorie der Gegenfarben und die Farbenblindheit,' Bericht 
uber den I. Kongress fur experimentelle Psychologic in Giessen, 1904, pp. 6-10. 
The conception of a central deficiency was used as early as 1868 by Niemetschek to 
explain color blindness {Prager Vierteljahrschrift, 100, p. 224). 
2 The results of these experiments have as yet been published only in part. (See 
"An Experimental Study of the Fusion of Colored and Colorless Light Sensations. 
The Locus of the Action," Journ. of Philos., Psychol, and Scientific Methods, 1911, 8, 
pp. 294-297.) 
The fuller publication has been delayed for one reason because we have felt the 
need of giving a somewhat exhaustive resume of the work that has yet been done on the 
subject. This work has been so scattered and unsystematic and so much of it appears 
hidden under titles that give no indication that it is there, that the work of compiling 
has been somewhat time consuming. 302 
C. E. FERREE AND GERTRUDE RAND 
blindness to one color over a whole or part of the retina, it 
would seem that white light should always be sensed by the 
subject in the tone complementary to that for which the 
blindness exists.1 Or to put the matter more conservatively 
in terms of color-blindness testing, it would seem impossible 
ever in such cases to match a full spectrum gray to the color 
to which the subject is blind, or to all combinations of comple- 
mentary colors, or to any in fact except the pair to one member 
of which the defect exists. 
In the study of the eye as a recording instrument it is 
always a helpful feature to take our start from the physical 
recording instruments which respond to light. Not only are 
the characteristics of a given one of these instruments more 
accessible to study than the eye, but the instrument itself 
can be changed and the effect produced be noted. Also differ- 
ent types of instrument are accessible to study. Just as the 
simpler work on the study of the physical instruments serves 
as a helpful methodological guide in the experimental deter- 
mination of the characteristics of response of the eye, so may 
we get methodological helps from the study of these instru- 
ments which may be of service in forming our conceptions of 
the actions and functional relations of the cerebro-retinal 
structure. 
One of the characteristics of the instruments which respond 
to radiant energy is with the exception of the photographic 
plate a surface or layer in which the energy of the light-wave is 
transformed into an effect which it is the function of another 
part of the apparatus to record. This transformation, more- 
over, in case of some of these instruments: the selenium cell, 
the photoelectric cell and the photographic plate, is selective; 
1 It is possible that an alternative explanation of the above point may be found 
in some of the other types of modification of existing theories, e. g., Fick's and Leber's 
modification of the Helmholtz theory to explain the variations in the color sensitivity 
of the peripheral retina (see Fick, A., 'Arbeiten aus dem physiol. Laborat. der Wiirz- 
burger Hochschule,' pp. 213-217; Leber, T., Klin. Monatsblatter f. Augenheilk., 1873, 
II, pp. 467-473). The conception of different functional levels, however, should 
at least be recognized as one of the possibilities of explanation for monochromatic 
deficiencies, and as a very plausible and perhaps necessary assumption to explain 
phenomena of the kind described by Schumann and by the present writers in this 
paper. AREAS OF COLOR BLINDNESS 
303 
i. e., it is different in amount for the different wave-lengths1 
and the selectiveness varies with the intensity of light used. 
There is an analogue to this in the selectiveness of the achro- 
matic response of the eye to wave-length and the variation 
of this selectiveness with change of intensity of light. (In 
case of the achromatic response this is known as the Purkinje 
phenomenon.) Some of these instruments also show like the 
eye a lag in coming to their maximum of response, a fatigue 
effect, and an after-effect. All of these effects are characteristic 
of the receiving part of the instrument, not of the recording 
mechanism. The final form into which the response of the 
instrument is put is, however, a function of the recording 
part of the apparatus. Moreover, either one of these parts 
of the apparatus may, we scarcely need point out, be separately 
deficient without the impairment of the other. In view of 
the general similarity in characteristics of response shown 
between the eye and these instruments, it would not seem 
entirely unreasonable to suppose,2 therefore, even in advance 
of a decisive amount of the evidence which seems to be ac- 
cumulating, that the visual apparatus consists of receiving 
and recording parts or levels both of which are necessary to 
the final response, but either of which may be separately 
deficient without impairment of the function of the other. 
1 Griffith, I. {Phil. Mag., 1907, 14, (6), p. 297) working with ultra-violet radiations 
and Dember, H. {Ber. d. kgl. sachs. Akad. d. Wiss., 1912, 64, p. 266) working with the 
visible spectrum both claim that the photoelectric cell is selective in its response to 
intensity of light. Elster and Geitel {Phys. Z., 1913, 14, p. 741; 1914, IS, p. 610) 
however, found a constant relation between intensity of light and the response of the 
cell except in case of very intense light. For a fuller discussion of this point see 
Ferree and Rand, Psychological Review Monographs, 1917, xxxir, (5), p. 46. 
* It will be understood that an analogy between the eye and the physical recording 
instruments is attempted here primarily as illustrative rather than as argumentative. THE POWER OF THE EYE TO SUSTAIN CLEAR AND COMFORTABLE 
SEEING WITH DIFFERENT ILLUMINANTS. 
By C. E. Ferree, Ph. D., and G. Rand, Ph. D. 
BRYN MAWR COLLEGE. 
In previous papers1, a study has been made of the effect on the eye of differences 
in the way in which light is delivered to it from a given type of illuminant. In 
the work of the present paper a series of tests is begun on the effect of the illuminant 
itself. Eleven of the more common illuminants have been tested with the same condi- 
tions of installation, shading, etc., and a correlation has been made between the lighting 
effects obtained and the power to sustain clear and comfortable seeing. 
popular and semi-technical publications 
on the effect on the eye of the color 
value of light, of which subject we do 
not wish to make a special point prior 
to experimentation, these are only a 
few of the more familiar statements of 
opinion that may be cited in evidence 
that there is a need for testing the ef- 
fect on the eye of the light of the older 
illuminants (more especially the kero- 
sene flame) as compared with the more 
modern illuminants with the intensity, 
conditions of shading, installation and 
use, etc., the same in each case. 
Two divisions may be made of this 
comparison: (a) with the illuminants 
compared used for the purpose of gen- 
eral illumination, and (b) with these 
illuminants adapted to local, reading 
table or desk lighting. In the first of 
these cases differences in result would 
perhaps be more apt to occur, because 
of the greater number and complexity 
of the factors present and the greater 
difference in difficulty in protecting the 
eye from unfavorable conditions relat- 
ing to' a set of factors which we have 
hitherto called the distribution factors. 
It is quite probable also that a com- 
parative rating of illuminants made on 
the basis of local lighting, in which 
case it is not difficult, for example, to 
eliminate high brilliancies from the 
field of view, will not hold for general 
INTRODUCTION. 
The belief seems to prevail among 
laymen and not a few technical and 
medical men that the kerosene flame as 
a source of light possesses, advantages 
for the eye not to be had by other 
illuminants, more particularly the in- 
candescent solids. At one of the earlier 
meetings of the American Medical As- 
sociation's Sub-committee on the Hy- 
giene of the Eye, the belief was ex- 
pressed and quite favorably received 
that of all of the common illuminants 
the kerosene flame gives the best light 
for the eye and that it should be taken 
as our model for hygienic lighting. 
Ap eminent ophthalmologist writes3: "It 
has been shown by experiment that the 
light which gives the maximum of illu- 
mination with the minimum of irrita- 
tion to the eye is composed of the yel- 
lowish rays of the middle of the spec- 
trum. For this reason the old fash- 
ioned candle and kerosene lamp have 
never gone entirely out of fashion." In 
a more recent article4 in the same 
journal we find a section on "Simulating 
Old Illuminants," and in the last paper 
read before the Philadelphia section of 
the Illuminating Engineering Society a 
growing sentiment for the older illumi- 
nants was noted.5 
Leaving out of consideration the 
many things that have been said in 
1 lighting, in which case the chief diffi- 
culty seems to be to protect the eye 
from high brilliancies. 
Because, however, of the greater dif- 
ficulty in getting comparable installa- 
tions for general lighting, we have 
chosen to make the first series of tests 
with local lighting given by a single 
unit, a one-burner student lamp of the 
standard type with modifications suit- 
able for the different illuminants em- 
ployed. We have been led to choose 
this particular type of unit in part be- 
cause the belief in the superiority of 
the kerosene flame for the eye is in the 
minds of those we have questioned 
associated largely with the lighting ef- 
fects given by the student lamp; and in 
part because this lamp is well adapted 
for the control of conditions under 
which we wish the first series of tests 
tO' be made. 
CONDITIONS TESTED. 
Two series of experiments were con- 
ducted. In the first series the illumi- 
nants tested were a kerosene flame; a 
50-watt, clear, metallized filament 
(Gem) carbon lamp; a 15-watt, clear, 
"mazda, type B" tungsten lamp (round 
bulb) ; a 60-watt, clear, "mazda, type 
B" lamp; a 75-watt "mazda, type C" 
lamp; and a 75-watt "mazda, type C-2" 
lamp.* The kerosene flame (Luster- 
lite kerosene) was burned at a height 
of 3 inches and had a horizontal candle- 
power of 15.8. For the sake of com- 
parison with the kerosene flame it 
might have been desirable to have con- 
ducted the tests with the other illumi- 
nants equal to it photometrically, or ap- 
proximately so, as well as with an 
equally illuminated reading page and 
test object. This was, of course, im- 
practicable in case of the "mazda, type 
C" lamps. For this reason two "mazda, 
type B" lamps were used,-one as 
nearly as possible equal in candlepower 
to the kerosene flame, the other to the 
two "type C" lamps. 
In choosing the sources, care was 
taken also to have them all as nearly 
as possible of the same size or to have 
a check condition on this factor analo- 
gous to that described above; and to 
adjust the position of the lamp so as to 
sustain approximately the same rela- 
tion to the shade. The bottom of the 
shade was, for example, in each case 
2.5 cm. below the center of the lumi- 
nous source. 
The lamp was placed behind an d to 
the left of the observer in the position 
that was judged by several observers to 
give the most favorable conditions for 
reading. This position may be roughly 
specified as follows: The angle with 
the median plane of the observer made 
by a plane passing vertically thru the 
center of the unit was approximately 21 
degrees; and the line in the latter plane 
connecting the bottom of the shade 
with the center of the reading page 
formed an angle of approximately 38.5 
degrees, with the horizontal plane pass- 
ing through the center of the reading 
page. The reading page was supported 
by a rack fastened to the upright to 
which was attached the mouth-board 
used by the observer in taking the 3- 
min. record before and after work. 
This rack was inclined at an angle of 
approximately 30 degrees with the ver- 
tical. To insure that the same amount 
of light fell on the reading page in each 
case, the brightness of the page was 
measured before and after work by 
means of a Sharp-Millar illuminometer, 
with the test plate removed and cali- 
brated to give readings directly in 
candlepower per square inch. 
The changes needed to give equaj.-c^ 
of illumination on the reading page 
were made by changing the distance 
of the lamp from the page. These 
changes in case of the first three illu- 
minants were very small. For the re- 
mainder, owing to the greater differ- 
ence in the candlepower of the lights 
used, the equalization required that a 
greater difference in the distance of the 
lamp from the reading page be em- 
ployed. This meant a slightly greater 
difference in the amount of general 
illumination given and a slightly 
*Trade definitions: Gas-filled, daylight (blue) glass incandescent lamp-Mazda C-2. Gas- 
filled, clear glass incandescent lamp-Mazda C. Vacuum, clear glass incandescent lamp- 
Mazda B. 
2 greater difference in the brightness of 
the surroundings. That is, the lamps 
of higher candlepower, placed at a 
greater distance from the reading page, 
illuminated a larger field than the 
lamps of lower candlepower. In mak- 
ing these changes of distance care was 
taken to keep the angle at which the 
ligt^ fell on the page in all cases the 
same. Some difficulty was given also 
"mazda, type B" lamp and an Ivanhoe- 
Regent steel reflector of the intensive 
type, aluminum lined, were used, 
placed in front and to one side of the 
test object at the distance and angle 
needed to give the required illumina- 
tion. In order that the test object alone 
should be illuminated and not the sur- 
rounding wall, objects, etc., the open- 
ing of the reflector was covered, and an 
Test Room showing position of source of light (student lamp) reader and page, in measuring the 
power of the eye to sustain clear and comfortable vision. 
FIG. 1. 
by the difference in the length of the 
lamps employed. For example, the 
long stems of the "type C" lamps made 
it necessary that the shade be raised if 
the filaments were to have approxi- 
mately the same position in the shade 
as were had by the kerosene flame and 
the filaments of the shorter lamps. To 
take care of the needed adjustment in 
the height of the shade an extension 
shade holder was used. 
Owing to the angle of direction of 
the light and the distance of the lamp, 
the test object had to be illuminated 
from a separate source. For this a 
oblong aperture was cut of the size and 
shape needed to give the cross section 
of light desired. The position of this 
aperture in the opening of the reflector 
was chosen with reference to giving 
the most even illumination of the test 
object. That is, the light was not taken 
directly from the lamp but from the 
most favorable part of the inner sur- 
face of the reflector. The test object 
was made to match the reading page 
both in brightness and color value. 
The match in color value was secured 
by means of thin gelatin filters cover- 
ing all or a part of the aperture. If only 
3 a part of the aperture was covered, the 
filter was used as a diaphragm with an 
opening similar in shape to the original 
aperture. There was, for example, 
enough difference in the color value of 
the illuminants that without this 
match a colored after-effect was given, 
distinctly different from the reading 
page. This would have necessitated 
that the final 3-min. record be taken in 
part at least with a test object having a 
coloration complementary to the read- 
ing page, which would not have been 
compatible with the purpose of the 
test. Before beginning each test of the 
series, the eye was allowed the cus- 
tomary adaptation period without work 
under the illumination to be tested. 
The choice of the length of adaptation 
period was empirical, based on a series 
of acuity tests, the object being to' de- 
termine a period the prolongation of 
which gave no further change in acuity. 
In the first series of tests with the? 
illuminants mentioned above, the ordi- 
nary green shade of the student lamp 
was used. However, as the work 
progressed, the results seemed to indi- 
cate more and more clearly that differ- 
ence in color value must be added to 
the list of factors which are considered 
to affect the power of the eye to sustain 
clear seeing for a period of work. In 
fact, as the tests were conducted, color 
value was the only variable of any mag- 
nitude present from series to series. In 
any event it was considered advisable to 
repeat the tests with the color value 
proper to the illuminant, unmodified by 
the light which filtered thru the shade, 
even tho the position of the lamp was 
such that a very small part of the light 
which fell on the reading page was of 
this origin. From this time on, there- 
fore, an opaque shade of the same size 
and design and with a neutral lining 
was substituted for the green shade. 
The results for the neutral shade only 
will be given in this paper, altho no sig- 
nificant difference in result between the 
green and the neutral shade was found. 
The reading page illuminated by the 
diffe ent lights had the following color 
values: the "mazda, type B" lamp, an 
unsaturated reddish yellow; the kero- 
sene flame, reddish yellow with a 
greater proportion of red and more sat- 
urated; the carbon lamp, reddish yel- 
low with less red than the kerosene 
flame and more than the "type B" 
lamp; the "type C" lamp, unsaturated 
yellow, nearly white; and the "type 
C-2" lamp, noticeably bluish. These 
estimates of color value are base4 in 
part on a direct .comparison, in pirt 
on the filters that had to be used, to 
make the color match between the test 
object illuminated by the "type B" 
mazda lamp and the reading page 
lighted by the illuminant to be tested. 
We have not as yet made a standard 
colorimetric or spectrophotometric com- 
parison. 
The tests were conducted in a room 
16 ft. 6 in. (5.03 m.) long, 11 ft. 9 in. 
(3.58 m.) wide and 9 ft. 6 in. (2.98 m.) 
high. A photograph of the room with 
an observer, lamp and recording appa- 
ratus in position is shown in Fig. 1. 
The recording apparatus and the fix- 
tures for lighting the test object are, it 
will be noted, screened from the ob- 
server's view. 
In the selection and use of observers 
for all of our work care has been exer- 
cised in the first place to choose only 
those who had already shown a satis- 
factory degree of precision in other 
work in physiologic optics and whose 
clinic record showed no uncorrected 
eye defects of consequence. All have 
been under 30 years of age. Before be- 
ing allowed to take part in the actual 
work of testing, each observer was 
trained to a satisfactory degree of " > 
cision in the 3-min. records unde.* a 
given lighting condition and in the 3- 
hour test under several conditions. In 
the actual work of testing, the results 
were compiled from a number of obser- 
vations and the precision was checked 
up by the size of the mean variation. 
No results were accepted as significant 
unless the variation produced by 
changing the condition to be tested was 
largely in excess of the mean error or 
mean variation for each condition 
tested. This, the accepted check on the 
influence of variable extraneous factors 
in work of this kind, was carefully ap- 
plied at each step in the work. A fuller 
statement of the precautions that have 
4 Type of Illuminant 
Dominant Color 
Brightness 
<L> 
g 
H 
Working distance 
(cm.) 
Total time clear 
(sec.) 
Total time blurred 
(sec.) 
Total time clear 4- 
total time blurred 
Ratios reduced to 
common standard 
* 
<D 
'6 
£ 
<D 
O 
tn 
cn 
O 
pressed in percent- 
age change of ratio 
Mean 
variation 
(per 
cent.) 
Test object P 
sq. in.) 
v 
bfi 
01 
a 
bo 
c 
It 
<L> 
Based on 3.5 
Based on re 
suit sough 
(drop in 
ratio) 
Mazda lamp, Type C 
. Unsaturated yellow, 
nearly white. 
0.003168 
0.003344 
9 A. M. 
12 M. 
60 
60 
144.27 
142.67 
35.73 
37.33 
4.038 
3.822 
3.50 
3.313 
5 
34 
0 
19 
3 
58 
Mazda lamp, Type B, 60 W... 
. Unsaturated yellow, 
slightly reddish. 
0.003168 
0.003344 
9 A.M. 
12 M. 
60 
60 
139.0 
136.7 
41.0 
43.3 
3.39 
3.157 
3.50 
3.26 
6 
86 
6 
43 
6 
25 
Mazda lamp, Type B, 15 W... 
. Unsaturated yellow, 
slightly reddish. 
0.003168 
0.003344 
9 A. M. 
12 M. 
60 
60 
138.6 
136.2 
41.4 
43.8 
3.348 
3.11 
3.50 
3.251 
7 
ii 
6 
503 
7 
07 
Carbon lamp (metallized fila- 
ment) . 
Reddish-yellow .... 
0.003168 
0.003344 
9 A. M. 
12 M. 
60 
60 
142.5 
140.0 
37.5 
40.0 
3.80 
3.50 
3.50 
3.224 
7 
89 
6 
371 
4 
7i 
Kerosene flame 
. Orange-Yellow .... 
0.003168 
0.003344 
9 A. M. 
12 M. 
60 
60 
139.17 
136.33 
40.83 
43.67 
3.408 
3.122 
3.50 
3.2063 
8 
39 
6 
323 
3 
84 
Mazda lamp, Type C-2 
• Unsaturated blue .. 
0.003168 
0.003344 
9 A. M. 
12 M. 
60 
60 
138.75 
134.12 
41.25 
45.88 
3.364 
2.923 
3.50 
3.04 
13 
14 
6 
60 
4 
57 
TABLE II. 
Showing a comparison of the tendency of the different illuminants used to cause loss of visual efficiency and to produce ocular discom- 
fort. The tendency to produce discomfort is estimated by the time required for just noticeable discomfort to be set up. 
£ M 
EE 
(D 
> tn 
an 
nS C 
cu 
<D 
ft E 
o 
tn 
tn 
O 
c 
.2 
cd 
rQ W 
o.c ~ 
C 
.2 
cd 
T! O O 
O CX 
-d ft 
o t 
Type of Illuminant 
Dominant Color 
.. 
cd 
, bo 
Etc 
cd 
o 
JS . M 
5 & 
° 5 
a c 
£.2'5 
C cd 
C C 
i> 
bo 2 0 
cd 
2 o U 
T tn CX 
b o 
<D O 
E 
<u Q 
Pm 
H 
s 
u 
Mazda lamp, Type C 
.Unsaturated yellow, nearly white 
0.003344 
5.34 • 
0.19 
116.5 
1.30 
Mazda lamp, Type B, 60 W 
.Unsaturated yellow, slightly reddish.. 
0.003344 
6.86 
0.43 
98.5 
1.53 
15.45 
Mazda lamp, Type B, 15 W 
.Unsaturated yellow, slightly reddish.. 
0.003344 
7.11 
0.503 
98.5 
0.51 
0 
Carbon lamp (metallized filament) 
. Reddish-yellow 
0.003344 
7.89 
0.371 
93.0 
0.72 
5.58 
Kerosene flame 
. Orange-yellow 
0.003344 
8.39 
0.323 
90.0 
1.11 
3.23 
Mazda lamp, Type C-2 
.Unsaturated blue 
0.003344 
13.14 
0.60 
54.5 
3.70 
39.44 
TABLE I. 
Showing the tendency of the different illuminants used to cause loss of visual efficiency, or power to sustain clear seeing. 
5 been used in this and previous work to 
secure reproducibility of results has 
been given in various places in preced- 
ing papers.6 
The results for the effect on the eye 
are given in Table I. The values 
given in this table are averaged in each 
case from the results of a number of 
three hour tests. In order to show the 
reproducibility of the results obtained 
and to determine whether the varia- 
tions produced by the changes in light- 
ing effects are safely in excess of the 
variations in the test itself, subject to 
all of the variable factors which may 
influence it, the mean variation from 
the average result has been computed 
in each case. The value of these in per 
cent is given in columns 12 and 13 in 
Table I. This value has been estimated 
in two ways. In column 13 it is based 
on the result sought, namely, the mean 
value of the drop in ratio of time seen 
clear to time seen blurred. Computed 
in this way the results indicate whether 
or not each individual determination 
has been made with an acceptable de- 
gree of precision as compared with 
other work of its class. In column 12 
it is based on 3.50, the value of the 
ratio time clear to time blurred, which 
has been chosen empirically as the 
standard of performance of the eye in 
the 3-min. record before work. Com- 
puted in this way, the results appear ini 
a form from which it can readily be de- 
termined whether or not the work has 
been done with a degree of precision 
which is acceptable for the compara- 
tive work which is the special purpose 
of these experiments. That is, to be 
acceptable in this regard, the variations 
of the drop in ratio caused by changing 
the conditions to be tested, must in 
each case be safely in excess of the 
mean variation. To make this compari- 
son convenient, the drop in ratio and 
the mean variation have both been es- 
timated in per cent on the same baye, 
3.50. 
In Fig. 2 a graphic representation is 
made of the results of Table I. In con- 
structing this chart the total length of 
the test period is plotted along the 
abscissa and the ratio of the time the 
test object is seen clear to the time it 
is seen blurred is plotted along the ordi- 
nate. Each numbered division shown 
along the abscissa represents one hour 
of the test period; and along the ordi- 
nate, an integer of the ratio. 
In former papers another method of 
evaluating the results of the test was 
employed in addition to the one used 
above. In this method the ratio of the 
time seen clear to the total time of the 
observation is taken as the measure of 
the eye's ability to sustain clear seeing 
at the time the test was taken. For the 
sake of again comparing this method of 
evaluation with the one used above, a 
chart has been prepared (omitted from 
this paper because of lack of space) in 
.which the ratio of time clear to total 
time of observation is plotted against 
(length of test period. A comparison of 
this chart with that given in Fig. 2 
shows the same order of rating of the 
illuminants, but a slight difference in 
the position in the scale given to soi 
of them. For the purpose of discover- 
ing what is the best way of treating the 
(1) The data given in this paper were obtained from the observer whose results have 
been given in the preceding papers on the effect of different conditions of lighting on the 
eye. In case of the present paper we have not as yet, for lack of time, been able to check 
up these results with those obtained from other trained observers. We have, however, 
in the work on the distribution factors always found the results of this observer to be 
typical of the group of observers used. Whether or not this will be the case for work 
in which the distribution factors are not the sole or principal variable, remains yet to be 
determined. In this regard it is perhaps only fair to say that the characteristics of re- 
sponse of the eyes for which these results are given, have been very widely investigated. 
They have been chosen especially for their normality and practiced precision of beha- 
vior, and have been used in these experiments under conditions of control based on a very 
unusual and widely tested knowledge of the factors which influence their steadiness of 
response. Data on their characteristics of response may be found in more than forty ar- 
ticles. Their spectral luminosity curve, for example, agrees very closely with the aver- 
age curve obtained by Nutting for 21 observers.5 
The results of the above tests are now being checked up on other trained observers. 
6 results of the tests, several methods 
have been employed. Up to and in- 
cluding the present paper, however, 
only three of them have been given in 
print: ratio of time clear to time 
blurred, ratio of time clear to total time 
of observation, and the per cent drop 
in the ratio time clear to time blurred.1 
w . ultimate decision with regard to 
what is the best method of treating the 
results has not yet been reached and 
determinations were made and a dis- 
cussion of the method that was used 
has been given in a previous paper. 
The results are shown in Table II. In 
this table are given, also, for the sake 
of comparison, results expressing the 
tendency of each type of illuminant to 
cause loss of ability to sustain clear 
seeing. 
The results of this work, so far as it 
has been carried, more particularly 
Fig. 2. 
Graphic representation of results of Table 1, each numbered division at the bottom represents one 
of the test period. The lines start from the assumed base of 3.5, showing loss in ratio of time clear 
to time blurred. 
for the purposes of this work is perhaps 
not necessary. From the data given 
any one of them may be used. 
As formerly, the work was con- 
cluded by determining for the different 
illuminants used the relative tendencies 
to produce ocular discomfort with the 
eye at work. A description of how the 
those to be presented in a later paper 
which cover a range of color values 
greater in amount and apparently more 
significant in direction, seem to indicate 
that the tests for the effect of color 
value of light on the power of the eye 
to sustain clear and comfortable seeing 
should be carried further. In the work 
(1) A comparison of this chart with those of the preceding papers shows that the or- 
der of magnitude of loss in power to sustain clear seeing for the kerosene flame and 
the Type B mazda lamp (student lamp unit), was about the same as for the best of the 
opaque inverted reflectors (Type B mazda lamp); and for the Type C-2 lamp as for the 
best of the translucent inverted reflectors. The effect for the Type C mazda lamp was 
slightly better than for the best of the opaque inverted reflectors. 
7 so far we have found that in case of a 
given color this power decreases with 
increase of saturation of color; but that 
independent of saturation some colors 
affect the eye more than others. The 
worst effects thus far have been ob- 
tained with colors towards the short 
wave-length end of the spectrum. The 
reading of black letters or other char- 
acters on a page which presents any 
considerable degree of coloration is a 
peculiarly baffling experience. There 
is an unclearness which is not the blur- 
ring of bad focusing or of faulty fix- 
ation, but which seems to be a matter 
of the ease or, rather, lack of ease, with 
which the details of the retinal picture 
are discriminated. Unclearness or diffi- 
culty of discrimination from any cause 
whatsoever leads reflexly to muscular 
effort towards a corrective readjust- 
ment which of course in the cases un- 
der consideration comes to naught ?» 
only induces fatigue. The effect of 
color value of light on the power of the 
eye to hold itself up to a satisfactory 
standard of performance thru a period 
of work should, we believe, receive 
attention. 
BIBLIOGRAPHY. 
1. Ophthalmology, v. x, p. 622; v. 12, p. 593; Annals of Ophthalmology, v. 25, p. 447, etc. 
2. Trans. Illuminating Engineering Society, March, 1913, p. 132. 
3. Trans. Illuminating Engineering Society, 1915, pp. 1027-1033. 
4. See also Electrical Rev. and W. E., July 24, 1915, p. 161. 
5. Philos. Mag., v. 29, (6), p. 
6. Trans. Illuminating Engineering Society, 1915, pp. 1122-1130; etc. 
8 SECTION XI MISSING [Reprinted from the Psychological Bulletin, December, 1918, Vol. 15, No, 12.i 
We wish to make the following corrections of Dr. Troland's 
review of work published by us during the year 1917-1918. 
1. He says of our work on "The Power of the Eye to Sustain 
Clear Seeing under Different Conditions of Lighting": "Semi- 
indirect reflectors of high density seem to be most conducive to eye 
comfort." While this conclusion would perhaps be more agreeable 
and satisfactory to certain commercial and professional factions in 
lighting circles, it was not drawn by us nor can it be drawn from our 
results. Of the commercial reflectors tested by us thus far, un- 
modified by any experimental device for the improvement of their 
effect on the eye (cf. Opaque Direct Reflectors, Trans, of Ilium. 
Eng. Soc., 1917, 12, pp. 466-468), the best results have without 
question been obtained with the totally indirect reflectors. 
2. Of our article, "Some Areas of Color Blindness of an Unusual 
Type in the Peripheral Retina," he says: "Ferree and Rand report 
observations which show that areas can be found in the peripheral 
visual fields of many persons which are relatively blind to red, 
green, yellow or blue but which are not correspondingly deficient 
in the complementary after-image and other related reactions." 
There is no ground for using the term "relatively blind" here. 
It was shown in the article reviewed that stimuli so intense as to 
carry the sensitivity to red, yellow and blue out to the limits of 
white light vision were not sensed as color in these areas. Further 
in order to leave no doubt as to their color blindness the series was 
made to include also stimuli so intense as to give colors greatly 
reduced in saturation when viewed in central vision. If areas so 
tested are to be called "relatively blind" it is difficult to under- 
stand why the term color blind should ever be used. What we 
actually reported was that both types of areas, color blind and 
color deficient, showing no detectable loss in the cancelling and 
after-image functions were to be found in the peripheral retina. 
3. Our discussion of the "Needs and Uses of Energy Measure- 
ments in Psychological Optics" is represented in a way which we 
do not care to have stand uncorrected in the year's reviews of 
work. Our argument was that if we are to determine the sensi- 
tivity of the eye in a way that is comparable with the determination 
of the sensitivity of the physical recording instruments we should 
be able to compare numerically both our amounts of response and 
A NOTE ON VISION-GENERAL PHENOMENA 452 
DISCUSSION 
amounts of stimulus. The question whether st muli should be 
equated subjectively or in energy terms for investigations bearing 
on various points of theory or doctrinal conception was quite aside 
from the main purpose of the paper. Which type of equation 
should be used depends on the nature of the problem in hand, as 
was made clear as a feature of minor importance in the article 
reviewed. The need of a method logically sure for the determina- 
tion of sensitivity was the particular point of emphasis and it was 
with reference to this point especially, which we believe is at present 
the most important in the laying of the groundwork of a more 
scientific psychophysics of vision, that our recommendation of 
energy measurements was made. In this connection it may be 
noted further that the article reviewed is in part a criticism of the 
reviewer's own advice that the eye may be used as a substitute for 
the nonselective instruments in the measurement of light energy; 
and that our criticism of this recommendation was not by any 
means based alone on the inadequacy of the "extant visibility 
data." These points are discussed in greater detail in two articles 
(Titchener Commemorative Volume, pp. 230-308 and Psychological 
Monographs, No. 103) which were not reviewed by Dr. Troland. 
C. E. Ferree, 
Gertrude Rand 
Bryn Mawr College CHROMATIC THRESHOLDS OF SENSATION FROM 
CENTER TO PERIPHERY OF THE RETINA AND 
THEIR BEARING ON COLOR THEORY [Reprinted from The Psychological Review, Vol. XXVI., No, 1, Jan. 1919.] 
CHROMATIC THRESHOLDS OF SENSATION FROM 
CENTER TO PERIPHERY OF THE RETINA AND 
THEIR BEARING ON COLOR THEORY 
PART I. 
BY C. E. FERREE AND GERTRUDE RAND 
Bryn Matvr College 
Introduction 
In the work reported in this paper a determination of the 
chromatic thresholds of red, green, blue and yellow in energy 
terms has been made at near-lying points from the center to 
the periphery of the retina. The incentive for making this 
study has been twofold: (i) We have wanted to make an 
investigation of the chromatic sensitivities of the central and 
peripheral retina that would be more nearly quantitative 
than those that have previously been attempted. (2) A de- 
tailed investigation of the sensitivity gradient for the four 
colors red, green, blue, and yellow from center to periphery 
of the retina has an important bearing on certain points of 
color theory. Two of these points will be considered in the 
second part of this paper. 
As a part of a general investigation of retinal sensitivities, 
we had planned several years ago (1) to make determinations 
both of the achromatic and chromatic sensitivities to wave- 
length that would be quantitative up to the standard ac- 
cepted for the physical recording instruments. One of the 
requirements for such a determination is, as was pointed out 
at that time, that the stimuli shall be rated in units that can 
be compared. This requirement could not be met until 
measuring instruments were obtained which were sufficiently 
senstitive for work in the visible spectrum and which were 
non-selective in their response to wave-length. It was fur- 
ther stated that if a rating is to be made which may fairly be 
considered as quantitative, it must also be possible from the 
16 CHROMATIC THRESHOLDS OF SENSATION 
17 
data at hand to compare numerically the amounts of response 
as well as the amounts of stimuli used to arouse the response. 
That is, while a radiometric rating of the stimuli is neces- 
sary for this purpose, an equally important point to be con- 
sidered is what amounts of response can be employed with 
sureness of principle in meeting the quantitative require- 
ment. In this regard it was pointed out that of the amounts 
of response that have at different times been used or sug- 
gested for the determination of sensitivity-namely, equal 
amounts, equal sense differences, the liminal threshold, the 
just noticeable difference, and the average error-perhaps 
only the first two can by common agreement be regarded as 
numerically comparable; and that the validity of the use of 
the others for the more strictly quantitative work should be 
tested by checking the results against those obtained when 
the rating is based, for example, on equal amounts taken as 
standard. This interchecking of results for achromatic sen- 
sitivity is now in progress in our laboratory, in which case 
all of the determinations mentioned above may be made. It 
can not be done, however, in case of chromatic sensitivity 
until it is first determined whether the judgment of equal 
saturations can be made with an acceptable degree of pre- 
cision. Furthermore, a determination of comparative sen- 
sitivities in the peripheral retina, based on equal amounts of 
response, while not impossible, is neither very convenient nor 
very feasible. For the purpose of the present paper, there- 
fore, we have been content to deal with the determination 
of the chromatic threshold for the wave-lengths in question 
and of their variation from the center to the periphery of the 
retina, which work has an interest of its own independent of 
its bearing on a determination of comparative sensitivities, 
and to reserve a report on the more strictly quantitative fea- 
tures of the general problem for a later paper. In making 
these determinations it was our intention to work at near- 
lying points from the center to the periphery of the retina in 
several meridians. The work was interrupted, however, by 
the pressure of other investigations when the determinations 
had been made in only two meridians, the temporal and the 18 
C. E. FERREE AND GERTRUDE RAND 
nasal. Results can be given at this time, therefore, for only 
these two meridians. 
Conditions under Which the Work was Done 
The determinations were made under the following con- 
ditions. (1) The colored lights used were taken from the 
spectrum. There are two reasons for this in an investiga- 
tion of the kind here undertaken. (<2) The stimuli should be 
as homogeneous with regard to the visible wave-lengths as 
possible,1 and (&) they should be free from the infra-red and 
ultra-violet radiations which would affect the thermopile 
used to measure the intensity of light, but not the eye. The 
stimuli employed were a narrow band of red in the region of 
670 w, of yellow in the region of 581 w, of green in the region 
of 522 w; and of blue in the region of 468 /z/z. The breadth 
of analyzing slit used in isolating these bands was maintained 
constant at 0.5 mm. The range of wave-lengths obtained was 
approximately 660-680^; 575-587 518-526 /z/z; and 468 
-474 /z/z. The spectrum was gotten and the different wave- 
lengths were presented to the eye by means of the apparatus 
described in the Journal of Experimental Psychology, 1916, 
1, pp. 247-284: ' A Spectroscopic Apparatus for the Investi- 
gation of the Color Sensitivity of the Retina, Central and 
Peripheral.' In every case the light was examined for im- 
purities at the analyzing slit by means of a small Hilger direct 
vision spectroscope provided with an illuminated scale. 
When found, impurities were absorbed out by thin gelatines 
selected so as to cut out as little of the useful light as pos- 
1 The presence of the alien visible wave-lengths affects the results of a determina- 
tion of chromatic sensitivity in two ways: (a) through physiological inhibitions and 
interactions it decreases the amount of the color response, and (Z>) it increases the energy 
measurement. In their work on the determination of the visibility of radiation in the 
red end of the visible spectrum, Hyde and Forsythe {Astrophysical Journal, 1915, 
44, p. 289) found impurities in the prismatic spectrum to the value of about 20 per 
cent., at 0.76 n. In our own work both on achromatic and chromatic sensitivity 
determinations made with and without provisions for absorbing the scattered light, 
show differences in result which are great enough to be considered of significance. 
This is true in particular for determinations of chromatic sensitivity, in which case the 
chromatic response may be reduced quite appreciably as a result of the physiological 
interactions produced by the alien wave-lengths. In some cases, for example, even 
the complementary wave-lengths may be present. CHROMATIC THRESHOLDS OF SENSATION 
19 
sible. These gelatines were placed over the analyzing slit 
and were held in position by short clips fastened to the front 
surface of the jaws the edges of which formed the slit. 
(2) The determinations of the threshold were made in 
energy terms. Measurements were made at two places: at 
the analyzing slit and at the eye. In making the threshold 
determinations of the stimulus light it was found to be con- 
venient first to make the colors all equal in energy value. 
The reductions needed for the equalization were made by 
appropriate adjustments of the collimator slit. Since the 
blue represents the smallest amount of energy of any of the 
colors employed, they were all made equal in energy to the 
blue of the spectrum used, namely, the prismatic spectrum 
of a Nernst filament operated by 0.6 ampere of current. 
From this intensity they were reduced to the threshold by 
means of the especially constructed sectored discs described 
in an earlier paper,1 and the energy values computed from 
the simple law of the disc. These discs, it will be remem- 
bered, were cut from hard sheet aluminum, No. 20 B. and 
S. gauge, 0.9 mm. thick, and of two sizes for just noticeable 
difference determinations, 19.5 and 17 cm. in radius. The 
total variation of range of open aperture is from o° to 
348.75°. A strong objection to the use of sectored discs 
when fine changes are needed such as are required, for exam- 
ple, in threshold and just noticeable difference work, is the 
difficulty of obtaining and measuring accurately sufficiently 
small amounts of change. Such discs are ordinarily con- 
structed with two or more open sectors and a change in one 
is multiplied as many times as there are open sectors. More- 
over, an error made in the measurement of one sector is 
multiplied by the number of open sectors. This latter diffi- 
culty becomes especially significant in working with intens- 
ities at or near the threshold, where a small error may repre- 
sent a high percentage of the total open sector. We have 
sought to overcome these difficulties in three ways. (1) 
Our discs for a low total aperture are so constructed that one 
sector may be varied at a time. (2) The sector is moved by 
1 Journal of Experimental Psychology, 1916, 1, pp. 271-274. 20 
C. E. FERREE AND GERTRUDE RAND 
means of a micrometer screw. This device for minute changes 
in the value of the open sector is so constructed as to be readily 
attached and removed from the disc. And (3) a special pro- 
tractor has been designed fitted with a movable arm carry- 
ing a knife edge and Vernier scale graduated to read to min- 
utes. Obviously some such precise means of making and 
measuring small changes in the disc are of prime importance 
in the work of determining the threshold and just noticeable 
difference. If such means are not at hand the average error 
of setting and measurement is apt to exceed that of the sense 
judgment. 
The method of making the energy measurements by means 
of a thermopile has already been described in a previous 
paper. However, because the procedure is as yet somewhat 
unfamiliar it may not be out of place to give again brief 
description of how the measurements are made. A descrip- 
tion at one of the places at which they were made, namely 
the analyzing slit, will be sufficient to show in a general way 
the method we have employed. The thermopile to be used 
was placed in position immediately behind the slit and a 
blackened aluminum shutter was interposed in the path of 
the beam of light between the slit and the end of the objec- 
tive tube of the spectroscope. Preliminary to the exposure 
of the thermopile to the light to be measured, the current 
sensitivity of the galvanometer was tested by means of a 
special device1 provided for this purpose in the construction 
of the galvanometer. With regard to this procedure it may 
be pointed out that the current sensitivity of the galvanom- 
eter varies with the period or time of the single swing of its 
needle system. Since it is not possible to control the field so 
as to get this period always the same, it is necessary, if re- 
sults are to be compared, to take some sensitivity as standard 
and to convert all readings into deflections for the standard 
sensitivity by means of a correction factor determined at 
each sitting. For a detailed description of the method of de- 
1 This device consists of a special galvanometer coil, dry battery circuit, and switch 
board with finely graduated resistance. For a description of this device, see 'Radio- 
metric Apparatus for Use in Psychological and Physiological Optics,' Psychol. Rev. 
Monog., 1917, 24, No. 2, pp. 63-65. CHROMATIC THRESHOLDS OF SENSATION 
21 
termining this factor, see Psychol. Rev. Monog., 1917, 24, 
No. 2, pp. 60-65. 
The thermopile was next connected with the galvanom- 
eter and the light allowed to fall on its receiving surface until 
a temperature equilibrium was reached (ca. 3 sec. for our 
thermopile). The deflections were read by means of the tele- 
scope and scale and the readings are corrected to standard 
sensitivity by means of the factor previously determined. 
The final step in the process of measuring was the calibration 
of the apparatus, i. e., the value of 1 mm. of deflection in 
radiometric units was determined for the area of thermopile 
exposed. To do this a radiation standard, the value of the 
radiations from which is already known, had to be employed. 
The standard used by us was a carbon lamp specially sea- 
soned and prepared for the purpose by W. W. Coblentz (4) 
of the radiometric division of the Bureau of Standards. This 
lamp was placed on a photometer bar 2 meters from the ther- 
inopile and operated at one of the intensities for which the 
calibration was made, in our case 0.40 ampere. The ther- 
mopile was exposed to its radiations with the same area of 
receiving surface as was used in case of the lights measured, 
and the galvanometer deflection was recorded. From the 
deflections obtained the value of 1 mm. of deflection, or the 
radiation sensitivity of the apparatus under the conditions 
given, was computed from the known amount falling on the 
surface of the thermopile. Having the factor expressing the 
radiation sensitivity of the apparatus, the deflections pro- 
duced by the wave-lengths of light measured were readily 
converted into energy units. The radiation sensitivity of 
the linear thermopile used by us was computed in a given 
case, for example, from the following data. The energy 
value of the radiations per sq. mm. at a distance of 2 m. from 
the standard lamp operated by 0.40 ampere was 90.70 X io-8 
watt. The deflections of the galvanometer produced by this 
intensity of radiation falling on the same area of receiving sur- 
face as was used in measuring the lights employed as stimuli, 
when corrected («) to a sensitivity of i = 1 X io~10 ampere, 
and (b) for the absorption of the glass cover of the thermo- 22 
C. E. FERREE AND GERTRUDE RAND 
pile, was 346.870 mm. The area of the surface exposed was 
4.400 sq. mm., and the time of exposure was 3 sec. The 
sensitivity of the instrument per sq. mm. of receiving surface 
was, therefore, 115 X io-10 watt. By means of this factor 
the galvanometer readings produced by the different wave- 
lengths of light may readily be converted into the energy 
value of light falling on the receiving surface of the thermo- 
pile. 
