[From the American Journal of Science and Arts, Vol. XIV, Aug., 1877.] 
A PROPOSED NEW METHOD IN SOLAR SPECTRUM ANALYSIS. 
By S. P. LANGLEY. [From the American Journal of Science and Arts, Vol. XIV, Aug., 1877.] 
A PROPOSED NEW METHOD IN SOLAR SPECTRUM ANALYSIS.* 
S. P. LANGLEY. 
No observation of modern Physical Astronomy is more strik- 
ing in its conception, than that which attempts to determine the 
motion of a celestial body, by the altered wave-length of its 
light, and none has attracted more general attention. It is 
popularly understood, I think, that the proper motion of certain 
stars in the line of sight, has been thus completely demonstra- 
ted, but those particularly engaged in such studies know how 
far astronomers have till very lately been from the certainty 
attributed to them. Only last April, Father Secchi presented 
a communication to the Comptes Rendusf in which he recalled 
the denial on theoretical grounds, by Van der Willigen, of the 
possibility of any such observation, pointed out the extreme 
danger of instrumental error in such spectroscopic work, and 
gave experiments of his own, to show that its causes were so 
numerous and so subtle that it was difficult to be certain of any 
result. 
Without discussing the replies, which aim to show that lately 
precautions have been taken which it may be hoped are effect- 
ive, it may be observed that there is another mode of investi- 
gating the reality of the phenomena, not liable to as many 
difficulties as in the case of the stars, though presenting a 
formidable one of its own. I mean, of course, the observation 
of the different wave-lengths of light coming from the east and 
west limbs of the sun, which, owing to that body’s rotation on 
its axis, have equatorial velocities that together make up nearly 
two and one-half English miles a second. This speed, enormous 
in itself, is most insignificant compared with the velocity of 
light, and this relative smallness constitutes the difficulty in 
* Communicated to the meeting of the National Academy, held in Washington, 
April, 1877. 
f C. R., tome lxxxii, pp. 761-812. 141 
S. P. Langley—New method in Solar Spectrum Analysis. 
attempting the solution of the problem by means of the sun, 
for the whole displacement due to it, is (as Professor Young 
has remarked in illustration), but on e-seven ty-seventh of the 
distance between the D lines, or between one-twelfth and one- 
thirteenth of one division of Angstrom’s scale. Zollner, Secchi 
and Hastings have believed that they, nevertheless, detected a 
change in the refrangibility of the light, and Yogel* using Zoll- 
ner’s reversion spectroscope obtained a displacement of from 
•08 to 015 of one of Angstrom's units. In August last, Profes- 
sor Young gave the results of his own measurements with one 
of Mr. Rutherfurd’s gratings, showing an equatorial velocity of 
l,m42. Professor Young was unable to find any displace- 
ment of the atmospheric lines. This last researchf being much 
more systematic than its predecessors, and given in satisfactory 
detail, has turned the weight of scientific opinion in favor of 
the view, that the change due to motion of the luminous body 
is fairly proven. It can hardly, however, be deemed superflu- 
ous to still offer upon so important a question, the results of an 
independent method of measurement, and one which renders 
errors from instrumental displacement, on the danger of which 
so much stress has been deservedly laid, in the sense in which 
the word is here used, not only unlikely but impossible. 
In the course of a research upon the selective absorption of 
the solar atmosphere, I arranged in 1875, means for com- 
paring homogeneous lights from different parts of the disc. 
The apparatus was too complex for description here, but 
it consisted essentially, in the provision of two pair of right- 
angled prisms of total reflection, so disposed in connection with 
a spectroscope, that the spectra could be formed side by side, 
of light from different parts of the sun, and of a photometric 
apparatus by which the relative intensity of the lights at differ- 
ent parts of these spectra could be compared. The results of 
this research, with an improved form of the instrument, will I 
hope be soon ready for publication. It was not intended, pri- 
marily, for the comparison of individual spectral lines, for 
which purpose optical arrangements, not very essentially differ- 
ent, had already been used by Lockyer, Hastings and perhaps 
others, but the investigation of Professor Young suggested to 
me another and cognate method of testing the principle of 
Doppler, to which this apparatus is especially applicable. The 
theory of the proposed method is very simple. Let two spectra 
be formed side by side, the one of light from one edge of the 
sun, the other of light from a point 180° distant. The instru- 
ment being in adjustment, if these points be in the neighbor- 
hood of the solar poles which are relatively at rest, all the lines 
* Beobachtungen auf der Sterawache zu Borhkamp. 
| American Journal of Science and Arts, vol. xii, Nov., 1876. S. P. Langley—jS'cw method in /Solar /Spectrum Analysis. 