(3) The field surrounding the stimulus, and the preexpo- 
sure were always maintained as nearly as possible at the same 
brightness as the stimulus at the threshold value of sensation. 
For want of better pigment materials these surfaces were 
made from the Hering standard gray papers. It was found 
to be necessary to change the brightness of the surrounding 
field and preexposure frequently for each stimulus because the 
brightness value of the color at the chromatic threshold 
changed quite rapidly from the center to the periphery of the 
retina. There were two causes for this change, (a) The 
intensity of the light had to be increased quite a great deal 
from center to periphery to give the chromatic threshold 
from point to point; and (b) the achromatic value of the colors 
does not remain the same from the center to the periphery 
of the retina. The gray that matched the stimulus in achro- 
matic value at each point was determined by the equality of 
brightness method.1 With reference to these determinations, 
it may be said that with the stimulus reduced to the chro- 
matic threshold the color difference between the stimulus and 
the gray was so small that the equality of brightness judg- 
ment was not difficult to make. The match was made in 
every case for the part of the retina under investigation. It 
had to be attained by a series of approximations. That is, 
the threshold was first obtained at the given point with no 
especial control of brightness of preexposure and surrounding 
1 The Hering papers did not present a sufficiently wide range of reflection coeffici- 
ents to match the brightness range of the threshold value of the stimulus light from the 
center to the periphery of the retina. For example, at the threshold of sensation at 
the center of the retina, the darkest of the Hering papers illuminated by the rather strong 
light of the room was brighter than the stimulus light; and in the far periphery of the 
retina through a zone varying in breadth from 2 to 13 degrees for the different colors, 
the lightest of this series of papers was darker than the stimulus color. CHROMATIC THRESHOLDS OF SENSATION 
23 
field. The brightness of preexposure and surrounding field 
was then made to match this value of the stimulus and the 
threshold was redetermined. This procedure was repeated 
until a threshold was obtained that required no further 
change in these surfaces to match it in brightness. In order 
to make the specification of the brightness of the preexposure 
and the surrounding field independent of the illumination 
of the room and of the variability of the reflection coef- 
ficients of different issues of the Hering papers, the brightness 
was in each case determined in candlepower per sq. in. This 
determination was made by means of a Sharp-Millar portable 
photometer with the test plate removed. The instrument 
was calibrated against a magnesium oxide surface obtained 
by depositing the oxide from the burning metal. By this 
method the reflecting surfaces were used as detached test 
plates. The readings were converted into candlepower per 
sq. in. by the following formula: Brightness = -1 
For the sake of reproducibility of conditions from time to 
time in a given laboratory or in different laboratories, it is 
obvious that a photometric specification should be given in 
all cases not only of the general illumination of the room but 
of the brightness of all surfaces for which precision of control 
is of importance to the results of the work. 
(4) The illumination of the room was kept at a constant 
value. Two features are necessary for this control, (a) A 
means must be had of detecting small changes of illumination. 
This may be accomplished by a portable photometer of the 
Sharp-Millar or Macbeth type, for example, furnished with 
a daylight screen, or of the simpler type described by the 
writers in a previous article (5). And (b} a means must be 
had also of producing small variations in the illumination of 
the room, else the changes due to fluctuations in the external 
light can not be compensated for with the precision and min- 
uteness of control that is needed. This is accomplished in 
Our optics room by two systems of thin white curtains running 
1 By multiplying these values in turn by 486.8 they may be converted into milli- 
lamberts, a term frequently used by engineers to specify small brightness quantities. 24 
C. E. FERREE AND GERTRUDE RAND 
on spring rollers beneath the skylight. One of the systems 
of white curtains and the light-proof curtain run lengthwise 
of the room; the other system of white curtains runs across 
the room. By means of the white curtains either small local 
or small general changes can be produced in the illumination 
of the room; and by means of the light-proof curtain larger 
changes may be produced ranging from full illumination to 
the darkness of a moderately good dark room. The light- 
proof curtain is of a breadth equal to that of the room and 
runs in a deep light-tight boxing. The white curtains are 
narrower and are made to overlap at the edges. These cur- 
tains run on wire guides so distributed as to prevent any 
sagging or wrinkling. Above these curtains are pivoted 
two large diffusion sashes of glass ground on one side com- 
pletely filling the skylight opening. These sashes diffuse the 
light in the room giving an even distribution of illumination 
and rendering, because of that fact, an even and precise con- 
trol easier to accomplish. In a careful specification of the 
conditions under which the work is done a very important 
item is to give a photometric specification of the illumination 
of the room. This may be done in foot or meter-candles as 
desired. If the illumination is uneven it should be done sys- 
tematically throughout the room. If, on the other hand, it 
is pretty uniform, it is usually sufficient to give its value in 
three or more directions at the point of work. In case of the 
present work, for example, the value of the horizontal com- 
ponent was 30.49 foot candles; the vertical component, 121.95 
foot candles; and the 45 degree component, 82.97 foot candles. 
(5) The amount of light entering the eye was made inde- 
pendent of variations in the size of the pupil. Independence 
of change in size of pupil was especially needed in this work 
because of the large variations in the intensity of light used. 
Such control is very easy to accomplish with the means of 
presenting the light to the eye that is used in our apparatus. 
All that is needed is to keep the image that falls on the pupil 
of a constant size and smaller than the pupil throughout its 
entire range of variations in the given series of experiments. 
Not only can this variation be determined in preliminary CHROMATIC THRESHOLDS OF SENSATION 
25 
experiments as a guide to the size of the image that is needed, 
but the image itself can be compared with the pupil at each 
observation. For details of the method of exercising this 
control see 'A Substitute for an Artificial Pupil,' Psychol. 
Rev., 1916, 23, 380-383.1 
Results 
A statement of the results of the investigation is given 
in Tables I.-VIII. The nearness of the points investigated 
to each other was determined by the rapidity with which the 
sensitivity decreased in the meridian in question. In those 
regions in which the decrease was gradual the determinations 
were made at points separated by as much as 5 degrees. In 
regions, however, where the rate of decrease was rapid or un- 
usual features were present, the determinations were made 
at points separated only by 1 degree. A graphic represen- 
tation of the results of these tables is given in Charts I.-IV. 
In these charts degree of eccentricity is plotted along the 
abscissa and the value of the threshold in watts (io7 ergs per 
sec.) is plotted along the ordinate. In Charts I. and II. the 
values of the threshold from the center of the retina to the 
limits of sensitivity are plotted. In case of the red, yellow 
and blue, it will be remembered from statements made in 
former papers that the limits of sensitivity for lights of high 
intensity coincide with the limits of the field of white light 
vision. This, however, was not the case for the green stimu- 
lus. By no increase of intensity were we able to make the 
limits of green sensitivity coincide with the limits of white 
light vision. In Charts III. and IV. the above values from 
the center of the retina through the region of gradual de- 
crease of sensitivity are plotted on a larger scale. This is 
done because when plotted on the scale used in Charts I. and 
1 Since the article referred to above was published we have devised and constructed 
a very convenient attachment for our analyzing slit by means of which the length 
of aperture of the slit may be varied by half millimeter steps. Space will not be taken 
here for a detailed description of this device. In brief, it consists of two knife edged 
jaws moving in the vertical, operated by means of a ratchet and spring. Still finer 
control could be secured, of course, by means of a micrometer screw. Constructed in 
this latter form the device would be very serviceable as a means of producing finely 
graded changes of intensity. 26 
C. E. FERREE AND GERTRUDE RAND 
IL the curves fall so closely together that the relative sensi- 
tivities to the four colors are not clearly represented. That is, 
the range of the values for the threshold from the center to 
the extreme periphery of the retina is so great that in Charts 
I. and IL, in which the entire range is represented, a scale 
value had to be chosen which is so large as almost to obscure 
the smaller differences in relative sensitivity to the different 
colors in the region of gradual decrease in sensitivity. 
Table I 
Chromatic Thresholds for Red, Nasal Meridian 
In this table are given the values of the threshold for red (670 mm) at 22 points in 
the nasal meridian. The intensity of light at the analyzing slit was 94.27X io-8 watt. 
Degree of 
Excentricity 
Surrounding 
Field and Pre- 
exposure. 
(Candle-power 
per Sq. In.) 
Value of Threshold 
Degrees 
Open Sector 
Total Amount 
of Light at 
Campimeter 
Opening and 
at Eye 
(Watt X io-12) 
Density of 
Light at 
Campimeter 
Opening 
(Watt X io~12 
per Sq. Mm.) 
Density of 
Light at Eye 
(W att X io-12 
per Sq. Mm.) 
O 
O.OOO764 
0-375 
30.80 
0.174 
9-34 
5 
O.OOO764 
0-375 
30.80 
0.174 
9-34 
IO 
O.OOO764 
0.438 
36.OO 
O.2O3 
10.90 
14-17 
Blind Spot 
20 
O.OOO764 
0-75 
61.70 
o-349 
18.68 
25 
O.OO2I25 
1.00 
82.20 
0.465 
24.90 
30 
O.OO3744 
2.50 
205.50 
1-163 
62.25 
35 
0.005088 
3-75 
308.30 
1-744 
93-38 
40 
O.OO5O88 
4.00 
328.80 
1.860 
99.60 
45 
O.OO52IO 
4-50 
369-9O 
2.093 
112.05 
50 
O.OO52IO 
5-oo 
411.00 
2.325 
124-50 
55 
O.OO59O2 
6.00 
493-20 
2.790 
149.40 
60 
O.OO6268 
8.00 
657.60 
3-720 
199.20 
65 
O.OO6838 
12.25 
1007.00 
5-696 
305-00 
70 
O.OO74O8 
20.00 
1644.00 
9-300 
498.00 
75 
O.OII4O 
23.00 
1890.60 
10.695 
572.70 
80 
O.OI262 
26.00 
2137.20 
12.090 
647.40 
82 
O.OI587 
34-oo 
2794.80 
15.810 
846.60 
85 
0.021X6 
50.00 
4110.00 
23-250 
1245.00 
87 
0.02686 
114.00 
9370.80 
53-oio 
2838.60 
88 
O.O3O93 
180.00 
14796.00 
83-700 
4482.00 
9° 
0.05088 
270.00 
22194.00 
125-550 
6723.00 
92 
O.O5O88 
338-00 
27783.60 
I57.I7O 
8416.20 
In this and the following tables the light was taken from the spectrum of a Nernst 
filament operated by 0.6 ampere of current. In all cases the preexposure and surround- 
ing field were made as nearly as possible the same brightness as the stimulus. 
With reference to the results given in Tables L-VIIL, 
the following points may be noted, (i) The characteristics 
of response of the eyes for which results are given have been CHROMATIC THRESHOLDS OF SENSATION 
27 
very widely investigated for both central and peripheral vis- 
ion. They have been chosen for this and other work espe- 
cially because of their normality and practised precision of 
behavior. Their spectrum luminosity curve, for example, 
agrees (6) very closely with the average curve obtained by 
Table II 
Chromatic Thresholds for Red, Temporal Meridian 
In this table are given the values of the threshold for red (670 up) at 25 points 
in the temporal meridian. The intensity of light at the analyzing slit was 94.27 X 
io-8 watt. 
Degree of 
Excentricity 
Surrounding 
Field and Pre- 
exposure. 
(Candle-power 
per Sq. In.) 
Value of Threshold 
Degrees 
Open Sector 
Total Amount 
of Light at 
Campimeter 
Opening and 
at Eye 
(Watt X io-12) 
Density of 
Light at 
Campimeter 
Opening 
(Watt X io-12 
per Sq. Mm.) 
Density of 
Light at Eye 
(Watt X xo-12 
per Sq. Mm.) 
0 
O.OOO764 
Q-375 
30.80 
0.174 
9-34 
5 
O.OOO764 
o-375 
30.80 
0.174 
9-34 
10 
O.OOO764 
0.438 
36.OO 
0.203 
10.90 
i5 
O.OOO764 
0-75 
61.70 
0-349 
18.68 
20 
O.OO2I25 
1.00 
82.20 
O.465 
24.90 
25 
O.OO3867 
2-75 
226.10 
1-279 
68.48 
3° 
O.OO59O2 
8.00 
657-6O 
3-720 
199.20 
33 
O.OO6431 
14.00 
1150.80 
6.510 
348.60 
35 
O.OO74O8 
17.00 
1397-4° 
7-905 
423-3O 
40 
O.OI4O4 
29.00 
2383.80 
I3-485 
722.10 
44 
O.OI79I 
38.00 
3123.60 
17.670 
946.20 
45 
0.02686 
104.00 
8548.80 
48.360 
2589.60 
46 
O.O3O93 
160.00 
13152.00 
74.400 
3984-00 
47 
O.O3663 
216.00 
17755-20 
100.440 
5378.40 
48 
O.O3663 
216.00 
17755-20 
100.440 
5378.40 
49 
O.O3663 
216.00 
17755-20 
100.440 
5378.40 
50 
O.O3663 
216.00 
17755-20 
100.440 
537840 
5i 
O.O3663 
216.00 
17755-20 
100.440 
5378.40 
52 
O.O3663 
216.00 
17755-20 
100.440 
5378.40 
53 
O.O3663 
216.00 
17755-20 
100.440 
5378.40 
54 
O.O3663 
216.00 
17755-20 
100.440 
5378.40 
55 
O.O3663 
245.00 
20139.00 
II3-925 
6100.50 
58 
O.O5O88 
270.00 
22194.00 
125-55° 
6723.00 
60 
0.05088 
300.00 
24660.00 
139-50° 
7470.00 
61 
O.O5O88 
315-0° 
25893.00 
I46-475 
7843-5O 
Nutting (7) for 18 observers. Also their normality of chro- 
matic response has, at various times, been checked up by a 
number of observers. Data on this point may be found in 
nearly all of the work that has been published from this lab- 
oratory. In case of the present work the systematic point 
by point determination has been made only upon the one Chart I. Chromatic Thresholds for the Four Colors, Nasal Meridian. In this 
chart and Chart II., degree of excentricity in the field of vision is plotted along the 
abscissa and the value of the threshold along the ordinate. Of the three values of the 
threshold given in the tables, the total amount of light at the campimeter opening and 
at the eye is represented. CHROMATIC THRESHOLDS OF SENSATION 
29 
Chart II. Chromatic Thresholds for the Four Colors, Temporal Meridian. Chart III. Chromatic Thresholds for the Four Colors, Nasal Meridian. In this 
chart and Chart IV., the values represented in Charts I. and II. respectively, from the 
center of the retina through the region of gradual decrease of sensitivity are plotted 
on a larger scale. This is done because when plotted on the scale used in Charts I. 
and IL, the curves fall so closely together that the relative sensitivities are not clearly 
represented. CHROMATIC THRESHOLDS OF SENSATION 
31 
Chart IV. Chromatic Thresholds (enlarged scale) for the Four Colors, Temporal 
Meridian. 32 
C. E. FERREE AND GERTRUDE RAND 
observer. However, one or more check observers have been 
used on points of the work of especial importance to theory, 
such as the irregular distribution of sensitivity to the pairs 
of colors, the criss-crossing or interlacing of limits, the areas 
analogous in type to the Schumann case of color-blindness, 
the deficiency of sensitivity of the far periphery of the retina 
to green, and the inability to get a red or green that is stable 
in color tone for all parts of the retina. 
Table III 
Chromatic Threshold for Yellow, Nasal Meridian 
In this table are given the values of the threshold for yellow (581 /qu) at 23 points 
in the nasal meridian. The intensity of light at the analyzing slit was 93.10 X io-8 
watt. 
Degree of 
Excentricity 
Surrounding 
Field and Pre- 
exposure 
(Candle-power 
per Sq. In.) 
Value of Threshold 
Degrees 
Open Sector 
Total Amount 
of Light at 
Campimeter 
Opening and 
at Eye 
(Watt X io-12) 
Density of 
Light at 
Campimeter 
Opening 
(Watts X 10-12 
per Sq. Mm.) 
Density of 
Light at Eye 
(Watt X 10-12 
per Sq. Mm.) 
O 
O.OO2I25 
O.2O 
16.30 
O.O92 
4-94 
S 
O.OO2I25 
O.25 
2O.4O 
0.II5 
6-175 
IO 
O.OO2I25 
0-375 
30.56 
O.172 
9-263 
14-17 
Blind Spot 
20 
O.OO3378 
0-75 
6l.I3 
0.346 
18.525 
25 
O.OO3867 
1.25 
101.88 
0-576 
30.875 
3° 
O.OO643 I 
2-375 
I93-56 
I.O94 
58-663 
35 
O.OO6919 
2.875 
234-31 
1-325 
71.013 
4° 
O.OO6956 
3-5° 
285.25 
I.614 
86.450 
45 
O.OO7326 
4-5° 
366.75 
2.074 
in.150 
5° 
O.OO74O8 
4-875 
397-31 
2.247 
120.413 
55 
O.OO74O8 
4-875 
397-31 
2.247 
120.413 
60 
O.OO7977 
5-oo 
4O7-5O 
2.3O5 
123.500 
65 
O.OO8547 
5-125 
4I7-69 
2.363 
126.588 
70 
O.OO9768 
5-75 
468.63 
2.651 
142.025 
75 
O.OII4O 
6-75 
550.13 
3-II2 
166.725 
80 
O.OI384 
8.00 
652.00 
3-688 
197.600 
85 
O.OI4O4 
10.75 
876.13 
4-956 
265.525 
87 
O.OI587 
22.00 
1793-00 
IO.I42 
543-400 
88 
O.O2523 
38.00 
3097-00 
I7-5I8 
938.600 
89 
O.O3O93 
144.00 
11736.00 
66.384 
3556.8oo 
90 
0.05088 
216.00 
17604.00 
99-576 
5335-200 
9i 
O.O5O88 
270.00 
22005.00 
I24.47 
6669.000 
92 
0.05088 
330.00 
26895.00 
I52-I3O 
8151.000 
(2) Up to 15-20 degrees the accuracy of the threshold 
values is doubtless lessened by the difficulty of making and 
reading the adjustments for such small open sectors, even with 
our micrometer device for making the adjustment and the CHROMATIC THRESHOLDS OF SENSATION 
33 
Vernier scale for reading it. On this account it may well be 
that the threshold at the center is somewhat high for all of 
the colors, inasmuch as open sectors of less than one-tenth of 
a degree are rather infeasible to use. For the exact deter- 
Table IV 
Chromatic Thresholds for Yellow, Temporal Meridian 
In this table are given the values of the threshold for yellow (581 pp) at 29 points 
in the temporal meridian. The intensity of light at the analyzing slit was 93.10 X 
io~® watt. 
Degree of 
Excentricity 
Surrounding 
Field and Pre- 
exposure. 
(Candle-power 
per Sq. In.) 
Value of Threshold 
Degrees 
Open Sector 
Total Amount 
of Light at 
Campimeter 
Opening and 
at Eye 
(Watt X 10-12) 
Density of 
Light at 
Campimeter 
Opening 
(Watt X io-22 
per Sq. Mm.) 
Density of 
Light at Eye 
(Watt X 10-U 
per Sq. Mm.) 
O 
O.OO2I25 
0.20 
i6-3° 
O.O92 
4-94 
S 
O.OO2I25 
0.25 
20.38 
0.II5 
6.175 
IO 
O.OO3175 
0.50 
40.75 
O.23I 
12.350 
IS 
O.OO3378 
O.625 
50-94 
0.288 
15438 
20 
0.003456 
0-75 
6l.I3 
0.346 
18.525 
2S 
O.OO3867 
0.875 
71-31 
0.403 
21.613 
30 
O.OO52IO 
1.125 
91.69 
O.5I9 
27.788 
33 
O.OO5779 
1.50 
122.25 
O.692 
37-050 
34 
O.OO7326 
2-75 
224.13 
1.268 
67-925 
35 
O.OO9768 
4-50 
366.75 
2.075 
in.150 
36 
0.01140 
7.00 
S70.50 
3-227 
172.900 
37 
0.01262 
9.00 
733-5° 
4-149 
222.300 
38 
0.01262 
11-25 
916.88 
5.186 
277.875 
39 
O.OI262 
12.50 
1018.75 
5763 
308.750 
40 
0.01262 
11.00 
896.50 
5-071 
271.700 
4i 
0.01262 
10.00 
815.00 
4.6lO 
247.000 
42 
0.01262 
9-50 
774-25 
4.38O 
234-650 
43 
0.01262 
9.00 
733-50 
4-149 
222.300 
44 
0.01262 
11-50 
937-25 
5-302 
284.050 
45 
O.OI4O4 
26.00 
2119.00 
II.986 
642.200 
46 
O.OI79I 
42.00 
3423-00 
I9.362 
1037.400 
47 
0.02686 
106.00 
8639.00 
48.866 
2618.200 
48 
O.O5O88 
180.00 
14670.00 
82.980 
4446.000 
SO 
0.05088 
240.00 
19560.00 
IIO.64O 
5928.000 
52 
O.O5O88 
270.00 
22005.00 
I24.47O 
6669.000 
55 
O.O5O88 
304.00 
24776.00 
I40.144 
7508.800 
58 
O.O5O88 
330-00 
26895.00 
I52.I3O 
8151.000 
60 
O.O5O88 
340.00 
27710.00 
156.740 
8398.000 
61 
O.O5O88 
350-00 
28525.00 
l6l.3SO 
8645.000 
mination of the absolute values of the threshold in this part 
of the retina, the discs can be used with more accuracy when 
a substantial reduction is first made by means of some other 
device. In considering the absolute values given above, the 
state of general adaptation of the eye must also be kept in 34 
C. E. FERREE AND GERTRUDE RAND 
mind. The horizontal component of illumination at the 
point of work was, it will be remembered, 30.49 foot-candles; 
the vertical component, 121.95 foot-candles; and the 45 de- 
gree component, 82.97 foot-candles. 
Table V 
Chromatic Thresholds for Green, Nasal Meridian 
In this table are given the values of the threshold for green (522 mm) at 23 points 
in the nasal meridian. The intensity of light at the analyzing slit was 91.50'X io-8 
watt. 
Degree of 
Excentricity. 
Surrounding 
Field and Pre- 
exposure 
(Candle-power 
per Sq. In.) 
Valne of Threshold 
Degrees 
Open Sector 
Total Amount 
of Light at 
. Campimeter 
Opening and 
at Eye 
(Watt X io-12) 
Density of 
Light at 
Campimeter 
Opening 
(Watt X io-12 
per Sq. Mm.) 
Density of 
Light at Eye 
(Watt X ro~1J 
per Sq. Mm.) 
0 
O.OO2I25 
O.IO 
8.00 
O.O45 
2.425 
5 
O.OO2I25 
0.10 
8.00 
O.O45 
2.425 
IO 
O.OO2I25 
O.I25 
10.00 
0.057 
3.O3I 
14-17 
Blind Spot 
20 
O.OO3175 
0.167 
13-33 
O.O76 
4.042 
25 
O.OO3378 
0-375 
30.00 
0.170 
9-094 
30 
O.OO3744 
0.625 
50.00 
O.283 
I5.I56 
35 
O.OO3867 
0-75 
60.00 
O.34O 
I8.I88 
40 
O.OO52IO 
1.00 
80.00 
0-453 
24.250 
45 
O.OO5779 
1.25 
100.00 
O.566 
30.313 
48 
O.OO5983 
1-50 
120.00 
0.680 
36.375 
50 
O.OO6268 
1-875 
150.00 
O.849 
45468 
5i 
O.OO6919 
6.50 
520.00 
2-945 
I57.625 
52 
O.OO74O8 
10.00 
8OO.OO 
4-530 
242.5OO 
55 
O.OO7977 
13-50 
1080.00 
6.1l6 
327-375 
57 
O.OO8954 
16.00 
1280.00 
7.248 
388.OOO 
59 
O.OIOI8 
22.50 
1770.00 
IO.I93 
545 625 
61 
O.OI425 
40.00 
3200.00 
18.120 
970.000 
63 
0.02116 
94.00 
7520.00 
42.582 
2279.500 
65 
0.02686 
212.00 
1696O.OO 
96.036 
5141.000 
66 
O.O3O93 
238.00 
I9O4O.OO 
IO7.814 
577I.5OO 
67 
O.O3663 
263.00 
2IO4O.OO 
II9-I39 
6377.750 
68 
0.05088 
315-00 
25200.00 
I42.695 
7638.750 
69 
0.05088 
338.00 
27O4O.OO 
I53-II4 
8196.500 
(3) The numerical values in Column I of these tables 
represent, of course, points in the field of vision. What the 
corresponding points on the retina are, could not be deter- 
mined without knowing the net displacement, if there be such 
displacement, of the image towards the principal axis of the 
refracting system as the beam of light enters the eye more 
and more obliquely. Just what the factors are that might 
contribute to this displacement in a refracting system so CHROMATIC THRESHOLDS OF SENSATION 
35 
complex as that of the eye, is somewhat difficult to deter- 
mine. The following are perhaps worthy of consideration: 
the difference in the optical density of the media traversed 
by the incident and the emergent rays, and the greater com- 
bined refracting power of the cornea and the anterior surface 
of the lens than of the posterior surface of the lens. Another 
Table VI 
Chromatic Thresholds for Green, Temporal Meridian 
In this table are given the values of the threshold for green (522 mm) at 19 points 
in the temporal meridian. The intensity of light at the analyzing slit was 91.50 X 
IO~8 watt. 
Degree of 
Excentricity 
Surrounding 
Field and Pre- 
exposure. 
(Candle Power 
per Sq. In.) 
Value of Threshold 
Degrees 
Open Sector 
Total Amount 
of Light at 
Campimeter 
Opening and 
at Eye 
(Watt X ro-12) 
Density of 
Light at 
Campimeter 
Opening 
(Watt X ro-12 
per Sq. Mm.) 
Density of 
Light at Eye 
(Watt X ro-12 
per Sq. Mm.) 
0 
O.OO2I25 
0.10 
8.00 
O.O45 
2.425 
S 
O.OO2I25 
O.I2S 
10.00 
0.057 
3-O3I 
IO 
O.OO2I25 
0.125 
10.00 
0.057 
3-O3I 
IS 
O.OO3175 
O.167 
13-33 
O.O76 
4.O42 
20 
O.OO3378 
O.25 
20.00 
O.II3 
6.063 
25 
O.OO3456 
0-375 
30.00 
0.170 
9-094 
26 
O.OO5779 
1.50 
120.00 
0.680 
36.375 
28 
O.OO74O8 
10.00 
800.00 
4-53° 
242.500 
30 
O.OIOI8 
32.00 
256O.OO 
14.496 
776.000 
31 
O.OII4O 
36.00 
2880.00 
16.308 
873.000 
33 
O.OI384 
65.00 
5200.00 
29-445 
1576-250 
34 
0.02116 
100.00 
8000.00 
45-300 
2425.000 
35 
O.O2523 
144.00 
II52O.OO 
65-232 
3497.000 
37 
0.02686 
180.00 
I44OO.OO 
81.540 
4365.000 
40 
0.02686 
201.50 
l6l2O.OO 
91.280 
4886.375 
4i 
O.O3O93 
226.00 
I8O8O.OO 
102.378 
5480.500 
43 
O.O3663 
270.00 
21600.00 
122.310 
6S47-5OO 
44 
0.05088 
305-00 
244OO.OO 
138.165 
7396.250 
45 
0.05088 
335-oo 
26800.00 
I5I-755 
8123.750 
thing which should be taken into consideration in the use of 
the campimeter, is the fact that the front of the cornea, and 
not the optic center of the refracting system of the eye 
(roughly speaking), is the point from which the graduations 
on the instrument are laid out. From considerations such 
as these, it can be understood, perhaps, why an object of an 
excentricity of 920 as read on the campimeter scale, or approx- 
imately 2° back of the plane tangent to the anterior surface 
of the cornea at or near the point of entrance of the line of 36 
C. E. FERREE AND GERTRUDE RAND 
regard, is still visible. In this connection the difference in 
the curvature of the cornea of different eyes should also be 
taken into account. 
Table VII 
Chromatic Thresholds for Blue, Nasal Meridian 
In this table are given the values of the threshold for blue (468 mm) at 26 points 
in the nasal meridian. The intensity of light at the analyzing slit was 91.97 X io-8 
watt. 
Degree of 
Excentricity 
Surrounding 
Field and Pre- 
exposure. 
(Candle-power 
per Sq. In.) 
Value of Threshold 
Degrees 
Open Sector 
Total Amount of 
Light at Campi- 
meter Opening 
and at Eye 
(Watt X io~12 
per Sq. Mm.) 
Density of 
Light at 
Campimeter 
Opening 
(Watt X io-12 
per Sq. Mm.) 
Density of 
Light at Eye 
(Watt X io-12 
per Sq. Mm.) 
O 
O.OOO764 
O.IO 
8.06 
O.O46 
2.44O 
5 
O.OOO764 
O.IO 
8.06 
O.O46 
2.44O 
IO 
O.OO2I25 
O.167 
13-40 
.07$ 
4.067 
I4-I7 
Blind Spot 
20 
O.OO2I25 
0.25 
20.15 
0.114 
6.100 
25 
O.OO2I25 
0.25 
20.15 
0.114 
6.100 
3° 
O.OO3175 
0.25 
20.15 
0.114 
6.100 
35 
0-003378 
0-375 
30.23 
0.171 
9-I5O 
40 
O.OO3456 
0-375 
30.23 
0.171 
9-I5O 
45 
O.OO3744 
0.50 
40.30 
0.228 
12.200 
50 
O.OO3744 
0.625 
50.38 
0.285 
15-250 
55 
O.OO3867 
1.125 
90.68 
0-513 
27-450 
60 
O.OO3867 
i-5° 
120.90 
0.684 
36.600 
65 
O.OO3867 
1.50 
120.90 
0.684 
36.600 
70 
O.OO3867 
i-75 
I4I-O5 
0.798 
42.700 
75 
O.OO3867 
1-875 
I5I-I0 
0-855 
45-750 
76 
O.OO3989 
3-oo 
241.80 
1.368 
73.200 
77 
O.OO52IO 
12.00 
967.20 
5-472 
292.800 
78 
0-005779 
29.00 
2337-4° 
13-224 
707.600 
80 
O.OO6956 
82.00 
6609.20 
37-392 
2000.800 
83 
O.OI262 
116.00 
9349-60 
52.896 
2830.300 
85 
O.O3O93 
183.00 
14749-80 
83-448 
4465.200 
86 
O.O3O93 
198.00 
I5958-8O 
90.288 
4831.200 
87 
O.O3663 
234.00 
18860.40 
106.704 
5709.600 
88 
O.O3663 
277.00 
22326.20 
126.312 
6758.800 
90 
0.05088 
295.00 
23777.OO 
134-520 
7198.000 
92 
0.05088 
328.00 
26436.80 
I49-568 
8003.200 
That there is this wide extension of the field of vision in 
the temporal meridian seems to be well established by other 
investigators. Baas, for example, quoted by de Schweinitz 
(8), finds the average limit for ten observers to be 990. Tra- 
quair (9) and others also find the limits to extend beyond 900. 
In our own work we have found it to be carried out as far 
as 920 in a number of cases. For the explanation of this CHROMATIC THRESHOLDS OF SENSATION 
37 
Table VIII 
Chromatic Thresholds for Blue, Temporal Meridian 
In this table are given the values of the threshold for blue (468 wt) at 26 points 
in the temporal meridian. The intensity of light at the analyzing slit was 91.97 X 
io-8 watt. 
Degree of 
Excentricity 
Surrounding 
Field and Pre- 
exposure. 
(Candle-power 
per Sq. In.) 
Value of Threshold 
Degrees 
Open Sector 
Total Amount 
of Light at 
Campimeter 
Opening and 
at Eye 
(Watt X xo-12) 
Density of 
Light at 
Campimeter 
Opening 
(Watt X io-12 
per Sq. Mm.) 
Density of 
Light at Eye 
(Watt X io-12 
per Sq. Mm.) 
O 
O.OOO764 
0.10 
8.06 
O.O46 
2.44O 
5 
O.OOO764 
0.10 
8.06 
O.O46 
2.44O 
IO 
O.OOO764 
0.10 
8.06 
O.O46 
2.44O 
15 
O.OO2I25 
O.167 
I3.4O 
O.O76 
4.067 
20 
O.OO2I25 
O.25 
20.15 
O.II4 
6.100 
25 
0-003175 
0-375 
3O-23 
0.I7I 
9.150 
30 
O.OO3378 
O.625 
50.38 
O.285 
15-250 
33 
O.OO3456 
0-75 
60.45 
0-342 
18.300 
35 
O.OO3456 
I.OO 
80.60 
0.456 
24.400 
40 
O.OO3744 
1-25 
IO7.5O 
0-570 
30.500 
41 
O.OO3867 
2.25 
181.35 
1.026 
54-900 
42 
O.OO5O88 
4.00 
322.4O 
I.824 
97.600 
43 
0-005779 
7.00 
564.20 
3-192 
170.800 
44 
O.OI262 
80.00 
6448.OO 
36.480 
1952.000 
45 
O.O3O93 
120.00 
9672.OO 
54-720 
2928.000 
46 
O.O3O93 
164.OO 
I3218.4O 
74-784 
4001.600 
46-5 
O.O3663 
270.00 
21762.OO 
123.120 
6588.000 
47-48 
Blind to Blue 
49 
O.O3663 
270.00 
21762.00 
123.120 
6588.000 
50 
O.O3663 
234.OO 
18860.40 
106.704 
5709.600 
52 
O.O3663 
180.00 
14508.00 
82.080 
4392.000 
55 
O.O3663 
23O.OO 
18538.00 
104.880 
5612.000 
56 
O.O3663 
25O.OO 
20150.00 
114.000 
6100.000 
57 
O.O5O88 
270.00 
21762.00 
123.120 
6588.000 
58 
0.05088 
29O.OO 
23374-00 
132.240 
7076.000 
59 
0.05088 
306.00 
24663.60 
I39-536 
7466.400 
61 
0.05088 
338.00 
27242.80 
154.128 
8247.200 
fact there is of course the alternative possibility of a loss of 
exact fixation, which possibility we have not ignored in our 
own thinking. However, if there is a loss of fixation, we have 
not been able to detect it by any means we have as yet been 
able to devise. 
Discussion of Results 
From data such as are presented in these tables and charts 
the following comparisons may be made in so far as the thresh- 
old may be regarded as a measure of sensitivity: (a) the sen- 
sitivity to a given range of wave-lengths from point to point 38 
C. E. FERREE AND GERTRUDE RAND 
in the same meridian; (b} the sensitivity at corresponding 
points in different meridians; and (c) the sensitivity to the 
different wave-lengths in any given meridian. From the lat- 
ter comparison, if properly made, an estimate of the selec- 
tiveness of the chromatic response at the threshold of sensa- 
tion may be had. In making these comparisons several 
points of interest may be noted. 
(i) The Irregularities in the Curve of Sensitivity for the 
Different Colors in a Given Meridian.-The rate of decrease in 
sensitivity after a certain degree of excentricity has been 
reached shows a great irregularity for the individual colors. 
This irregularity is greater in the temporal than in the nasal 
meridian. The following are some of the more notable ex- 
amples. (a) The plateau in the curve for red in the tem- 
poral meridian. That is, the curve representing the amounts 
of energy required to arouse the just noticeable red sensation 
in this meridian rises almost imperceptibly from the center 
of the retina to about 20 degrees. From there it rises sharply 
to about 44 degrees. From 44 to 47 degrees the curve is almost 
vertical. Between these two points the energy required to 
arouse just noticeable sensation changes from 31.236 X io~10- 
I77-552X io-10 watt. From 47 to 54 degrees occurs the pla- 
teau referred to. Between these two points there is prac- 
tically no change in sensitivity, (b) The hump in the curve 
for yellow in the temporal meridian occurring between 33 and 
43 degrees. From 36 to 39 degrees there is a sharp drop in sen- 
sitivity and from there to 43 degrees almost as sharp a rise. 
Between 36 and 39 degrees the energy required to arouse just 
noticeable sensation changed from 5.705 X io-10 to 10.188 
X io-10 watt. This is an area of deficiency or partial blind- 
ness to yellow. There is in this area no corresponding loss 
in sensitivity to the blue stimulus nor to the other colors. 
Moreover, there is no detectable change in the cancelling or 
after-image reactions to yellow. That is, this area shows the 
characteristics of the Schumann case of color blindness, (c) 
The quick rise in the curve for all of the colors near the limits 
of sensitivity. This feature is very marked both in the nasal 
and temporal meridians, but more marked in the nasal than CHROMATIC THRESHOLDS OF SENSATION 
39 
in the temporal meridian. In the temporal meridian the 
rise begins around 45 degrees for red, yellow and blue and at 
26 degrees for green. In the nasal meridian it begins around 
75 to 85 degrees for red, yellow and blue; and at 51 degrees 
for green. In this region were found the limits of sensitivity 
in our previous work with the Hering pigment papers with 
the degree of illumination, etc., employed. And (W) the area 
of total blindness to blue from 47 to 49 degrees in the temporal 
meridian. On all sides of this area is a border of lowered 
sensitivity, the sensitivity falling off quite sharply as the 
area is approached. This area also shows the characteristics 
of the Schumann case of color blindness. That is, there is 
no corresponding decrease in sensitivity to yellow, red or 
green and no corresponding change in the cancelling or after- 
image reactions. 
(2) The Great Difference in Sensitivity at Corresponding 
Points in the Temporal and Nasal Meridians, more Especially 
in the more Remote Portions of the Retina.-This difference is 
already so well known in a general way as to need no especial 
discussion here. Our data, however, may be considered as 
contributive in two regards: (<2) the large number of points 
investigated and (ff) the rating of the stimuli in units that can 
be compared numerically. In comparing the distribution of 
sensitivity at corresponding points in the two meridians it 
may be of interest among other things to note the difference 
in the order in which the sensitivity to the different colors 
falls off in case of red, yellow and blue as the limits of sensi- 
tivity are approached. So far as the relative sensitivity to 
red and green is concerned, this point bears upon the changes 
in the color tone of red in the two meridians in passing from 
the center to the periphery of the retina. As will be noted 
later in a more detailed statement of color tone changes, the 
red stimulus employed in the investigation was sensed as 
red in the nasal meridian from the center to 60 degrees, from 
60 to 86 degrees as yellowish red or orange, and 86 to 92 de- 
grees as red. In the temporal meridian it was sensed as red 
from the center to 30 degrees, from 30 to 47 degrees as yel- 
lowish red or orange and from 47 to 61 degrees as red. That 40 
C. E. FERREE AND GERTRUDE RAND 
is, the zone in the far periphery of the retina in which the 
weakly aroused yellow component of the excitation is below 
the threshold in red, is relatively broader in the temporal 
than in the nasal meridian, as the relation of the curves of 
sensitivity to red and yellow would indicate should be the 
case. 
(3) The Striking Absence of Uniformity of Ratio of Sensi- 
tivity to the Pairs of Colors, Red and Green, and Blue and Yel- 
low, from Center to Periphery of the Retina.-From the center 
to about 27 degrees in the temporal meridian and about 51 
degrees in the nasal meridian, the sensitivity to green is 
greater than to red. At these points a radical reversal of 
sensitivity takes place and the sensitivity to red is much 
greater than to green. A reversal of sensitivity occurs also 
for blue and yellow; but at a point farther removed from the 
center of the retina, namely, at about 44 degrees in the tem- 
poral meridian and at about 77 degrees in the nasal meridian. 
Even up to the points of reversal with their radical deviations 
in relative sensitivity, an inspection of the charts, especially 
III. and IV., will show how very little ground there is for 
any claim that constancy of ratio exists between the colors 
red and green, and blue and yellow from center to periphery 
of the retina. (A further discussion of this point will be 
given in the second part of this paper: 'Bearing of Results 
on Color Theory.') 
And (4) the Correspondence of the Distribution of Sensitivity 
to Red, Green and Yellow with what might be expected from the 
Changes in the Color Tone of Red and Green in Passing from 
the center to the periphery of the retina.-For example, in pass- 
ing from the center to the periphery of the retina the red stim- 
ulus used in making the threshold determinations given in 
the foregoing curves was sensed as red from the center to 
about 60 degrees; from 60 degrees to about 86 degrees it was 
sensed as a yellowish red or orange; and from 86 degrees to 
the limits of sensitivity it was sensed again as red. Corre- 
sponding to this, it will be noted that there is in this meridian 
a fairly close agreement in sensitivity to red and yellow from 
the center to about 60 degrees, at which point there is a rela- CHROMATIC THRESHOLDS OF SENSATION 
41 
tively sharp decrease in sensitivity to red. That is, from 
about 60 to 86 degrees there is much less sensitivity to red 
than to yellow. At about 86 degrees there is a sharp decrease 
in sensitivity to yellow, and from this point on to the limits 
of sensitivity a fairly close agreement again in sensitivity to 
the two colors. In the temporal meridian the red stimulus 
was sensed as red from the center of the retina to about 30 
degrees; from there to about 47 degrees as yellowish red or 
orange; and from 47 degrees to the limits of sensitivity as red. 
Similarly in this meridian there is a fairly close agreement in 
sensitivity to red and yellow from the center to about 30 de- 
grees; from 30 degrees to about 47 degrees there is consider- 
ably greater sensitivity to yellow than to red; and from this 
point on to the limits greater sensitivity to red than to yellow. 
In case of green, in the nasal meridian the greater loss in sens- 
itivity to green as compared with yellow begins at about 51 
degrees; and in the temporal meridian at about 26 degrees. 
Correspondingly at these points the green stimulus began to 
be sensed as yellowish green and continued to be sensed in 
this tone until the limits of sensitivity to green were reached, 
from which point on for a short distance it was sensed as 
yellow. 
Bibliography 
I. Ferree, C. E., and Rand, G. A Note on the Determination of the Retina's Sen- 
sitivity to Colored Light in Terms of Radiometric Units. Amer. Jour, of Psy- 
chol., 1912, 23, pp. 328-332. 
2. Ferree, C. E., and Rand, G. A Spectroscopic Apparatus for the Investigation 
of the Color Sensitivity of the Retina, Central and Peripheral. Jour, of Exper. 
Psychol., 1916, 1, pp. 271-274. 
3. Ferree, C. E., and Rand, G. The Selectiveness of the Achromatic Response of 
the Eye to Wave-length and its Change with Change of Intensity of Light. 
Studies in Psychology. Titchener Commemorative Volume, 1917, Worcester, 
Mass., pp. 285-287. 
4. Coblentz, W. W. Measurements on Standards of Radiation in Absolute Value. 
Bulletin Bureau of Standards, 1914, 11, pp. 87-100. 
5. Ferree, C. E., and Rand, G. A Simple Daylight Photometer. Amer. Jour, of 
Psychol., 1916, 27, pp. 335-34O- 
6. Ferree, C. E., and Rand, G. Op. cit. Titchener Commemorative Volume, pp. 
304-306. 
7. Nutting, P. G. The Visibility of Radiation. Philos. Mag., 1915, 29 (6), pp. 301- 
309- 
8. de Schweinitz, G. E. Diseases of the Eye. 8th ed., W. B. Saunders Co., 1916, 
p. 81. 
9. Traquair, H. M. British Journal of Ophthalmology, 1917, 1, p. 216. [Reprinted from The Psychological Review, Vol. 26, No. 2, March, 1919.] 
CHROMATIC THRESHOLDS OF SENSATION FROM 
CENTER TO PERIPHERY OF THE RETINA 
AND THEIR BEARING ON COLOR 
THEORY-PART II 
BY C. E. FERREE AND GERTRUDE RAND 
Bryn Mawr College 
As was stated in Part L, one of the incentives to this 
investigation was to clear up two points in relation to color 
theory. These points are as follows: 
I. The claim has been made by followers of the Hering 
theory that the sensitivity of the retina to the pairs of colors: 
red and green, and blue and yellow, falls off in a constant 
ratio from the center to the periphery of the retina. This 
claim, it will be remembered, was made first by Hess (1) on 
the grounds of an investigation of the relative limits of sen- 
sitivity with colors equalized both in cancelling power and 
brightness; and was given a great deal of importance by 
Hering (2) in a companion article in refutation of a revision 
of the Young-Helmholtz theory made independently by 
Fick (3) and Leber (4) to explain the color blindness of the 
peripheral retina. Fick, for examp e, assumed that from the 
middle towards the periphery of the retina the relative ex- 
citability of the three nerve fibers to lights of the various 
wave-lengths constantly alters in such a way that at a certain 
distance from the fovea, namely, in the zone called by Helm- 
holtz red-blind, the red sensing fibers possess the same excit- 
ability as the green sensing fibers towards lights of all wave- 
lengths; and that further towards the extreme periphery all 
differences between the relative excitability of the three 
fibers diminish and finally disappear. In the red-blind zone, 
then, the intensity curves for the red and green sensing fibers 
coincide, and in the totally color-blind zone, the curves 
for all three coincide. Curves drawn in accord with these 
assumptions will, it is contended by Fick, explain the types 
150 151 
C. E. FERREE AND GERTRUDE RAND 
of color-blindness found in the peripheral retina without 
violating any of the fundamental principles of the Young- 
Helmholtz theory. Helmholtz accepts the essential points 
of this modification and incorporates them in his theory in 
his later edition of the 'Physiologische Optik' (5). With 
stimuli equalized in cancelling power and brightness, however, 
Hess claimed to find a coincidence of the limits for the pairs 
of colors used, and contended therefrom that the sensitivity 
to the pairs of antagonistic colors falls off uniformly from the 
center to the periphery of the retina or that a constant ratio 
of sensitivity to these colors obtains throughout the retina. 
(See also in this connection the papers of Bull (6), Hegg (7), 
and Baird (8).) 
Prior to the presentation of a direct disproof of Hess's 
conclusion that a constant ratio of sensitivity to the paired 
colors obtains throughout the retina, in the form of results 
obtained in a detailed investigation of sensitivity from center 
to periphery, we had pointed out in a previous paper (9) that 
his conclusion was not warranted by the work and results 
on which it was based. It was based, it will be remembered, 
on the twofold assumption that if in passing from center to 
periphery sensitivity ends at the same point of the retina for 
two stimuli which have equal power to arouse sensation at 
the center, (<2) they must still have equal power to arouse 
sensation at the periphery and (b} sensitivity must have fallen 
off as much for the one as for the other and evenly and uni- 
formly from point to point. This assumption in the first 
place begins with a fallacy, for the stimuli were not equalized 
in power to arouse sensation but in cancelling power. Can- 
celling power and the power to arouse sensation are, as we 
have already pointed out, not at all equivalent (10). In the 
second place the assumption is itself incorrect; for because 
of the abrupt decrease in sensitivity with stimuli of medium 
and high intensities as the limits are approached, the rela- 
tive sensitivity to the two colors may have changed greatly, 
even assuming an even grading in the loss of sensitivity for 
each color from the center out, and still the deviations from 
equal sensitivity not make a difference of as much as 1 de- CHROMATIC THRESHOLDS OF SENSATION 
152 
gree in the limits for the two colors. We have found, for 
example, that working with pigment colors of good satura- 
tion under an illumination of 390 foot-candles, it takes, 
varying with the color and the meridian investigated, 90 to 
120 degrees of color mixed with a gray of the brightness of 
the color to make a difference of I degree in the limits. But 
an even grading in the loss of sensitivity can not be assumed 
as the results given in this study show; hence even if it could 
be demonstrated that the same ratio holds at the limits or 
at any point well removed from the center, as at the center, 
the conclusion could not be drawn that constancy of ratio 
obtains between these points. On this question it is obvious 
that a conclusion is not warranted unless a point to point 
investigation is made, and such an investigation shows that 
striking irregularity and not constancy and uniformity char- 
acterizes the changes in sensitivity from the center to the peri- 
phery of the retina. And in the third place, when the results 
of Bull, Hess and Baird who all claimed coextensive limits are 
examined in detail, it is found that they show the same sort of 
deviation from coincidence from meridian to meridian as were 
obtained by Kirschmann and by us, who have made a point 
of lack of coincidence when stimuli of the same order of 
intensity are employed. Baird, for example, who determined 
the limits for red, green, blue, and yellow stimuli in eight 
meridians in the dark room by means of a perimeter, concludes: 
"The results show that the zone of stable red is coincident 
with that of stable green and that the zone of stable yellow is 
coextensive with that of stable blue" (11). An inspection 
of his results shows, however, that the coincidence is extremely 
rough. In case of the results for every observer it is found 
that in some meridians the green field is narrower than the 
red by 1, 2 or 3 degrees; in other meridians there is coincidence 
of limits; and in still other meridians the red field is narrower 
than the green by 1, 2 or 3 degrees. The same is true of blue 
and yellow, the deviations from coincidence ranging from 
l°-5°. Hess and Bull's results show similar variations, in 
some cases even greater in amount. It seems probable from 
their conclusions concerning the coincidence of limits, that 153 
C. E. FERREE AND GERTRUDE RAND 
they regarded these variations as insignificant. But it should 
be borne in mind that .2 or 3 degrees of difference in limits is 
not insignificant when conclusions with regard to the relative 
sensitivity of the peripheral retina to the complementary colors 
are to be drawn from the results. Because of the abrupt 
falling off in sensitivity before the limits are reached with 
stimuli of medium and high intensities, a difference of 2 or 3 
degrees in the limits represents quite a large difference in sen- 
sitivity. For example, according to our results with the 
Hering standard papers under 390 foot-candles of illumination 
(vertical component), a difference of 2 degrees in the limits 
represents a difference in sensitivity sufficient to raise the 
threshold for yellow 120 degrees; for green, 100 degrees; for 
red, 160 degrees; and for blue 160 degrees. And a difference 
of 3 degrees represents sufficient difference in sensitivity to 
raise the threshold for yellow 210 degrees; for green, 215 
degrees; for red, 210 degrees; and for blue, 215 degrees. 
Our results with the more intense spectrum lights show, as 
might be expected, that a difference of 2 or 3 degrees in the 
limits represents a still greater difference of sensitivity. This 
should be quite obvious from the curves we have given in 
Part I. It is scarcely needful to note in this connection also 
that the weaker are the stimuli employed, the nearer will 
the limits be to the center of the retina; and it should be clear 
from the curves we have given that the nearer the limits are 
to the center of the retina, the closer will be the approximation 
to coincidence. Bull, Hegg, Hess and Baird in their attempts 
to equalize their stimuli both in cancelling power and bright- 
ness must have worked with colors of comparatively low 
saturation, hence with stimuli unduly favorable to coincidence 
of limits. Their results, therefore, are the product of a 
special method of working rather than are representative of 
the relations of sensitivity actually existing in the more remote 
periphery of the retina even so far as these relations can be 
judged from a determination of limits alone. Unfortunately 
no specification of the intensity of their stimuli was given, 
but the narrowness of the zones of sensitivity obtained indi- 
cate that stimuli of low color arousing power were used. CHROMATIC THRESHOLDS OF SENSATION 
154 
It may be of some interest also to note in this connection 
that the present writers have never been able to secure red, 
green, blue and yellow stimuli of spectrum purity, all equal 
in brightness and at the same time to have the pairs of com- 
plementary colors in cancelling proportions. To conceive 
that the spectrum colors can be equalized at one and the same 
time with regard to these two independent variables would 
seem a -priori to be a logical impossibility; and the task of 
making this twofold equalization has as yet proved too diffi- 
cult for us as a practical problem. It might perhaps be done 
if a variable weighting factor, namely, colorless light, were 
introduced in the right proportions into the composition of a 
part or all of the stimuli, but that would scarcely be com- 
patible with the purpose of the investigation. It is, in fact, 
difficult to understand why such an equation should ever have 
been attempted in the first place in an investigation of chro- 
matic sensitivities. An equation in the power to arouse the 
chromatic response is, so far as we can see, the only subjective 
equation that could be rightfully given a place in the determi- 
nation of the relative limits of chromatic sensitivity, and 
this only in a determination of whether or not the same ratio 
of sensitivity holds for the limits as for the center or other 
point at which the equation was made. That is, if it does 
hold, the limits would be coincident, and if it does not hold, 
they would not be coincident; but no definite knowledge 
would be gained of the amount of deviation from equality of 
ratio, nor would any inference be justified with regard to the 
relative values of the ratio between the point at which the 
equation was made and the limits. Just what would be 
accomplished by an equation in cancelling power which 
neither equalizes the stimuli in intensity nor in power to 
arouse the chromatic response, is far from clear. Had the 
object of the investigation been a determination of whether or 
not constancy in cancelling proportions holds for all parts of 
the retina, the verdict would be different. For one type of 
investigation of this sort, then, the equation would be of 
service, but for an investigation of constancy of ratio of 
sensitivity, it is obviously irrelevant. There seems also to 155 
C. E. FERREE AND GERTRUDE RAND 
be no more experimenal justification and little if any more 
a -priori plausibility for the equation in brightness for a de- 
termination of the relative limits of chromatic sensitivity; 
for («) it does not equate the stimuli in power to arouse the 
chromatic response, (Z>) neither does it equate them in in- 
tensity (the equation is merely of the very selective achromatic 
response to the stimuli), and (c) so far as the effect of the 
achromatic on the chromatic component of the excitation is 
concerned (the final variable factor that might be considered), 
it has already been shown by one of us in a previous article 
(12) that there is not enough difference in this effect for the 
colors used to change the limits of sensitivity by a detectable 
amount. However, the irrelevancy and the positive dis- 
advantage of such equations, as they appear to us, have been 
discussed in detail in the previous paper. The question is 
raised again here only because the results of our point to 
point investigation throw additional light on the effect that 
the attempt to treat the stimuli in this way would have on 
determinations of the type under consideration, namely, 
compelling the use of stimuli of such low color arousing power 
as to make the conditions unduly favorable for a coincidence 
of limits. 
But of much greater importance than all of this as a general 
consideration, is the realization that the determination of the 
relative or apparent limits for the purpose of ascertaining 
whether or not a constant ratio of sensitivity to the paired 
colors obtains throughout the retina, falls far short of its 
objective. The information sought can be obtained only by 
a point to point investigation of sensitivity. With the 
passing then of the belief in any especial significance of the 
determination of the relative limits, which after all is only a 
very inadequate way of comparing sensitivities at a limited 
number of points and which was given undue importance in 
relation to theory by the failure of Hess and his followers to 
realize that great irregularity and not uniformity charac- 
terizes the decrease of sensitivity from the center to the 
periphery of the retina, will doubtless pass also any feeling of 
need to be concerned about the reasons that may have in- 
fluenced these writers to treat their stimuli as they did. CHROMATIC THRESHOLDS OF SENSATION 
156 
2. The point to point investigation also has an important 
bearing on the question of stability of color tone. The 
results given in Part I make it easy to understand why it is 
not possible to find a red and a green stimulus that are in- 
variable in color tone from the center to the periphery of the 
retina in all meridians. That is, the conception of a red and 
a green that are stable in tone presupposes a regularity in the 
relative rate of decrease in sensitivity to red and green on the 
one hand and to yellow on the other which the point to point 
investigation shows is very far removed from fact. 
The claim to a stable red and a stable green was first 
made by Bull and later by Hess, Hegg and Baird in investi- 
gations of the relative limits of sensitivity to the paired colors. 
Working with pigment papers Bull added blue to his red and 
green stimuli in order that he might get colors that would 
not be sensed as yellowish in the peripheral retina.1 His 
purpose in doing this, he states, was to find the physiologically 
pure red and green. Passing over the fact that the addition 
of blue in sufficient amounts to cancel the peripheral yellow 
gives an excess of blue in the more central portions of the 
retina (even outside of the macula) which is scarcely com- 
patible with the tenet of introspective simplicity, this method 
of obtaining a stable red and green would presuppose, as we 
have already stated, a regularity in the relative rates of 
decrease of sensitivity to red and green on the one hand and 
to yellow on the other in the different meridians of the retina 
which is far removed from fact. For example, if the sen- 
sitivity to red fell off at the same rate in all meridians of the 
retina and the sensitivity to red and yellow in a constant ratio, 
the amount of blue which is required in a given meridian to 
neutralize the yellow component in sensation would suffice 
for this purpose in all meridians. Since neither of these 
essential conditions is present in the relative distribution of 
sensitivities from center to periphery of the retina in the 
different meridians, the futility of the search for the stable 
red and the stable green by the method proposed by Bull is 
1 Speaking of the gelatines used for his red stimulus, Baird (op. cit., p. 60) says: 
"The red stimulus transmitted no part of the visible spectrum." 157 
C. E. FERREE AND GERTRUDE RAND 
obvious. Moreover, it is perhaps just as obvious that 
stability of tone throughout the retina is by no means a 
necessary corollary of a four-color theory of the type proposed 
by Hering, and therefore that its use by the aforementioned 
writers for the purpose of searching out or isolating the four 
physiological processes was questionable even on a priori 
grounds. That is, the assumption that our color processes 
are conditioned by four physiological processes, the action of 
any one of which alone would give a sensation which is intro- 
spectively simple, should not by any means carry with it also 
the assumption that stimuli can be found to which one alone 
of these processes responds. For in the first place, such a 
narrowness of selectivity of response is not needed to explain 
our experience of the introspectively simple sensation; 
secondly, it is not as a general case characteristic of selective- 
ness of action; and thirdly, it is quite out of keeping with the 
change of tone of red and green in passing from the center to 
the periphery of the retina. The explanation of this pheno- 
menon seems to have given not a little concern to the followers 
of the Hering theory who have apparently, in some cases at 
least, thought that if corresponding to the four simple sen- 
sations there are four simple physiological processes, it should 
be possible to find stimuli that would arouse one of these 
processes alone, which stimuli should of course be invariable 
in color tone for all parts of the retina. But, as we have al- 
ready pointed out, this is neither a necessary nor perhaps 
even a plausible corollary of the fundamental assumption of 
four processes, the action of any one of which alone should 
give a simple sensation, and besides detracts needlessly from 
its explanatory value. 
Hering's own criteria for the selection of the Urfarben 
were (i) introspective simplicity at or around the maximal 
saturation (13) and (2) following the lead of Fick, no change 
of color tone with change of intensity of the stimulus light 
(14). Whether or not constancy of tone can be expected for 
all intensities of light would again seem to depend on whether 
constant ratios of chromatic sensitivity obtain for all inten- 
sities of light; in other words, upon whether or not the select- CHROMATIC THRESHOLDS OF SENSATION 
158 
iveness of the chromatic response of the eye varies with the 
intensity of light, as does the achromatic response. If it 
does, it is too much to expect that this second criterion of 
Hering's will be of any especial service for the purpose for 
which he used it; for, depending upon the variations of ratio 
of sensitivity or relative amounts of the selectiveness of the 
chromatic response, one spectrum band may give the intro- 
spectively simple sensation at one intensity and a mixed 
sensation at a different intensity. Also the effect of the 
varying strength of the achromatic component on the color 
tone of the sensation aroused can not be left out of considera- 
tion. The attempt, therefore, to label, so to speak, the simple 
physiological process with a wave-length or spectrum specifica- 
tion, presupposes a simplicity in the eye's reactions which 
very probably does not exist. The investigation of the 
selectiveness of the chromatic response of the eye in relation 
to intensity of light is, as we have already stated, now in 
progress in this laboratory. 
To explain the experience of introspectively simple red and 
green at the center of the retina and its change in tone in passing 
towards the periphery in terms of four physiological processes, 
the action of any one of which alone would give a simple 
sensation, it is necessary only to call attention to three factual 
considerations1 the application of which to the point in ques- 
1 With regard to the first of these considerations it may be of interest to note that 
Hering did not himself seem to regard the simple physiological processes as narrowly 
selective in their response to wave-length, and that he recognized, implicitly at least, 
that all of the color processes exert an inhibitive action on each other. This conclusion 
may be derived from the following passages and from others in the article (15) from 
which they they are quoted: "Da alle sechs Processe fortwahrend gleichzeitig, wenn 
auch mit sehr verschiedener Starke in der Sehsubstanz stattfinden, so sind auch immer 
alle sechs Grundempfindungen, gleichzeitig gegeben. Jede Gesichtsempfindung ist 
daher eigentlich ein Gemisch aus den sechs Grundempfindungen, doch sind darin immer 
nur einige von den Grundempfindungen deutlich, die andern unter der Schwelle. Die 
Deutlichkeit, mit welcher die eine oder die andere der Grundempfindungen sich in der 
Gesammtempfindung zeigt, hangt von dem Verhaltniss ab, in welchem die Starke des, 
dieser Empfindung correlaten Processes zur Starke der fiinf ubrigen steht. Ist z.B. 
der schwarze Process sehr stark im Vergleich zu alien andern, so tritt die schwarze 
Empfindung mit besonderer Deutlichkeit hervor, wobei die fiinf ubrigen so undeutlich 
werden konnen, dass sie nicht mehr einzeln wahrnehmbar oder, wie man zu pflegt, 
unter der Schwelle sind. Wir nennen dann die Gesammtempfindung schwarz. Sind 
Z.B. die beiden Grundempfindungen Griin und Blau besonders deutlich, so nennen wir 
die Empfindung griinblau u.s.w. ..." 159 
C. E. FERREE AND GERTRUDE RAND 
tion apparently has not always been clearly kept in mind. 
(1) The selectiveness of the eye's response to wave-length is 
not complete. Apparently it ranges from a minimum to a max- 
imum. In case of the red and yellow processes, for example, the 
maximum is reached respectively at certain points in the red 
and yellow portions of the spectrum and the minimum at a 
point in the orange. (2) There is an inhibitive action of the 
non-complementary colors1 on each other as well as of the 
complementary colors. That is, the threshold of one of these 
colors in another is high and its value differs from color to 
color. And (3) the distribution of sensitivity is very irregular 
from point to point over the retina. That is, a certain range 
of wave-lengths in the red portion of the spectrum acting on 
the center of the retina arouses both the red and the yellow 
processes, the red strongly and the yellow weakly. The 
yellow does not come to sensation because it is below the 
threshold of yellow in red. When, however, the same stimulus 
acts on the peripheral retina at points where the red process 
is relatively undeveloped as compared with the yellow, the 
yellow is no longer subliminal in the red but becomes a com- 
ponent of the sensation of a value depending upon the ratio 
of sensitivity to red and yellow at the points in question, the 
On pp. 79-81 he continues: "Ausser der weissen Valenz, welche alien Licht- 
strahlen gemeinsam ist, kommen nun den einzelnen Strahlenarten verschiedene farbige 
Valenzen zu. Alle Strahlen von aussersten Roth oder vom Anfange des Spectrums bis 
zu jenem im Tone reinen Grun, welches eine Grundfarbe ist und welches wir das 
Urgriin nennen wollen, haben eine gelbe, alle Strahlen vom Urgriin bis zum violetten 
Ende des Spectrums eine blaue Valenz. Demnach theilen wir das Spectrum in eine 
gelbwerthige und eine blauwerthige Halfte, wenn auch beide nicht gleich lang erschienen. 
Am Anfange des Spectrums ist die gelbe Empfindung so schwach, dass sie gegeniiber 
der deutlicheren rothen unter der Schwelle bleibt; ebenso tritt sie in der Nahe des 
Urgriin wieder mehr und mehr hinter der griinen Empfindung zuriick. Nur in einem 
schmalen Streifen erscheint uns das tonreine Gelb oder das Urgelb, welches der Grund- 
farbe entspricht. Analoges gilt vom Urblau, welches nach dem Urgriin hin immer 
mehr gegeniiber der griinen Empfindung zuriick tritt, nach dem Ende des Spectrums 
hin aber sich mehr und mehr mit rother Empfindung mischt. 
1 We do not wish to be understood as suggesting here that the inhibitive action of 
the non-complementary colors on each other requires the same mechanism for its 
explanation as the complementary or even that it takes place at the same functional 
level. We are inclined rather to believe that the inhibitive action of the achromatic 
on the chromatic excitation and possibly also of the non-complementary colors on each 
other takes place at a level posterior to the inhibiting action of the complementary 
colors. CHROMATIC THRESHOLDS OF SENSATION 
160 
mutually inhibitive actions of the excitations upon each other, 
etc. A similar explanation holds for the green. Why an 
analogous phenomenon of change of tone due to these factors 
should not occur in case of blue and yellow is obvious. That 
is, even though a subliminal red or green excitation were 
aroused in the center of the retina by the blue or yellow wave- 
lengths, it would become still more subliminal as the periphery 
of the retina is approached because of the more rapid decrease 
in sensitivity to red and green, and would not come to sensa- 
tion. The changes that do take place in the color tone of 
yellow in passing from the center to the periphery of the 
retina we have already explained as an effect of the achromatic 
upon the chromatic component of the excitation. The 
demonstration of the validity of this explanation will be the 
work of a later paper. 
Comments 
The discussion relative to color theory, so far as we wish 
to consider theory at this time, may perhaps be summed up 
in the following comments. 
i. An explanation of the color changes of red and green 
in passing from the center to the periphery of the retina may 
be found in three factual considerations: (a) the absence of 
complete selectivity of response of the eye to the red and green 
wave-lengths of light; (b) the inhibitive action of the non- 
complementary colors on each other; and (c) the relative 
distribution of sensitivity to red and green on the one hand 
and to yellow on the other in the periphery of the retina. 
The color tone of red and green seems to be very little de- 
pendent on the achromatic conditions of stimulation in any 
part of the retina. 
2. The changes in the color tone of yellow (also of blue, 
so far as they occur), in passing from the center to the peri- 
phery of the retina seem to be largely, if not entirely, an 
effect of the achromatic conditions,-in physiological terms 
an effect of the achromatic upon the chromatic component 
of the excitation, the state of achromatic adaptation of the 
eye, etc. Of the four principal colors, the color tone of blue 
and yellow is as a general case the most dependent on the 161 
C. E. FERREE AND GERTRUDE RAND 
achromatic conditions. Moreover, given the same achro- 
matic conditions there is a striking agreement in the effect in 
all parts of the retina. The relative distribution of chromatic 
sensitivities apparently plays little if any part in the changes 
of color tone of blue and yellow in passing from the center to 
the periphery of the retina. 
3. The claim that a red, green, blue, and yellow stimulus 
may be found to which the eye will give a response invariable 
in color tone in all parts of the retina is based on a very in- 
complete and inadequate investigation of the eye's possibili- 
ties of response. Also the importance of the bearing of such 
a possibility on color theory seems to the present writers in 
many instances at least to have been very wrongly stressed. 
(Our own conclusions with regard to the possibility of ob- 
taining stability of tone for these colors, for example, are 
based on a very minute investigation on a number of ob- 
servers in sixteen meridians of the retina. Moreover, some 
of the more important findings of this investigation have 
been confirmed year by year in the work of the undergraduate 
laboratory.) 
4. There is no basis of fact for a claim that a constant 
ratio of sensitivity to the pairs of colors red and green, and 
blue and yellow obtains in all parts of the retina; nor is it 
apparent that such a claim is of any considerable consequence 
to the fundamental postulates of theories of the Hering type. 
It is more important, for example, (a) that wherever one of 
these pairs of processes be found, the other shall also be found; 
and (b) that a constancy of ratio of cancelling proportions 
for the pairs of colors obtains in all parts of the retina. (The 
power to arouse sensation and the power to cancel the an- 
tagonistic or complementary color are not, as we have already 
shown, equivalent.) With regard to the first of these points 
it does seem to be of rather serious consequence that we have 
not been able to get even approximately coextensive zones of 
sensitivity to red and green, for we have no means of knowing 
where the color sensing substances are except by the responses 
aroused. With intensive stimuli, it will be remembered, we 
have found the limits of sensitivity to red, blue, and yellow to CHROMATIC THRESHOLDS OF SENSATION 
162 
coincide with the limits of white light vision, but the limits 
of sensitivity to green fall far short of this even with the great- 
est spectrum intensities we have as yet been able to obtain. 
5. The constancy in the cancelling proportions of the 
paired colors from center to periphery of the retina in all 
meridians as contrasted with the great irregularity in the 
distribution of sensitivity, obviously presents a problem to 
theories of the Hering type; for while there is not and should 
not necessarily be in terms of theory an equivalence in the 
power to arouse sensation and to cancel the complementary 
color, some degree of constancy of relation between the two 
functions might be expected. Perhaps the easiest solution 
is to be found in the conception that more than one level of 
activity is involved in the process of arousing sensation and 
that the locus of the deficiencies which cause the irregularity 
in the distribution of sensitivity to the paired colors is pos- 
terior to the level at which the cancelling action takes place. 
While an explanation of this type meets with less inertia of 
acceptance, perhaps, for the occasional and sporadic de- 
ficiency such as the small areas of the Schumann type in the 
peripheral retina (16), than for deficiencies and anomalies of 
the order of magnitude here considered, still the need for it 
or some similar concept to explain these anomalies and de- 
ficiencies is no less insistent. Even the extensive deficiency in 
the sensitivity to green noted above is contradictory to the 
concept of paired processes only on the assumption that the 
deficiencies which affect sensitivity may occur at only one 
functional level; for again it may be that the deficiencies 
which prevent the green stimulus from arousing sensation 
are posterior to the level of the cancelling action. If this 
were true, the cancelling proportions between the paired 
colors could be constant from the center to the periphery of 
the retina even though the sensitivity to one of the colors 
had fallen off a great deal or disappeared entirely, as seems to 
be the case when either a partial or full spectrum gray is 
sensed as colorless from the center to the periphery of the 
retina without any change in the composition of the stimulus 
to compensate for the extensive deviation from regularity in 163 
C. E. FERREE AND GERTRUDE RAND 
the distribution of sensitivities. As in the previous paper, 
however, our comments with reference to theory are meant 
to be only tentative and suggestive (17). Our purpose has 
been primarily to call attention to the lack of adequate 
explanatory concepts to meet the needs of our growing know- 
ledge of the visual phenomena. Theories whose especial 
fitness is for the explanation of the fundamental facts of 
positive sensation, the after-image, and contrast can scarcely 
be considered as final and complete. 
BIBLIOGRAPHY 
I. Hess, C. Ueber den Farbensinn bei indirectem Sehen. A.f.O., 1889, 35 (4), 
pp. 1-62. 
2. Hering, E. Ueber die Hypothesen zur Erklarung der peripheren Ferbenblind- 
heit. A.f.O., 1889, 35 (4), pp. 63-83. 
3. Fick, A. Zur Theorie der Farbenblindheit. Arbeiten aus dem physiol. Laborat. 
der Wiirzburger Hochschule, 1873, pp. 213-217. 
4. Leber, T. Ueber die Theorie des Farbenblindheit und uber die Art und Weise, 
wie gewisse, der Untersuchung von Farbenblinden entnommene Einwande gegen 
die Young-Helmholtz' sche Theorie sich mit derselben vereinigen lassen. Klin. 
Monatsblatter f. Augenheilk., 1873, 11, pp. 467-473. 
5. Helmholtz, H. Handbuch der physiologischen Optik, 2d Ed., 1896, p. 373. 
6. Bull, O. Studien uber Lichtsinn und Farbensinn. A.f.O., 1881, 27, pp. 54-154. 
7. Hegg, E. Zur Farbenperimetrie. A.f.O., 1892, 38, (3), pp. 145-168. 
8. Baird, J. W. The Color Sensitivity of the Peripheral Retina. Carnegie Insti- 
tution of Washington, 1905, pp. 80. 
9. Rand, G. The Factors the Influence the Sensitivity of the Retina to Color: a 
Quantitative Study and Methods of Standardizing, Psychol. Rev. Monog., 1913, 
15, pp. 28-31. 
IO. Ibid., p. 65. 
11. Baird, J. W. Op. cit., p. 61. 
12. Rand, G. Op. cit., pp. 97-110. 
13. Hering, E. Zur Lehre vom Lichtsinne. Wien, 1878, pp. 108-109. 
14. Hering, E. Zur Erklarung der Farbenblindheit aus der Theorie der Gegenfarben. 
Lotos, Jahrbuch fur N aturwissenschaft, 1880, 1, pp. 81-82. 
15. Ibid., p. 77. 
16. Ferree, C. E., and Rand, G. Some Areas of Color Blindness of an Unusual 
Type in the Peripheral Retina. Jour, of Exper. Psychol., 1917, 2, pp. 295-304. 
17. Ibid., pp. 300-304. [Reprinted from The Psychological Review, Vol. 27, No. 1, January, 1920.] 
THE ABSOLUTE LIMITS OF COLOR SENSITIVITY 
AND THE EFFECT OF INTENSITY OF LIGHT 
ON THE APPARENT LIMITS 
BY C. E. FERREE AND GERTRUDE RAND 
Bryn Mawr College 
Introduction 
In describing a general plan of investigating the chro- 
matic sensitivity of the peripheral retina in an earlier paper 
(1) the following were mentioned as two of the problems 
which we wished to take up: («) a point to point determination 
of comparative sensitivities to the different colors from the 
center to the periphery, and (^) an investigation of the limits 
of sensitivity. The former of these problems has been made 
the subject of a recent paper (2). The latter will be treated 
of here. 
The investigation of the limits of sensitivity may be 
considered from two points of view. As indicated in the 
first of the papers referred to above, (a) it may be made a 
part of the investigation of the comparative sensitivities of 
the peripheral retina to the different colors; and (^) it may 
be considered more specifically in relation to points of theory. 
In the former case the limits should be obtained with stimuli 
equalized in energy. The results will then represent positions 
on the retina at which the stimuli for one of the intensities 
which it is possible to employ have the same or nearly the 
same threshold value.1 In the latter case the problem con- 
1 Strictly speaking the threshold value may be considerably less at this point 
than the intensity of the stimulus employed, because the stimulus may be increased 
much above the threshold value in the far periphery of the retina without changing the 
limits by a detectable amount. That is, the stimulus value of the just noticeable 
1 2 
C. E. FERREE AND GERTRUDE RAND 
sidered in relation to its historical development divides into 
two,-a determination of the relative or apparent limits and a 
determination of the absolute limits. In the second of the 
papers referred to above it was shown that the determination 
of the apparent limits was given an undue importance in 
relation to theory by Hess and his followers because of their 
failure to realize that great irregularity and not uniformity 
characterizes the decrease of sensitivity from the center to 
the periphery of the retina. The details of that demonstra- 
tion need not be repeated here. The determination of the 
absolute limits of sensitivity, however, does sustain an im- 
portant relation to theory, especially to theories of the 
paired process type; for if it be found that sensation can 
be aroused farther out from the center of the retina for one 
of the paired colors than for the other, that fact must tell 
against the theory unless some supplementary concept is 
provided to explain the discrepancy. For one thing we have 
undertaken, therefore, to determine the limits of sensitivity 
with stimuli any further increase in the intensity of which 
tends to decrease rather than to increase the chromatic 
component of the response. For another we have determined 
the effect of a given range of variation of intensities on the 
apparent limits. Our reason in part for doing this was to 
supplement at higher intensities the work of former papers (3) 
in which we called attention to the large variations that are 
required in any of the factors influencing the chromatic 
response (intensity being the most effective of these) to 
change the limits of sensitivity as much as I degree especially 
when a certain degree of excentricity has been reached, 
pointing out in particular with regard to the work of previous 
writers, (a) the importance of taking into account deviations 
of 1-3 degrees from coincidence of limits when conclusions 
with regard to comparative sensitivities are to be drawn 
from the results, and (&) the futility of making a brightness 
equalization of the stimuli, with its attendant disadvantages, 
difference in limits is much greater than the stimulus value of the difference limen of 
intensity. In other words, a given point in the peripheral retina may be considered 
the limit for a range of stimulus intensities, varying in magnitude with its degree of 
excentricity. ABSOLUTE LIMITS OF COLOR SENSITIVITY 
3 
for the determination of the limits with lights of medium and 
high intensities and perhaps for any but intensities so low as 
to give very narrow limits. For the latter point three 
reasons may be given, (i) The brightness equation does 
not equalize the stimuli in power to arouse the chromatic 
response, the only subjective equation, so far as we can see, 
that could rightly be given a place in a determination of the 
limits of chromatic sensitivity, and this only in a determina- 
tion of whether or not the same ratio of sensitivity holds for 
the limits as for the center or other point at which the equa- 
tion was made. (2) It does not equate them in intensity 
(the equation is merely of the very selective achromatic 
response to the stimuli). And (3) so far as the effect of the 
achromatic on the chromatic component of the excitation is 
concerned (the final variable that might be considered), it 
has already been shown by one of the writers in a previous 
paper (4) that there is not enough of this effect for the colors 
ordinarily used to change the limits by a detectable amount. 
The particular bearing of the present work on this question 
is to give a clearer and more definite idea of just how much 
difference in intensity or equivalent influence is required to 
change the limits by detectable amounts in the mid and far 
periphery of the retina. In the beginning this was in fact 
our chief incentive to undertake the work. 
The Problem 
The investigation was given the following form. (1) An 
attempt was made to find out whether by means of our 
spectroscopic apparatus, which was designed especially to 
give high intensities of light, stimuli could be obtained which 
could be sensed as color to the limits of white light vision. 
(2) The effect on the extension of the limits of sensitivity of 
varying the stimuli through quite a wide range of intensities 
was investigated. And (3) the determination of the limits 
was made in 16 meridians with all of the lights made equal 
in energy to the blue of the prismatic spectrum employed 
and with 1/32 of this amount of energy. 4 
C. E. FERREE AND GERTRUDE RAND 
Conditions under which the work was done 
The conditions under which the work was done fall 
under five headings: (i) the wave-lengths of light employed 
and the means used of getting greater purity of light than is 
found in the prismatic spectrum; (2) the energy content of 
the stimuli used and the method of measurement; (3) the 
control of the brightness of preexposure and surrounding 
field; (4) the control of the general illumination of the optics 
room; and (5) the method of rendering the amount of light 
entering the eye independent of variations in the size of the 
pupil, without the use of an artificial pupil. These conditions 
are so nearly identical with those used in the work of the 
immediately preceding papers that at the request of the 
Editor space has not been taken for their repetition in the 
present paper. For a description of the conditions the 
reader is referred to 'Chromatic Thresholds of Sensation 
and their Bearing on Color Theory, Part I.,' this journal, 
1919, 26, pp. 18-25. 
The stimulus used was the circular aperture of the cam- 
pimeter, 15 mm. in diameter, filled with light by the focusing 
lens. At a distance of 25 cm. from the pupil of the eye, on 
which the light from the objective slit of the spectroscope was 
focussed, this aperture subtended a visual angle of 30 26'. The 
time of exposure was 1 sec. and the interval between exposures 
varied between 3-5 min. depending on circumstances and 
the need for precautionary measures. If the stimulus was 
sensed in its proper color at any time during the 1 sec. in- 
terval of exposure, the retina was called color sensitive at 
that point. (At the limits of white light vision the red stimu- 
lus, for example, of the intensity used was sensed as a tint of 
red.) The field in the 16 meridians was always mapped for 
one color before the work on another color was begun. 
Systematic results were obtained for all of the points of 
the work for only one observer. This was the observer 
whose results were published in the immediately preceding 
papers: 'Chromatic Thresholds, etc., Parts I. and II.' For 
data with regard to the various ways in which the normality of 
both the chromatic and achromatic sensitivity of this observer, ABSOLUTE LIMITS OF COLOR SENSITIVITY 
5 
central retina, and chromatic sensitivity, peripheral retina, 
has been confirmed, the reader is referred to pp. 26-32 of the 
first of the papers noted above. Data on additional points, 
important in a general specification of the ocular condition 
of the observer, have also been published in various places: 
e.g., on the dioptric or refraction condition and power to 
sustain acuity in Trans. Ilium. Eng. Soc., 1915, 10, p. 1128, 
and in other papers by us on lighting in relation to the eye; 
on muscle strength, muscle balance, muscle lag, photopic 
acuity, near point, range of accommodation, and refraction 
condition (more recent), Trans. Amer. Ophthal. Soc., 1918, 
66, pp. 142-163; and on scotopic acuity and amount and 
rapidity of scotopic adaptation, Trans. Amer. Ophthal. Soc., 
1919, 67 (in press). The more important points such as 
the coincidence of the limits of red, yellow and blue with 
the limits of white light vision; the narrower limits for green; 
the interlacing of limits for stimuli of medium intensity of 
equal energy, or of the same general order of intensity; and 
the large differences in amount of light required to change 
the limits of sensitivity by a detectable amount in the mid 
and far peripheral portions of the retina, have been con- 
firmed in a less detailed and systematic way by one or more 
check observers. 
Results 
The following results were obtained. (1) It was quite 
easy to obtain an intensity of light for the red, yellow and 
blue wave-lengths that could be sensed to the limits of white 
light vision. In fact these wave-lengths in the spectrum 
employed were considerably above the threshold at the 
limits of white light vision in the sixteen meridians investi- 
gated. The limits of the green of this spectrum, however, 
fell far short of the limits for white light; nor could the zone 
of sensitivity be widened as much as 1 degree by increasing 
the current in the Nernst filament from 0.6 to 0.8 ampere. 
The energy entering the eye from the spectrum of the Nernst 
filament operated by 0.6 ampere of current with the width 
of collimator slit employed was for the red 9096.639 x io~10 6 
C. E. FERREE AND GERTRUDE RAND 
watt; for the yellow, 4065.624 x io-10 watt; for the green, 
1562.388 xio~10 watt; and for the blue, 882.025 x io-10 watt. 
The energy value of the threshold at the limits of white 
light vision in the nasal meridian, for example, was for the 
red 277.836 x io~10 watt; for the yellow, 268.95 x to-10 watt; 
and for the blue, 264.368 x io~10 watt. The intensity of 
light for these colors in the 0.6 ampere spectrum was, there- 
fore, strongly supra-liminal at the limits of white light 
vision, as is stated above. In the 0.6 ampere spectrum, 
the energy of the green light, it will be noted, was greater 
than the energy of the blue, but less than the energy of the 
red and yellow. It was, however,- nearly six times as great 
as the threshold value of these colors at the limits of white 
light vision. Moreover, when the current was raised to 0.8 
ampere this value was considerably increased, but there 
was still no detectable extension of the limits. Since 
then the sensitivity to green at the center of the retina 
and for several degrees towards the periphery is approxi- 
mately the same as to blue and considerably greater than 
to red and to yellow, and since so large an increase in the 
energy value of the stimulus made no detectable difference 
in the limits and any further increase lessened rather than 
increased the chromatic component of the response, it seems 
highly improbable that the limits could by any means what- 
soever be extended the 20-35 degrees needed to make them 
coextensive with the limits of white light vision. It seems 
fairly certain, therefore, that while the far periphery of the 
retina is only deficient in its chromatic sensitivity to red, 
yellow and blue, the blindness to green for the observers used 
is absolute. 
(2) In the investigation of the effect of changes of in- 
tensity on the limits of sensitivity eight intensities were 
used, sustaining to each other the following relations: I, 1/2, 
1/4, 1/8, 1/16, 1/32, 1/64 and 1/128. The highest intensities 
were taken respectively from the prismatic spectrum of a 
Nernst filament operated by 0.6 ampere of current and 
from a spectrum made equal in energy to the blue of this 
spectrum. These spectra will be designated as Spectrum ABSOLUTE LIMITS OF COLOR SENSITIVITY 
7 
A and Spectrum B. The reductions were produced by means 
of an aluminum sectored disc of 180, 90, 45, etc., degrees 
open sector. The energy values of the different intensities 
of light, as has already been stated, were obtained by radio- 
metering the highest intensities and computing the lower 
from the simple law of the disc. It had been our intention 
to make the investigation systematically with the eight dif- 
ferent intensities in the sixteen meridians of the retina. 
However, for the purpose of the present paper a briefer 
substitute plan has been adopted. A preliminary investi- 
gation was made with the eight intensities of Spectrum A 
in two meridians of the retina, the nasal and the temporal, 
which meridians represent opposite extremes with regard to 
breadth of zone of sensitivity, in order to get some idea of 
the amounts of reduction that would be needed to be effective 
in changing the limits. It was found, for example, that a 
reduction of the red light to 1/32 of its value at intensity A 
was not sufficient to narrow the limits in the nasal and 
temporal meridians, the meridians designated in the tables 
and charts as 90 degrees. At this value the stimulus was 
still slightly supra-liminal in these meridians at the limits of 
white light vision. This amount of reduction, however, was 
sufficient to narrow the limits for the other stimuli by quite 
considerable amounts. Also a further investigation showed 
that it was enough to narrow the limits for red in 12 out of 
the 16 meridians employed. It was decided, therefore, to 
make the final determinations in the 16 meridians with the 
full intensities A and B and with 1/32 A and B. The amount 
of narrowing for the yellow of the prismatic spectrum in the 
different meridians produced by this reduction ranged from 
3-11 degrees; for the green from 5-17 degrees; for the blue, 
from 10-18 degrees; and for the red, from 0-8 degrees. For 
the equal energy spectrum the amount of narrowing for the 
yellow ranged from 5-18 degrees; for the green, from 5-15 
degrees; for the blue, from 4-18 degrees; and for the red from 
3-25 degrees. 
(3) In case of the equal energy spectrum of the higher 
intensity, all of the lights with the exception of the green 8 
C. E. FERREE AND GERTRUDE RAND 
were seen in their proper color to the limits of white light 
vision in each of the 16 meridians. Made equal in energy 
to the blue of the prismatic spectrum (882.025 x io-10 watt) 
the red and yellow were considerably less in energy value than 
was the green of the prismatic spectrum, still the red and yellow 
were sensed to the limits of white light vision while the green 
which represented a considerably greater amount of energy 
fell short of those limits by amounts varying from 20-35 
degrees in the different meridians. There can be no reason- 
able doubt, we believe, that the difference found here repre- 
sents an actual difference in sensitivity. It obviously can 
not be attributed to the relative intensities of the stimuli 
employed. 
Landolt has also investigated the effect of high intensities 
on the extension of the limits of sensitivity. Writing of this 
work (5), he says: "In ein absolut dunkles Zimmer fiel nur 
durch eine kleine Offnung im Finsterladen directes Sonnen- 
licht. Dieses wurde auf das ausserste Ende des Perimeter- 
bogens gelenkt. Wahrend wir unser Auge ins Centrum des 
Bogens setzen, bracht man in die kleine, intensive beleuchtete 
Stelle farbige Papiere von moglichster Intensitat der Farbung. 
Nun bewegtet sich das Auge langsam vom entgegengesetzen 
Ende des Bogens nach Scheitelpunkte zu und es zeigte sich 
dabei, dass wenigstens mit der innern Netzhautpartie alle 
Farben schon bei 900 erkannt wurden. Die Grosse des 
Objectes betrug weniger als 1 cm2. 