142 
will be continuous in both spectra. But if the instrument is 
rotated till the light comes from points on the eastern and west- 
ern sides of the sun, which are in relative motion, not only will 
the solar lines be discontinuous, in the two spectra : as though 
the one receiving light from the advancing or eastern side had 
been slid past its neighbor toward the violet; but any mal- 
adjustments of the instrument, which simulate this effect, can be 
with certainty detected by a means to be shortly described. 
The actual quantity by which we may anticipate from theory 
that the spectra are displaced, is, as has been observed, extremely 
small, and to produce the desired result we need not only very 
great dispersion in the spectroscope, but very delicate work- 
manship in the cutting and mounting of the prisms which are 
to bring the light to the slit, and in general it will be evident 
that more than common skill is to be desired of the instrument 
maker. 
I wish to acknowledge the obligations I am under in these 
respects to Mr. F. Walther, of Philadelphia, the optician, and 
to Mr. William Grunow, the maker, of New York. Not to 
dwell on those difficulties of detail which always come between 
the conception and its embodiment in practice, I may say that 
more than six months have passed in experiment and modifica- 
tion till the instrument has attained its present form, in which 
I have been finally able within the past few days to see my 
anticipations justified. That it has been possible to me to 
undertake this research at all at present, is due to Mr. Ruther- 
furd, who has given me choice specimens of his gratings, which 
are so generally known and valued, that it is unnecessary that 
I should describe them. 
It is desirable that a very clear mental picture should be 
present to the observer, of the amount of displacement he is to 
expect, for though it is well that he should be ignorant, if possi- 
ble, of the anticipated direction, this knowledge of the amount 
will prevent him at the outset from confusing apparent displace- 
ments due to mal-adjustment of the instrument with those to 
which his attention should be directed.* I have taken advan- 
tage of a table prepared by Professor Pickering, showing the 
* The following method of statement appears to present material for such a 
mental picture. In the 3d spectrum of the 11280 line grating the D lines, viewed 
under the magnifying power most favorable in my instrument to clear vision, 
appear like two sharply-defined parallel lines \ inch apart viewed by the naked 
eye at the distance of 10 inches. One of the units of Angstrom’s scale appears 
to fill rather more than inch at the same distance. The displacement here due 
to rotation is as is remarked rather more than 4s or ’08 of a unit, or, referred to 
the distance of distinct vision, very near of an inch. It is known that we 
can with the naked eye distinguish -sifoo of an inch or less, in the form of discon- 
tinuity of two lines on a vernier plate, which is essentially our present case. 
Owing to defective light and other causes, we cannot do quite this on the sun, as 
yet, but can still count on detecting by this method a discrepancy of somewhat 
less than ‘03 units, which again is less than half the amount in question. 143 
S. P. Langley—New method in Solar Spectrum Analysis. 
results of different instruments on the fine lines of the E group, 
to compare mine with them, and I observe that I can discrim- 
inate more, and more delicate lines, with either the 8,000 or 
17,000 line grating than observers have been able to do with 
the most powerful spectroscopes of the common construction. 
Where Angstrom, using a grating of Nobert’s, has delineated 
twelve lines, the grating supplied me by Mr. Kutherfurd shows 
thirty-one, and one of these pairs, though separated by but dis- 
tances of one-sixth of a unit of Angstrom’s scale, is yet clearly 
divided. The displacement to be expected from the solar rota- 
tion, under favorable circumstances, is about one-half of this, 
and hence if it exist, it is clearly within the power of the instru- 
ment to exhibit, since the two spectra are formed in juxtaposi- 
tion as nice as that of a vernier plate to its circle, a circumstance 
under which, as every one knows, the eye judges of the contin- 
uity of lines with the most accuracy. 