"Ais dieselben Priifungen auch mit Spectralfarben zu 
machen, entwarfen wir ein Sonnenspectrum im sonst dunkeln 
Zimmer und liessen es durch eine achromatische Linse auf 
einen Ende des Perimeters befindlichen Schirm fallen. 
Dieser hatte eine verandliche Spalte, mittelst welcher man 
die einzelnen Farben aus dem Spectrum isolieren konnte. 
Wahrend wir nun wiederum nach langer Adaptation, und 
bei verbundenem zweiten Auge das eine Ende des Bogens 
fixierten, wurde von einem Assistenten irgendeine Farbe des 
Spectrums auf die Spalte gelenkt, und wir drehten nun, 
unter stehter Fixation unserer Fingerspitze, welche sich auf 
dem Bogen bewegte, das Auge allmahlig der Farbe entgegen. ABSOLUTE LIMITS OF COLOR SENSITIVITY 
9 
Es zeigte sich auch hier wiederum dass alle Farbe schon bei 
900 erkannt werden, wenn sie intensiv genug sind." 
Landolt's investigation was made, it will be noted, in a 
dark room while ours was made in a light room. We have 
Table I 
A. The Effect of Intensity of Stimulus on the Limits of Sensitivity, 
Prismatic Spectrum 
In this table are given the results of a preliminary investigation in two representa- 
tive meridians to show how much reduction is needed to produce a significant change 
in the limits of sensitivity. Eight intensities of stimulus were used: A, 1/2A, 1/4AJ 
1/8A, etc. 
Meridian Investi- 
gated 
Stimulus 
Limits of Sensitivity for 
Inten- 
sity 
A 
Inten- 
sity 
u 
Inten- 
sity 
}A 
Inten- 
sity 
iA 
Inten- 
sity 
AA 
Inten- 
sity! 
AA 
Inten- 
sity 
b^A 
Inten- 
sity 
r^sA 
Nasal 
Red 
(670 mm) 
Yellow 
(581 mm) 
Green 
(5 22 mm) 
Blue 
(468 mm) 
Red 
(670 mm) 
Yellow 
(581 mm) 
Green 
(5 22 mm) 
Blue 
(468 mm) 
92 
92 
69 
92 
6l 
6l 
45 
61 
92 
92 
69 
92 
61 
61 
45 
61 
92 
92 
69 
87 
61 
61 
45 
56 
92 
92 
66 
83 
61 
61 
42 
45 
92 
91 
63 
79 
61 
55 
34-5 
43-5 
92 
89 
62 
78 
6l 
47 
32.5 
43 
88 
88 
59 
77 
46 
46 
3° 
43 
86 
88 
56 
76 
44 
45 
29 
43 
Temporal. . .u . 
B. The Energy Values of the Stimuli Used 
Total energy of light at campimeter opening and at eye 
(watt X io-10) 
Intensity 
Red (670^/x) 
Yellow (sSxpg) 
Green (522/xju) 
Blue (468^) 
A1 
9096.639 
4065.624 
1562.388 
882.025 
1 The energy values of 1/2, 1/4, 1/8, 1/16, 1/32, 1/64 and 1/128 A may be obtained 
by dividing the above values by the appropriate factor. 
The energy density at the campimeter opening (watt per sq. mm.) may be ob- 
tained by multiplying the above values by 0.005659; the energy density at the eye, 
by multiplying them by 0.303. 
not as yet had opportunity to repeat the work of the present 
paper with the dark adapted eye. However, determinations 
somewhat rougher and less detailed than those described 10 
C. E. FERREE AND GERTRUDE RAND 
here have sufficed to show that for our observers the far per- 
iphery of the retina is color-blind to green also with the dark 
adapted eye. With reference to the relative insensitivity 
of the peripheral retina to green, it may further be noted 
that in our results with the Hering papers with a different 
set of observers the limits for green fell much nearer to the 
center of the retina than for red, yellow and blue. The 
results represented in Fig. 5, for example, were taken from 
this series of observations. That the limits for green are 
narrower than for red, yellow and blue with stimuli of the 
same order of intensity has, moreover, been verified many 
times in the work of our undergraduate laboratory. 
In Table I., A, are given the results of the preliminary 
investigation in the nasal and temporal meridians to find 
out whether an intensity of light may not be gotten sufficiently 
high to make the limits of color sensitivity coincide with the 
limits of white light vision, and once this intensity is attained 
how much reduction is needed to produce a significant 
narrowing of the limits. We have already indicated in this 
and in previous papers the large changes of intensity that 
are needed to change the limits by a significant amount when 
a certain degree of excentricity has been reached. How very 
large these changes have to be for the far periphery of the 
retina is shown in this table. 
In Table I., B, is given a specification of the energy values 
of the stimuli used in making the determinations represented 
in Table I., A. Four energy values may perhaps be consid- 
ered of importance for each determination: The total value 
at the campimeter opening, the density per sq. mm. at the 
campimeter opening, the total energy entering the eye, and 
the density per sq. mm. at the eye. For the sake of brevity, 
however, only one of these values is given in the table, 
namely, the total energy entering the eye; and the factors 
needed to convert this value into density at the eye and at 
the campimeter opening are appended in a footnote. Since 
all of the light from the campimeter opening is focused into 
the image on the pupil, the figures expressing the total energy 
at the eye and at the campimeter opening are the same. ABSOLUTE LIMITS OF COLOR SENSITIVITY 
11 
The most important of the four specifications noted are 
probably the total amount of light entering the eye and the 
density at the campimeter opening. The latter value, for 
example, sustains a fixed but unknown ratio to the density 
of the image formed on the retina. 
Table II 
The Brightness Values of Preexposure and Surrounding Field 
In this table are given the brightness values of preexposure and campimeter 
screen in candlepower per square inch1 for the determination of limits given in Table I. 
In all cases in which it was possible the brightness of the preexposure and campimeter 
screen was made equal to that of the stimulus at the limits of sensitivity. 
Meridian 
Stimulus 
Brightness Value of Preexposure and Campimeter Screen for 
Intensity 
A 
Intensity 
JA 
Intensity 
iA 
Intensity 
iA 
Intensity 
AA 
Intensity 
AA 
Intensity 
th A 
Intensity 
isk A 
Nasal.... 
Red 
(670 mm) 
Yellow 
0.05088 
O.O5O88 
0.05088 
0.05088 
O.O5O88 
0.05088 
O.O3O93 
O.O2Il6 
(581 mm) 
0.05088 
O.O5O88 
0.05088 
0.05088 
0.05088 
0.05088 
O.O3663 
0.02686 
Green 
(522 mm) 
0.05088 
O.O5O88 
0.05088 
O.O3663 
O.O3663 
0.02686 
0.02116 
O.OI384 
(468 mm) 
0.05088 
0.05088 
O.O5665 
0.02686 
0.01262 
O.OII4O 
O.OO643 
O.OO578 
Temporal 
Red 
(670 mm) 
0.05088 
O.O5O88 
0.05088 
0.05088 
O.O5O88 
0.05088 
0-03093 
O.OI79I 
Yellow 
(581 mm) 
0.05088 
O.O5O88 
0.05088 
0.05088 
O.O5O88 
0.05088 
0.02686 
0.02116 
(522 mm) 
0.05088 
O.O5O88 
0.05088 
O.O3663 
0.02686 
O.OI384 
O.OII4O 
O.OII4O 
Blue 
(468 mm) 
0.05088 
0.05088 
O.O3663 
O.O3O93 
0.01262 
O.OII4O 
O.OO716 
O.OO643 
In Table II. are given the brightness values of the pre- 
exposure and campimeter screen for the work represented in 
Table I. As stated earlier in the paper the preexposure and 
the campimeter screen were selected from the Hering series 
of standard papers. In case of the higher intensities of light 
used, No. I of this series (the standard white, coefficient of 
reflection about 75 per cent.) reflecting the light of the room 
was not as bright as the stimulus light. These cases may be 
identified in this and the following tables by the brightness 
value of this paper under the illumination of the room, namely, 
0.05088 cp. per sq. in. 
1 The above values may be converted into millilamberts by multiplying by 486.8. 12 
C. E. FERREE AND GERTRUDE RAND 
In Table III. are given the limits of sensitivity in 16 
meridians of the retina for the highest intensity of light used 
for the work of Table I., intensity A, and for 1/32 A, an 
intensity representing the order of reduction needed for all 
of the colors to produce any considerable narrowing of the 
Table III 
The Effect of Intensity of Stimulus on the Limits of Sensitivity, Prismatic 
Spectrum 
In this table are given the limits of sensitivity in 16 meridians of the retina for 
intensity A and 1/32A of Table II. The upper vertical meridian is numbered o° 
and the lower vertical 1800. Reading down to right or left they are 250, 450, 700, 
900, 1100, 135°, 1550, and 1800. 
Meridian Investi- 
gated 
Limits of Sensitivity for 
Intensity A 
Intensity 
Red 
(670^) 
Yellow 
(58w) 
Green 
(522^) 
Blue 
(468/xM) 
Red 
Yellow 
(581^1 
Green 
(522^) 
Blue 
(468gp) 
0° 
65 
65 
36 
65 
64 
57 
29 
50 
Nasal 250 
86 
86 
49 
86 
83 
79 
36 
72 
45° 
9° 
90 
52 
9° 
89 
83 
38 
76 
700 
92 
92 
67 
92 
92 
86 
59 
77 
900 
92 
92 
69 
92 
92 
89 
62 
78 
IIO° 
9i 
9i 
68 
9i 
90 
87 
59 
78 
135° 
88 
88 
61 
88 
87 
84 
49 
75 
155° 
86 
86 
47 
86 
84 
78 
35 
67 
1800 
57 
57 
36 
57 
53 
47 
27 
42 
Temporal 250 
65 
65 
44 
65 
64 
55 
3i 
52 
45° 
65 
65 
50 
65 
57 
56 
33 
5i 
700 
62 
62 
46 
62 
55 
50 
35 
44 
900 
61 
61 
45 
61 
61 
47 
32.5 
43 
IIO° 
58 
58 
37 
58 
57 
49 
3i 
48 
135° 
55 
55 
32 
55 
55 
48 
27 
20 
*55° 
54 
54 
30 
64 
So 
43 
25 
4i 
limits at the extreme periphery of the retina. The 16 
meridians used are designated as follows. The upper vertical 
meridian is marked o and the lower vertical 180 degrees. 
Beginning with o and reading down to left or right they are 
o, 25, 45, 75, 90, no, 135 and 180 degrees. The specification 
of the energy of the stimuli at intensity A and 1/32 A is 
given, it will be noted, in Table I., B. For all of the stimuli 
at intensity A, No. 1 of the Hering series of papers was used 
as preexposure and campimeter screen, as has already been 
noted. This paper illuminated by the light of the room was too 
dark (0.05088 cp. per sq. in.) for all of the four colors at the ABSOLUTE LIMITS OF COLOR SENSITIVITY 
13 
limits of sensitivity. However, for the want of a suitable 
pigment surface of higher reflection coefficient it was used. 
For intensity 1/32 A, it was darker than the yellow stimulus 
Table IV 
A. The Effect of Intensity of Stimulus on the Limits of Sensitivity, Equal 
Energy Spectrum 
In this table are given the limits of sensitivity in 16 meridians of the retina for 
stimuli all made approximately equal in energy to the blue of the prismatic spectrum 
used in Table I., and for 1/32 of this intensity,-intensity B and 1/32 B. 
Meridian Investi- 
gated 
Limits of Sensitivity for 
Intensity B 
Intensity ^jB 
Red 
(670^) 
Yellow 
(sSigg) 
Green 
(522MM) 
Blue 
(468^) 
Red 
Yellow 
(58imm) 
Green 
(5230g) 
Blue 
(<68gg) 
O° 
65 
65 
34 
65 
62 
56 
29 
50 
Nasal 250 
86 
86 
47 
86 
74 
70 
34 
72 
45° 
90 
90 
50 
90 
74 
78 
36 
76 
700 
92 
92 
63 
92 
78 
85 
5° 
77 
900 
92 
92 
69 
92 
81 
87 
59 
78 
110° 
91 
9i 
64 
9i 
79 
85 
58 
78 
135° 
88 
88 
44 
88 
74 
80 
42 
75 
155° 
86 
86 
36 
86 
61 
74 
32 
67 
1800 
57 
57 
31 
57 
46 
39 
25 
42 
Temporal 250 
65 
65 
39 
65 
57 
54 
30 
52 
45° 
65 
65 
39 
65 
47 
55 
32 
5i 
7°° 
62 
62 
42 
62 
46 
48 
3° 
44 
900 
61 
61 
4i 
61 
4° 
45 
3° 
43 
IIO0 
58 
58 
36 
58 
45 
45 
29 
48 
135° 
55 
55 
28 
55 
49 
4i 
25 
40 
155° 
54 
54 
26 
54 
47 
38 
24 
4i 
B. The Energy Values of the Stimuli Used. 
Total energy of light at campimeter and at eye 
(watt X io"10) 
Intensity 
Red (670 nn) 
Yellow (581 
Green (522! 
Blue (468 
B1 
891.050 
882.510 
884.946 
882.025 
and approximately equal to the green and red. For the blue 
stimulus Nos. 9-14 of the Hering series (0.01404-0.0114 cp. 
per sq. in.) were used as needed in the different meridians. 
The photometric values for these intensities and intensity 
1 The energy value of 1/32 B may be obtained by dividing the above values by 
the appropriate factor. 
The energy density at the campimeter opening (watt per sq. mm.) may be ob- 
tained by multiplying the above values by 0.005659; the energy density at the eye, 
by multiplying them by 0.303. 14 
C. E. FERREE AND GRETRUDE RAND 
B and 1/32 B have not been given in detailed tabular form 
because of the large number of repetitions that occur. 
In Table IV., A, are given the limits of sensitivity in 
16 meridians of the retina for the four stimuli all made equal 
in energy to the blue used in Table I. and for 1/32 of this value. 
These values are, as we have already indicated, designated in 
Fig. 1. The effect of intensity of stimulus on the limits of sensitivity, prismatic 
spectrum. In this chart are represented the limits of sensitivity for intensity A of 
Table L: red 9096.639, yellow 4065.624, green 1562.388, and blue 882.025 watt X io-10. 
the table as intensity B and 1/32 B. In Table IV., B, are 
given the energy values of the stimuli used for Table IV., A. 
For the higher intensity of these stimuli, intensity B, No. 1 
of the Hering series of papers (0.05088 cp. per sq. in.) was 
used for the preexposure and campimeter screen. Again it 
was darker than all of the four colors at the limits of sensitivity. 
For intensity 1/32 B it was slightly darker than the yellow. ABSOLUTE LIMITS OF COLOR SENSITIVITY 
15 
For the green of this intensity the no. 2 gray of this series was 
used (0.0366 cp. per sq. in.); for the red nos. 10-14 (0.01384- 
0.0114 cp. per sq. in.) varying for the different meridians; 
and for the blue, nos. 7-14 (0.01791-0.0114 cp. per sq. in.). 
A graphic representation of the results of Table III. is 
given in Figs. 1 and 2. In Fig. 1 are shown the limits of 
Fig. 2. The effect of intensity of stimulus on the limits of sensitivity, prismatic 
spectrum. In this chart are represented the limits of sensitivity for intensity 1/32 A 
of Table I.: red 284.27, yellow 127.051, green 48.825, and blue 27.563 watt X io-10 
sensitivity to the four stimuli in the 16 meridians for the inten- 
sities represented in the prismatic spectrum A. The limits 
for the red, yellow and blue stimuli at this intensity are, 
it will be remembered, coincident with the limits of white 
light vision. All four limits may be represented, therefore, 
by a single tracing, an unbroken line in black. The limits for 
green are represented by a broken line. In Fig. 2 are 16 
C. E. FERREE AND GERTRUDE RAND 
represented the limits of sensitivity for the four stimuli at 
the intensities represented in the prismatic spectrum, 1/32 A. 
In this case the zone of sensitivity to blue is outlined by an 
unbroken line and the zones for the other colors by broken 
lines as indicated in the figure. An inspection of this figure 
will show that the degree of excentricity of the limits is in the 
order of the intensity of the stimuli. In discussing the 
Fig. 3. The effect of intensity of stimulus on the limits of sensitivity, equal 
energy spectrum. In this chart are represented the limits of sensitivity for intensity 
B of Table IV.: red 891.05, yellow 882.51, green 884.946 and blue 882.025 watt X io-1*. 
significance of the crisscrossing or interlacing of the limits 
obtained with the Hering pigment papers in previous work, 
this is what was pointed out would occur if there were a 
significant difference in the intensity of the stimuli. That is, 
if the zone of sensitivity to red, for example, is in one meridian 
wider and in another narrower than to green, etc., it can not ABSOLUTE LIMITS OF COLOR SENSITIVITY 
17 
be due to any difference in the intensity of the stimuli; for 
such a difference, if significant, would make one zone con- 
sistently wider or narrower than the other in all meridians. 
A graphic representation of the results of Table IV. is 
given in Figs. 3 and 4. In Fig. 3 are shown the limits of 
sensitivity to the four stimuli in the 16 meridians for the 
intensities represented in the equal energy spectrum B. 
Fig. 4. The effect of intensity of stimulus on the limits of sensitivity, equal energy 
spectrum. In this table are represented the limits of sensitivity for intensity 1/32 B 
of Table IV.: red 27.845, yellow 27.578, green 27.655, and blue 27.563 watt X io-10. 
Again the limits for the red, yellow and blue stimuli coincide 
with the limits of white light vision and are represented by a 
single tracing, the unbroken line in black. The limits for 
green are represented by a broken line. In Fig. 4 are shown 
the limits for the four stimuli at the intensities represented 
in the equal energy spectrum 1/32 B. With regard to this 18 
C. E. FERREE AND GERTRUDE RAND 
figure the following points may be noted. (1) With stimuli 
of equal energy the limits of no one of the colors, red, yellow 
and blue, are consistently wider than the others. That is, 
their limits are characterized by frequent crisscrossing or 
interlacing. The limits for all three colors, however, are 
consistently wider than for green. And (2) Fig. 4 sustains a 
Fig. 5. The limits of sensitivity to red, yellow, green and blue of the Hering 
series of pigment papers, intensity of illumination, vertical component, 390 foot-candles. 
somewhat striking general similarity to the charts obtained 
for the Hering pigment papers. One of these showing the 
limits with a surrounding field and preexposure of the bright- 
ness of the colors employed is given in Fig. 5. While no 
conclusion can be drawn from this similarity with regard to 
the relative energies of the wave-lengths dominantly reflected 
by these papers; still it suggests that they may all, roughly 
speaking, be somewhere near the same order of value, at ABSOLUTE LIMITS OF COLOR SENSITIVITY 
$ 
19 
least much more nearly so than are these colors in the pris- 
matic spectrum. The red, yellow, green and blue of the 
prismatic spectrum gave, it will be remembered, rather 
widely concentric, not crisscrossing limits. 
In this general connection it may be of interest also to 
note the close correlation which obtains between the results 
of this investigation and those of the previous investigation 
of the sensitivity of the peripheral retina by the threshold 
method. That is, wherever the thresholds are found to be 
low the limits are found to be wide, and conversely wherever 
the thresholds are high the limits are found to be corre- 
spondingly narrow. Some interesting results follow from 
this. For example, in a given meridian the threshold curve 
for a given color is found to be very irregular, rising in some 
places slowly, in others quickly, and still in others dropping 
and rising again. These fluctuations in the curve are, 
moreover, different in the different meridians. This means, 
of course, that the shape of the zones of sensitivity for this 
color should change with the intensity of the stimulus em- 
ployed, which is found to be the case. Furthermore, in the 
same meridian the threshold curves for the different colors 
differ from each other widely in the direction and amount of 
the irregularity: and this difference in turn varies from 
meridian to meridian. The result of this is that a crisscrossing 
or interlacing of limits must take place whenever stimuli of 
such relative intensities are used that the limits are of the 
same general order of excentricity. In other words, as was 
pointed out in our discussion of this phenomenon in earlier 
papers, crisscrossing can mean only that there is a lack of 
uniformity in the relative sensitivity to the different colors in 
the different meridians. For example, when it occurs in the 
limits for blue and yellow, it indicates that the ratio of sen- 
sitivity to blue and yellow changes in passing from meridian 
to meridian. In short any investigation at all comprehensive 
either of the thresholds or limits of sensitivity shows that 
striking irregularity and not uniformity characterizes the 
distribution of chromatic sensitivity in the peripheral retina. 
This is in direct opposition, it will be remembered, to the 20 
C. E. FERREE AND GERTRUDE RAND 
claim made by Hess (6) that constancy of ratio of sensitivity 
to the paired colors prevails throughout the retina which 
claim, it will be remembered, was advanced by Hering (7) in 
support of his own theory and in refutation of Fick's (8) 
and Leber's (9) modifications of the Helmholtz theory to 
explain the color blindness of the peripheral retina. So far 
as we are able to determine no one intensity or set of conditions 
will give coincidence of limits in all meridians for any two 
colors inside the limits of white light vision. 
In conclusion it may not be out of place to point out the 
bearing of these results on the work of the clinic. In the prac- 
tice of perimetry as applied to diagnosis it is. commonly accepted 
that the field of vision for the normal eye may be divided con- 
centrically from periphery to center in the following order: 
white light and form, blue, red and green. It is obvious from 
the fore-going results (a) with stimuli taken from the prismatic 
and equal energy spectra and (b} from the effects obtained by 
varying the intensity of the stimuli that the responsibility for 
such a rating of the color fields rests for the greater part with 
the relative intensities of the pigment stimuli used in the 
work of the clinic. That is, the limits of sensitivity to 
red, yellow, blue and white light for stimuli of high intensities 
are coincident; for stimuli of lower intensities taken from the 
prismatic spectrum they are rather widely concentric; and 
for stimuli of equal energies of the order of intensity of 
27.563 x io-10 watt they are interlacing. 
Another feature of interest is the claim that has been 
made by certain clinicians, but not generally accepted, we 
believe, that the interlacing of the limits for blue and red 
indicates a pathological disturbance in the relative distribu- 
tion of sensitivities. While we are not disposed to dispute 
this conclusion because of a too meager knowledge of all of 
the data that should be taken into consideration in its evalu- 
ation, still we do think it fair to note that pathological 
disturbances are only one set of factors that may contribute 
to such a result and that widely different results may be 
gotten with the same eye with no greater differences in the 
test conditions than may occur from time to time in the same ABSOLUTE LIMITS OF COLOR SENSITIVITY 
21 
clinic or laboratory unless a clear understanding is had of 
the factors which affect the apparent powers of response of 
the peripheral retina and adequate means are exercised for 
their control. These factors are, so far as we are able to list 
them, composition, area, intensity of the stimulus and dura- 
tion of the stimulation, breadth of pupil, the intensity of the 
general illumination and the state of adaptation of the retina, 
and the brightness of the preexposure and surrounding field. 
Obviously if the determination of the apparent limits is to 
be given clinical significance the work should be done under 
conditions of work which have been most carefully standard- 
ized, for the apparent limits are a resultant of these conditions 
as well as of the actual distribution of sensitivities. 
The degree of importance that is attributed by at least 
one clinician to the absolute and relative distribution of 
sensitivities over the retina may be indicated by the following 
quotation from a recent work on perimetry. " Contraction 
of the form fields shows the degree of disease of the visual 
tract. It is better evidence of the real condition of the 
visual path than an ophthalmoscopic study can possibly 
furnish. The evidence is minute and analytical. The color 
fields and color changes moreover furnish a more delicate test 
in the early stages of the disease and at times furnish a clue 
to the seat of the trouble before an appreciable change has 
taken place in the form field" (10). 
Summary of Results and Conclusions 
The more significant features of the above results may be 
summarized briefly as follows: 
I. The far periphery of the retina is not blind to red, 
blue and yellow. It is merely deficient in sensitivity to 
these colors. That is, with stimuli of sufficient intensity 
the limits of red, blue and yellow coincide with the limits of 
white light vision. The blindness to green, however, is for 
our observers absolute. 
2. The amount of change of intensity required to produce 
a detectable change in the apparent limits of sensitivity in 
the more remote parts of the retina is very great. This C. E. FERREE AND GERTRUDE RAND 
amount changes very irregularly from center to periphery of 
the retina in a given meridian and from meridian to meridian 
as might be expected from the great irregularity in the 
distribution of sensitivity in the peripheral retina. (Cf. 
'Chromatic Thresholds of Sensitivity from Center to Peri- 
phery of the Retina and their Bearing on Color Theory,' 
Psychol. Rev., 1919, 26, pp. 16-42.) 
3. Two other important phenomena may also be men- 
tioned as a result of this irregularity, (a) The shape of the 
zone of sensitivity to a given color changes with the intensity 
of the stimulus employed in making the determination. 
And (b) when stimuli of equal or of the same order of intensity 
are used the limits for red, yellow and blue are found to 
interlace or crisscross each other irregularly rather than to 
coincide in complementary pairs as was reported by Hegg, 
Hess and Baird in a more limited investigation of the retina's 
powers of response. The former phenomenon is the direct 
corollary of the difference in the rate of decrease of sensitivity 
to a given color in passing from the center to the periphery 
of the retina in the different meridians; the latter, to the 
change in the ratio of sensitivity to the different colors from 
meridian to meridian. The lack of uniformity of grading 
of function from point to point in the periphery of the retina, 
reported in this and previous papers, while striking, can 
scarcely be considered as surprising. It is in fact just what 
might be expected of those parts of a sense organ which are 
little used and poorly developed. 
4. The responsibility of the accepted clinic rating of 
limits in the order from widest to narrowest of blue, red and 
green doubtless for the greater part rests with the relative 
intensities of the pigment stimuli used in the work of the 
clinic. With stimuli of high intensity the limits for red, 
yellow and blue coincide with the limits of white light vision; 
for stimuli of lower intensities, taken from the prismatic 
spectrum, they are rather widely concentric; and for stimuli 
of equal energies of medium intensities they are interlacing. 
5. The interlacing of limits for red and blue is a normal 
result for stimuli of equal energy of medium intensities. It ABSOLUTE LIMITS OF COLOR SENSITIVITY 
23 
may not therefore be due to pathological disturbances in the 
distribution of sensitivities as has been claimed by certain 
clinicians. In all responsible work on the determination of 
the apparent limits it is obviously of great importance to 
bear in mind that the results are dependent both upon the 
actual distribution of sensitivity and the numerous factors 
which affect the apparent powers of response of the peripheral 
retina. 
Bibliography 
i. Ferree, C. E., and Rand, G. A Note on the Determination of the Retina's 
Sensitivity to Colored Light in Terms of Radiometric Units. Amer. Jour, of 
Psych., 1912, 23, pp. 328-332. 
2. Ferree, C. E., and Rand, G. Chromatic Thresholds of Sensation from Center 
to Periphery of the Retina and their Bearing on Color Theory. Psychol. Rev., 
1919, 26, pp. 16-41; 150-163. 
3. Ibid., pp. 152-153; Rand, G. The Factors that Influence the Sensitivity of the 
Retina to Color: A Quantitative Study and Methods of Standardizing. Psychol- 
Mon., 1913, 15, No. 1, pp. 117 ff. 
4. Rand, G. Psychol. Mon., 1913, 15, No. 1, pp. 97-110. 
5. Landolt und Snellen. Ophthalmometrologie. Handbuch der ges. Augenheilk. 
von Graefe und Saemische, 1874, 3> P- 7°. 
6. Hess, C. Ueber den Farbensinn bei indirectem Sehen. A.f.O., 1889, 35, pp. 1-62. 
7. Hering, E. Ueber die Hypothesen zur Erklarung der peripheren Farbenblindheit. 
A.f.O., 1889, 35, pp. 63-83. 
8. Fick, A. Zur Theorie der Farbenblindheit. Arbeiten aus dem physiol. Laborat. 
der Wurzburger Hochschule, pp. 213-217. 
9. Leber, T. Ueber die Theorie der Farbenblindheit und uber die Art und Weise, 
wie gewisse, der Untersuchung von Farbenblinden entnommene Einwande 
gegen die Young-Helmholtz'sche Theorie sich mit derselben vereinigen lassen. 
Klin. Monatsblatter f. Augenheilk., 1873, 11, pp. 467-473. 
10. Peter, L. C. The Principles and Practice of Perimetry. N. Y., 1916, pp. 97-98. [Reprinted from The Psychological Review, Vol. 27, No. 5, September, 1920.] 
THE LIMITS OF COLOR SENSITIVITY: EFFECT 
OF BRIGHTNESS OF PREEXPOSURE AND 
SURROUNDING FIELD 
BY C. E. FERREE AND GERTRUDE RAND 
Bryn Mawr College 
Introduction 
The difficulty of getting reproducible results in deter- 
minations of the color sensitivity of the peripheral retina is a 
common complaint among clinic workers. This difficulty 
is so great as to lead many seriously to question the value 
of such determinations in the work of diagnosis. Their value 
in diagnosing and in checking up the course of some of the 
most serious affections of the eye is readily conceded, how- 
ever, provided the needed precision can be attained. The 
need of greater precision of working in the laboratory, while 
less important to human welfare, is no less insistent. These 
combined needs led us several years ago to make a study 
of the variable factors which influence the chromatic response, 
the details of which are still in progress. Some of these 
factors pertain to the control of the stimulus, some are peculiar 
to the response of the eye itself. All may be standardized 
and controlled. The normal eye is highly sensitive and 
complex in its responses but not inherently erratic. While 
the abnormal eye may be more erratic, one of the symptoms 
it may be of its abnormality, there should be so far as we can 
see no essential difference in the technique of the study and 
of the testing of its functioning. In fact a characteristic 
difference in this regard, which can be determined with 
certainty only when other variable factors are controlled, 
may well be found to serve as a clue to an early diagnosis of 
its abnormality. 
The variable factors which influence the chromatic 
response of the retina are, so far as we have discovered, the 
wave-length and the purity of the stimulus, the intensity of 
377 378 
C. E. FERREE AND GERTRUDE'RAND 
the stimulus and the visual angle, length of exposure of the 
eye, accuracy and steadiness of fixation, general illumination 
and state of adaptation of the retina, breadth of pupil, and 
the brightness of the preexposure and of the field surrounding 
the stimulus. We have already published considerable data 
on the effect of these factors in earlier papers (1). It will 
be the special purpose of the present paper to deal with the 
last two, the brightness of the preexposure and of the sur- 
rounding field. A detailed explanation of the effect of these 
two factors on the amount of the chromatic response has 
been given in the second of the papers referred to above (1). 
A brief explanation and statement of principles will suffice 
here. 
I. When a small colored stimulus, surrounded by a field, 
for example, of white or black is viewed, a sensation is given 
which consists of the color mixed with black or white, due 
to a contrast sensation induced from the surrounding field. 
The effect of fusing a color with white or black is twofold, 
(fl) There is a quantitative effect due to the inhibition of the 
chromatic excitation by the achromatic. In general, in the 
central retina at medium and high illuminations, white 
inhibits color the most, the grays in order from light to dark 
next, and black the least. Also the amount of the inhibitive 
action varies with the different colors, with the part of the 
retina at which the stimulation takes place, and the state of 
brightness adaptation of the retina. The amount of induc- 
tion depends upon the difference in brightness between the 
stimulus and the surrounding field; it increases with the 
distance from the fovea and with decrease in the general 
illumination; and, with a given difference in brightness 
between the stimulus and the surrounding field, it is greater 
with a white than with a black field-also the amount of 
increase of induction with decrease of illumination and with 
increase of distance from the fovea is greater with a white 
than with a black field. And (b) there is also a qualitative 
effect. The hue of certain colors is changed by the action 
of the achromatic excitation. The change is greatest when 
the stimuli are blue and yellow. For example, yellow when THE LIMITS OF COLOR SENSITIVITY 
379 
mixed with black gives a greenish yellow which with the 
right proportion of components may become an olive green; 
and blue when mixed with white or light gray gives a sensa- 
tion of reddish blue. 
2. When making the color observation in the peripheral 
retina, the observer is given a short period of preparation 
before the stimulus is exposed, in which to obtain and hold a 
steady and accurate fixation. This introduces the factor of 
preexposure for, during this period of preparation, the area 
which is to be stimulated by color receives a previous stimu- 
lation. This previous stimulation, when it differs in bright- 
ness from the color, gives a brightness after-image which 
mixes with the color sensation and both reduces its saturation 
and modifies its color tone. If the preexposure is lighter 
than the stimulus color, it adds by after-image a certain 
amount of black to the succeeding color impression; if 
darker, it adds a certain amount of white. Since both white 
and. black as after effect reduce the sensitivity to color, 
the eye is rendered more sensitive when no after-image is 
given, that is when the preexposure is of the same bright- 
ness as the color. The preexposure should, therefore, be a 
gray of the brightness of the color. No brightness after- 
image will then be added to the succeeding color impression 
to modify either its saturation or color tone. The only 
brightness change acting upon it will be due to the slight 
adaptation to this gray during the short time of preexposure. 
Even closing the eye, as is frequently done before stimulating, 
is equivalent to giving a black preexposure. 
The general principle then is clear. There remains only 
to explain why in the peripheral retina the short preexposure 
which takes place while the eye is obtaining a steady fixation 
has so much effect on the color stimulation immediately 
following. Two reasons are found for this, (a) The after- 
image reaction of the peripheral retina is extremely quick. 
While some slight variation is found at different angles of 
excentricity, in general the peripheral after-image seems to 
reach its maximal intensity with a few seconds of stimulation. 
This amount of time is usually consumed in obtaining fixa- 380 
C. E. FERREE AND GERTRUDE RAND 
tion and preparing for the stimulation, hence in each observa- 
tion there is fused with the color sensation about as strong a 
brightness after-image as can be aroused. For this reason 
alone it is readily seen why the brightness of the preexposure 
is of so much greater consequence in the peripheral than in 
the central retina, where the maximal intensity of after- 
image is, roughly speaking, obtained from a stimulation of 
40-60 seconds or longer, (b) There is apparently no latent 
period in case of the peripheral after-image. It flashes out 
at full intensity immediately upon the cessation of the stimu- 
lation. Thus there is no possibility of escaping the full 
effect of the brightness after-image on the stimulus color as 
might happen in the central retina where the latent period 
obtains, if there were a very short exposure to stimulus color. 
Conditions under which the Work was Done 
The determinations were made in an optics room of the 
type described in previous articles (2). The illumination 
was kept constant at a value at the point of work of 42 foot- 
candles, vertical component; 31.2 foot-candles, 45 degree 
component; and 12.5 foot-candles, horizontal component. 
Three investigations were conducted. 
1. A determination was made of the effect on the apparent 
limits of color sensitivity of variations in the brightness of 
the field surrounding the stimulus. Three fields were used: 
the standard white of the Hering series, giving a surface 
brightness at the intensity of illumination used of 0.0209 
candle-power per sq. in.; the standard black of the series, 
giving a surface brightness of 0.00094 candle-power per sq. in. 
and grays of the brightness of the color at the limits of sensi- 
tivity in each of the meridians investigated. These grays 
ranged in brightness in the different meridians from 0.00350 
to 0.00395 CP- Per scl- in- f°r red5 0.01445 to 0.0189 f°r 
yellow; 0.01058 to o.01185 for green; and 0.00289 to 0.00366 
for blue. In order to study the effect of brightness of sur- 
rounding field in separation, the preexposure was in each 
case made of the brightness of the color at the point of 
investigation. THE LIMITS OF COLOR SENSITIVITY 
381 
2. A determination was made of the effect on the apparent 
limits of sensitivity of varying the brightness of the preex- 
posure. Again three brightnesses were used: the standard 
Hering white; the standard Hering black; and grays of the 
brightness of the color at the limits of sensitivity in each of 
the meridians investigated. The photometric value of the 
white, black and the range of grays for each of the colors are 
given in I above. In this series of experiments the surround- 
ing field was made in each case of the same brightness as the 
color at the point of investigation. 
3. A determination was made of the combined effect of 
preexposure and surrounding field on the apparent limits of 
sensitivity. The same three brightnesses were used as in 
the preceding investigations. In these cases, however, the 
surrounding field and preexposure were both made of the 
same brightness, i.e., both white, black or grays of the 
brightness of the color at the limits of sensitivity in the 
meridians investigated. 
Since the results obtained were meant only to be com- 
parative of the effect of varying given factors, it was deemed 
sufficient to make the determinations with pigment stimuli. 
So obtained the results are moreover more nearly what may 
be expected in the work of the clinic. The standard red, 
yellow, green and blue of the Hering series of papers were 
used. The work was done with the rotary campimeter 
described in previous papers (3). With the control of 
surrounding field afforded by the campimeter, this apparatus 
combines the rotary features of the perimeter. Without 
some apparatus combining both of these features we have 
not found it possible to make a determination of the apparent 
limits of sensitivity with an adequate control of the bright- 
ness of the surrounding field and of the preexposure. The 
need of an apparatus in the clinic by means of which this 
control may be accomplished is obvious. Not only is it 
impossible to secure an adequate control of these two im- 
portant factors by means of the standard perimeter, but a 
very great practical difficulty is encountered in daylight work 
in getting an equal illumination of the pigment stimulus at 382 
C. E. FERREE AND GERTRUDE RAND 
different points in the field of vision and a constant illumina- 
tion from sitting to sitting. In case of artificial illumination 
the latter difficulty can perhaps be eliminated with care; 
but the task of securing an equal effective illumination of the 
stimulus from point to point in the same meridian and of 
corresponding points in different meridians is practically 
impossible in case of any perimeter now in use, because of 
the unequal shading of the moving stimulus by the observer, 
the varying inequalities of the incident and reflecting angles, 
etc. In case of the instrument used by us these difficulties 
are minimized by using a stationary pigment surface, 20 x 20 
cm. placed with special reference to evenness of illumination 
at some constant distance (in the present work 45 cm.) 
behind the stimulus opening in the campimeter and by 
securing the excentric stimulation by shifting the fixation 
from point to point along an arm specially constructed for 
the purpose. For other points of criticism of the perimeter 
as an instrument of precision for either light or dark room 
work the reader is referred to former papers. The preex- 
posure was secured by inserting the appropriate pigment 
surface between the stimulus card and the stimulus opening 
in the campimeter. The duration of the preexposure was 
kept constant at 2 seconds. The stimulus opening in the 
campimeter was 15 mm. in diameter. At the eye, 25 cm. 
distant, this subtended a visual angle of 30 26'. 
The more important results given in this paper have been 
confirmed repeatedly both in the graduate and under- 
graduate work in our laboratory. The determination of the 
effect of the brightness of preexposure and surrounding field 
on the apparent limits of color sensitivity has in fact formed 
a part of the drill work in the undergraduate laboratory for 
several years. Space will be taken here for the results of 
only one observer-the observer whose results have been 
given in the preceding studies on the color sensitivity of the 
peripheral retina. 
As has already been indicated, the effect of brightness of 
the preexposure and of the surrounding field falls under the 
general heading of the inhibitive action of the achromatic THE LIMITS OF COLOR SENSITIVITY 
383 
excitation on the chromatic. This action takes place how- 
ever the achromatic excitation is aroused-by the admixture 
of white light, by after-image, by contrast, etc. It may be 
strikingly and conveniently demonstrated to large numbers 
at once in the following lecture room experiments, (a) Set 
up side by side on three color mixers discs made up of 180 
degrees of color, e.g. blue, and 180 degrees of white, 180 
degrees of blue and 180 degrees of gray of the brightness of 
the blue, and 180 degrees of blue and 180 degrees of black. 
When mixed, although the eye receives the same amount of 
colored light from each set of discs, the mixture with black 
seems to have lost but very little, if any, color; the mixture 
with white is a lavendar with but little color; and the mix- 
ture with gray of the brightness of the color3 in this case a 
very dark gray, is less saturated than the mixture with black. 
When different grays are used the saturation decreases appar- 
ently in graded steps as white is approached. The demon- 
stration can be made on a single color mixer by compounding 
the color disc with white, black and gray discs of different 
breadths or radii. When rotated this gives the effect of a 
surface made up of three concentric zones or rings, one in 
which the color is mixed with white, one with gray and the 
other with black. The demonstration may be made roughly 
quantitative by determining the proportions of color required 
to give the chromatic threshold in black, white and the grays; 
also by determining the proportions of color and the achro- 
matic series to give equal saturations. 
(Z>) Prepare a preexposure surface, half white and half 
black, 50x60 cm. Expose the eye 15-20 seconds and pro- 
ject the after-image on a colored surface, e.g., blue, of the 
same dimensions. The half of the field preexposed to black 
will appear a very pale unsaturated lavendar, while the half 
preexposed to white will be a dark strongly saturated blue, 
although the eye receives the same amount of light from 
both halves of the field. As the after-image dies away the 
two halves of the field become more and more nearly alike 
in saturation and color tone. If desired, the preexposure 
surface may be made of white, black and a series of graded 384 
C. E. FERREE AND GERTRUDE RAND 
grays, appropriately arranged. When this is done the 
graded loss in saturation due to the different brightnesses of 
the after-image may be observed. This demonstration also 
may be made quantitative by finding the threshold of color 
after the eye has been preexposed for 15-20 seconds to white, 
black and the grays. 
(c) Prepare contrast discs with narrow rings of color and 
inside and outside surfaces of black, white and a gray of the 
brightness of the color, respectively. Set up on color mixers 
side by side and rotate to smooth out all margins. The 
colors are lightened and darkened respectively by contrast 
induced by the black and white fields. The effect of these 
achromatic excitations on the hue and saturations of the 
colors is similar to that obtained in the former experiments. 
A more striking effect is produced if a mixed color, e.g., 
orange, is used. The quantitative features noted above can 
also be utilized in this demonstration by employing for the 
contrast ring in each case a gray of the brightness of the 
color and enough of the color to give the threshold of color 
sensation when acted upon by the white and black inductions. 
The effect of induction and after-image, it will be remem- 
bered, are not nearly so striking in the central as in the 
peripheral retina. Much more induction with a given bright- 
ness difference between the inducing and the contrast field, 
for example, is produced in the peripheral retina, and only a 
short period of preexposure (2-3 seconds) is required to give 
a strong after-image with no latent period. 
Results 
The following results were obtained: (1) The widest 
angular limits of the color zones were obtained when the 
preexposure and surrounding field were of the same bright- 
ness as the color. (2) When the brightness of preexposure 
and surrounding field were different from that of the color, 
the effect of surrounding field was less than that of preexpo- 
sure; and the effect of either is always less than the com- 
bined effect of both. (3) In some meridians the effect of 
surrounding field alone narrowed the limits as much as THE LIMITS OF COLOR SENSITIVITY 
385 
II degrees; the effect of preexposure alone, as much as 17 
degrees; and the combined effect of preexposure and sur- 
rounding field, as much as 20 degrees. 
(4) The amounts the limits were narrowed for red, 
yellow, green and blue, respectively, by a white preexposure 
alone ranged in the different meridians1 from 4-15 degrees, 
2-17 degrees, 3-15 degrees, and 4-12 degrees; by a black 
preexposure from 3-11 degrees, 3-10 degrees, 4-13 degrees, 
and 2-12 degrees; by a white surrounding field 1.5-10 de- 
grees, 2-9 degrees, 2-11 degrees, and 2-10 degrees; by a 
black surrounding field 1-8 degrees, 1-8 degrees, 2-10 
degrees, and 1.5-9 degrees; by a combined white preexposure 
and white surrounding field 5-19 degrees, 2-20 degrees? 
4-20 degrees, and 5-17 degrees; by a combined black pre- 
exposure and black surrounding field 4-17 degrees, 5-12 
degrees, 7-18 degrees and 5-18 degrees. When the effect 
of a white or black surrounding field alone was wanted, the 
preexposure was made of the same brightness as the color 
at the point of investigation; similarly when the effect of a 
white or black preexposure was wanted, the surrounding 
field was made of the same brightness as the color at the 
point of investigation. The value of the limits with a pre- 
exposure and surrounding field of the same brightness as the 
color served in each case as the standard value in terms of 
which to estimate the amounts the limits were narrowed by 
the white and black preexposures and surrounding fields and 
their combinations. 
These values, it will be remembered were obtained with a 
very precise control of the illumination of the working 
surfaces. It is obvious that a much greater variability of 
result should be expected had there been no better control 
of the constancy of illumination than is ordinarily exercised 
in office and clinic work, and too often in laboratory work- 
The effect on both the limits and hue of the color of such 
variations in the daylight illumination of the working surfaces 
as are apt to occur over long periods of time when no especial 
control is exercised, will be given in a later paper. 
1 In the order shown in the tables. 386 
C. E. FERREE AND GERTRUDE RAND 
In order to realize how profoundly the powers of chromatic 
response must have been affected to change the limits of 
sensitivity by the amounts represented by the above figures 
one must bear in mind how abruptly sensitivity falls off in 
the far periphery of the retina. A determination of the 
thresholds of color in the temporal meridian with preexposure 
Limits of Color Field for Red 
Showing the Effect of Brightness of Pre exposure, Brightness of Surrounding Field, and 
the Combined Effect of Brightness of Preexposure and Surrounding 
Field on the Apparent Limits for Red 
Table I 
ft 
[eridian 
Effect of Preexposure1 
Effect ot Surrounding 
Field2 
Combined Effect of 
Preexposure and Sur- 
rounding Field 
Gray of 
Brightness 
of Color 
White 
Black 
Gray of 
Brightness 
of Color 
White 
Black 
Gray of 
Brightness 
of Color 
White 
Black 
Upper 
0° 
58 
45 
47 
58 
48 
50 
58 
4° 
41 
Nasal 
25° 
49 
43 
43 
49 
46 
46 
49 
41 
39 
CC 
45° 
49 
43 
41 
49 
46 
44 
49 
38.5 
37-5 
cc 
70° 
47 
43 
42-5 
47 
45-5 
44-5 
47 
41 
40 
cc 
90° 
43 
38 
37 
43 
4i-5 
40 
43 
38 
38 
110° 
47 
42 
42 
47 
43 
43 
47 
41 
42-5 
cc 
135° 
5° 
46 
45 
5° 
48 
47 
50 
45 
44 
cc 
155° 
51 
47 
47 
5i 
48-5 
48.5 
51 
46 
46 
Lower 
1800 
60 
53 
56 
60 
55 
57 
60 
52 
56 
Temporal 250 .. 
73 
59 
68 
73 
66 
70 
73 
55 
62 
CC 
45 •• 
79 
64 
74 
79 
70 
76 
79 
60 
72 
cc 
700 .. 
85 
75 
80 
85 
80 
82 
85 
69 
78 
cc 
900 .. 
89 
83 
85 
89 
85 
88 
89 
80 
84 
cc 
IIO° .. 
89 
82 
85 
89 
84 
86 
89 
80 
83 
c. 
135° 
85 
78 
82 
85 
81 
83 
85 
77 
81 
cc 
155° •• 
75 
62 
65 
75 
65 
68 
75 
60 
64 
and surrounding field of the same brightness as the color for 
red, yellow, green and blue at 5 degrees, 3 degrees, 2 degrees 
and 1 degree respectively from the limit shows the following 
values: for red, 132, 150, 250 and 320 degrees; for yellow, 
100, 150, 240 and 330 degrees; for green 130, 145, 260 and 
345 degrees; and for blue 130, 145, 200 and 310 degrees. 
1 In determining the effect of the different brightnesses of preexposure, the bright- 
ness of the surrounding field was made equal to that of the color at the point of in- 
vestigation. 
2 In determining the effect of the different brightnesses of surrounding field, the 
brightness of the preexposure was made equal to that of the color at the point of 
investigation. THE LIMITS OF COLOR SENSITIVITY 
387 
For red thus there was an increase of 172.7 per cent, in the 
threshold in passing to the limit from a point 5 degrees from 
the limit; for yellow an increase of 260 per cent.; for green 
an increase of 207.7 Per cent.; and for blue an increase of 
207.7 per cent. For a more detailed experimental analysis 
of the effect of preexposure, surrounding field, intensity of 
Table II 
Limits of Color Field for Yellow 
Showing the Effect of Brightness of Preexposure, Brightness of Surrounding Field, and 
the Combined Effect of Brightness of Preexposure and Surrounding 
Field on the Apparent Limits for Yellow. 
Meridian 
Effect of Preexposure1 
Effect of Surrounding 
Field2 
Combined Effect of 
Preexposure and Sur- 
rounding Field 
Gray of 
Brightness 
of Color 
White 
Black 
Gray of 
Brightness 
of Color 
White 
Black 
Gray of 
Brightness 
of Color 
White 
Black 
Upper o° 
Nasal 250 
" 45° 
" 700 
" 9°° 
" IIO° 
" 135° 
" 155° 
Lower i8o° 
Temporal 2$° .. 
45° 
" 700 .. 
" 900 .. 
" iio° .. 
" 135° •• 
" 155° •• 
47 
42 
42 
46 
44 
46 
5° 
48 
59 
65 
73 
87 
89 
89 
87 
72 
41 
39 
37 
42 
42 
42 
46 
44 
Si 
48 
63 
70 
75 
81 
80 
60 
37-5 
38 
36 
4° 
40 
38 
45 
44 
54 
55 
70 
84 
85 
86 
84 
65 
47 
42 
42 
46 
44 
46 
50 
48 
59 
65 
73 
87 
89 
89 
87 
72 
41 
40 
4° 
44 
42 
43 
48 
46 
53 
58 
68 
79 
80 
83 
82 
63 
39 
39 
38 
42 
4i 
4i 
47-5 
46 
56 
61 
72 
86 
87 
87 
85-5 
67 
47 
42 
42 
46 
44 
46 
5° 
48 
59 
65 
73 
87 
89 
89 
87 
72 
38 
38-5 
37 
42 
42 
4i 
46 
43 
47 
45 
62 
69 
72 
80 
78 
59 
36 
37 
35-5 
39 
38.5 
37 
45 
43 
52 
53 
67 
80 
84 
85 
84 
63 
the illumination of the visual field, amounts of induction 
with different brightness relations of surrounding field to 
stimulus at different intensities of illumination, etc., and the 
effect of all of these on the thresholds of color and the limits 
of sensitivity the reader is referred to the first two papers 
cited in the appended bibliography (1). 
5. In those meridians in which the limits are wide there 
is a general tendency for the white preexposure and surround- 
ing field to narrow the limits more than a black preexposure 
1 Brightness of Surrounding Field: gray of the brightness of yellow. 
2 Brightness of Preexposure: gray of the brightness of yellow. 388 
C. E. FERREE AND GERTRUDE RAND 
and a black surrounding field. We have stated in our intro- 
duction that the amount of inhibition of the chromatic by 
the achromatic excitation varies with the color, the part of 
the retina stimulated and the state of adaptation of the 
retina. This statement applies also to the relative effects 
of white and black. In the central retina at medium and 
high illuminations white inhibits color much more than black. 
Table III 
Limits of Color Field for Green 
Showing the Effect of Brightness of Preexposure, Brightness of Surrounding Field, and 
the Combined Effect of Brightness of Preexposure and Surrounding 
Field on the Apparent Limits for Green 
Effect of Preexposure1 
Effect of Surrounding 
Fipld 2 
Combined Effect of 
Preexposure and Sur- 
Meridian 
rounding Field 
Gray of 
Gray of 
Gray of 
Brightness 
White 
lack 
Brightness 
White 
Black 
Brightness 
White 
Black 
of Color 
of Color 
of Color 
Upper o° 
36 
26 
29 
36 
28 
31 
36 
27 
22 
Nasal 250 
35 
30 
27 
35 
31 
29 
35 
26 
21 
" 45° 
38 
30 
28 
38 
32 
3° 
38 
29 
24 
" 70° 
39 
34 
31 
39 
36 
31 
39 
32 
27 
" 900 
39 
35 
33 
39 
37 
35 
39 
33 
28 
" no° 
37 
3i 
3i 
37 
33 
33 
37 
3i 
30 
" I3S° 
37 
32 
29 
37 
34 
3i 
37 
3i 
25 
" 155° 
33 
3° 
29 
33 
3i 
30 
33 
29 
26 
Lower 1800 
37 
32 
3i 
37 
34 
33 
37 
28 
26 
Temporal: 250 . . 
37 
30 
30 
37 
34 
35 
37 
28 
26 
45° •• 
42 
34 
36 
42 
39 
40 
42 
30 
33 
" 700 . . 
61 
5i 
53 
61 
56 
57 
61 
47 
50 
" 900 .. 
69 
5<> 
60 
69 
60 
62 
69 
50 
53 
" no0 .. 
65 
53 
56 
65 
58 
61 
65 
46 
5° 
" 135° •• 
57 
42 
44 
57 
46 
47 
57 
37 
39 
" 155° •• 
44 
39 
37 
44 
41 
39 
44 
35 
34 
At these illuminations therefore a black preexposure and 
surrounding field are much more unfavorable than white. 
At lower illuminations this difference in effect becomes less 
pronounced. In the far periphery of the retina the following 
are some of the conditions which contribute to make black as 
preexposure and surrounding field give wider limits of 
sensitivity, {a} A condition of low illumination and a state 
of low illumination adaptation, (^) A darkening of all of the 
1 Brightness of Surrounding Field: gray of the brightness of green. 
2 Brightness of Preexposure: gray of the brightness of green. THE LIMITS OF COLOR SENSITIVITY 
389 
colors, particularly red and yellow (the Purkinje shift of the 
peripheral retina). This brings the brightness of the color 
nearer to black than to white and the stronger relative 
darkening of red and yellow than of their neutral or colorless 
preexposures and surrounding fields, increases the contrast 
Table IV 
Limits of Color Field for Blue 
Showing the Effect of Brightness of Preexposure, Brightness of Surrounding Field, and 
the Combined Effect of Brightness of Preexposure and Surrounding 
Field on the Apparent Limits for Blue 
Meridian 
Effect of Preexposure1 
Effect of Surrounding 
Field2 
Combined Effect of 
Preexposure and Sur- 
rounding Field 
Gray of 
Brightness 
of Color 
White 
Black 
Gray of 
Brightness 
of Color 
White 
Black 
Gray of 
Brightness 
of Color 
White 
Black 
Upper o° 
52 
40 
46 
52 
42 
48 
52 
35 
34 
Nasal 250 
45 
39 
39 
45 
41 
42 
45 
38 
36.5 
" 45° 
48 
44 
46 
48 
45 
46 
48 
40 
40 
" 700 
46 
41 
4i 
46 
44 
44 
46 
41 
41 
" 900 
52 
42 
42 
52 
48 
47 
52 
42 
4° 
" 110° 
So 
46 
45 
50 
47-5 
47 
50 
44 
43 
" 135° 
52 
47-5 
47-5 
52 
50 
49 
52 
46 
46 
" 155° 
58 
48 
46 
58 
5i 
49 
58 
43 
42 
Lower 1800 
70 
63-5 
62 
70 
66 
64-5 
70 
61 
59 
Temporal: 25" . . 
70 
62 
65 
70 
65 
68.5 
70 
56 
59 
" 45° • • 
79 
7i 
73 
79 
73 
75 
79 
65 
69 
" 700 . . 
86 
78 
82 
86 
80 
84 
86 
77 
80 
900 . . 
9i 
86 
85 
9i 
89 
89 
9i 
84 
84 
" IIO° . . 
9i 
85 
85 
9i 
88 
88 
9i 
83 
83 
" 135° •• 
89 
84 
83 
89 
86 
85 
89 
83 
83 
" 155° •• 
80 
75 
75 
80 
77 
77 
80 
75 
74 
and after-image effect for white and decreases it for black. 
The darkening of red and yellowr in passing to the far periphery 
of the retina is very great. In the nasal half of the retina 
with its wide limits, the effect of this darkening on the results 
of our determinations was, of course, the most pronounced. 
As colors darken, there is, when a certain point in the process 
is reached, varying with the color, a tendency for them to 
lose their saturation very rapidly, (c) Achromatic induction 
increases very strongly with decrease of illumination and 
therefore increases in passing from the center to the periphery 
1 Brightness of Surrounding Field: gray of the brightness of blue. 
2 Brightness of Preexposure: gray of the brightness of blue. 390 
C. E. FERREE AND GERTRUDE RAND 
of the retina. It increases much faster for white than for 
black. 
In the meridians in which the limits are narrower the 
situation is more nearly as it is in the central retina. Here 
the tendency is for the limits to be narrowed more by a black 
Fig. i. Effect of brightness of preexposure on the limits of the color field. In 
this chart are shown the apparent limits for red with preexposures respectively of white, 
black, and gray of the brightness of the color at the point of investigation, surrounding 
field in each case gray of the brightness of the color at the point of investigation. 
than by a white preexposure and surrounding field. In 
some meridians the amount of narrowing is approximately 
equal for both. Another factor which tends to make the 
effect more nearly the same in these meridians for all back- 
grounds and preexposures is the more abrupt falling off in 
sensitivity. That is, more effect on sensitivity is required 
here to change the limits by a detectable amount than is THE LIMITS OF COLOR SENSITIVITY 
391 
required in those portions of the retina where the sensitivity 
grades off more slowly. 
A detailed representation of the results is given in Tables 
I-IV. and a graphic representation of a part of the results 
in Figures 1-6. In the tables results are given separately 
Fig. 2. Effect of brightness of surrounding field on the limits of the color field. 
In this chart are shown the apparent limits for red with a surrounding field respectively 
of white, black, and gray of the brightness of the color at the point of investigation, pre- 
exposure in each case gray of the brightness of the color at the point of investigation. 
for the effect of preexposure, surrounding field and com- 
bined effect of preexposure and surrounding field for each 
of the four colors: red, yellow, green and blue. In case of 
the figures, however, -space has been taken to represent 
separately the effect of preexposure and surrounding field 
for only one of the colors, red-Figs. 1-3. Figs. 3-6 show 392 
C. E. FERREE AND GERTRUDE RAND 
the combined effect of preexposure and surrounding field on 
each of the four colors. In our previous papers the repre- 
sentation of results has been in terms of position on the retina. 
In this paper the representation has been made in terms of 
field of vision. 
Fig. 3. The combined effect of brightness of preexposure and surrounding field 
on the limits of the color field. In this chart are shown the apparent limits for red 
with both preexposure and surrounding field respectively of white, black, and gray of 
the brightness of the color at the point of investigation. 
Conclusion 
It is quite obvious from the preceding data that repro- 
ducible results can not be hoped for in perimetric or campi- 
metric determinations of the sensitivity of the peripheral 
retina unless the variable effects of preCxposure and surround- 
ing field be eliminated from the conditions of work. This 
can be done completely only by making the brightness of THE LIMITS OF COLOR SENSITIVITY 
393 
the preexposure and surrounding field in each case the same 
as that of the color employed and working under constant 
intensity of illumination. Among the effects of a variable 
intensity of illumination on the results of a perimetric or 
campimetric determination the following two may be men- 
Fig. 4. The combined effect of brightness of preexposure and surrounding field 
on the limits of the color field. In this chart are shown the apparent limits for yellow 
with both preexposure and surrounding field respectively of white, black, and the gray 
of the brightness of the color at the point of investigation. 
tioned. (a) When the color stimulation is given by light 
reflected from pigment stimuli of a given coefficient of reflec- 
tion the amount of colored light obtained depends upon the 
intensity of light incident on the reflecting surface. And (b) 
a. brightness match of preexposure and surrounding field with 
the stimulus surface will not hold at different illuminations 
(the Purkinje phenomenon). 394 
er E. FERREE AND GERTRUDE RAND 
We have worked out in previous papers the conditions 
under which the desired standardization of intensity and 
color value of illumination and control of brightness of pre- 
exposure and surrounding field may be obtained in labor- 
Fig. 5. The combined effect of brightness of preexposure and surrounding field 
on the limits of the color field. In this chart are shown the apparent limits for green 
with both preexposure and surrounding field respectively of white, black, and gray of the 
brightness of the color at the point of investigation. 
atory campimetry (4). These conditions however are scarcely 
feasible for the work of the office or clinic. We have there- 
fore more recently devised and constructed a perimeter by 
means of which equal illumination of the stimulus is received 
at every point on the perimeter arm in all meridians and the 
effect of brightness of preexposure and surrounding field can 
be eliminated with an ease and speed of manipulation which THE LIMITS OF COLOR SENSITIVITY 
395 
should be feasible for office and clinic work and with a com- 
pleteness of result that should be adequate for this type of 
work. We have in fact constructed two types of perimeter 
either one of which provides for the uniform illumination of 
the arm of the perimeter. The perimeters will be described 
in a later paper. 
Fig 6. The combined effect of brightness of preexposure and surrounding field 
on the limits of the color field. In this chart are shown the apparent limits for blue 
with both preexposure and surrounding field respectively of white, black, and gray of the 
brightness of the color at the point of investigation. 
Comment 
A much more detailed study of the quantitative relations 
of the chromatic and achromatic components of the visual 
sensation for different intensities of stimulus and for different 
states of the reacting eye is needed. There are many im- 
portant practical bearings of the knowledge that would be 396 
C. E. FERREE AND GERTRUDE RAND 
gained by such a study. For example, it is often deemed 
sufficient to give a colorimetric specification of a light at one 
intensity alone in spite of the fact that the saturation, even 
the hue of the color, changes with the intensity as well as 
the composition of the light. We are all familiar in a general 
way with the fact that even the sensation aroused by a 
spectrum band of light begins as achromatic or colorless at 
very low intensities, passes through saturation and hue 
changes with increase of intensity of light and finally becomes 
colorless again at high intensities. We have pointed out 
many times in connection with problems of lighting (5) 
that while a specification of the composition of light is in- 
dependent of intensity, a true colorimetric specification 
may not, depending on the method used, be definite un- 
less it is accompanied also with a specification of intensity. 
Filters designed to give a certain coloration of light can not 
be depended upon to give this subjective coloration at all 
intensities even though the wave-lengths transmitted are 
in the same proportions. Indeed when used in connection 
with the same intensity of source the coloration of the il- 
lumination of an object as seen by the eye, particularly 
the saturation, will vary at different distances from the 
source. The lack of realization of this dependence of the 
color 'of light on its intensity as well as its composition has 
doubtless played no small part in the popular confusion 
which exists as to the comparative color values of different 
artificial lights and of the closeness of approximation of 
certain artificial lights to daylight. The surface of a Wels- 
bach mantle, 0.7 per cent, ceria, viewed directly is, for 
example, whitish; but the reading page illuminated by it to 
ordinary working brightness appears distinctly yellowish 
green. Again the illumination given by the blue bulb lamp 
may be judged of different color values depending upon the 
intensity of light falling on the illuminated object. Comple- 
mentary colors combined to gray at medium or high intensi- 
ties may not be seen as colorless at low illuminations, e.g., 
the gray produced by combining the Hering standard blue 
and yellow under daylight of good intensity becomes dis- THE LIMITS OF COLOR SENSITIVITY 
397 
tinctly lavendarish under the same light at low intensities. 
Daylight itself is popularly said to become bluish at low 
intensities. Examples may thus be multiplied indefinitely 
of the apparently peculiar complexity of the selectiveness of 
the eye's chromatic response to intensity. 
In addition to the practical bearings of the shifting of the 
quantitative relations of the achromatic and chromatic 
components in the visual sensation, with no change in the 
composition of light, there is the interesting problem of 
explanation. Many factors, it may be, are operative in the 
production of this phenomenon: a selectiveness of response 
to intensity, perhaps even a change in the range of the eye's 
chromatic response to wave-length with change of intensity, 
in case of spectrum lights; this and slight variations for 
change of intensity, in the cancelling proportions of the com- 
plementary colors and in the mutually inhibitive actions of 
the non-complementary colors, in case of mixed lights; a 
direct action of the achromatic excitation on the chromatic, 
for both simple and mixed lights; etc. It seems not only 
reasonable but necessary to infer this latter action because 
the same type of effect is produced on the color when the 
achromatic component of the sensation is varied in all of the 
following ways: by keeping the composition of the light the 
same and varying its intensity, by adding colorless light, by 
adding white oi>black to the sensation as after-image or con- 
trast, and by the achromatic changes in adaptation. No 
other explanation seems possible when the phenomenon is 
produced as an effect of preexposure and surrounding field or 
as we commonly say by after-image and contrast, as has been 
the case in the work reported in this paper. 
Bibliography ' 
I. Rand, Gertrude. The Effect of Changes in the General Illumination of the 
Retina upon its Sensitivity to Color, Psychol. Rev., 1912, 19, 462-491; 
The Factors that Influence the Sensitivity of the Retina to Color: A Quanti- 
tative Study and Methods of Standardizing, Psychol. Monog., 1913, 15, No. 
62, 166+xl; Ferree, C. E. and Rand, G., The Absolute Limits of Color 
Sensitivity and the Effect of Intensity of Light on the Apparent Limits, Psy- 
chol. Rev,, 1920, 27, 1-24. 398 
C. E. FERREE AND GERTRUDE RAND 
2. Ferree, C. E. and Rand, G., An Optics Room and a Method of Standardizing its 
Illumination, Psychol. Rev., 1912, 19, 364-373; A Simple Daylight Photom- 
eter, Amer. J. of Psychol., 1916, 27, 335-340; Chromatic Thresholds of 
Sensation from Center to Periphery of the Retina and their Bearing on Color 
Theory, Part I, Psychol. Rev., 1919, 26, 16-42. Ferree, C. E., Rand, G. 
and Haupt, I. A., A Method of Standardizing the Color Value of the Day- 
light Illumination of an Optics Room, Amer. J. of Psychol., 1920, 31, 77-87. 
3. Ferree, C. E., Description of a Rotary Campimeter, Amer. J. of Psychol., 
1912, 21, 449-453. Ferree, C. E. and Rand, G., A Spectroscopic Appar- 
atus for the Investigation of the Color Sensitivity of the Retina, Central and 
Peripheral, J. of Exper. Psychol., 1916, 1, 246-283. 
4. Op. cit.: also Ferree, C. E. and Rand, G., A Substitute for an Artificial Pupil, 
Psychol. Rev., 1916, 23, 380-383. 
5. Ferree, C. E. and Rand, G., Some Experiments on the Eye with Different Illum- 
inants-Part I, Trans. Illuminat. Eng. Soc., 1918, 13, 1-18; Part II, ibid., 
1919, 14, 107-133; etc. Reprinted from the Transactions of the American Ophthalmological Society, 1920. 
THE CAMPPERIMETER-AN ILLUMINATED PER- 
IMETER WITH CAMPIMETER FEATURES. 
C. E. FERREE, PH.D., AND G. RAND, PH.D., 
Bryn Mawr College. 
(By invitation.) 
This apparatus was devised in response to a request from 
a committee appointed by this Society to work out a better 
standardization of the illumination of perimeters and test- 
charts. The request was for a feasible means of illuminating 
the perimeter arm with light of a good intensity and quality, 
so that every point on the arm in any meridian in which it 
might be placed would receive equal intensities of light. 
Intensity and quality of illumination, however, are only two 
of the factors which influence the results of the perimetric 
determination. In devising the instrument described in this 
paper it has been the purpose of the writers to provide a 
control also of other factors which are of importance in the 
work of the office and clinic. 
The variable factors which influence the apparent limits 
of color sensitivity are, so far as we have been able to dis- 
cover, the wave-length and purity of the stimulus, the in- 
tensity of the stimulus and the visual angle, length of ex- 
posure of the eye. the method of exposure (moving or sta- 
tionary stimulus), accuracy and steadiness of fixation, the 
intensity of the general illumination of the retina and its 
state of adaptation, breadth of pupil, and the brightness of 
the preexposure and of the background or surrounding field. 
The most important of these, from the standpoint of the 
office or clinic, are perhaps the intensity of the stimulus, the 
brightness of the preexposure and surrounding field, the 
1 2 
Ferree and Rand: The Campperimeter. 
intensity of the general illumination, and the accuracy and 
steadiness of fixation. 
(1) Intensity of Stimulus.-By a sufficiently wide variation 
in this factor alone the fields of color sensitivity may be made 
to have almost any breadth within the limits of the field of 
vision and to differ radically in shape. When pigment sur- 
faces of a given coefficient of reflection are used as stimuli, 
the illumination of the perimeter arm determines the in- 
tensity of the stimulus light. Two methods are proposed for 
securing an even illumination of the stimulus at every point 
on the perimeter arm 
and of reproducing this 
illumination from time 
to time. 
Method 1.-When the 
source of light is inlaid in 
the surface of the arm 
or its continuation, the 
illumination on this sur- 
face will be equal for 
approximately 180° on 
either side of the source. 
The value of this illu- 
mination at every point 
will be equal to the nor- 
mal flux of light from the luminous surface divided by four 
times the square of the radius of curvature of the perimeter 
arm. A perimeter embodying this principle of illumination is 
being constructed in the following way: A lamp-house is fast- 
ened on the continuation of a 90° arm in such a position that 
an opening in its surface facing the observer lies in the con- 
tinuation of the surface of the perimeter arm. This opening 
is filled in with diffusing glass bent to take the curvature of 
the arm, and shaded in such a way as to shield the eye of the 
physician, and the observer, without changing or interfering 
Fig. 1. Ferree and Rand: The Campperimeter. 
3 
with the distribution of light to the perimeter arm. The lamp- 
house rotates with the arm and thus illuminates it uniformly 
at every point in all meridians. 
The principle by which an even illumination of the perime- 
ter arm is secured by this method may be demonstrated as 
follows: Let S, Fig. 1, be the source of light inlaid in the 
surface of the arm; P, any point in the perimeter arm that 
is to be illuminated; x, the distance from S to P; y, the 
radius of curvature of the arm or. the distance of the eye 
from the arm; a, the angle of emission of the light from the 
source, S; and 9, the angle of incidence of this light at the 
point P. Then the intensity of the light at P will be inversely 
as the square of the distance of P from S, or inversely as x2, 
and directly as the cosine of the angle of emission a, and 
the angle of incidence 9, also That is: 
1=1 x -2 x - x - = I 
x2 y y x2y2 4y2' 
in which I is the intensity of light at P, and I the intensity 
at S. From this equation is derived the law of illumination 
of the arm, the intensity of light at any point on the arm is 
equal to the normal flux of light from the source, divided 
by four times the square of the radius of curvature of the 
arm. 
The method has the following objections: (1) The difficul- 
ties in construction are not easy to overcome. (2) Evenness 
of illumination requires an approximately perfectly diffusing 
glass. This glass is difficult to obtain and prepare, and its 
transmission is apt to be low. Moreover, the light incident 
on the perimeter arm should approximate daylight in com- 
position. The selective absorption required to correct the 
light from the lamp to this composition further reduces the 
intensity enormously. This double loss, first by absorption 
and second by the somewhat wasteful type of distribution 4 
Ferree and Rand: The Campperimeter. 
employed, renders it difficult to get an adequate intensity 
of illumination of the perimeter arm. 
Method 2.-When the source of light lies in the perpen- 
dicular to the plane of the perimeter arm at its center of 
curvature, it will be equidistant from every point on the arm; 
also the angles of emission and incidence of the beam of light 
will be equal for every point on the arm. A perimeter has 
been constructed embodying this principle of illumination. 
This perimeter is shown in Fig. 2 Two arcs of the same 
radius of curvature were constructed at right angles to each 
other-one, a 180° arc, the perimeter arm; the other, a 90° 
arc, the lamp arm at the end of which is placed the source of 
light. In order that the source of light shall sustain a fixed 
relation to the perimeter arm for all positions of that arm, 
the two arms are fastened together.at the center of rotation. 
About the source is a housing which was designed in such a 
way as to shield the eye of the patient and the physician 
without interfering with the distribution of light to the 
perimeter arm. This housing is made of black japanned iron 
and is painted a mat black on the inside, in order that, as 
nearly as possible, all the light which passes to the perimeter 
arm shall radiate directly from the lamp filament. Its 
dimensions are 4^4 by 4^4 by 5 inches. A rectangular aper- 
ture 2}4 inches wide was cut out of the side of the housing 
facing the perimeter arm, at the bottom, and for 3 inches 
back on the two adjacent sides. The relation of the lamp to 
this aperture is such that the light- radiates freely without 
shadows from the filament to every point on the perimeter 
arm. In order that the lamp may be removed when desired, 
the bottom of the housing is hinged at the back and is held 
in place by a latch on either side. To prevent overheating 
the housing is well ventilated by specially designed light- 
tight ventilators-four in the sloping roof of the lamp house 
and four on each of the sides at the bottom. 
Provisions are made in the construction of the lamp-house Fig. 3. 
Fig. 2. Ferree and Rand: The Campperimeter. 
5 
for filtering the light to daylight quality. A well-seasoned 
type C Mazda lamp, operated by ammeter and rheostat 
control, is used as the source of light. The instrument is 
designed to run on any 110-volt circuit. In order that the 
lamp-cord which connects the lamp with the line shall be 
well out of the road and shall not interfere with the rotation 
of the perimeter arm, it passes up through the hollow stem 
which supports the perimeter arm to two copper brushes 
fastened to a small hard-rubber base 2^4 by % inches at 
the top of the stem. These brushes are in contact with two 
circular, insulated metal strips on the back of the brass disc to 
the face of which is fastened the perimeter arm. From these 
strips the circuit is continued to the lamp by two insulated 
wires which thread in and out of the short braces which 
reenforce the lamp arm. 
This perimeter is not difficult to construct or to operate. 
It provides for a uniform illumination of the perimeter arm 
in all meridians with light of a good intensity and quality; 
and with it a precision of control is possible which is com- 
parable with the work of the physical laboratory. Of the 
two instruments proposed, it is without doubt much the more 
feasible, and it is also very probably the more correct in 
actual practice. Both instruments are correct in theory. 
(2) The Brightness of the Preexposure and the Surrounding 
Field.-The brightness of the surface to which the eye is 
preexposed may change the apparent limits in certain 
meridians as much as 17° to 20°. A preexposure lighter than 
the color gives a dark, and one darker than the color a light, 
after-image. These after-images change profoundly the 
saturation of the color sensation, also its hue. A background 
or surrounding field lighter or darker than the color produces 
a similar effect on the limits, but not so great. In this case 
the disturbing chromatic effect is due to physiologic induc- 
tion or contrast. The variable effect of brightness of pre- 
exposure and of surrounding field can be eliminated only by 6 
Ferree and Rand : The Campperimeter. 
making both a gray of the same brightness as the stimulus 
color. Here again a precise control of the intensity of the 
illumination for all points of the perimeter arm becomes im- 
portant. That is, the shade of gray which is needed to match 
the color in brightness changes with change of illumination; 
therefore the selection of a gray which will match the color 
in brightness for all points of work presupposes constancy 
and uniformity of illumination. A further advantage is 
gained by making the background of the same brightness 
as the color. That is, when color and background are of the 
same brightness, the stimulus disappears completely when 
the limit of sensitivity to that color is reached, instead of 
turning into a gray concerning the colorlessness of which 
the patient is apt to be in doubt. This gives the effect of 
the disappearance type of photometer, and like it, adds 
greatly to the ease and certainty of making the judgment. 
The control of brightness of preexposure and surrounding 
field is provided for in the perimeter shown in Fig 2 as fol- 
lows: To the stimulus carriage is attached an aluminum 
holder, No. 19 B. and S. gage, grooved to hold a card 5 by 
6 inches. These cards are covered on one side respectively 
by grays of the brightness of the four colors, red, yellow, 
green, and blue of the Hering standard series of pigment 
papers, as seen in the peripheral retina. At the center of 
each of these cards is pasted a disc of the appropriate color, 
subtending a visual angle of 1°. To provide for the control 
of the preexposure for the stationary method of giving the 
stimulation, cards identical with the background cards are 
provided, covered also on one side with a gray of the bright- 
ness of the color. The stimulation is given by this method 
as follows: The stimulus is placed at the point to be tested 
and covered with the preexposure card. The observer is 
told to take his fixation. At a given signal the stimulus is 
uncovered for one second and recovered. In case the moving Ferree and Rand: The Campperimeter. 
7 
stimulus method is used the surrounding field serves also as 
the preexposure. 
The perimeter arm and body are painted gray, of a shade 
which is approximately mid-gray to the blue and yellow, the 
darkest and lightest of the stimuli employed. In our own 
laboratory the perimeter is used on a table painted with the 
same gray and stands before a gray screen. These latter 
precautions, however, are not necessary. 
When the perimeter is supplied to the profession, provision 
will be made that the stimulus and preexposure cards, a 
seasoned lamp, and all other perishable parts can be pur- 
chased in the quantities desired. 
(3) The Accuracy and Steadiness of Fixation.-All are 
familiar with the disturbing effect of inaccuracy and un- 
steadiness of fixation. If correct and reproducible results 
are to be obtained, the eye must be accurately placed at the 
center of the sphere, in the surface of which lies the perimeter 
arm, and the line of sight must not shift from the fixation- 
point while the color observation is being made. As an aid 
to the correct placement of the eye and a check on its steadi- 
ness of fixation, a small circular mirror is used as a fixation 
object, in which the observer sees the image of his own eye. 
When the eye is correctly placed with the line of sight normal 
to the surface of the mirror at its central point, the fact is 
indicated to the observer by the position of the image of his 
pupil and iris as seen in the mirror. Not only is this simple 
device of service in determining the correct position of the 
eye, but it aids the observer greatly in holding a steady 
fixation. 
However, while the image in the mirror will indicate to 
the observer when the line of sight is normal to the surface of 
the mirror at its central point, there are two important 
features in the correct adjustment of the eye over which it 
exercises no control: (a) The distance of the eye from the 
mirror, and (6) the agreement of the meridians of the field 8 
Ferree and Rand: The Campperimeter. 
of vision as read on the perimeter with the meridians of the 
retina. In order that it may be known when the eye is at 
the correct distance from the perimeter arm, a light measur- 
ing rod 33 cm. in length is provided, to one end of which is 
fastened at right angles a small metal disc. In making the 
adjustment for distance one end of this rod is placed against 
the mirror at its center, and the distance of the observer's 
eye is changed until the closed lid is just in contact with the 
metal disc. Perhaps the simplest device for insuring a con- 
stancy of relation of the meridians of the retina to the meri- 
dians of the field of vision, as laid off by the perimeter arm, 
in other words, for guarding against a slight tilting of the 
head to one side or the other, is a mouth-bit. We have de- 
signed a mouth-bit of light wood to be changed for each 
observer, so shaped that it cannot be bitten too far forward 
or back, and thus the distance of the eye from the mirror be 
changed, or too far to one side or the other. 
In order quickly and conveniently to locate the patient's 
eye at the center of the perimeter system, three adjustments 
are provided: a rack and pinion to raise or lower the head, a 
second rack and pinion to shift the head to the right or left, 
and a coarse screw adjustment to change the distance of the 
eye from the perimeter arm. In the process of getting the 
eye in position the patient bites the mouth-bit, the eye is 
brought to the level of the mirror with the first rack and pinion, 
its image is centered in the mirror with the second rack and 
pinion, and its correct distance from the mirror is obtained by 
means of the screw adjustment and the measuring rod al- 
ready referred to. Once these adjustments are made for an 
eye they need not be made again during the process of taking 
the fields for that eye; that is, the re-biting of the mouth-bit 
guarantees that the eye always returns to the same position 
for which the original adjustments were made. These ad- 
justment devices were not completed in time to be shown in 
Figs. 2 and 3. Ferree and Rand : The Campperimeter. 
9 
The steadiness of fixation is greatly influenced by the 
method of giving the stimulation. One of the serious ob- 
jections to a moving stimulus is the difficulty of holding a 
steady fixation while the object to be observed is moving. 
The alternative procedure is the use of a stationary stimulus. 
That is, the stimulus is placed at the desired point on the 
perimeter arm and covered with the preexposure card. The 
observer takes his fixation, and at a given signal the stimulus 
is exposed and re-covered. By this method of giving the 
stimulation more time is consumed, but a much greater pre- 
cision of result is possible. A compromise procedure is 
recommended. That is, the approximate location of the 
limit is determined with the moving stimulus and the exact 
location with the stationary stimulus. By this compromise 
but very little more time is required and there is no sacrifice 
of precision. 
In order to provide for the mapping of the normal blind 
spot and for the quick detection and mapping of central and 
paracentral scotomata, it has been deemed advisable to add 
to the perimeter recommended a tangent screen subtending 
a visual angle of 60 or more degrees. Provision is made so 
that this screen can be quickly and conveniently attached 
to the stimulus carriage and moved into position. Cards of 
white or black, as may be desired, with the field laid off on 
the tangent scale, are provided for mapping the area deficient 
in the light sense, and of grays of the brightness of the colors 
for mapping the color deficiencies. This is shown in Fig. 3. 
With the controls provided in the perimeter recommended 
a careful worker can without difficulty reproduce the limits 
of sensitivity within one or two degrees. Reprinted from the Transactions of the American Ophthalmological Society, 1920. 
FACTORS WHICH INFLUENCE THE COLOR SENSI- 
TIVITY OF THE PERIPHERAL RETINA. 
C. E. FERREE, PH.D., AND G. RAND, PH.D., 
Bryn Mawr College. 
(By invitation.) 
Introduction.-1The difficulty of getting reproducible re- 
sults in determinations of the color sensitivity of the periph- 
eral retina is a common complaint among laboratory and 
clinic workers. The actual distribution of retinal sensitivi- 
ties is only one of the factors which influence the results of 
the perimetric or campimetric determination. By varying 
the conditions under which the work is done the fields of 
color sensitivity, beyond a certain degree of eccentricity, 
may be made to have almost any extent within the limits of 
the field of vision, and to vary radically in shape. 
The difficulty of obtaining reproducible results is so great 
as to lead many seriously to question the value of the peri- 
metric or campimetric determination in the work of diag- 
nosis. Their value in diagnosing and checking up the course 
of some of the most serious affections of the eye is readily 
conceded, however, provided the needed precision can be ob- 
tained. The need of greater precision of work in the labora- 
tory, while less important to human welfare, is no less insis- 
tent. These combined needs led us several years ago to make 
a study of the variable factors which influence the chromatic 
response, the details of which are still in progress. Some of 
these factors pertain to the control of the stimulus; some are 
peculiar to the response of the eye itself. All may be stan- 
dardized and controlled. The normal eye is highly sensitive 
and complex in its responses, but not inherently erratic. 
1 2 
Ferree and Rand : Color Sensitivity of Retina. 
While the abnormal eye may be more erratic,-one of the 
symptoms, it may be, of its abnormality,-there should be, 
so far as we can see, no essential difference in the technic of 
the testing and study of its functioning. 
The variable factors which influence the apparent limits 
of color sensitivity are the wave-length and purity of the 
stimulus, the intensity of the stimulus and the visual angle, 
length of exposure of the eye, the method of exposure (mov- 
ing or stationary stimulus), accuracy and steadiness of fixa- 
tion, the intensity of the general illumination of the retina 
and its state of adaptation, breadth of pupil, and the bright- 
ness of the preexposure and of the background or surrounding 
field. Only a few of these can be considered here. For the 
fuller treatment a bibliography is appended. The most 
important of these factors, from the standpoint of the work 
of the office and clinic, are perhaps the intensity of the 
stimulus and the precision of its control, the brightness of the 
preexposure and of the surrounding field, the intensity of 
the general illumination, and the accuracy and steadiness 
of fixation. 
The Intensity of the Stimulus.-By a sufficiently wide 
variation in this factor alone the fields of color sensitivity 
may be made to have almost any breadth within the field of 
vision, and to differ radically in shape. With very high 
intensities the limits of red, yellow, and blue are coincident 
with the limits of white light vision. Green cannot be made 
to have so wide an extent. With stimuli of medium intensi- 
ties of equal energy the limits of red,*blue, and yellow inter- 
lace or criss-cross. The limits for green again are narrower. 
The conventional clinic rating of limits from widest to nar- 
rowest in the order blue, red and green is, with the exception 
of green, a function of the relative intensities of the stimuli 
employed. A decrease of intensity of the stimuli not only 
narrows the limits but, because of the irregular distribution 
of sensitivities in the different meridians, causes a marked Ferree and Rand : Color Sensitivity of Retina. 
3 
change in the shape of the fields of sensitivity. Without 
great precision in the control of intensity, it is obvious that 
reproducibility of result cannot be obtained and little signifi- 
cance can be attached either to extent or shape of the fields 
of sensitivity or to variations from time to time or from 
person to person in these important features. 
The effect of changes in the intensity of the stimulus, both 
on the extent and shape of the color fields, varies with the 
order of magnitude of intensity employed. For medium 
and low intensities the effect of a given amount of change 
is very much greater than for high intensities. This is an 
obvious corollary of the type of distribution of sensitivities 
found in the peripheral retina. That is, in passing from the 
center toward the periphery the decrease in sensitivity is 
comparatively slow and gradual in the paracentral retina, 
it is much faster and more abrupt in the mid-periphery, and 
very abrupt in the far periphery. It requires, therefore, 
comparatively large changes of intensity in stimuli of high 
intensity, which carry the limits of sensitivity into the far 
periphery, to produce a significant change in the limits; not 
so great a change in stimuli of medium intensity; and a still 
smaller change in stimuli of low intensity. This effect varies 
greatly, however, for the same color in the different meridi- 
ans, and for different colors in the same meridian. For 
stimuli of the medium and low intensities used in the office 
and clinic the effect of change of intensity is very marked 
indeed both on the extent and shape of the fields of sensi- 
tivity. 
The Chromatic Thresholds of Sensation from Center to 
Periphery of the Retina.-In order to show the irregularity 
of decrease in sensitivity in passing from the center toward 
the periphery of the retina, we have determined the threshold 
(the amount of light required just to arouse the color sensa- 
tion) for the different colors in the different meridians. A 
graphic representation of the results of these determinations 4 
Ferree and Rand : Color Sensitivity of Retina. 
for two meridians-the temporal and nasal-is given in 
Charts I-IV. In these charts the degree of eccentricity is 
plotted along the horizontal coordinate and the energy or 
intensity values of the threshold in watts (107 ergs per second) 
are plotted along the vertical coordinate. The representa- 
tion is retinal, not field of vision. In case of red, yellow, and 
blue, it will be remembered from statements made earlier in 
the paper, the limits of sensitivity for lights of high intensi- 
ties coincide with the limits of white light vision. This, 
however, was not the case for the green stimulus. By no 
increase of intensity were we able to make the limits of green 
sensitivity coincide with the limits of white light vision. 
In Charts III and IV the above values from the center of 
the retina through the region of gradual decrease of sensi- 
tivity are plotted on a larger scale. This was done because, 
when plotted on the scale used in Charts I and II, the curves 
fall so closely together that the relative sensitivities to the 
four colors are not clearly represented. That is, the range 
of values for the threshold from the center to the extreme 
periphery of the retina is so great that in Charts I and II, in 
which the entire range is represented, a scale value had to 
be chosen for the vertical coordinate, which is so large as 
almost to obscure the smaller differences in relative sensi- 
tivity to the different colors in the paracentral retina, the 
region of gradual decrease in sensitivity. 
Space cannot be taken here for a full discussion of the 
results of these determinations. For a more detailed state- 
ment of results and a fuller discussion of their bearing on 
points of theoretic and practical importance, the reader is 
referred to "Chromatic Thresholds of Sensation from Center 
to Periphery of the Retina and Their Bearing on Color The- 
ory," Psychol. Review, 1919, xxvi, pp. 16-42; 150-163. 
The following points, however, may be noted in connection 
with the present paper: 
(a) Among other things, the foregoing charts show the Ferree and Rand: Color Sensitivity of Retina. 
5 
great irregularity in the decrease of sensitivity to each color 
which is found in passing from the center to the periphery 
of the retina. This irregularity, moreover, differs greatly 
in the different meridians. From these irregularities it is 
Chart I.-Chromatic thresholds for the 
four colors, nasal meridian. In this chart 
degree of eccentricity in the retinal field is 
plotted along the horizontal coordinate 
and the energy or intensity value of the 
threshold is plotted along the vertical co- 
ordinate. The threshold values are in 
terms of total energy entering the eye. 
Chart II.-Chromatic thresh- 
olds for the four colors, temporal 
meridian. 
obvious why the shape as well as the extent of the fields of 
sensitivity changes with the change of intensity of the stimu- 
lus light. That is, depending upon the different rates of 
decrease of sensitivity in the different meridians, a given in- 
crease or decrease of intensity of the stimulus light causes 6 
Ferree and Band : Color Sensitivity of Retina 
respectively different amounts of extension or contraction 
of the fields in these meridians. The result is, of course, a 
change of shape of the field proportionate to the amount of 
Chart III.-Chromatic thresholds for the 
four colors, nasal meridian. In this chart 
and Chart IV the values represented in 
Charts I and II respectively, from the cen- 
ter of the retina through the region of grad- 
ual decrease of sensitivity, are plotted on a 
larger scale. This is done because when 
plotted on the scale used in Charts I and 
II, the curves fall so closely together that 
the relative sensitivities are not clearly 
represented. 
Chart IV.-Chromatic thresh- 
olds for the four colors, temporal 
meridian. 
irregularity of the distribution of sensitivity in the different 
meridians. 
(6) A further obvious corollary of irregularity of distribu- 
tion of sensitivity in the different meridians is the interlacing Ferree and Rand : Color Sensitivity of Retina. 
7 
or criss-crossing of the limits when the stimuli are so graded 
in intensity as to give the limits for all the colors approxi- 
mately the same degree of -eccentricity. That is, a condition 
of coincident or concentric limits would presuppose regu- 
larity of distribution of sensitivity from meridian to meri- 
dian; irregularity inevitably leads to an intersection or criss- 
crossing when the conditions under which the determinations 
are made are such that the fields of sensitivity have approxi- 
mately the same breadth. A frequent criss-crossing of the 
limits, it will be noted in Fig. 4, occurs for red, yellow, and 
blue when stimuli of medium intensity and equal energies 
are used. 
(c) A third point which may be noted in passing' is the 
correspondence of the changes in the hue of red and green, 
in passing from the center to the periphery of the retina, 
with the relative distribution of sensitivities to red, green, 
and yellow. That is, the red and green wave-lengths of 
light have the power to arouse not only the sensations of red 
and green, but also weakly the sensation of yellow. In the 
center of the retina, which is fully sensitive to red and green, 
the weakly aroused yellow sensation is below the threshold, 
i.e., too weak to be sensed in the presence of the strongly 
aroused red and green sensations. However, in those parts 
of the periphery of the retina in which the loss of sensitivity 
to red and green is greater than to yellow, the yellow com- 
ponent of the sensation comes above the threshold and the 
red and green stimuli are sensed as yellowish-red and yel- 
lowish-green. For example, in passing from the center to 
the periphery of the retina in the nasal meridian, the red 
stimulus was sensed as red from the center to about 60°; 
from 60° to about 86° it was sensed as yellowish-red or 
orange; from 86° to the limits of sensitivity it was sensed 
again as red. Corresponding to this it will be noted that 
there is in this meridian a fairly close agreement in sensi- 
tivity to red and yellow from the center to about 60° (stimu- 8 
Ferree and Rand : Color Sensitivity of Retina. 
lus sensed as red), at which point there is a relatively sharp 
decrease in sensitivity to red. That is, from 60° to 86° there 
is much less sensitivity to red than, to yellow (stimulus sensed 
as yellowish-red or orange). At about 86° there is a sharp 
decrease in sensitivity to yellow, and from this point on to 
the limits, a fairly close agreement again in sensitivity to 
the two colors (stimulus sensed as red). In the temporal 
meridian the red stimulus was sensed as red from the center 
of the retina to about 30°; from there to about 47° as yel- 
lowish-red or orange; and from 47° to the limits of sensi- 
tivity, as red. Similarly in this meridian there is a fairly 
close agreement in sensitivity to red and yellow from the 
center to about 30°; from 30° to about 47° there is a con- 
siderably greater sensitivity to yellow than to red; and from 
this point on to the limits greater sensitivity to red than to 
yellow. In case of green in the nasal meridian the greater 
loss in sensitivity to green as compared with yellow begins 
at about 51°; and in the temporal meridian at about 26°. 
Correspondingly at these points the green stimulus began 
to be sensed as yellowish-green and continued to be sensed 
in this hue until the limits of sensitivity to green were 
reached, from which point on for a short distance it was 
sensed as unsaturated yellow. 
These changes of hue are the normal changes for spec- 
trum or pure red and green in passing from the center to the 
periphery of the retina when there is no achromatic effect of 
preexposure and surrounding field, that is, when the pre- 
exposure and surrounding field are a gray of the same bright- 
ness as the color and when the general illumination of the 
room is held constant. Later in the paper the changes pro- 
duced by a preexposure and surrounding field lighter or 
darker than the color and by a variable illumination will be 
given. 
In Chart II, temporal meridian, a gap will be noted in the 
curve for blue not present in the curves for the other colors. Ferree and Band : Color Sensitivity of Retina. 
9 
In this area is represented a peculiar type of color-blindness, 
small areas or spots of which are found, so far as we have 
been able to discover, in all or most peripheral retinas. In 
these spots the blindness is to one color alone. The spot is 
fully sensitive to all of the other colors. Moreover, it shows 
no deficiency in the canceling and after-image and contrast 
reactions of the color in question. That is, in this area the 
stimulus, although it is not sensed as color, has just as much 
power to cancel the complementary color and to arouse the 
after-image as it has in the immediately surrounding retina. 
These spots seem not to be subject to change and apparently 
are not pathologic. An examination of a large number of 
observers show that eyes may differ widely with regard to 
the number and size of these spots, their location and the 
color affected. A successful search of the peripheral retina 
for the presence of such spots requires a means of making a 
rather minute investigation of the retina from center to 
periphery in a number of meridians. A more detailed report 
of the study of this phenomenon may be found in "Some 
Areas of Color Blindness of an Unusual Type," Jour, of 
Exper. Psych., 1917, ii, pp. 295-324. F. Schumann: "Ein 
ungewohnlicher Fall von Farbenblindheit," Bericht uber den 
I. Kongress fur experimentelle Psychologic in Giessen, 1904, 
pp. 10-13, reports a case (his own) in which the whole 
retina is affected by this type of color blindness. 
Conditions Under Which the Thresholds were Determined - 
The determinations were made under the following condi- 
tions: (1) The colored lights were taken from the spectrum. 
There are two reasons for this in an investigation of the kind 
here undertaken: (a) The stimuli should be as homogeneous 
as possible with regard to the visible wave-lengths. The 
presence of alien visible wave-lengths affects the results of a 
determination of chromatic sensitivity in two ways. Through 
physiologic inhibitions and interactions it decreases the 
amount of the color response; and it increases the energy 10 
Ferree and Rand : Color Sensitivity of Retina. 
measurement. In both of these ways the value of the 
threshold is raised by the presence of impurities in the stimu- 
lus light. And (6) the stimuli should be free from the infra- 
red and ultra-violet radiations which would affect the 
thermopile used to measure the intensity of light, but not 
the eye. 
The stimuli employed were a narrow band of red in the 
region of 670 mm J of yellow in the region of 581 yp,; of green 
in the region of 522 pp; and of blue in the region of 468 pp. 
The breadth of analyzing slit used in isolating these bands 
was maintained constant at 0.5 mm. The ranges of wave- 
lengths obtained were approximately 660-680 mm; 575-587 
mm; 518-526 pp', and 468-474 mm- Th® spectrum was gotten 
and the different wave-lengths were presented to the eye by 
means of our apparatus described in the Journal of Experi- 
mental Psychology, 1916, i, pp. 247-284: "A Spectroscopic 
Apparatus for the Investigation of the Color Sensitivity of 
the Retina, Central and Peripheral." In every case the 
light was examined for impurities at the analyzing slit by 
means of a small Hilger direct vision spectroscope provided 
with an illuminated scale. When found, impurities were 
absorbed by thin gelatins selected so as to cut out as 
little as possible of the useful light. These gelatins were 
placed over the analyzing slit and were held in position by 
short clips fastened to the front surface of the jaws the edges 
of which formed the slit. 
(2) The determinations of the threshold were made in 
energy terms. Measurements were made at two places: 
at the analyzing slit and at the eye. In making the threshold 
determinations it was found convenient first to make the 
colors all equal in energy value. The reductions needed for 
this equalization were made by appropriate adjustments of 
the collimator slit. Since the blue represents the smallest 
amount of energy of any of the colors employed, they were 
all made equal in energy to the blue of the spectrum used, Ferree and Rand : Color Sensitivity of Retina. 
11 
namely, the prismatic spectrum of a Nernst filament oper- 
ated by 0.6 ampere of current. From this intensity they 
were reduced to the threshold by means of the especially 
constructed sectored discs described in a former paper,* and 
the energy values computed from the simple law of the disc. 
The method of making the energy measurements by means 
of a thermopile has been described in previous papers.f 
(3) The field surrounding the stimulus and the preexposure 
were always maintained as nearly as possible at the same 
brightness as the stimulus at the threshold value of sensa- 
tion. These surfaces were made from the Hering standard 
gray papers. It was found to be necessary to change the 
brightness of the surrounding field and preexposure fre- 
quently for each stimulus because the brightness value of 
the color at the chromatic threshold changed quite rapidly 
from center to periphery of the retina. There were two 
causes for this change: (a) The intensity of light had to be 
increased a very great deal from center to periphery to give 
the chromatic threshold from point to point; and (6) the 
achromatic value of the colors does not remain the same 
from the center to the periphery of the retina (the Purkinje 
shift of the peripheral retina). The gray that matched the 
stimulus in achromatic value at each point was determined 
by the equality of brightness method. The match was made 
in every case for' the point of the retina under investigation. 
In order to make the specification of the brightness of the 
preexposure and the surrounding field independent of the 
illumination of the room and of the variability of the reflec- 
tion coefficients of different issues of Hering papers, the 
* Journal of Experimental Psychology, 1916, i, pp. 271-274. 
t Radiometric Apparatus for Use in Psychological and Physiological 
Optics, Psych. Rev. Monog., 1917, xxiv, 66 pp. +xvi; Chromatic Thresholds 
of Sensation from Center to Periphery of the Retina, Part I, Psych. Rev., 
1919, xxvi, pp. 16-42; Selectiveness of the Achromatic Response of the Eye 
to Wave-length and Intensity of Light. Studies in Psychology, Titchener 
Commemorative Volume, Published by L. N. Wilson, Worcester, Mass., 
1917, pp. 280-307. 12 
Ferree and Rand : Color Sensitivity of Retina. 
brightness was in each case determined in candle-power per 
sq. in. 
(4) The illumination of the room was kept at a constant 
value. Two features are necessary for this control: (a) A 
means must be had of detecting small changes of illumina- 
tion. This may be accomplished by a portable photometer 
of the Sharp-Millar or Macbeth types, furnished with a 
daylight screen, or of the simpler type described by the 
writers in a previous paper.* And (6) a means must be had 
also of producing small variations in the illumination of the 
room, else the changes due to fluctuations in the external 
light can not be compensated for with the precision and 
minuteness of control that is needed. This is accomplished 
in our optics roomsf by two systems of thin white curtains 
running on spring rollers beneath the skylight. Large 
changes are produced by a light-proof curtain. One of the 
systems of white curtains and the light-proof curtain run 
lengthwise of the room; the other system of white curtains 
runs across the room. By means of the white curtains either 
small local or small general changes can be produced in the 
illumination of the room; and by means of the light-proof 
curtain larger changes may be produced ranging from full 
illumination to the darkness of a moderately good dark room. 
The light-proof curtain is of a breadth equal to that of the 
room and runs in a light-tight boxing. The white curtains 
are narrower and are made to overlap at the edges. These 
latter curtains run on wire guards so distributed as to prevent 
sagging or wrinkling. Above these curtains are pivoted two 
large diffusion sashes of glass ground on one side completely 
filling the skylight opening. These sashes diffuse the light 
* A Simple Daylight Photometer, Amer. Jour. Psych., 1916, xxvii, pp. 335- 
340. 
f An Optics Room and a Method of Standardizing Its Illumination, Psych. 
Rev., 1912, xix, pp. 364-373; A Method of Standardizing the Color Value 
of the Daylight Illumination of an Optics Room, Amer. Jour. Psych., 1920, 
xxxi, pp. 77-87. Ferree and Rand : Color Sensitivity of Retina. 
13 
in the room giving an even distribution of illumination and 
rendering, because of that fact, an even and precise control 
easier to accomplish. In a careful specification of the con- 
ditions under which the work is done, a very important item 
is to give a photometric specification of the illumination of 
the room. This may be done in foot- or meter-candles as 
desired. If the illumination is uneven it should be done 
systematically throughout the room. If, on the other hand, 
it is pretty uniform, it is usually sufficient to give its value 
in three or more directions at the point of work. In case of 
the present work, for example, the value of the horizontal 
component was 30.49 foot-candles; the vertical component, 
121.95 foot-candles; and the 45° component, 82.97 foot- 
candles. 
(5) The results were made independent of the size of the 
pupil. Breadth of pupil affects the results of a determination 
of the sensitivity of the peripheral retina in the following 
ways: It influences the clearness of imaging, the amount of 
light entering the eye, and, by limiting the angle at which 
the beam of light may enter the eye, the degree of eccentricity 
at which an image may be formed on the retina. Inde- 
pendence of change in size of pupil was especially needed in 
this work because of the large variations in the intensity of 
light used. Such control is very easy to accomplish with 
the means of presenting the light to the eye that is used in 
our spectroscopic apparatus. All that is needed is to keep 
the image that falls on the pupil of a constant size and smaller 
than the pupil throughout its entire range of variations in 
the given series of. experiments. Not only can this variation 
be determined in preliminary experiments as a guide to the 
size of image that is needed, but the image itself can be 
compared with the pupil at each observation. For details 
of the method of exercising this control, see "A Substitute 
for an Artificial Pupil," Psych. Rev., 1916, xxiii, pp. 380-383. 
The Effect of Variations of the Intensity of the Stimulus on 14 
Ferree and Rand: Color Sensitivity of Retina. 
the Breadth of the Color Fields.-Only the results obtained 
with stimuli of very high intensities will be given in this 
paper. In later work the effect of variations of intensity on 
the size and shape of the color fields will be given for stimuli 
of medium and low intensities. 
Five intensities of light were used, the prismatic spectrum 
of a Nernst filament operated by 0.6 ampere of current, In- 
tensity A; the colors of this spectrum reduced to 1/32 of 
their original intensity, 1/32 A: an equal energy spectrum 
in which the four colors were all made equal in intensity to 
the blue (the color of least energy) at intensity A, intensity 
B; the colors of this spectrum reduced to 1/32 of their origi- 
nal intensity, intensity 1/32 B; and the standard Hering 
pigments red, yellow, green and blue illuminated by 390 
foot-candles of light, vertical component. The fields for 
these stimuli are given in Figs. 1-5. 
The energy values of the red, yellow, green, and blue wave- 
length at intensity A were respectively 9096.639, 4065.624, 
1562.388, and 882.025 watt x 10"10; at intensity B, 891.05, 
882.51, 884.946, and 882.025 watt x 10-10. These intensities 
were much greater than were needed to make the limits 
of red, yellow, and green coincide with the limits of white 
light vision. This fact should, therefore, be borne in mind 
in considering how much the limits were narrowed by reduc- 
ing the intensity to 1/32 A and 1/32 B. That is, the narrow- 
ing of the limits shown in Figs. 1-4 was produced by only a 
small part of each of these reductions, as may be seen by 
comparing the following values of the threshold at the limits 
of white light vision with the intensity values of 1/32 A and 
1/32 B. The energy values of the threshold in the nasal 
meridian at the limits of white light vision were for red, 
yellow, and blue respectively 227.836, 268.95, and 264.368 
watt x 10~10; in the temporal meridian, 258.93, 285.25, and 
272.428 watt x 10-10. The energy values for red, yellow, and 
blue at intensity 1/32 A were respectively 284.27, 127.051, Fig. 2.-The effect of intensity of 
stimulus on the limits of sensitivity, 
prismatic spectrum. In this chart 
are represented the limits of sensi- 
tivity for intensity 1/32 A: red, 
284.27, yellow, 127.051, green, 48.825, 
and blue, 27.563 watt X 10~10. 
Fig. 1.-The effect of intensity of 
stimulus on the Emits of sensitivity, 
prismatic spectrum. In this chart 
are represented the limits of sensi- 
tivity for intensity A: red, 9096.639, 
yellow, 4056.624, green, 1562.388, 
and blue, 882.025 watt X 10~10. 
In Figs. 1-5 the representation is 
retinal, not field of vision. 
Fig. 3.-The effect of intensity of 
stimulus on the limits of sensitivity, 
equal energy spectrum. In this 
chart are represented the limits of 
sensitivity for intensity B (equated in 
energy to the blue, intensity A): red, 
891.05, yellow, 882.51, green, 884.946, 
and blue, 882.025 watt X 10-1°. 
Fig. 4.-The effect of intensity of 
stimulus on the limits of sensitivity, 
equal energy spectrum. In this 
chart are represented the limits of 
sensitivity for intensity 1/32 B: red, 
27.845, yellow, 27,578, green, 27,655, 
and blue, 27.563 watt X 10-10. 
15 16 
Ferree and Rand: Color Sensitivity of Retina. 
and 27.563 watt x IO-10; at intensity 1/32 B, 27.845, 27.578, 
and 27.563 watt x 1O"10. At intensity 1/32 A, therefore, the 
red was not reduced to its threshold value at the limits of 
white light vision; the yellow a little more than 1/2, and the 
blue not greater than 1/10 of this value; at intensity 1/32 B 
all these-stimuli were reduced not quite to 1/10 of their 
threshold value at the limits of white light vision. A more 
detailed and exact knowledge of the effect of change of in- 
tensity on the limits from center to periphery of the retina 
in the temporal and nasal meridians may be had from an 
examination of the threshold curves in Charts I and II. In 
these curves the effect of 22-26 changes of intensity are 
shown for the nasal meridian, and 19-29 changes for the 
temporal meridian. That is, the threshold values are given 
for that number of points respectively in these two meridians, 
selected and spaced with special reference to the steepness 
or pitch of the sensitivity gradient. An inspection of these 
charts shows that in the nasal meridian from the center of 
the retina to 85°, only very small reductions of intensity are 
required to narrow the limits for yellow by significant 
amounts; for blue this region extends to 75°; for red to 60°- 
85°; and for green to 50°. In the temporal meridian it ex- 
tends for yellow to 35°-45°; for blue to 40°; for red 25°-40°; 
and for green to 25°. Beyond these points sensitivity falls 
off much more abruptly, i. e., much larger changes of inten- 
sity are required to produce significant changes in the limits. 
Conditions Under Which the Limits were Determined.-The 
spectrum was obtained and the different wave-lengths pre- 
sented to the eye by means of the same apparatus employed 
in the preceding threshold determinations. The stimuli were 
again narrow bands of red, green, yellow, and blue in the 
regions respectively of 670 w, 581 w, 522 and 468 w. The 
control of purity, intensity, brightness of preexposure and 
surrounding field, general illumination, breadth of pupil, 
etc., also were so nearly identical with those described in Ferree and Rand : Color Sensitivity of Retina. 
17 
connection with the threshold determinations that space 
will not be taken for their further discussion here. 
The stimulus used was the circular aperture of the camp- 
imeter (rotary), 15 mm. in diameter, filled with light by the 
focusing lens. At a distance of 25 cm. from the pupil of the 
eye, on which the light from the objective slit of the spectro- 
scope was focused, this aperture subtended a visual angle of 
3° 26'. The time of exposure was one second and the interval 
between exposures varied between three and five minutes, 
depending upon circumstances and the need for precaution- 
ary measures. If the stimulus was sensed in its proper color 
at any time during the one second interval of exposure, the 
retina was called color sensitive at that point. (At the limit 
of white light vision the red stimulus^ for example, of the 
intensity used was sensed as a tint of red.) The field in the 
16 meridians was always mapped for one color before the 
work on another color was begun. The maps shown in Figs. 
1-5 represent points on the retina, not field of vision. 
For a fuller statement and discussion of the conditions 
under which the work was done and a detailed statement and 
discussion of results the reader is referred to "The Absolute 
Limits of Color Sensitivity and the Effect of Intensity of 
Light on dhe Apparent Limits," Psych. Rev., 1920, xxvii, 
pp. 1-23. 
The Brightness of the Preexposure and the Surrounding 
Field.-The brightness of the surface to which the eye is pre- 
exposed may change the apparent limits in certain meridians 
as much as 17°. A preexposure lighter than the color gives 
a dark after-image; a preexposure darker than the color 
gives a light after-image. These after-images change pro- 
foundly the saturation of the color sensation, also its hue. 
In a given series of experiments in which all other factors 
were carefully controlled the limits for red were narrowed by 
a white preexposure by amounts varying in the different 
meridians from 4°-15°, for yellow 2°-17°, for green 3°-15o, 18 
Ferree and Rand : Color Sensitivity of Retina. 
and for blue 4°-12°; by a black preexposure, red 3°-ll°, 
yellow 3°-10°, green 4°-13°, and blue 2°-12°. In these ex- 
periments the surrounding field was a gray of the brightness 
of the color at the point of investigation, and the results 
obtained with the preexposure of a gray of the brightness of 
the color were taken as the standard in terms of which to 
estimate the amount the field was narrowed by the white 
and black preexposures. 
A background or surrounding field lighter or darker than 
the color produces a similar effect on the limits, but not so 
great. In this case the disturbing achromatic effect is pro- 
duced by physiologic induction or contrast. In some meridi- 
ans the effect of surrounding field alone narrowed the limits 
as much as 11°. A white surrounding field narrowed the 
limits in the different meridians for red by amounts ranging 
from 2°-10°, for yellow 2°-9°, for green 2°-ll°, and for blue 
2°-10°; the black surrounding field narrowed the limits for 
red l°-8°, for yellow l°-8°, for green 2°-10°, and for blue 
2°-9°. 
The combined effect of surrounding field and preexposure 
is, of course, greater than either alone. A preexposure and 
surrounding field of white narrowed the limits for red by 
amounts varying in the different meridians from 4°-17°, for 
yellow 5°-12°, for green 7°-18°, and for blue 5°-18°; a pre- 
exposure and surrounding field of black narrowed the field 
for red from 5°-19°, for yellow 2°-20°, for green 4°-20°, and 
for blue 5°-17°. 
The effect of preexposure and surrounding field on the 
breadth of the color-fields is shown in Figs. 6-8. Although 
the effect of both factors has been determined separately for 
the different colors, a separate graphic representation has 
not been made for each in this paper. In order to save space 
only the combined effect of both has been shown, and for 
only three of the colors-red, green, and blue. 
The Intensity of the General Illumination.-The variable Ferree and Rand : Color Sensitivity of Retina. 
19 
effects both of the preexposure and surrounding field are 
strongly influenced by changes in the intensity of the illumi- 
nation. The results cited above were obtained with a pre- 
cise control of the intensity of illumination. When the 
Fig. 5.-The limits of sensitivity 
to red, yellow, green, and blue of 
the Hering series of pigment papers, 
intensity of illumination, vertical 
component, 390 foot-candles. 
Fig. 6.-The combined effect of 
brightness of preexposure and sur- 
rounding field on the limits of the 
color field. In this chart are shown 
the apparent limits for red with both 
preexposure and surrounding field 
of white, black, and gray of the 
brightness of the color at the point 
of investigation. 
Fig. 7.-The combined effect of 
brightness of preexposure and sur- 
rounding field on the limits of the 
color field. In this chart are shown 
the apparent limits for green with 
both preexposure and surrounding 
field of white, black, and gray of the 
brightness of the color at the point 
of investigation. 
Fig. 8.-The combined effect of 
brightness of preexposure and sur- 
rounding field on the limits of the 
color field. In this chart are shown 
the apparent limits for blue with 
both preexposure and surrounding 
field of white, black, and gray of the 
brightness of the color at the point 
of investigation. 20 
Ferree and Rand: Color Sensitivity of Retina. 
results are obtained under such ranges of change of illumi- 
nation as may occur during the course of a given day or from 
day to day, the variability in effect is greatly increased, 
reaching in some meridians as much as 28°-30°. Further 
important effects of surrounding field as influenced by change 
of illumination are the changes in hue which the color under- 
goes in passing towards the periphery of the retina. For 
example, on bright, dark, and medium days with different 
brightnesses of surrounding field, the red stimulus may be 
seen in any of the following hues in passing from the center 
towards the periphery of the retina: red, red-orange, orange- 
red, orange-yellow, yellow, and dark red; the green stimulus 
as green, blue-green, pale blue, and yellow; the yellow 
stimulus as yellow, orange, orange-yellow, red-orange, and 
pale yellow. Under some conditions of surrounding field 
and illumination the color is seen in the central retina in its 
proper hue, in the mid-periphery in the modified hue, and 
near the limits again in its proper hue. This is true in par- 
ticular of red on a dark day with a white surrounding field 
and of yellow on a dark day with a black surrounding field. 
To those who assign as the limit the first point at which the 
color is no longer seen in its proper hue, such phenomena 
would afford considerable annoyance especially if in deter- 
mining the limit the stimulus were moved from within out 
and from without in. In the cases mentioned the limits for 
red obtained by the two procedures would have differed in 
some meridians as much as 24°, and for yellow as much as 
59°. 
In Figs. 9-19 the effects of such ranges of change of il- 
lumination as are represented by bright and dark days in 
conjunction with preexposures and surrounding fields of 
white, black, and gray of the brightness of the color are 
represented. In these maps the effect, both on the breadth 
of the fields and the hue in which the color was sensed, is 
represented. That is, not only is the total field mapped for Fig. 9.-Field for red. Bright 
day, gray surrounding field and pre- 
exposure. 
Fig. 10.-Field for red. Dark 
day, gray surrounding field and pre- 
exposure. 
Fig. 11.-Field for red. Bright 
day, white surrounding field and 
preexposure. 
Fig. 12.-Field for red. Dark 
day, white surrounding field and 
preexposure. 
Fig. 13.-Field for red. Bright 
day, black surrounding field and 
preexposure. 
Fig. 14.-Field for red. Dark 
day, black surrounding field and 
preexposure. 
21 Fig. 15.-Field for green. Bright 
day, white surrounding field and 
preexposure. 
Fig. 16.-Field for green. Dark 
day, white surrounding field and 
preexposure. 
Fig. 17.-Field for green. Bright day, gray surrounding field and preexposure. 
Fig. 18.-Field for green. Bright 
day, black surrounding field and 
preexposure. 
Fig. 19.-Field for green. Dark 
day, black surrounding field and 
preexposure. 
22 Ferree and Rand : Color Sensitivity of Retina. 
23 
each stimulus, but also the subdivisions in which the different 
hues were sensed. The rotary campimeter, by means of 
which the fields were taken, was so constructed that the 
different hues in which the stimulus was sensed at any point 
in the field could be matched in color and brightness in cen- 
tral vision on a small electric color mixer. The equations 
representing these hues in terms of the Hering standard 
colors and white and black are shown in the maps referred 
to above. Space has been allowed for showing the effect 
on only two of the colors-red and green. 
In order to realize how profoundly the powers of chromatic 
response must have been affected to change the limits of 
sensitivity by the amounts represented in the foregoing 
figures and charts, one must bear in mind how abruptly 
sensitivity falls off in the far periphery of the retina. A 
determination of the threshold of color in the nasal meridian 
with preexposure and surrounding field of the same bright- 
ness as the color shows that for red 173 per cent, more colored 
light was required just to be sensed at the limits than 5° 
inside the limits; for yellow, 260 per cent.; for green, 208 
per cent.; and for blue, 208 per cent. 
In those meridians in which the limits were wide there is a 
general tendency for the white preexposure and surrounding 
field to narrow the limits more than a black preexposure and 
surrounding field. In explanation of this effect it should be 
stated that the amount of inhibition of the chromatic by 
the achromatic excitation varies with the color, the part of 
the retina stimulated, and the state of adaptation of the 
retina. In the central retina at medium and high illumina- 
tions white inhibits color more than black. At these illumi- 
nations a black preexposure and surrounding field are, there- 
fore, much more unfavorable than white. At lower illumi- 
nations this difference in effect becomes less pronounced. 
In the far periphery of the retina, the following are some of 
the conditions which contribute to make black as preexposure 24 
Ferree and Rand: Color Sensitivity of Retina. 
and surrounding field give the wider limits of sensitivity: 
(a) A condition of low illumination and a state of low illumi- 
nation adaptation. (6) A darkening of all of the colors, 
particularly of red and yellow (the Purkinje shift of the 
peripheral retina). This brings the brightness of the color 
nearer to black than to white; and the stronger relative 
darkening of red and yellow than of their neutral or colorless 
preexposures and surrounding fields increases the contrast 
and after-image effect for white and decreases it for black. 
The darkening of red and yellow in passing to the far periph- 
ery of the retina is very great. In the nasal half of the 
retina, with its wide limits, the effect of this darkening on 
the results of our determinations is, of course, the most 
pronounced. As colors darken, there is, when a certain 
point in the process is reached, varying with the color, a 
tendency for them to lose their saturation very rapidly, (c) 
Achromatic induction increases very strongly with decrease 
of illumination and, therefore, increases in passing from the 
center to the periphery of the retina. It increases much 
faster for white than for black. 
In the meridians in which the limits are narrow the situa- 
tion is more nearly as it is in the central retina. Here the 
tendency is for the limits to be narrowed more by a black 
than by a white preexposure and surrounding field. In some 
meridians the amount of narrowing is approximately equal 
for both. Another factor which tends to make the effect 
more nearly the same in these meridians for all backgrounds 
and preexposures is the more abrupt falling off in sensitivity. 
That is, more effect on sensitivity is required here to change 
the limits by a detectable amount than is required in those 
portions of the retina where the sensitivity grades off more 
slowly. 
In conclusion, a word may not be out of place further in 
explanation of the effect of brightness of the preexposure 
and the surrounding field on the limits of color sensitivity. Ferree and Rand : Color Sensitivity of Retina. 
25 
As has already been indicated, this effect falls under the 
general heading of the inhibitive or canceling action of the 
achromatic excitation on the chromatic. In addition to 
this quantitative action there is also a qualitative effect. 
That is, the hue of certain colors is changed by the action 
of the achromatic excitation. This hue change is greatest 
when the stimuli are blue and yellow. For example, yellow 
when mixed with black gives for central vision a greenish- 
yellow which, with the right proportions of components, may 
become an olive-green; and blue when mixed with white or 
light gray gives a sensation of reddish-blue or lavender. 
These actions, both quantitative and qualitative, take place 
however the achromatic excitations are aroused-by the 
admixture of white light, by aften-image, and by contrast. 
It may be strikingly and conveniently demonstrated in the 
following lecture-room experiments: 
(a) Set up side by side on three color mixers discs made up of 
180° of color, e. g., 180° of blue, and 180° respectively of white, 
gray of the brightness of the color, and black. When mixed, 
although the eye receives the same amount of colored light 
from each set of discs, the mixture with black seems to have 
lost but very little, if any, color; the mixture with white is 
a lavender with but little color; and the mixture with gray 
of the brightness of the color, in this case a very dark gray, 
is less saturated than the mixture with black. When differ- 
ent grays are used the saturation decreases apparently in 
graded steps as white is approached. The demonstration 
can be made on a single color mixer by compounding the 
color disc with white, black, and gray discs of different 
breadths or radii. When rotated, this gives the effect of a 
surface made up of three concentric zones or rings one in 
which the color is mixed with white, one with gray, and the 
other with black. The demonstration may be made roughly 
quantitative by determining the proportions of color re- 
quired to give the chromatic thresholds in black, white, and 26 
Ferree and Rand : Color Sensitivity of Retina. 
the grays; also by determining the proportions of color and 
white, black, and gray respectively required to give equal 
saturations. 
(6) Prepare a preexposure surface, half white and half 
black, 60 x 70 cm. Expose the eye fifteen to twenty seconds 
and project the after-image on a colored surface, e. g., blue, 
of the same dimensions. The half of the field preexposed to 
black will appear a very pale unsaturated lavender, while 
the half preexposed to white will be a dark strongly saturated 
blue, although the eye receives the same amount of light 
from both halves of the field. As the after-image dies away 
the two halves of the field become more and more nearly 
alike in saturation and color tone. If desired, the preex- 
posure surface may be made of white, black, and a graded 
series of grays, appropriately arranged. When this is done, 
the graded loss in saturation due to the different brightnesses 
of the after-image may be observed. This demonstration 
also may be made quantitative by finding the threshold of 
color after the eye has been preexposed for fifteen to twenty 
seconds to white, black, and the grays. 
(c) Prepare contrast discs with narrow rings of color and 
inside and outside surfaces respectively of black, white, and 
a gray of the brightness of the color. Set up on color mixers 
side by side, and rotate to smooth out the margins. The 
colors are lightened and darkened respectively by contrast 
induced by the black and white fields. The effect of these 
achromatic excitations on the hue and saturation of the 
colors is similar to those obtained in the former experiments. 
The quantitative features noted above can also be utilized 
in this demonstration by employing for the contrast ring in 
each case a gray of the brightness of the color and enough 
of the color to give the threshold of color sensation when 
acted upon by the white and black induction. The effects 
of induction and after-image, it will be remembered, are not 
nearly so striking in the central as in the peripheral retina. Ferree and Rand: Color Sensitivity of Retina. 
27 
Much more induction with a'given brightness difference 
between the inducing and the contrast fields, for example, 
is produced in the peripheral retina, and only a short period 
of preexposure (two to three seconds) is required to give a 
strong after-image with no latent period. 
For these reasons it is easy to understand why it is so much 
more necessary to control the factors of preexposure and 
surrounding field in the peripheral than in the central retina. 
With a given brightness difference between stimulus and 
surrounding field,the brightness induction is much increased; 
and the after-image reaction to the preexposure in the peri- 
pheral retina is strong and extremely quick. With a very short 
preexposure the after-image flashes out in full intensity im- 
mediately on the cessation of the stimulation. Thus there 
is no possibility of escaping the full effect of the brightness 
after-image on the stimulus color, as might happen in the 
central retina where a latent period obtains, if there were a 
very short exposure to the color. 
BIBLIOGRAPHY. 
C. E. Ferree and G. Rand: Ueber die Bestimmung der Sensibilitat der Retina 
fiir farbiges Licht in radiometrischen Einheiten, Z. f. Sinnesphysiol., 
1911, xlvi, 225-228. 
C. E. Ferree and G. Rand: An Experimental Study of the Fusion of Colored 
and Colorless Light Sensation. The Locus of the Action, Jour. Philos., 
Psych, and Scientific Methods, 1911, viii, 294-297. 
C. E. Ferree: Description of a Rotary Campimeter, Amer. Jour. Psych., 
1912, xxiii, 449-453. 
C. E. Ferree and G. Rand: An Optics Room and a Method of Standardizing 
Its Illumination, Psych. Rev., 1912, xix, 364-373. 
G. Rand: The Effect of Changes in the General Illumination of the Retina 
upon its Sensitivity to Color, Psych. Rev., 1912, xix, 463-490. 
C. E. Ferree and G. Rand: A Note on the Determination of the Retina's 
Sensitivity to Colored Light in Terms of Radiometric Units, Amer. 
Jour. Psych., 1912, xxiii, 328-332. 
C. E. Ferree and G. Rand: Colored After-Image and Contrast Sensations 
from Stimuli in Which No Color is Sensed, Psych. Rev., 1912, xix, 195- 
239. 
C. E. Ferree and G. Rand: The Spatial Values of the Visual Field Immedi- 
ately Surrounding the Blind Spot and the Question of the Associative 
Filling In of the Blind Spot, Amer. Jour. Physiol., 1912, xxix, 398-417. 
C. E. Ferree: A Note on the Rotary Campimeter, Psych. Rev., 1913, xx, 373- 
377. 28 
Ferree and Rand : Color Sensitivity of Retina. 
G. Rand: The Factors that Influence the Sensitivity of the Retina to Color. 
A Quantitative Study and Methods of Standardizing, Psych. Monog., 
1913, xv, (1), pp. 178. 
C. E. Ferree and G. Rand: A Spectroscopic Apparatus for the Investigation 
of the Color Sensitivity of the Retina Central and Peripheral, Jour. 
Exper. Psych., 1916, i, 247-283. 
C. E. Ferree and G. Rand: A Simple Daylight Photometer, Amer. Jour, of 
Psych., 1916, xxvii, 335-340. 
C. E. Ferree and G. Rand: A Substitute for an Artificial Pupil, Psych. Rev., 
1916, xxiii, 380-382. 
C. E. Ferree: The Retinal Sensibilities Related to Illuminating Engineering 
(Discussion), Trans. Illuminating Engineering Soc., 1916, xi, 131-137. 
C. E. Ferree and G. Rand: Radiometric Apparatus for Use in Psychological 
and Physiological Optics-Including a Discussion of the Various Types 
of Apparatus That Have Been Used for measuring Light Intensities, 
Psych. Monog., 1917, xxiv, 66 pp. +xvi. 
C. E. Ferree and G. Rand: The Selectiveness of the Achromatic Response of 
the Eye to Wave-length and its Change with Change of Intensity of 
Light. Studies in Psychology, Titchener Commemorative Volume, Pub- 
lished by L. N. Wilson, Worcester, Mass., 1917, 280-308. 
C. E. Ferree and G. Rand: Some Areas of Color Blindness of an Unusual 
Type in the Peripheral Retina, Jour. Exper. Psych., 1917, ii, 295-304. 
C. E. Ferree and G. Rand: A Note on the Needs and Uses of Energy Measure- 
ments for Work in Psychological and Physiological Optics, Jour. Philos. 
Psych, and Scientific Methods, 1917, xiv, 457-463. 
C. E. Ferree and G. Rand: Chromatic Thresholds of Sensation from Center to 
Periphery of the Retina and their Bearing on Color Theory, Part I, 
Psych. Rev., 1919, xxvi, 16-42; Part II, ibid., 150-163. 
C E. Ferree and G. Rand: The Absolute Limits of Color Sensitivity and the 
Effect of Intensity of Light on the Apparent Limits, Psych. Rev., 1920, 
xxvii, 1-23. 
C. E. Ferree and G. Rand: The Limits of Color Sensitivity: The Effect of 
the Brightness of the Preexposure and of the Surrounding Field, 
Psych. Rev., 1920, xxvii. 
C. E. Ferree, G. Rand, and I. A. Haupt: A Method of Standardizing the Color 
Value of the Daylight Illumination of an Optics Room, Amer. Jour. 
Psych., 1920, xxxi, 77-87. (Reprinted from the Psychological Bulletin, April, 1920, Vol. XVII, No 4 I 
A NOTE ON THE SELECTIVENESS OF THE ACHRO- 
MATIC RESPONSE OF THE EYE TO WAVE-LENGTH 
AND ITS CHANGE WITH CHANGE OF INTEN- 
SITY OF LIGHT 
BY C. E. FERREE AND G. RAND 
Bryn Mawr College 
In Dr. Troland's review of the article by us on this subject 
(Psychol. Bull., 16, 121) errors of statement occur on two im- 
portant points.. (1) "Unfortunately the writers do not speak of 
the size of field employed, so that one can not feel certain that 
strictly foveal stimulation was secured." The sizes of the fields 
employed are given on pp. 283 and 284, and again on p. 304, of 
the article reviewed. (2) "Ferree and Rand have made careful 
measurements of visibility curves at a number of intensity levels 
and find that the form of the curve varies radically with intensity 
even for intensity levels similar to those utilized in previous elab- 
orate investigations by others. . . . The changes in visibility for 
certain wave-lengths due to intensity amount to many hnudred 
per cent." The form of our curves did not vary radically with 
intensity at the levels which have been used in recent determinations 
of standard visibility curves, meant to be independent of intensity. 
The differences at these intensities were significant but not radical. 
The radical differences were for much lower intensities 5 meter- 
candles and under. The impression certainly should not be left 
that the differences amount to many hundred per cent, at the higher 
intensities. It is doubtful if the Purkinje shift can be estimated in 
per cent from our data in the sense referred to by Troland. What 
was given in our tables was a comparison of the photometric and 
radiometric evaluations of our stimuli at the different intensities. 
The deviation of these from exact correspondence of ratio as the 
intensity was changed ranged for the different parts of the spectrum 
used from 10 to 21 per cent, for a change of 75 to 50 meter-candles; 
from 14 to 48 per cent, for 75 to 25 meter-candles; and from 16 to 
50 per cent, for 75 to 12.5 meter-candles. 
134 135 
DISCUSSION 
We are not aware of published results of "elaborate investiga- 
tions of others" contradictory to our own. The recent belief that 
the Purkinje shift ceases at 25 meter-candles or thereabouts seems 
to refer back to statements made by Ives and Nutting. Ives 
claimed that at "approximately 25 meter-candles" (300 meter- 
candles falling on a pupillary aperture of I sq. mm.) the achromatic 
response is practically, if not entirely free from Purkinje effects; 
and Nutting, that an illumination of 350 meter-candles falling on a 
pupillary aperture of 1.465 sq. mm. is safely outside the range of 
the Purkinje effects. Neither man cites results in support of his 
claim. Moreover the photometric determinations in the work in 
which these statements occur were made by the flicker method; 
ours were made by the equality of brightness method, or as the 
eye normally sees its brightnesses. There are good physiological 
reasons, also experimental data, for not expecting agreement by 
the two methods. Helmholtz and others of the earlier writers 
(Chodin, Sammlung phys. Abhandl. v. Preyer, 1877, I, p. 33, fl., 
Briicke, Sitzungsber. der Wiener Akad., Math.-Natur. Klasse, 
1878, (3), p. 63; etc.) believed that the eye changes its selectiveness 
of response to wave-length of light at the higher as well as at the 
lower intensities. This conclusion is drawn from a statement made 
by them that beginning with a spectrum of fully saturated colors 
and increasing the intensity of light, all the colors tend towards 
white and in so doing change their luminosities at different rates. 
The question with regard to the intensity at which the Purkinje 
shift ceases, if at all, should be carefully studied before still more 
effort is expended on determining visibility curves meant to be 
used at 25 meter-candles and over. A METHOD OF STANDARDIZING THE COLOR 
VALUE OF THE DAYLIGHT ILLUMINATION 
OF AN OPTICS ROOM 
By C. E. Ferree, G. Rand, and I. A. Haupt, Bryn Mawr College. 
In a previous article (Psychol. Rev. 1912, XIX, pp. 364- 
373) we have given a method of standardizing the intensity of 
light in an optics room illuminated by daylight and have de- 
scribed provisions for keeping constant the intensity selected 
as standard. If still more exact conditions are wanted for 
conducting work on color sensitivity another variable, namely, 
the changing color of daylight, should perhaps be taken into 
account. It is the purpose of this paper to describe a method 
of standardizing the color value of the daylight illumination 
of a room and of correcting the changing values to the stand- 
ard value in such optics rooms as we have in our laboratory. 
We mean by standardizing here only a means of matching 
the daylight selected as desirable and of reproducing at any 
given time as a standard the color value which matches the 
daylight selected. Our plan does not include primarily a 
means of determining what is the most desirable color value of 
daylight to use, although that might well be made a valuable 
feature of the plan. To be of the greatest service in psycho- 
logical optics the plan should be both reasonably simple of 
accomplishment and have a fair degree of precision. In devis- 
ing a method we have endeavored to keep both of these re- 
quirements in mind. Obviously the first step to accomplish 
is to have a source of light the color value of which can be 
kept constant within acceptable limits. A well seasoned tung- 
sten light operated by a constant current gives a light of suffi- 
cient constancy of composition to serve our purpose.1 It has 
1 In reply to an inquiry about the constancy of color value of the 
light emitted by the tungsten filament operated at a constant voltage, 
also its constancy as compared with the light emitted by the carbon 
filament, we have the following from the Bureau of Standards. "With 
regard to the constancy of the color from lamps operated at a con- 
stant voltage, it can be said in general that almost any seasoned in- 
candescent lamp will burn a considerable time without any appreciable 
change in the color of the light, unless the filament is operated at a 
temperature above the normal. In comparing carbon with tungsten 
filaments the relative constancy will depend on the conditions. If the 
two are set at voltages which will give the same color, the tungsten 
will change much more slowly than the carbon In fact a tungsten 
filament operated at the normal color of carbon lamps will usually 
77 78 
FERREE, RAND, AND HAUPT 
the advantage too that its changes of color with change of 
voltage are roughly similar in direction to those which occur 
in daylight. A second step is to have a means of changing 
the color value of the light to match the daylight selected as 
desirable and a comparison surface on which to make the 
match. The first of these purposes can be accomplished by 
means of thin gelatine filters of known spectrum transmis- 
sion, properly selected and combined to give the color 
values needed; and the second by means of any photom- 
eter head which presents a sufficiently good field for 
the comparison. We have already described a simple and 
inexpensive photometer for daylight work which with the 
proper selection of filters can also be made to serve the present 
purpose very well. For a description of this photometer the 
reader is referred to this Journal, 1916, XXVII, pp. 335-340. 
When the proper filters are inserted on the side next to the 
standard lamp this photometer can be used to make both the 
photometric and colorimetric comparison with no change either 
in the apparatus or the adjustment of the lamp. With the in- 
strument set for the color and intensity of light selected as 
standard it can be determined at a glance whether the illumi- 
naton of the room fulfills the desired requirements with regard 
to both of these features. 
We have stated that the method employed should have a 
satisfactory degree of precision and sensitivity. In order to 
get some estimate of the sensitivity of the method or what 
amounts of color change can be detected at different intensities 
of illumination, the following experiments were conducted. 
(1) The light from two well seasoned tungsten lamps of equal 
watt value (40-watt, type B Mazda) was brought to a color 
and brightness match at the photometer head. The change in 
voltage required to produce a just noticeable change in color 
tone was then determined. The work was conducted in a 
dark room and for convenience in getting the proper range of 
intensities the photometer head was removed from its place 
in the box photometer and mounted on an ordinary photo- 
meter bar. In making the determinations one lamp, A, was 
operated at 107 volts and set at the desired distance from the 
photometer head. The other lamp, B, was operated at such 
a voltage and set at such a distance from the head as was 
needed to give an exact color and brightness match. When 
burn thousands of hours with no appreciable change in color. Whether 
different lamps of the same kind give light of the same color depends 
on how closely the lamps are kept uniform in manufacture. As fur- 
nished on the market at present different tungsten lamps of the same 
size will be found to be more nearly alike than carbon lamps." THE COLOR VALUE OF DAYLIGHT 
79 
the match was obtained the voltage of lamp A was lowered 
until a just noticeable difference in color tone could be detected 
at the screen. A measure was thus had of the amount of color 
change that could be detected by the instrument in terms of 
the amount that is produced by a given variation of the 
voltage of a tungsten lamp the specification and the con- 
ditions of operation of which are known. When the voltage 
of lamp A was lowered the intensity as well as the color value 
of the light at the photometer head was changed. This re- 
quired a resetting of the lamp B to bring the two photometric 
fields to equal brightness. However, as a check on the judg- 
ment of just noticeable difference in color tone each observa- 
tion was made under more than one brightness relation be- 
tween the two photometric fields. The judgment was not 
difficult to make. Its precision in fact compares very favor- 
ably with that of the photometric judgment. The determina- 
tion was made with the lamps at different distances from the 
photometer head,-14, 20, 30, 40 and 50 cm. to give the differ- 
ences in intensity at which it was desired to make the deter- 
minations of colorimetric sensitivity. For the sake of a com- 
parison of the colorimetric sensitivity of the photometer heads 
more commonly used, the determinations were repeated with 
Lummer-Brodhun heads of the contrast and disappearance 
types. The results of these determinations are given in 
Table I. 
TABLE I* 
A Comparison of the Sensitivity of Different Photometer Heads 
To Change in Color of Light of Tungsten Lamp (Madza Type B) 
Operated at 107 Volts 
Voltage Change of Lamp to Give Just Noticeable 
Change of Color Tone 
Distance of Lamp 
From Photo- 
meter Head 
(cm.) 
Lummer Brodhun 
Head Contrast 
Type 
Lummer Brodhun 
Head Disappear- 
ance Type 
Bunsen Head 
14 
0.25 
0.5 
1.0 
20 
0.50 
1.0 
1.5 
30 
0.75 
1.5 
2.5 
40 
1.00 
3.0 
3.0 
50 
1.25 
3.5 
3.0 
*The results of this and the following tables are for Observer R. The 
main points were verified by a check observer, H, 80 
FERREE, RAND, AND HAUPT 
(2) In the second series of experiments a comparison was 
made of the sensitivity to change of color tone with the three 
photometer heads, the Bunsen and the two types of Lummer- 
Brodhun, when the photometric surfaces were illuminated by 
the tungsten lamps, natural color, and when they were illumi- 
nated by the light from these lamps filtered through gelatines 
so selected as to match the daylight in one of our optics rooms 
at 11 A. M. on a clear day. The intensity of light at the 
photometer head was made the same for both kinds of illumi- 
nation. The comparison was made at three intensities of 
illumination, corresponding to those given by the tungsten 
lamps in the preceding set of experiments when placed at 30, 
40 and 50 cm. from the photometer head. The comparison 
was not made for the other intensities because they could not 
be obtained on our photometer bar on account of the reduction 
in intensity produced by passing the light through the daylight 
filters. That is, the lamps could not be brought nearer to the 
photometer head than 14 cm. and to match the tungsten lamp 
at 30 cm. from the head, for example, the lamp in front of 
which the filter was placed had to be set at a distance of 14.3 
cm. The results of this comparison are given in Table II. 
TABLE II 
A Comparison of the Sensitivity of Different Photometer Heads 
to Change in Color of Light of Tungsten Lamp (Madza Type B) 
Operated at 107 Volts and of This Light Filtered to 
Match Daylight 
Distance of Lamp 
From Photometer 
Head Giving Equal 
Illumination for 
Each Intensity for 
Voltage Change of Lamp to Give Just Noticeable Change of 
Color Tone 
Type B Mazda Lamp 
Type B 
I 
Mazda Lamp With 
laylight Filter 
Type B 
Type B 
Lummer 
Brodhun 
Lummer 
Brodhun 
Bunsen 
Head 
Lummerj 
Brodhun 
Lummer 
Brodhun 
Bunsen 
H ead 
Mazda 
Lamp 
(cm.) 
30 
Mazda 
Lamp 
With 
Daylight 
Filter 
(cm.) 
13.5 
Head 
Contrast 
Type 
0.75 
Head 
Disap- 
pearance 
Type 
1.5 
2.5 
Head 
Contrast 
Type 
0.25 
Head 
Disap- 
pearance 
Type 
0.50 
0.75 
40 
18.8 
1.00 
3.0 
3.0 
0.25 
0.50 
1.00 
50 
25.2 
1.25 
3.5 
3.0 
0.25 
0.75 
1.50 
(3) A somewhat limited comparison was made of the^sen- 
sitivity of the Bunsen head to change in color tone when the 
photometric surfaces were neutral and when they were 
colored. This was done because the color of some pigments is 
known to be very sensitive to changes in color value of the THE COLOR VALUE OF DAYLIGHT 
81 
light falling upon them. For the purpose of making this com- 
parison a number of such standard pigments as are commonly 
found in psychological laboratories were substituted for the 
white screen of the Bunsen head. Two cases were made of 
this investigation, (a) Six colored screens were used, selected 
from the Hering series of papers: the dark red, the orange, the 
yellow, the yellowish green, the blue-green and the violet; and 
three from the Milton-Bradley series: the red-violet, the red- 
violet, tint No. 1 and the orange-red, tint No. 1. The light 
from the 40-watt lamp was in each case passed through the 
daylight filter referred to above. The brightness of the photo- 
metric surfaces was kept constant in all cases at a value equal 
to that of the white screen illuminated by the filtered light of 
the tungsten lamp at a distance of 74 cm. This was the highest 
brightness that could be obtained with this lamp for the col- 
ored screen having the lowest coefficient of reflection. This 
intensity of illumination was selected in order that the colori- 
metric comparison should be made in all cases for the same 
brightness of surfaces compared. The results of this com- 
parison are given in Table III. (b) Since in the preceding 
TABLE III 
A Comparison of the Colorimetric Sensitivity of the Bunsen Photo- 
meter Head When Provided With White and Colored Fields- 
This Comparison was Made With the Intensities of Light 
Adjusted to Give in All Cases the Same Brightness of 
Photometer Fields, Namely, the Brightness of the Blue 
Pigment (Pigment With Lowest Coefficient of Re- 
flection) Illuminated With the Highest Intensity 
That Could Be Obtained With the Filtered Light 
of the Mazda Type B LampOperated at 107 Volts 
Photometer Field Distance of lamp (cm.) Voltage change of 
from photometer head lamp to give 
giving brightness of just noticeable 
colored fields equal to change of color 
that of white field il- tone, 
luminated by this lamp 
with daylight filter at 
74 cm. 
Yellowish-green 
43.0 
1.0 
Violet 
15.1 
2.0 
Red-violet 
22.6 
3.0 
Orange-red, tint 
No. 1 
40.0 
3.0 
Red-violet, tint 
No. 1 
33.5 
4.0 
Yellow 
50.8 
4.0 
Orange 
38.0 
5.0 
Dark red 
31.5 
6.0 
Blue 
13.8 
6.0 
Blue-green 
27.3 
8.0 
White 
74.0 
4.0 82 
FERREE, RAND, AND HAUPT 
experiments the pigments of the higher reflection coefficients 
had to be used at lower illuminations than the other pigments 
in order to fulfill the conditions that the colorimetric com- 
parison should be made in all cases on surfaces of equal 
brightness, it was decided to make a comparison of the most 
favorable of these colors with the neutral screen at illumina- 
tions approximately equal to that used for the color of lowest 
reflection coefficient in the former experiments. This, it will 
be remembered, was the highest that could be obtained with 
the filtered light of the 40-watt lamp. For this purpose the 
yellowish green, the yellow and orange-red, tint No. 1, screens 
were used. The results of this comparison are given in Table 
IV. These results, it scarcely need be pointed out, have a 
TABLE IV 
A Comparison of the Colorimetric Sensitivity of the Bunsen Photo- 
meter Head When Provided With White and the More Sensi- 
tive Colored Fields Used in Table III, at Intensities of Illum- 
ination Equal Approximately to the Highest That Could be 
Obtained With the Filtered Light of the Type B Madza 
Lamp Operated at 107 Volts 
Photometer Field Distance of lamp (cm.) Voltage change 
from photometer head of lamp to give 
giving equal brightness just noticeable 
of white and colored change of color 
fields for Mazda type B tone 
lamp with daylight filter 
Yellowish-green 
16.5 
0.25 
White 
26.5 
1.50 
Yellow 
13.5 
1.50 
White 
20.0 
1.0 
Orange-red, tint No. 
1 
15 
1.50 
White 
22 
1.25 
Violet 
15.1 
2.0 
White 
74 
4.0 
much more direct bearing on the problem of selecting screens 
for our standardizing instrument than those of the former 
comparison. That is, in the selection of screens for such in- 
strument we are concerned with relative sensitivities at equal 
illuminations, not equal brightnesses, of screen. Obviously in 
the final selection of screens for any particular instrument that 
screen should be chosen which shows the greatest sensitivity 
for the range of illuminations possible for the instrument or 
for the range that is likely to be used. THE COLOR VALUE OF DAYLIGHT 
83 
An inspection of the results in Tables I-IV shows the follow- 
ing points. 
(1) As might be expected smaller color changes could 
be detected at the higher than at the lower intensities of illu- 
mination. That is, at the higher intensities the color of the 
light was less saturated (the inhibitive action of the achro- 
matic on the chromatic component of the retinal excitation) 
therefore a smaller change was needed to be just noticeable. 
(2) The colorimetric sensitivity of the photometer heads 
employed is in the following order from greatest to least,- 
the Lummer-Brodhun, contrast type; the Lummer-Brodhun, 
disappearance type; the Bunsen. As bearing on the type of 
field that gives high colorimetric sensitivity such comparisons 
are of importance to colorimetry by the monochromatic 
method. 
(3) Smaller changes of voltage were required to produce 
a noticeable change in the color of the light filtered to match 
daylight than in the color of the unfiltered light of the Mazda 
type B lamp. 
(4) Of all the colored screens employed the yellowish- 
green, the violet, the yellow and the orange-red, tint No. 1, 
showed greater sensitivity to changes of color of light of the 
type produced in these experiments (changes similar to those 
which occur in daylight) than the white screen when the com- 
parison was made at equal brightness of screen. When, how- 
ever the comparison was made at illuminations approximately 
equal to the highest illumination that could be obtained with 
the filtered light of the 40-watt lamp only the green showed a 
greater colorimetric sensitivity than the neutral screen. The 
green screen seemed to be peculiarly sensitive to the changes 
which are produced by adding the longer wave-lengths to 
daylight. That is, the adding of the red wave-lengths ap- 
peared to decrease the saturation of the green and the adding 
of the yellow wave-lengths, to increase the yellow component 
already present. At the highest intensity of illumination used 
a change of only 0.25 volt was needed to cause a noticeable 
change in the color of the screen. Indeed at this intensity of 
illumination the colorimetric sensitivity of the Bunsen head 
was so much improved by the substitution of the green screen 
as to equal that of the more sensitive of the two Lummer- 
Brodhun heads with the white screen. 
As we have already indicated the work of standardizing 84 
FERREE, RAND, AND HAUPT 
for color value and intensity may be done at the same time. 
It may be accomplished as follows: A daylight is selected of 
the color value and intensity desired. On the side of the 
photometer head next to the standard lamp gelatine filters are 
inserted so chosen that when the light from this lamp run at 
a given voltage and set at the position on the bar needed to 
give the intensity match, is filtered through them, the photo- 
metric surfaces illuminated by this light match in color 
value the surface illuminated by the daylight. To reproduce 
this standard at any future time all that is needed is to reset 
the lamp in the same position on the bar and to reproduce the 
voltage. With the photometer set up at the point of work 
in the room or as near to it as possible, the process of check- 
ing up the illumination both as to color value and intensity 
becomes very simple. A glance at the photometric field is 
sufficient to give the desired information. Since it is not so 
easy to arrange for the correction of daylight to the color value 
chosen as standard as it is to correct the intensity, feasibility 
may dictate that the work be done within a certain range of 
variation of color value. The apparatus recommended may 
be used to standardize this range as follows: If the day- 
light chosen as desirable be that of skylight near the middle of 
a clear day the changes that are apt to occur during the course 
of the day or from day to day may be approximated roughly by 
lowering the voltage of the standard lamp or by lowering the 
voltage supplemented by the addition of one or more of the thin 
gelatine filters properly selected. When this is done and the 
lamp is reset to compensate for the change of intensity pro- 
duced, a range of variation of color value is fixed within which 
the daylight illumination must fall or be rejected for the partic- 
ular work in hand. In such a case the photometer is set so that 
the standard surface in the photometer head is illuminated 
with the light chosen as the limit towards which the color value 
may vary and still be accepted for the work in hand. It is very 
easy then at any particular time to judge whether the domi- 
nant color of the daylight incident upon the receiving surface 
of the photometer falls within the limiting value. 
Obviously a means of correcting the changing daylight to 
the color value chosen as desirable would be an advantageous 
supplement to the work of standardizing. The method we 
purpose for use for this is as follows: At a distance above 
the diffusion sash of ground glass installed beneath the sky- THE COLOR VALUE OF DAYLIGHT 
85 
light in our optics rooms, sufficient to give a good spread of 
light, will be installed opaque pendant reflectors of the dis- 
tributing type. These reflectors will be supplied with filters 
which transmit an excess of the short wave-lengths. They 
will be installed on separate circuits, one or more to the cir- 
cuit, so that a variable proportion of the artificial light may 
be used as is needed. Further to vary both the composition 
and the amount of artificial light used, wall rheostats will be 
included in as many of the circuits as is found to be advisable. 
Quite a wide range of composition and intensity of light can 
be obtained also by the use of lamps of different types and 
wattages. The artificial units may be installed well out of the 
road and the desired direction and throw of light be obtained 
by the shape of the reflector and the angle at which it is in- 
stalled. Further the amounts of daylight used in getting the 
desired composition may be controlled by a system of curtains 
placed under the skylight above the diffusion sash and the 
artificial units. The elaborate system of curtains which is 
already installed beneath the diffusion sash in our rooms will 
serve as they do now for fine gradations of intensity of light 
and may prove useful to some extent perhaps for the control 
of the composition. 
An alternative to a flexible system of correction of actual 
daylight for both composition and intensity is the utilization 
of some one of the artificial daylights which are now on the 
market. Of the unfiltered sources the carbon dioxide tube 
gives perhaps by common agreement the closest approximation 
to skylight. Its cost, however, is prohibitive for the greater 
number of laboratories. The same thing might well be said of 
the best of the filter units. The ones most familiar to us all and 
the most available from the standpoint of cost are the blue 
bulb lamps. With regard to these lamps, however, only a 
rough approximation to daylight is claimed. We have thought 
that it may be of interest to show here a spectrophotometric 
comparison of one of them, the type C-2 Mazda lamp, and of 
some of the closer approximations to daylight, with the black 
body at 5000 degrees absolute which is sometimes taken as 
the standard of average daylight. (See Fig. I.) For the 
( mparative curves given in this figure we are indebted to the 
Electrical Testing Laboratories, 80th street and East End 
avenue, New York City. The comparison here, it will be re- 
membered, is photometric, not colorimetric. 86 
FERREE, RAND, AND HAUPT 
Fig. I 
A-Black body at 5000 degrees absolute ("Average Daylight") 
B-Blue sky (Ives) I. E. S. Transactions, 1910, p. 208. 
C-Daylight glass with Mazda C lamp (Brady) I. E. S. Transactions. 
IQU. P- 952. . 
D-Bluish glass with Mazda C lamp (Sharp) I. E. S. Transactions, 
1915, p. 220. 
E-Mazda O lamp. 
F-Mazda B lamp (7.9 lumens per watt). 
G-Mazda C lamp (20 lumens per watt). 
H-Moore tube (Paper read before I. E. S. November 11, 1915). 
I-Trutint glass, (Luckiesh) I. E. S. Transactions, 1914, p. 839. 
J-Trutint glass, (Luckiesh) I. E. S. Transactions, 1914, p. 839. Reprinted from Journal of Experimental Psychology. Vol. Ill, No. i, Feb., iq2O.] 
AN APPARATUS FOR DETERMINING ACUITY AT 
LOW ILLUMINATIONS, FOR TESTING THE 
LIGHT AND COLOR SENSE AND FOR DETECT- 
ING SMALL ERRORS IN REFRACTION AND IN 
THEIR CORRECTION 
BY C. E. FERREE AND GERTRUDE RAND, 
Bryn Mawr College 
This apparatus was devised in response to a request by the 
Eye Division of the U. S. Naval Hospital for a means of making 
a quick and accurate test of acuity at low illumination. Expe- 
rience has shown, roughly speaking, that only 25-30 per cent, 
of the men on the battleships have a sufficiently keen acuity 
at low illuminations to qualify for all branches of the lookout 
and signal service work. The apparatus provides for a wide 
range of illumination in just noticeably different steps (be- 
ginning at 0.07 meter-candle or lower), with no change in the 
color value of the light and with a specification at each step 
of the intensity of light falling on the test-object. 
Among the requirements for an apparatus for determining 
acuity at low illuminations or the effect of change of illumina- 
tion, the following points may be mentioned. (1) A means 
of changing the illumination by small amounts over a wide 
range, beginning at or below the threshold for the test-object 
employed, without changing the color value of the light. 
If in making this change the color value of the light is altered 
it is obvious that another factor affecting the results is intro- 
duced. (2) A means of keeping constant for an indefinite 
length of time any desired intensity of illumination and of 
reproducing this intensity at will. Also the test-object must 
be uniformly illuminated. (3) A means of specifying accu- 
rately at any point in the scale the intensity of light falling 
on the test-object. And (4) it is desirable that the apparatus 
employed for controlling the illumination can be used with 
the test-objects already accepted in clinic practice. 
59 60 
C. E. FERREE AND GERTRUDE RAND 
The most difficult problem one has to face in constructing 
an apparatus for determining the min'mum amount of light 
that permits of the discrimination of a given test-object, more 
particularly if that object consists of a line of test letters, is 
to secure a uniform illumination of the line. This problem is 
relatively unimportant at the illuminations ordinarily used 
in acuity testing, because at these illuminations acuity changes 
sp slowly with chan e of intensity of light that the differences 
which may occur throughout the line of test letters do not 
ordinarily produce a detectable effect on the results of the 
test. However, if no more care is exercised at the threshold 
to secure uniformity of illumination than is ordinarily used 
at the higher illuminations, no single intensity at the source 
will serve for the discrimination of all of the letters of the line. 
We were able satisfactorily to meet this difficulty in only one 
way, namely, by selecting an aperture sufficiently small to 
permit of its uniform illumination and projecting a magnified 
image of this aperture on the test card. That is, an aperture 
was selected of such a size and shape that when magnified 
fivefold it gave a band of light which just blocked off one 
line of the test letters. It is obvious that this aperture could 
be made of different sizes and shapes, depending upon what 
is wanted in the projected image. For example, two or three 
lines of test letters could be blocked off if desired, or the 
whole card or any part of it could be illuminated, etc. There 
is no reason, moreover, why the aperture could not be made 
adjustable in size to suit the needs and preferences of the 
individual operator. In one model of the apparatus these 
apertures were cut in a series of slides which could be inserted 
in the projection tube just outside the lamp house in grooves 
in a light-tight boxing. A convenient means was thus pro- 
vided for changing the aperture, if desired, during a series of 
tests without having to open the lamp house. The source 
of light is a well-seasoned Mazda C lamp of the round bulb 
or stereopticon type of 100, 250 or 500 watts, depending upon 
the range of illumination that is desired. This lamp is 
installed vertically in the roof of the lamp house at such a 
height that its filament is well above the aperture which is AN APPARATUS 
61 
to be illuminated. In order to secure a uniform and diffuse 
illumination of the aperture the lamp house is lined with 
opal glass ground on one side. The aperture, 6 x I cm., is 
cut at the center of the cap covering the inner end of the pro- 
jection tube. Further to aid in the even illumination of the 
aperture, it is covered with a slide of ground glass. To 
prevent the overheating of the lamp house, a rather elaborate 
ventilating system is provided consisting of a light-tight 
ventilating hood at the top and a series of holes on two sides 
at the bottom of the housing, furnished with light-tight 
shields. The changes in the intensity of light are produced 
by means of an iris diaphragm. When such a diaphragm is 
placed either at the front or back surface of the focussing 
lens, changes in the flux of light can be produced without any 
alteration in the size or the shape of the image produced by 
the lens, just as happens, for example, in the action of the 
iris of the eye. At a suitable point in the circumference of the 
diaphragm is fastened a pointer which, as the diaphragm is 
opened and closed, moves over a translucent millimeter scale. 
This scale is mounted over a slot in the projection tube and 
receives its illumination from the light inside of the tube. 
The inside of the tube is painted a mat black. At the further 
end of the projection tube, 18.1 cm. from the illuminated 
aperture, in a brass ring and collar is mounted the focussing 
lens. This lens is 7.5 cm. in diameter and has a focal length 
of 14.8 cm. A different strength of lens could have been used 
and different relative distances of aperture and test card from 
the lens. With appropriate variations in these factors the 
distance of the lantern from the test card and the amount of 
magnification of the projected image may be varied. Any 
increase of magnification results of course in a decrease in the 
brightness of the image, hence an increase in the scale of 
brightness of image with no change in its size could have 
been obtained by increasing the size of the aperture and 
decreasing correspondingly the amount of magnification. In 
the construction of the present apparatus a 14.8 cm. focal 
length lens was used because it could be obtained the most 
readily on the market of the diameter needed. On the plat- 62 
C. E. FERREE AND GERTRUDE RAND 
form supporting the lamp house are mounted a small Weston 
ammeter and a small rheostat to guard against fluctuations 
in the current and consequent fluctuations in light intensities. 
In order that any line of the chart may be illuminated at 
will, the lamp house is mounted on the end of a rod which is 
raised and lowered by means of a rack and pinion. The 
test card is mounted at a distance of 81 cm. from the focussing 
lens. A photograph of the apparatus is shown in Fig. I. 
Fig. i. 
In order that the intensity of light used at any time may 
be known, a calibration chart is provided in which are given 
the readings on the millimeter scale and the equivalent meter- 
candle values at the test card. This calibration was accom- 
plished as follows: The lamp house was removed and mounted 
on a photometer bar at a distance from the photometer head 
equal to its original distance from the test card. The scale AN APPARATUS 
63 
was then gone over point by point and the meter-candle 
value of the light at the photometer head was measured. 
The calibration chart is shown in Fig. 2 a. In Fig. 2 b is 
shown the calibration curve in which the divisions of the 
millimeter scale are plotted against meter-candles at the test 
card. These values are corrected to conform at the center 
of the card to the cosine law. 
Diaphragm Setting 
Meter-candles 
Diaphragm Setting 
Meter-candles 
82-5 
9-19 
46.O 
2.46 
82.0 
9-04 
44.O 
2.23 
80.0 
8-54 
42.O 
2.02 
78.O 
8.06 
40.0 
I.8l 
76.O 
7-58 
38.O 
I.6l 
74-o 
7.12 
36.O 
1-43 
72.0 
6.67 
34-o 
1.27 
70.0 
6.23 
32-0 
1.12 
68.0 
5.81 
3°-o 
o-97 
66.0 
5-43 
28.0 
• 0.82 
64.0 
5.06 
26.0 
0.69 
62.0 
4-7i 
24.0 
0.58 
60.0 
4-36 
22.0 
0.48 
58.0 
4.02 
20.0 
0.38 
56.0 
3-72 
18.0 
0.28 
54-0 
3-45 
16.0 
0.21 
52-0 
3.18 
14.0 
0.16 
SO-o 
2-93 
12.0 
0.11 
48.0 
2.69 
10.0 
0.07 
Fig. 2a. Calibration Table. 
For our own use in the laboratory we have preferred to 
substitute for the Snellen chart a single test character, the 
broken circle (the international test object), which can be 
turned in different directions and the judgment of its direction 
rather than the recognition of the character be required of the 
observer as a test of discrimination. Our reasons for this 
preference are as follows: (i) A test letter may be recognized 
when it is not seen at all clearly. Recognition is too de- 
pendent on extraocular functions to be used with precision 
as a measure of ocular capacity. (2) The different letters of 
the Snellen chart set an unequal task for the resolving power 
of the eye. (3) An objective check 's had on the judgment. 
This is especially helpful in case of children, and the unin- 
telligent, untrained and subjective type of adult. (4) By 
the use of the same test character, turned in different direc- 64 
C. E. FERREE AND GERTRUDE RAND 
tions at will, all possibility of learning the test series is 
eliminated. Also the test-object becomes much more valu- 
able for the detection of astigmatisms. And (5) at low 
Fig. 2b. Calibration Chart. 
illuminations the eye fatigues very rapidly. Thus if the 
task is the reading of the whole line of letters the results 
obtained measure not only acuity, but the power to sustain 
acuity which may or may not be compatible with the purpose 
of the test. 
As stated in the introduction, the apparatus was designed 
to meet a specific testing need of the Navy. However, it has 
in addition the following laboratory and clinic uses, (i) 
Photopic acuity may be tested under the conditions of a 
constant and uniform illumination of known intensity. In 
case the test-object is a line or chart of letters, provision is 
made that each letter receives, within sensible limits, equal 
amounts of light. (2) Scotopic or twilight vision may be 
tested-also the amount and rate of scotopic adaptation. 
A precise and feasible means is thus afforded for testing the 
light sense insofar as it affects the power to see clearly. (3) AN APPARATUS 
65 
If the image of the aperture is projected on a blank white 
surface of good reflecting power, more particularly if its size 
and shape are changed to that of a square or circle of dimen- 
sions favorable for making a sensitive judgment, the apparatus 
can be used with an equal degree of precision and convenience 
for testing the light ^ense directly in terms of the amount of 
light required to arouse just noticeable sensation. That is, 
the threshold of sensation can be determined in terms of 
meter-candles of light falling on the test surface; or, knowing 
the coefficient of reflection of this surface and the breadth of 
pupil, in terms of the amount of light entering the eye. The 
testing of the light sense will probably always remain founda- 
tion work in the clinic routine. The testing of scotopic 
acuity, for example, is not sufficiently differential in all cases 
between refraction defects and the hemeralopias and other 
retinal deficiencies to serve as a satisfactory substitute. 
There is, at present, we are told on very good authority, no 
satisfactory instrument for testing the light sense available 
to the ophthalmologist. And (4) by making it possible to 
determine with great exactness the minimum of illumination 
at which the test-object can just be discriminated, the appa- 
ratus provides a very sensitive means for detecting small 
errors in refraction and in their correction, as will be demon- 
strated in a later paper. 
Sensitivity for detecting small errors in refraction and in 
their correction could also be added to the acuity test by 
using a test-object, small changes in the size of which could 
be made. However, no means has as yet been provided for 
making small changes in the size of any acuity object suffi- 
ciently complicated in form to test simultaneously the re- 
solving power of the eye in any great number of meridians 
which is, we believe, a very important feature in determining 
the exact location of an astigmatism or the exact amount and 
placement of its correction, more particularly when a cyclo- 
plegic is not used. 
We consider this an important feature in testing for an 
astigmatism because of our belief that the astigmatic eye in 
the attempt to compensate for its defect has in many cases 66 
C. E. FERREE AND GERTRUDE RAND 
at least acquired unusual powers and habits of accommoda- 
tion. Our belief in this is based on three sets of observations, 
(i) In the use of the astigmatic charts without a cycloplegic 
in cases of low astigmatism, one is frequently annoyed by the 
astigmatic indication shifting from one meridian to another. 
(2) We have found observers who could voluntarily, in some 
cases requiring considerable practice, shift the meridian show- 
ing the astigmatic indication. And (3) in our test of astigma- 
tism based on the relative speed of discrimination in the 
different meridians, American Journal of Ophthalmology, 19.18, 
I., pp. 3-16, the speed of discrimination for the least favorable 
meridian can be increased by practice almost to equal that 
in the most favorable meridian with an equal amount of 
practice. This result is so noticeable that in order to make 
the test sensitive we were compelled to eliminate as much as 
possible the opportunity for the practice effect. This was 
done in two ways. (1) The series was always begun below 
rather than above the minimum time of exposure required 
just to detect the direction in which the test-object was 
turned. And (2) meridians were inserted in the series clearly 
outside of the region of maximum astigmatic effect in order 
to break up any progressive tendency to accommodate espe- 
cially for the meridian showing the poorer resolving power. 
The fact that the eye can with long exposures discriminate a 
fineness of detail in its unfavorable meridian which it is 
utterly unable to master with short exposures and that this 
excess lag can be overcome in a considerable measure by 
practice seems to indicate that it has the power through its 
active accommodation to overcome in part the effect of 
meridianal inequalities in resolving power, at least when a 
test-object taxing the resolving power in only one meridian 
is turned successively into different meridians. In any event 
it seems only the part of sound procedure in testing without 
a cycloplegic to guard against the possibility of selective 
meridianal accommodation by the use of a test-object which 
taxes the resolving power of the eye in as many meridians 
as possible. 
But even if a test-object complicated in form and minutely AN APPARATUS 
67 
adjustable in size were available, a device for determining 
the minimum illumination at which the test-object subtending 
the standard visual angle can just be discriminated would 
afford a still more sensitive means for the detection of low 
astigmatisms and small errors in the amount and placement 
of their correction. This follows rather obviously from the 
fact that for all but very low intensities acuity changes slowly 
with change of illumination. That is, for all but very low 
intensities small differences in acuity correspond, compara- 
tively speaking, to large changes in illumination. Used as 
we have recommended, the illumination scale becomes in 
effect an amplified indicating scale by means of which the 
relatively slight differences in acuity, represented by the 
proper correction of an error in refraction and small deviations 
therefrom, may be detected with comparative ease and cer- 
tainty. It is not infrequent perhaps to find that in cases of 
low astigmatism, with the full illumination of a test-object 
presenting no smaller gradations in visual angle than are 
found in the Snellen chart, the observer is able to detect no 
difference in the ease or clearness of discrimination of the 
test character through a range of from 20-40 degrees in the 
placement of the correction. This difficulty is especially 
annoying in the case of children and the unintelligent, un- 
trained and subjective type of adult. In such cases the 
apparatus shown here is especially helpful. With it a mini- 
mum is left to the comparative and observational powers of 
the subject. All that he is required to do is to indicate the 
direction in which the test-object points, the most favorable 
amount and placement of the correction being determined by 
the minimum amount of illumination at which he is able 
correctly to give this indication. The apparatus possesses 
ample sensitivity, as our results will show, for the detection 
of an error of 5 degrees and less in the placement of the cor- 
rection of a low astigmatism or of 0.12 diopter and less in the 
amount of the correction. 
In a table to be given in a later paper it will be shown, for 
example, that an error of 5 degrees in the placement of the cor- 
rection of an astigmatism produced by a 0.25 diopter cylinder 68 
C. E. FERREE AND GERTRUDE RAND 
required as an average for five eyes 66.5 per cent, more light 
for the discrimination of the test-object than the correct 
placement; in case of an astigmatism produced by a 0.75 
diopter cylinder it required 107.2 per cent, more light than 
the correct placement. The large number of scale divisions 
between the settings of the light control for the correct and 
incorrect placements of the cylinder will be shown also in these 
results. In case of the 5-degree displacement of the correc- 
tion of 0.25 diopter astigmatism this difference averaged 9.6 
for the five eyes. Since the apparatus can readily be set to 
half divisions, 19 settings of the light control could have 
been made with precision between the values needed for the 
true correction and the 5-degree displacement. This shows 
that the sensitivity of the apparatus far exceeds the present 
possibilities of the precise manipulation of the correcting 
cylinders. 
In case of an error of 0.12 diopter in the amount of the 
correction, 54.6 per cent, more light was required for the 
least favorable meridian; and in case of an error of 0.25 
diopter, 108.9 Per cent. 
The relation of the illumination scale to the detection of 
small errors in refraction and in their correction may be 
stated briefly as follows: Insofar as the test-object is con- 
cerned, clearness of seeing depends upon the value of the 
visual angle subtended and the intensity of the illumination. 
It follows from this that either the illumination scale or the 
visual angle scale may be used for the detection of errors in 
refraction, i.e., in the diagnostic procedure either the illumina- 
tion may be held constant and the visual angle varied, or 
the converse. Since the visual angle scale sustains by con- 
vention a 1 : 1 relation to acuity while acuity changes slowly 
with change of illumination for all but very low intensities, 
the illumination scale possesses the greater sensitivity for 
the detection of small errors in refraction-also the greater 
ease and feasibility of contrivance and manipulation. Used 
in this way the illumination scale becomes in effect an amplify- 
ing scale--somewhat analogous to the use of the tangent 
scale in detecting small deflections in the magnet system of a AN APPARATUS 
69 
galvanometer-and has an advantage in sensitivity in pro- 
portion to the amplification. In clinic practice it has been 
shown to be of particular value in determining the exact 
amount and placement of the correction of astigmatisms. 
That is, if the eye has equal resolving power in all meridians, 
the amount of light required just to discriminate the test- 
object in all meridians will be the same; if the resolving power 
is not equal, the amount of light required will be different in 
the different meridians and different by an amount pro- 
portional to the amplification represented by the illumination 
scale. A more detailed discussion of a feasible and convenient 
means of using the illumination scale in office and clinic 
practice will be given later. 
Attachment for Testing the Light and Color Sense 
A consideration of the foundation principles of the acuity 
apparatus reveals at a glance that they lend themselves 
readily to light and color sense testing for clinic purposes. 
In order to convert the apparatus in the form described in 
this paper into a light sense tester three features are needed: 
(a) the choice of an aperture such that when magnified five- 
fold a stimulus is obtained of a size and shape suitable for a 
sensitive judgment of the threshold of sensation; (£) the 
provision of a suitable surface on which to project the magni- 
fied image of the aperture; and (c) a means of reducing the 
intensity of light from the acuity threshold to the light sense 
threshold, i.e., from the amount needed just to discriminate 
the standard acuity object to the amount needed just to 
arouse the light sensation. The iris diaphragm used in the 
present form of apparatus, range of pupil 5-65 mm., does 
not provide for this range of intensity without changing the 
source of light. It is obvious that an attachment for the 
further reduction of the light which does not interfere in any 
way with the use of the apparatus for the acuity work, 
would afford a more convenient means of securing the lower 
intensities than the changing of the source of light. Provision 
has been made for this in two ways: («) by neutral absorption 
screens or filters; and (&) by a Nicols prism (polarizer and 70 
C. E. FERREE AND GERTRUDE RAND 
analyzer). The attachment is made so that it will hold 
either of these reducing agencies, leaving the operator an 
option as to which shall be used. The filter holder is made 
from three grooved metal strips, 8 cm. long and of appropriate 
width and thickness, built into a three sided rectangular 
figure open at the top. It is fastened to a narrow collar 
which slips over the end of the projection tube of the acuity 
apparatus and is held in place by a set screw. The holder 
is provided with three grooves into which one, two or three 
filters 8x8 cm. may be inserted as desired, or the metal 
plate which holds the Nicols prism. The Nicols prism is 
mounted in telescoping tubes in the customary manner for 
reducing light intensities, one tube containing the polarizer 
and the other the ' analyzer. At the end of the tube con- 
taining the analyzer is a large milled head by means of 
which very small angles of rotation may be made. The 
angle of rotation is read by means of a graduated dial, 
6 inches in diameter, and an indicator with a Vernier scale, 
attached respectively to the tubes containing the polarizer 
and the analyzer on either side of their junction. The tube 
containing the polarizer is firmly mounted in a brass plate, 
8x9 cm., with its axis coincident with the normal to the 
plate at its center. When the Nicols is to be used instead 
of'the filters, this plate is inserted in one of the grooves of the 
filter holder. So inserted its axis is in the principal axis of 
the projection lens of the acuity lantern and the inner end of 
the polarizer is in contact with the outer surface of the pro- 
jection lens. When the filters are employed to reduce the 
light they are inserted in the holder to give the large initial 
cut down and the further graded reduction is made by the 
iris diaphragm of the acuity lantern. When the Nicols is 
employed, the iris diaphragm is set at its minimum aperture, 
5 mm., and the further reduction is made by the Nicols and 
read from its scale. 
The testing of the color sense is provided for by inserting 
color filters in the beam of light. These filters may be 
inserted at the illuminated aperture; in the filter holder in 
front of the iris and lens; or, with a slightly different con- AN APPARATUS 
71 
struction of projection tube, back of the lens as near to the 
iris as possible. The simplest of these possibilities, from the 
standpoint of the construction and operation of the apparatus, 
is to insert the filter in the holder immediately in front of 
the lens and cut down the light intensity by means of the 
iris diaphragm. If it should be desired or considered tech- 
nically more correct, however, to produce the changes in 
intensity after the light has been passed through the filter, 
this result can be accomplished either by inserting the filter 
at the illuminated aperture or anywhere in the projection 
tube back of the iris, or by placing it in the holder in front 
of the lens, with the iris held constant, and changing the 
intensity by means of the Nicols prism. 
Color sense apparatus for clinic purposes seems at present, 
so far as the central field is concerned, to be limited to the 
testing of such gross deficiencies as are classed as color blind- 
ness. They are of little use for detecting the smaller changes 
which mark the advance and recession of many pathological 
conditions. The present apparatus is designed for detecting 
and measuring the degree of deficiency in terms of the amount 
of light of a given range of wave-lengths which is required 
just to arouse the color sensation. [Reprinted from Journal of Experimental Psychology, Vol. Ill, No. 4, Aug., 1920.] 
THE USE OF THE ILLUMINATION SCALE FOR THE 
DETECTION OF SMALL ERRORS IN REFRAC- 
TION AND IN THEIR CORRECTION 
C. E. FERREE AND GERTRUDE RAND 
Bryn Mawr College 
There are doubtless many ways in which sensitivity can 
be added to the acuity test for the detection of small errors 
in refraction and in their correction. In connection with the 
problems which we have undertaken during the past eight 
years involving modifications and refinements in functional 
testing, three principles have come to light which can be 
used very effectively to this end. That is, the eye which 
suffers from an insufficient resolving power shows the following 
functional defects, (i) An undue lag or slowness of discrim- 
ination and of making the adjustments needed for clear seeing. 
(2) A marked loss in power to sustain the adjustments needed 
for clear seeing. And (3) an increase in the amount of light 
required just to discriminate details in the standard acuity 
object. The devising of test methods based on the first two of 
these principles has been treated of in former papers. The 
third alone will be considered here. 
The relation of the illumination scale to the detection of 
small errors in refraction and in their correction may be 
stated briefly as follows. Insofar as the test-object is con- 
cerned, clearness of seeing depends upon the value of the 
visual angle subtended and the intensity of the illumination. 
It follows from this that either the illumination scale or the 
visual angle scale may be used for the detection of errors in 
refraction, i.e., in the diagnostic procedure either the illumina- 
243 244 
C. E. FERREE AND GERTRUDE RAND 
tion may be held constant and the visual angle varied, or the 
converse. Since the visual angle scale sustains by convention 
a I : I relation to acuity while acuity changes slowly with 
change of illumination for all but very low illuminations, the 
illumination scale possesses the greater sensitivity for the 
detection of small errors in refraction-also the greater feasi- 
bility of contrivance and manipulation. Used in this way the 
illumination scale becomes in effect an amplifying scale- 
somewhat analogous to the use of the tangent scale in detect- 
ing small deflections in the magnet system of a galvanometer 
-and has an advantage in sensitivity in proportion to the 
amplification. In clinic practice it has been shown to be of 
particular value in determining the exact amount and place- 
ment of the correction of astigmatisms. That is, if the eye 
has equal resolving power in all meridians, the amount of 
light required just to discriminate the test-object in all 
meridians will be the same; if the resolving power is not equal, 
the amount of light required will be different in the different 
meridians and different in proportion to the amplification rep- 
resented by the illumination scale. This gain in sensitivity 
over the clinic methods is needed in particular to determine 
the exact amount of the correction in case of high astigmatisms 
and both the amount and exact placement of the correction in 
case of low astigmatisms. The checking up of a number of 
cases shows that the corrections by the clinic methods may be 
and frequently are off from 0.12-0.25 diopter in the strength 
of the cylinder and, in case of low astigmatisms, from 5-20 
degrees in the placement of the cylinder axis. While errors 
of this magnitude may or may not be troublesome in the 
ordinary uses of the eye-sometimes they are very trouble- 
some indeed and, perhaps always tend in time to increase the 
amount of the defect-they do constitute a much more serious 
handicap, perhaps an actual disqualification, for work or voca- 
tions requiring special ocular proficiencies, e.g., keen acuity, 
particularly keen acuity at low illuminations; the power to 
sustain acuity; speed in the use of the eye, especially speed of 
discrimination and of making the adjustments needed for clear 
seeing; etc. Moreover, it is safe to say that a considerably DETECTION OF SMALL ERRORS IN REFRACTION 
245 
greater amount of light is required as a comfortable and effi- 
cient working minimum by the poorly than by the well cor- 
rected eye. Indeed our experience with the tricornered 
relation of intensity of light, resolving power and retinal 
sensitivity to acuity has impressed us with the relative im- 
portance of resolving power in explaining the difference in the 
amount of light that is required by different people as a 
working minimum. 
The relation of the intensity of illumination to acuity 
may be illustrated by the curve shown in Fig. 1. This curve 
represents the average results for four observers, tested by 
Koenig.1 In this curve acuity is plotted along the ordinate 
and intensity of illumination along the abscissa. It will be 
noted, for example, that a change of from 1 to 9 meter-candles, 
an increase of 800 per cent, in the intensity of illumination, 
produced an increase of only 74 per cent, in acuity; and a 
change of 9 to 100 meter-candles, an increase of ion per 
cent, in illumination, produced an increase of only 28 per 
cent, in acuity. The amplification within the latter range of 
illumination is doubtless too great for feasibility of applica- 
tion. That is, too wide a range of illumination would have 
to be used to compensate for the difference between the 
resolving power in the poorest and best meridians in the 
ordinary run of astigmatisms. The range from 1-9 meter- 
candles is, however, quite feasible and the relation between 
the two scales gives abundant sensitivity. These values fall 
within the range given by the apparatus described in our 
former paper, 0.07-9.5 meter-candles. The testing of a large 
number of astigmatisms with this apparatus showed that in 
the majority of cases the minimum amount of light required 
for the discrimination of the opening in the broken circle 
(visual angle, 1 min.) in the most favorable meridian was of 
the order of 1-3 meter-candles; in the least favorable meridian, 
of the order of 6-9.5 meter-candles. 
A very convenient apparatus for using the illumination 
scale for detecting low astigmatisms and small errors in the 
1 "Ueber die Beziehung zwischen der Sehscharfe und der Beleuchtungsintensitat," 
Verhandl. der Physikal. Ges. in Berlin, 1885, 16, S. 79-83. 246 
C. E. FERREE AND GERTRUDE RAND 
amount and placement of their corrections was described in a 
formernumberof this journal, "An Apparatus for Determining 
Acuity at Low Illuminations, etc.," 1920, Vol. Ill, No. 1, 
pp. 59-71. In this apparatus, it will be remembered, uni- 
formity of illumination of the test surface was secured by 
Fig. I. Showing the relation of intensity of illumination to acuity (Koenig, 4 ob- 
servers). 
projecting upon it the image of an evenly illuminated aperture 
at the inner end of a projection tube of a lantern or lamp 
house. In order to secure a uniform illumination of this 
aperture, the lamp house was lined with opal glass ground on 
one side, and the aperature itself was covered with a slide 
of ground glass. The source of light was a round bulb, 
100-watt, type C Mazda lamp with its filament well above 
the aperture to be illuminated, and the changes of illumination 
were produced by an iris diaphragm placed immediately 
behind the focusing lens in the projection tube, which reduced 
the illumination without changing the size or shape of the 
image. The test-object was a broken circle fastened at the 
center of a graduated dial the opening of which (visual angle, 
I minute) could be turned into any meridian that was desired. 
The angle of turning could be read in terms of the divisions 
on the dial which was graduated to correspond to the readings 
on the trial frames used in office and clinic work. The 
results given in this paper were obtained with this type of 
apparatus. They are fairly representative of the large num- 
ber that have been obtained. 
In the testing and demonstration of the sensitivity and 
serviceability of the illumination method for determining the 
exact amount and placement of the correction of an astigma- 
tism the following types of material have been selected: (i) 
Artificial astigmatisms made with cylinders of low diopter DETECTION OF SMALL ERRORS IN REFRACTION 
247 
value. In choosing to include artificial astigmatisms in this 
work it should be understood that we did not consider the 
artificial astigmatism the precise functional equivalent of the 
natural astigmatism. We are too strongly impressed with 
the possibility that the astigmatic eye may progressively 
acquire power to compensate in part for its defect to be of this 
opinion. They were selected because we wished to have in 
one set of cases an exact knowledge of the amount and loca- 
tion of the defect as a check on the determinations made by 
the test. (2) Natural astigmatisms without a cycloplegic. 
(3) Office and clinic cases with a cycloplegic. The difference 
in result between the most and least favorable meridians 
or between a true and a false correction have thus far been 
of a considerably greater order of magnitude with than 
without a cycloplegic either in case of a natural or an artificial 
astigmatism. (4) Office and' clinic cases, submitted to us 
by experienced refractionists, in which the apparatus has 
been used merely to check up corrections already made by 
the clinic methods, objective and subjective. Among these 
cases it was comparatively rare to find one in which the mini- 
mum amount of light required to discriminate the test-object 
in the corrected meridian was even approximately equal to 
that required in the other meridians. Indeed in some cases 
the difference between the most and least favorable meridian 
exceeded the range of variation obtainable with the apparatus 
when provided with the 100-watt lamp. And (5) irregular 
astigmatisms. 
For the artificial astigmatisms three cases have been used. 
1. Low Astigmatisms Produced by Weak Cylinders.-In this 
case the minimum amount of light required to discriminate the 
test-object with the opening of the circle turned into the most 
and least favorable meridians has been determined; also 
when turned 5, 10 and 45 degrees from the most favorable 
meridian. In Table I. it will be noted that in case of an 
astigmatism produced by a 0.25 diopter cylinder the difference 
in the light required for the discrimination of the test-object 
in the worst and best meridians amounted to 107.2 per cent, 
as an average result for five eyes. At 5 degrees from the 248 
C. E. FERREE AND GERTRUDE RAND 
Observer 
Value of 
Cylinder 
Producing 
Astigmatism 
Minimum Illumination Required for Discrimination of 
Test-object (Meter-candles) 
Difference in Results Between Best and Other Meridians 
Best 
Meridian 
50 from 
Best 
Meridian 
io° from 
Best 
Meridian 
450 from 
Best 
Meridian 
Worst 
Meridian 
Meter-candles 
Per Cent. 
5° 
xo° 
45° 
Worst 
Meridian 
5° 
xo° 
45° 
Worst 
Meridian 
A 
0.25 
0.60 
0.88 
I.49 
1.62 
1.62 
0.28 
0.89 
1.02 
1.02 
46.7 
148.3 
170.0 
170.0 
B 
O.25 
1.12 
1-75 
2.405 
2.69 
2.69 
O.63 
1.285 
1-57 
1-57 
S6.3 
114.7 
I3I-5 
I3I-5 
C 
0.25 
O.46 
0.60 
O.76 
O.76 
O.76 
O.I4 
0.30 
0.30 
0.30 
30.4 
65-2 
65.2 
65-2 
D 
0.25 
I.42 
2.115 
2-49 
2-49 
2-49 
O.695 
1.07 
1.07 
I.07 
48.9 
75-4 
75-4 
75-4 
E 
0.25 
1.12 
1.88 
i-95 
2.17 
2.17 
O.76 
0.83 
1-05 
i-05 
67-9 
74-i 
93-8 
93-8 
Average. 
0.501 
0.875 
1.002 
1.002 
50.0 
95-5 
107.2 
107.2 
A 
o-75 
I-3O 
2.69 
3-°5 
3-05 
3-05 
i-39 
i-75 
i-75 
i-75 
IO6.9 
134-6 
134-6 
134-6 
B 
o-7S 
1.8l 
4.11 
4-39 
4-39 
4-39 
2.30 
2.58 
2.58 
2.58 
127.8 
142.5 
142.5 
142.5 
C 
o-75 
0.60 
i-75 
i-95 
2.32 
2.32 
i-i5 
i-35 
1.72 
1.72 
I9I-7 
225.0 
286.7 
286.7 
D 
0-75 
3-05 
6.895 
6.895 
7.60 
7.60 
3-845 
3-845 
4-55 
4-55 
126.1 
126.1 
149.2 
149.2 
Average. 
2.171 
2.381 
2.65 
2.65 
I38.I 
I57-I 
178.3 
178-3 
Table I 
Sensitivity of Apparatus for Locating Meridian of Astigmatism 
Astigmatisms produced by 0.25 and 0.75 diopter cylinders. DETECTION OF SMALL ERRORS IN REFRACTION 
249 
best meridian, this difference was 50 per cent.; at 10 de- 
grees, 95-5 per cent.; and at 45 degrees 107.2 per cent. In 
case of the 0.75 diopter astigmatism, the difference between 
the worst and best meridians was 178.3 per cent.; at 5 
degrees from the best meridian, 138.1 per cent.; at 10 degrees, 
157.1 per cent.; and at 45 degrees, 178.3 per cent. 
2. Small Errors in the Placement of the Correction.-In 
Table II. it will be noted that in case of an astigmatism pro- 
Table II 
Sensitivity of Apparatus for Detecting Small Errors in the Placement of 
the Correction of an Astigmatism 
Observer 
Value of 
Cylinder 
Producing 
Astigma- 
tism 
Minimum Illumination Required for 
Discrimination of Test-object 
(Meter-candles) 
Difference in Result for Correct 
and Incorrect Placements 
Exact 
Placement 
of Correc- 
tion 
50 Displace- 
ment 
ro° Dis- 
placement 
Scale Di- 
visions 
Meter- 
candles 
Per Cent. 
5° 
IO° 
5° 
IO° 
5° 
IO° 
A 
0.25 
0.60 
i-33 
1.62 
13-0 
17.0 
o-73 
1.02 
I2I.7 
170.0 
B 
0.25 
1.12 
2.25 
2.69 
17-5 
22.0 
I-I3 
i-57 
IOO.9 
140.2 
C 
O.25 
O.46 
0.60 
O.65 
3-o 
4.0 
0.14 
0.19 
30-4 
41-3 
D 
O.25 
I.42 
i-55 
1.88 
2.0 
7.0 
0.13 
0.46 
9.2 
32.4 
E 
O.25 
1.12 
1.91 
2.17 
12.5 
20.0 
o-79 
1.05 
7°-5 
93-8 
Average.. 
9.6 
14 
0.58 
66.5 
oc.r 
A 
0-75 
1-30 
2-49 
3-05 
17.0 
22.0 
1.19 
i-75 
91-5 
134.6 
B 
0-75 
1.81 
3-43 
4-39 
17.0 
22.0 
1.62 
2.58 
89-5 
142-5 
C 
0-75 
0.60 
1.42 
2.12 
14.0 
24.O 
0.82 
1-52 
136.7 
253-3 
D 
0.75 
3-05 
6-44 
6-44 
20.0 
20.0 
3-39 
3-39 
in.1 
in.1 
Average.. 
17.0 
22.0 
i-75 
2.31 
107.2 
160.4 
duced by a 0.25 diopter cylinder a displacement of the correc- 
tion 5 degrees from the true position required 66.5 per cent, 
more light for the discrimination of the test-object as an 
average result for five eyes; a displacement of 10 degrees, 
95-5 per cent, more light. In case of a 0.75 diopter astigma- 
tism a displacement of 5 degrees required 107.2 per cent, 
more light; and a displacement of 10 degrees, 160.4 per cent, 
more light. In connection with this table note also the large 
number of scale divisions between the correct and the in- 
correct placement of the cylinder. In case of a 5 degree dis- 
placement for the 0.25 diopter astigmatism, this difference 250 
C. E. FERREE AND GERTRUDE RAND 
averaged 9.6 for the five eyes. That is, since the diaphragm 
can be readily set to half divisions, 19 settings of the light 
control could have been made with precision between the 
values needed for the true correction and the 5 degree dis- 
placement. This shows that the sensitivity of the apparatus 
far exceeds the present possibilities of the precise manipulation 
of the correcting cylinders. 
3. Small Errors in the Amount of the Correction.-In the 
case of a 0.12 diopter error in the correction of an astigmatism, 
54.6 per cent, more light was required for the discrimination 
of the test-object in the worst meridian; in case of a 0.25 
diopter error, 108.9 and in case of a 0.75 diopter error, 178.25 
per cent, more light was required. These results are shown 
in Table III. 
Table III 
Sensitivity of Apparatus for Detecting Errors in the Amount of Correction 
of an Astigmatism 
Observer 
Minimum Illumination Required for 
Discrimination of Test-object with Different 
Errors in Amount of Correction 
(Meter-candles) 
Difference in Minimum Illumination to 
Discriminate Test-object in Most and 
Least Favorable Meridians 
o.i2 Diopter 
0.25 Diopter 
0.75 Diopter 
0.12 Diopter 
0.25 Diopter 
0.75 Diopter 
Best 
Merid- 
ian 
Worst 
Merid- 
ian 
Best 
Merid- 
ian 
Worst 
Merid- 
ian 
Best 
Merid- 
ian 
Worst 
Merid- 
ian 
Meter- 
can- 
dles 
Per 
Cent. 
Meter- 
can- 
dles 
Per 
Cent. 
Meter- 
can- 
dles 
Per 
Cent. 
A 
0.60 
0.88 
0.60 
1.62 
1-30 
3-05 
0.28 
46.7 
1.02 
170.0 
i-75 
134-6 
B 
1.12 
2.17 
1.12 
2.69 
1.81 
4-39 
1.05 
93-8 
i-57 
I4O.2 
2.58 
142.5 
C 
O.46 
0.65 
O.46 
O.76 
0.60 
2.32 
0.19 
41-3 
0.30 
65-2 
1.72 
286.7 
D 
I.42 
i-75 
I.42 
2-49 
3-05 
7.60 
o-33 
23-3 
1.07 
75-4 
4-55 
149-2 
E 
1.12 
1.88 
1.12 
2.17 
0.76 
67.9 
1-05 
93-8 
- 
Average. 
0.52 
54-6 
1.00 
108.9 
2.65 
178.25 
Of the large number of natural astigmatisms tested space 
will be taken here for the representation of only a few cases. 
Astigmatism (Without a Cycloplegic) 
Case I. {Age 13 Years') 
O.D.: Correction by clinic methods, 0.25 cyl., ax. 700. 
(Placement of axis could be varied over a range of about 450 
and cylinder could be changed to 0.12 diopter without notice- DETECTION OF SMALL ERRORS IN REFRACTION 
251 
able change in the results by these methods.) With this 
correction illumination required with opening of the circle in 
meridian of cylinder axis, 0.20 m.c.; at 90 degrees from this 
position, 0.55 m.c.; difference, 0.35 m.c. or 175 per cent. 
Correction by illumination method, +0.12 cyl., ax. 550. 
With this correction equal illumination (0.16 m.c.) was re- 
quired for the discrimination of the test-object in all meridians. 
Difference in amount of light required for discrimination 
of test-object in least favorable meridian for the two correc- 
tions, 0.39 m.c. or 244 per cent. 
O.S.: Correction by clinic methods, + 0.12 cyl., ax. 1800. 
(Placement of axis could be varied over a range of about 45 
degrees without change in result by these methods.) With 
this correction illumination required with opening of circle 
in meridian of cylinder axis, 0.12 m.c.; at 90 degrees from 
this position, 0.21 m.c.; difference, 0.09 m.c. or 75 per cent. 
Correction by illumination method, +0.12 cyl., ax. 150. 
With this correction equal illumination (0.105 m.c.) was 
required for discrimination of test-object in all meridians. 
Difference in amount of light required for discrimination 
of test-object in least favorable meridian for the two correc- 
tions, 0.105 m-c- or 100 Per cent. 
Case II. {Age 48 Years') 
O.D.: Illumination required before correction with open- 
ing of circle in most favorable meridian, 2.93 m.c.; at 90 
degrees from this position, 9.19 m.c.; difference, 6.26 m.c. 
or 214 per cent. 
Correction by illumination method, - 0.50 cyl., ax. 1050. 
With this correction, equal illumination (2.93 m.c.) was 
required for the discrimination of the test-object in all 
meridians. 
O.S.: Illumination required before correction with opening 
of circle in most favorable meridian, 2.35 m.c.; at 90 degrees 
from this position, 5.25 m.c.; difference, 2.90 m.c. or 123 
per cent. 
Correction by illumination method, + 0.37 cyl., ax. 1370. 
With this correction, equal illumination (2.35 m.c.) was re- 252 
C. E. FERREE AND GERTRUDE RAND 
quired for the discrimination of the test-object in all 
meridians. 
Irregular Astigmatism 
Case I. (Age 32 Years') 
O.S.: Illumination required with opening of circle turned 
right, left, and down, 0.97 m.c.; when turned up, 5.25 m.c.; 
difference for two halves of vertical meridian, 4.28 m.c. or 
441 per cent. 
Astigmatism (with Cycloplegic) 
Case I. (Age 23 Years) 
O.D.: Correction by clinic methods, +0.50 S., +0.37 
cyl., ax. 150. With this correction, illumination required 
with opening of circle in meridian of cylinder axis, 2.46 m.c.; 
at 90 degrees from this position, 9.19 m.c.; difference, 6.73 
m.c. or 274 per cent. 
Correction by illumination method, + 0.50 S., + 0.37 
cyl., ax. 300. With this correction, equal illumination (1.61 
m.c.) was required for the discrimination of the test-object 
in all meridians. 
Difference in amount of light required for discrimination of 
test-object in least favorable meridian for the two corrections, 
7.58 m.c. or 471 per cent. 
Case II. (Age 33 Years) 
O.D.: Corrections by clinic methods, - 0.62 cyl., ax. 
1800. With this correction, illumination required with open- 
ing of circle in meridian of cylinder axis, 2.32 m.c.; at 90 
degrees from this position, 9.19 m.c.; difference, 6.87 m.c. 
or 296 per cent. 
Correction by illumination method, - 0.75 cyl., ax. 1800. 
With this correction, equal illumination (2.09 m.c.) was re- 
quired for the discrimination of the test-object in all meridians. 
Difference in amount of light required for discrimination of 
test-object in least favorable meridian for the two corrections, 
7.10 m.c. or 339 per cent. DETECTION OF SMALL ERRORS IN REFRACTION 
253 
Astigmatism (Checking Up of Glasses) 
Case I. (Age 42 Years') 
O.D.: Correction by clinic methods, - .50 S., - .37 cyl. 
ax. io°. With this correction, illumination required with 
opening of circle in meridian of cylinder axis, 2.34 m.c.; at 
90 degrees from this position, 7.35 m.c.; difference, 5.01 m.c. 
or 214 per cent. 
Case II. (Age 45 Years) 
O.D.: Correction by clinic methods, - 0.25 S., - 0.50 
cyl., ax. 1250. With this correction, illumination required 
with opening of circle in meridian of cylinder axis, 2.02 m.c.; 
at 90 degrees from this position, 6.67 m.c. in one half of 
meridian, 7.82 m.c. in other half; difference, 4.65 m.c. (230 
per cent.) and 5.80 m.c. (287 per cent.). Astigmatism may 
be slightly irregular. 
O.S.: Correction by clinic methods, - 0.50 cyl., ax. 8o°. 
With this correction, illumination required with opening of 
circle in meridian of cylinder axis, 0.97 m.c.; at 90 degrees 
from this position, 5.62 m.c. in one half of meridian, 6.23 
m.c. in other half; difference, 4.65 m.c. (479 per cent.) and 
5.26 m.c. (542 per cent.). Astigmatism may be slightly 
irregular. 
In the above notes on cases we have, for the sake of 
brevity, used the term clinic methods, instead of specifying 
in greater particular the tests employed. Where we have 
made the comparison ourselves between the illumination 
method and the methods ordinarily employed in office and 
clinic work, we have used the acuity method, the astigmatic 
charts, the point of light test and in some cases the ophthal- 
mometer. We have not made frequent use of the retinoscope 
because of the need of a cycloplegic. The acuity method 
was used in different ways. In one, patterned after a pro- 
cedure much employed by the ophthalmologists, some char- 
acter difficult of discrimination and taxing the resolving power 
of the eye in as many meridians as possible, such as the letter 
B, was selected. It was brought to or near to the threshold of 
discrimination by fogging, by changing the visual angle, by 254 
C. E. FERREE AND GERTRUDE RAND 
the use of a graded scale of illumination, etc., in order to make 
the conditions favorable for a sensitive judgment; and the 
strength and placement of cylinder was determined which 
gave the maximum clearness of seeing. In order to decide 
between doubtful determinations other acuity tasks or tests 
were imposed. That is, we not only used the acuity test as 
employed by the practitioner but have endeavored in many 
ways to add to its sensitivity and precision without sacrificing 
its distinctive features. However, in collecting the data for 
the comparison we have preferred to lay the chief stress on 
the cases in which the clinic testing has been done by prac- 
ticing ophthalmologists, who have very willingly given us 
their cooperation, In all cases but one, which have been 
submitted to us for testing, the physician has himself accom- 
panied the patient, looked after the cycloplegia, and inspected 
the test procedure at every step, the principle of the apparatus 
and method having previously been made familiar to him. 
Due care was taken on both sides that a fair comparison of 
sensitivities was made. 
Doubtless the apparatus can be used in different ways 
depending upon the experience and preference of the operator. 
For example, the minimum amount of light required to dis- 
criminate the test-object could be determined for one meridian 
and the setting of the light control be held constant while the 
test-object is rotated into the different meridians, the observer 
being required to judge in each case whether the same or 
more or less light would be required for its discrimination. 
This would serve as a rough indication of whether or not the 
eye is astigmatic. The exact meridian of the defect, that is 
the meridian in which the greatest amount of light is required 
to discriminate the opening in the circle, could be determined 
through a series of settings of the test-object and the light 
control. The placement of the correction having been deter- 
mined, its amount could be found by the strength of cylinder 
required to render the minimum illumination needed to dis- 
criminate the test-object the same for all meridians, or more 
roughly speaking for the meridian of the defect and at 90 
degrees either way from this position. A quicker and more DETECTION OF SMALL ERRORS IN REFRACTION 
255 
feasible method, however, is first to make an approximate 
determination of the amount and placement of the correction 
by the clinic methods and employ the illumination method 
only for a more precise determination. In using this method 
as a refinement on the clinic methods, the procedure we 
ordinarily employ is as follows: The patient's eye is fitted 
with a cylinder of the strength and placement indicated by 
the clinic tests and the minimum amount of light required to 
discriminate the opening in the circle is determined in four 
positions, two in the meridian of the cylinder axis and two 
in the meridian at right angles to this. If the minima are 
not equal in these four positions, the cylinder axis is shifted 
and the determinations are made again, the four positions of 
the opening of the circle always being in the meridian of the 
cylinder axis and the meridian at 90 degrees from it. If no 
placement of the cylinder is found which gives equal minima 
for the four positions, the strength of the cylinder is changed. 
The strength and placement of cylinder which requires both 
equal and the smallest amounts of light for the four positions 
of the test-object is accepted as the final correction. 
The apparatus can also be used to advantage with astig- 
matic charts of the sunburst type, the radial lines of which are 
no more than 5 degrees apart, in the preliminary approximate 
determination of the axis of the defect. In this case the 
procedure would be to reduce the illumination until only one 
or perhaps two of the lines stand out clearly. This would give 
a sensitivity roughly speaking of about 5 degrees, and requires 
little more time than is usually consumed in the use of the 
astigmatic charts. 
In our own work we have found that the apparatus would 
be very helpful even if it were used only to check up the 
corrections made by the clinic methods and were not em- 
ployed further as an aid in finding out the exact amount and 
placement of the correction. For example, but a very few 
minutes are required to determine with it whether any given 
correction equalizes or levels up the resolving power of the 
eye in the different meridians. The advantage of a checking 
method which is definite and at the same time feasible, 256 
C. E. FERREE AND GERTRUDE RAND 
can readily be appreciated by any one who has tried to decide 
by the clinic methods in any wide range of cases just what 
should be the exact amount and placement of the correction 
of an astigmatism. The method has its chief value perhaps 
in those cases in which it is particularly difficult to make a 
decision by the clinic methods, that is, in determining the 
exact amount of the correction in cases of high astigmatism 
and both the amount and placement of correction in case of 
low astigmatisms. The simple character of the judgment, 
namely the mere indication of the direction in which the 
opening of the circle points instead of the more difficult task 
of deciding under the comparatively rough conditions of the 
office and clinic test whether this or that placement or strength 
of cylinder gives the clearer vision, together with the objective 
check on the correctness of each judgment, also contribute to 
make the method especially valuable in case of children, and 
the subjective, unintelligent and untrained type of adult. 
A further advantage of the method as worked out in connec- 
tion with the present apparatus is its great sensitivity for the 
detection of irregular astigmatisms. The lack of satisfactory 
tests for this troublesome defect is generally conceded. [Reprinted from Journal of Experimental Psychology, Vol. Ill, No. 5, Oct., 1920.] 
A STUDY OF OCULAR FUNCTIONS WITH SPECIAL 
REFERENCE TO THE LOOKOUT AND 
SIGNAL SERVICE OF THE NAVY 
BY C. E. FERREE, G. RAND AND D. BUCKLEY 
Bryn Mazur College 
The incentive for this work was the need for establishing a 
system of testing for those branches of service in the Navy 
requiring especially keen scotopic or low illumination acuity. 
The first step towards the accomplishment of this purpose 
was the devising of a suitable apparatus and test method. 
The request for an apparatus came to us from the head of the 
Eye Division of the United States Naval Hospital at Wash- 
ington. The apparatus was described in a former paper, 
"An Apparatus for Determining Acuity at Low Illuminations, 
etc.," this Journal, 1920, III., pp. 59-71. A further need 
was to find out what range of difference in scotopic acuity 
might be expected among eyes graded as fit on the basis of 
the tests of other functions and capacities. A consideration 
of this need has led us to make a preliminary survey of eyes 
graded as normal with regard to photopic acuity and other 
commonly tested functions in order to determine whether 
such eyes may be expected to show a significant difference in 
keenness of functioning at low illuminations. 
In a thorough test for vocations requiring keenness of 
discriminations at low illuminations, the following points 
should be taken into account: (1) the minimum amount of 
light required to discriminate the test-object before adapta- 
tion; (2) the minimum amount after a properly selected 
period of adaptation; and (3) the rapidity as well as the 
amount of gain in acuity in the process of adapting. Deter- 
minations covering all of these points have been made in 
this study. 
347 348 
C. E. FERREE, G. RAND AND D. BUCKLEY\ 
The Range of Illumination Required by Normal Eyes 
for the Discrimination of the Standard 
Test Object 
In making these determinations three test-objects were 
used: the Snellen chart and two single test-objects which 
could be rotated into different positions-the letter E and 
the international test-object, the broken circle, each sub- 
tending a 5-min. angle. In case of the latter two, the task 
required of the observer was to indicate roughly the direction 
in which the opening in the test character was turned, an 
objective check being had on the correctness of the judgment. 
The determinations were made at the beginning and end of a 
45-min. adaptation period. It is obvious that the results at 
low illumination should be influenced by the refraction condi- 
tion of the eye as well as by its light sensitivity and the 
individual differences in the effect of light sensitivity on 
acuity. In order that the observers could be chosen so that 
the results would represent the differences which may occur 
among eyes having normal or better than normal photopic 
acuity, each eye was refracted and the acuity was taken 
under 5 foot-candles (53.8 meter-candles) of light. In the 
first series of tests 22 observers were used ranging from 18 to 
28 years of age. Results were obtained for both eyes and for 
each eye separately. Of the eyes used, 75.7 per cent, would 
be rated in the Snellen scale as having 6/4 acuity; 13.5 per 
cent, as having 6/5 acuity; and 10.8 per cent, as having 6/6 
acuity. It was our intention to use throughout only eyes 
which could be ranked as Grade A with regard to photopic 
acuity. 
The results of these determinations show a greater range 
of individual difference for the broken circle than for the 
Snellen chart or the letter E (905 per cent, for the broken 
circle, 548 per cent, for the letter E and 357 per cent, for the 
Snellen chart). This superior showing for the broken circle 
is perhaps in accord with the general finding that the broken 
circle as a test-object picks up smaller differences in acuity 
than either of the other two test-objects employed. These 
differences too, it will be remembered, are amplified in the A STUDY OF OCULAR FUNCTIONS 
349 
present case by the fact that they are read on the illumination 
scale-an amplifying scale-and not on a scale which sustains 
a I : i relation to acuity. Inasmuch as the greater sensitivity 
was shown in these preliminary experiments by the broken 
circle as a test-object, 15 additional observers (photopic 
acuity, 6/4) were employed using this test-object alone. 
Space will be taken here only for a brief general statement 
of the results for this latter series of determinations alone. 
(1) The individual differences in the minimum illumination 
required for the discrimination of the test-object before the 
period of dark adaptation fell between 0.70 and 5.29 meter- 
candles, a range of 657 per cent.; after the period of dark 
adaptation it fell between 0.32 and 2.2 meter-candles, a range 
of 593 per cent. A greater range of individual difference, it 
will be noted, was found for the tests taken before the period 
of dark adaptation than for those taken after the 45-minute 
adaptation period. This was doubtless in part due to the 
lack of careful standardization of the initial sensitivity by a 
period of preadaptation to light of a fixed intensity and to 
small individual differences in photopic acuity revealed by the 
more sensitive method of testing; and in part to individual 
differences in the amount and rate of adaptation. A careful 
initial standardization of sensitivity was purposely avoided 
in this preliminary work with the apparatus in order more 
closely to approximate the rough conditions of testing which 
are apt to prevail in the selection of men with reference to 
vocational fitness. The results of these determinations are 
shown in Table I. 
(2) Thus far without exception the two eyes of the same 
observer have required a different amount of light just to 
discriminate the test-object. This difference has ranged after 
adaptation from 19 to 54 per cent, of the amount of light 
required for the better eye. 
(3) A question is often raised with reference to points of 
advantage of binocular as compared with monocular seeing. 
In 6 per cent, of the number of cases tested, the binocular 
result after adaptation was equal to or approximated the result 
for the poorer eye; in 88 per cent, of the cases it was better 350 
C. E. FERREE, G. RAND AND D. BUCKLEY 
Table I 
Showing the Amount of Light Required just to Discriminate the Test-object 
at the Beginning of Dark Adaptation and at the End of 
15, 30 and 45 Minutes (15 Observers) 
In these experiments there was no standardization of the initial sensitivity by a 
previous adaptation to an illumination of constant intensity. 
Observer 
Photopic 
Acuity 
Illumination in Meter-candles Required Just to 
Discriminate the Test-Object 
Difference in Illumination 
Required at Beginning 
and at End of 45 Minutes 
Beginning 
At End of 
15 Min. 
At End of 
30 Min. 
At End of 
45 Min- 
Meter-can- 
dles 
Per Cent. 
G 
6/4 
O.7O 
o-SS 
0-35 
O.32 
O.38 
II8.8 
M 
6/4 
1.00 
1.00 
0.82 
0.82 
O.l8 
2I.Q 
Me 
6/4 
1.24 
1.00 
I .OO 
I.OO 
O.24 
24.0 
R 
6/4 
1.36 
0.60 
0.50 
0.60 
O.76 
126.7 
L 
6/4 
1.55 
1.00 
0.88 
0.88 
O.67 
76.1 
S 
6/4 
i-75 
1.42 
0.88 
0.88 
O.87 
98.9 
Th 
6/4 
2.11 
0.82 
0.94 
o-94 
1.17 
124-5 
Y 
6/4 
2.11 
i-55 
i-49 
1.42 
O.69 
48.6 
St 
6/4 
2.40 
0.60 
0.60 
0.60 
I.80 
200.0 
Sw 
6/4 
3-43 
2.17 
2.20 
2.20 
1-23 
55-9 
K 
6/4 
3-90 
2.81 
2.40 
2.10 
I.80 
85-7 
T 
6/4 
3-97 
1.18 
0.82 
0.88 
3-°9 
35I-I 
Sm 
6/4 
4.10 
3-8° 
1.40 
1-3° 
2.80 
215-4 
W 
6/4 
4.20 
1.40 
0.76 
0.76 
3-44 
452-6 
Ba 
6/4 
5-29 
2.11 
2.02 
2.11 
3.18 
150.7 
than the result for either eye; and in the remaining 6 per cent, 
of the cases it was intermediate to the results obtained with 
the two eyes separately. In none of the cases tested separately 
was it equal or approximately equal to the result for the better 
eye. In the 88 per cent, of cases referred to, less light was re- 
quired for the discrimination of the test-object by the two eyes 
than by the better eye alone by amounts ranging from 14.5 to 
67.3 per cent. 
In order to get a rough idea of the grouping of the 15 
observers with reference to the minimum amount of light 
required to meet the standard of acuity imposed by the test, 
before and after the period of dark adaptation, they have in 
each case been divided into six equally spaced groups, each 
group for the work before adaptation covering a range of 1 
meter-candle and for the work after adaptation a range of 
0.4 meter-candle. For the tests before adaptation 13.3 per 
cent, fall in the first or best group; 26.7 per cent, in the 
second group; 20 per cent, in the third group; 20 per cent. A STUDY OF OCULAR FUNCTIONS 
351 
in the fourth group; 13.3 per cent, in the fifth group; and 
6.7 per cent, in the sixth group. For the tests after adapta- 
tion 6.7 per cent, fall in the first group; 20 per cent, in the 
second group; 40 per cent, in the third group; 13.3 per cent, 
in the fourth group; none in the fifth group; and 20 per cent, 
in the sixth group. A graphic representation of this grouping 
is shown in Fig. I. 
Meter-candles 
Meter-candles 
Fig. 1. Representing the relative distribution of 15 observers graded with 
reference to the minimum illumination required just to discriminate the test-object: 
A, before adaptation; B, after a 45-minute period of dark adaptation. 
Individual Variations in the Amount and Rate of 
Adaptation in Terms of Effect on Acuity 
The preceding experiments furnish data with regard to the 
minimum amounts of light required just to discriminate the 
different test-objects at the beginning and end of the 45- 
minute adaptation period. In case of the 15 observers tested 
with the broken circle in the final series of experiments at the 
beginning of dark adaptation and at the end of 15, 30 and 
45 minutes, the minimum difference in this amount of light 
was 0.18 meter-candle or 22 per cent, of the amount required 
at the end of the adaptation period; the maximum difference 
was 3-44 meter-candles or 452.6 per cent, of the amount 
required at the end of the adaptation period. These results 
are shown in Table I. 
In addition to these experiments a special adaptation 
series was run in which the minimum illumination required 
just to discriminate the test-object was determined at the 352 
C. E. FERREE, G. RAND AND D. BUCKLEY 
beginning of the adaptation period and at the end of 5, 10, 
1$, 25, 35 and 45 minutes. In order to standardize the initial 
sensitivity of the eyes of the observer, a preadaptation period 
of 20 minutes was given to 80 foot-candles of light (vertical 
component), the skylight illumination of an optic's room on a 
medium bright day. A few of the results obtained are repre- 
sented in Table II. and Fig. 2. These results, it will be 
remembered, represent adaptation only as it affects acuity, 
which is the effect of greatest importance to the special work 
for which the apparatus was devised and the effect with which 
we are the most concerned for the greater part of our working 
lives. A comparison of these results with those of similar 
Table II 
Showing the Amount of Light Required just to Discriminate the Test-object 
at the Beginning of Dark Adaptation and at the End of 
5, 10, 15, 25, 35, and 45 Minutes (6 Observers) 
In these experiments the initial sensitivity was standardized by 20 minutes pre- 
adaptation to 80 foot-candles of light (vertical component), the illumination of a sky- 
light optics room on a medium bright day. 
Obser- 
ver 
Pho- 
topic 
Acuity 
Illumination in Meter-candles Required Just to Discriminate 
the Test-object 
Difference in Illumi- 
nation Required at 
Beginning and at 
End of 45 Minutes 
Begin- 
ning 
At End 
of 5 
Min. 
At End 
of xo 
Min. 
At End 
of 15 
Min. 
At End 
of 25 
Min. 
At End 
of 35 
Min. 
At End 
of 45 
Min. 
Meter- 
candles 
Per Cent. 
I... 
6/4 
O.55 
O-5OS 
O.42 
0-35 
0.32 
o-35 
0-35 
0.20 
57-i 
II... 
6/4 
O.7O5 
O.42 
O.42 
O.32 
0-32 
0.32 
O.38 
O.32S 
85-5 
Ill... 
6/4 
1.06 
O.76 
0.60 
O.46 
0-35 
0.46 
O.46 
0.60 
130.4 
IV. .. 
6/4 
1.12 
0.82 
O.52 
O.32 
0.32 
0.38 
O.38 
o-74 
194-7 
V... 
6/4 
1.62 
1.12 
0-94 
0-55 
O.6o 
0.60 
o-55 
1.07 
194-5 
VI... 
6/4 
2.20 
1.12 
1.14 
1.18 
I.36 
1.24 
1.24 
0.96 
77-4 
series in which the object is to measure the increase in light 
sensitivity as a result of dark adaptation shows that just as 
acuity increases slowly with increase of illumination (except 
at very low illuminations) so also does it increase slowly with 
increase of sensitivity to light. That is, the eye does not 
gain in acuity by adaptation nearly so fast as it gains in light 
sensitivity. 
In Fig. 2 the actual amounts of illumination required just 
to discriminate the test-object are plotted against time of A STUDY OF OCULAR FUNCTIONS 
353 
adaptation. It thus affords a comparison of the position of 
the minima of the different observers in the illumination scale 
and comprehends data from which the following points can 
be determined: (a) their relative ranking with regard to 
Meter-candles 
Fig. 2. Curves showing the decrease in the amount of light required just to dis- 
criminate the test-object as the result of dark adaptation. In'order to standardize the 
initial sensitivity, the eye was preadapted in each case for 20 minutes to 80 foot- 
candles of illumination (vertical component). 
Minutes 
scotopic acuity before and after adaptation, rated on the illu- 
mination scale; (b) their light sensitivity before and after 
adaptation insofar as it affects the minimum amounts of light 
required to discriminate the test object; and (c) their relative 
amounts of change in sensitivity, measured in terms of effect 354 
C. E. FERREE, G. RAND AND D. BUCKLEY 
on acuity, read on the illumination scale, due to adaptation. 
All of these features are important for vocational and clinic 
classification. In order to make these results more directly 
comparable with reference to the last of these points, namely 
Percentage 
Minutes 
Fig. 3. Curves showing the increase in sensitivity as the result of dark adapta- 
tion, the reciprocal of the amount of light required just to discriminate the test-object 
being taken as the measure of sensitivity. Percentage increase in sensitivity is plotted 
against time of adaptation. The initial sensitivity of the eye was standardized in each 
case by 20 minutes of preadaptation to 80 foot-candles of light, vertical component. 
relative amounts of change in sensitivity due to adaptation, 
the ratios or percentages of increase in sensitivity are plotted 
in Fig. 3, the reciprocals of the minimum amounts of light 
required just to discriminate the test-object being taken as A STUDY OF OCULAR FUNCTIONS 
355 
the measure of sensitivity. That is, the ratio or percentage 
change in the value of these reciprocals is plotted against 
time of adaptation, the curves beginning at a common point 
or unit ratio. The relative ratings with regard to the second 
point could of course be represented by plotting the reciprocals 
themselves. Space will not be taken here for this repre- 
sentation. 
It will be noted that the greater part of these observers 
reach their maximum acuity at the end of 15 minutes of 
adaptation and that some even show a lower acuity if the 
series is continued beyond this time. The loss in the latter 
case is doubtless due to fatigue of the muscles of adjustment. 
That is, in case of the observers more susceptible to muscle 
fatigue, the loss of acuity due to fatigue more than compen- 
sated for the small gain in light sensitivity after the first 15-25 
minutes. In this connection it may be noted that the muscle 
strain imposed by taking the acuity at the minimum illum- 
ination is much greater than at the illuminations ordinarily 
used. Even with a 5-10 minute rest period between deter- 
minations and a 2-second interval between the individual 
observations making up one determination, a very noticeable 
fatigue was present at the end of the 45-minute series. 
In testing fitness for the lookout and signal service work 
of the Navy, night flying, and for other work and vocations 
that require the keen discrimination of objects when small 
amounts of light enter the eye, it was deemed better to make 
the tests in terms of acuity rather than of the light sense. 
Retinal sensitivity is only one of the factors in the eye's power 
to see its objects at low illuminations. For example, we 
frequently find observers with an excellent light sense whose 
scotopic acuity or power to discriminate objects at low 
illuminations is poor. Indeed, as shown in this and former 
papers, slight differences in resolving power correspond to 
relatively large differences either in illumination (except for 
very low illuminations) or retinal sensitivity in their effect on 
the eye's power to see clearly at low illuminations. Any test 
therefore for fitness for tasks or work requiring the power to 
see clearly at low illuminations is far from complete which 356 
E. FERREE, G. RAND AND D. BUCKLEY 
leaves out of account resolving power and the varying effect 
of changes in illumination or in light sensitivity on acuity. 
The acuity test, on the other hand, includes all of the factors 
involved in seeing and in the exact proportions in which they 
are contributory to seeing. Moreover, it is much better 
directly to determine the candidate's power to see clearly at 
low illuminations than to try to infer this from a light sense 
test and data on acuity taken at higher illuminations. This 
was, we may say, also the point of view of the Naval authori- 
ties under the auspices of whom we undertook to devise an 
apparatus suitable for testing acuity at low illuminations.