The instrument itself consists of the grating, with a collimat- 
ing and observing telescope, each of one and one-quarter inch 
aperture and fourteen and one-half inches focus, with positive 
eye-pieces magnifying from ten to thirty times. The slit is 
constructed so that its jaws are bevelled to knife-edges, where 
they meet, at the outer surface, and these owing to the excel- 
lent workmanship, still close so as to admit no light when held 
up against the sun. Over this slit, with their bases almost in 
contact with it, and their united edges crossing it at right angles, 
are the first pair of reflecting prisms. They are cut from the 
same piece of glass, and so fitted that their junction-line shows 
no more in the spectrum than a particle of dust, and in fact, 
the division between the two spectra is with difficulty distin- 
guishable from the ordinary dust lines. The other two prisms 
slide independently on ways, with verniers reading to thous- 
andths of an inch, permitting them to be put in any part of a 
solar image of from one and one-half inch, to five inches in 
diameter. Over them is a hood, carrying a screen, which 
receives the image projected by the equatorial telescope. This 
focal image is nearly one and three-quarters inches in diameter, 
and any two selected portions of it pass through adjustable 
apertures in the screen. By the construction of the instrument 
they are equally distant from the optical axis, and in every other 
optical condition, as far as possible, similar, except as they may 
differ at their origin in the sun itself. It shows the difficulties 
of the method of research successfully employed by Mr. Huggins, 
and at Greenwich, that when the instrument is turned directly 
on the sun (i. e. so that each spectrum receives light from all 
parts of the sun’s disc), the two sets of spectral lines will not 
ordinarily be continuous. Theoretically they should be, prac- 
tically, we find they are not, owing to numerous latent causes S. P. Langley—New method in Solar Spectrum Analysis. 
144 
of disturbance. A touch on the prisms, a movement of the slit, 
an adjustment of the eye-piece, will ordinarily disturb one 
spectrum relatively to the other, by a minute amount. But the 
whole change we are seeking is of a minuter order still; how 
then, can we discriminate it with certainty? In our ability to do 
this lies the advantage of the method I describe, which, grant- 
ing sufficient dispersive power, makes impossible the instru- 
mental error that has been dwrelt on by Secchi and others with 
justice, as shaking confidence in the result. To see how we 
are authorized to use this word “ impossible,” let us bear in 
mind that the solar spectrum consists of two distinct kinds of 
lines, one caused by absorption in the solar, the other by ab- 
sorption in the terrestrial atmosphere. These latter being 
formed by light from all parts of the sun are independent of its 
rotation. 
The prisms are adjusted, till, on looking on the sun directly, 
the lines are all continuous in both spectra, then the instrument 
is put in the telescope and the slit placed at such a position- 
angle that the light in spectrum A comes from the vicinity of 
the north solar pole, that in spectrum B from the south. On 
looking in, we see a very long and narrow spectrum, filled with 
dark lines and exhibiting the chromospheric lines on both sides. 
It is divided by what appears to be a fine dust line, in two 
exactly corresponding parts, and is in reality two distinct 
spectra, as we see by the opposed chromosphere lines, but as 
the sources of light for both spectra are relatively at rest, all 
the dark lines are still continuous. But now (without dis- 
turbing any adjustment), revolve the whole 90° about the 
optical axis passing through the center of the solar image, so 
that spectrum A is formed by light from the eastern or advan- 
cing edge of the sun; spectrum B by light from the western or 
retreating one. A curious change has taken place. By a very 
minute but perceptible quantity, spectrum A appears to have 
been slid past its neighbor, toward the violet end, so that every 
solar line in the first is “notched” at its junction with the 
second, while at the same time, the telluric lines are as unaltered 
as the fixed lines of a micrometer web would be, by moving a 
scale about in the field. The effect is the same as though the 
spectra were tangible things, like two engine-divided scales, 
whose numerous delicate divisions (represented by the solar 
lines), were all in exact juxtaposition a moment before, and are 
all now just perceptibly displaced, as when a vernier plate is 
moved till a coincidence is made at a new stroke on the limb. 
Moving the instrument 90° more, we come again into the 
axial line of the sun, and the coincidence should return; with 
still 90° more we are again in the equator, but now spectrum 
A is formed by light from the western edge, and this time it is 145 
S- P. Langley—New method in Solar Spectrum Analysis. 
moved the other way, as if it were a seale which had been slid 
by a very slight but distinctly perceptible amount toward the 
red end; while still the telluric lines retain their continuity, 
assuring us that no mal-adjustment has occurred. 
It will be admitted that this change is, if real, excellent ex- 
perimental evidence that the wave length is virtually different 
in light from the eastern and western limbs, as theory predicts. 
For, granting that the instrument is mal-adjusted in any un- 
known way or degree, any instrumental cause will affect solar 
and telluric lines alike, and we may in fact defy ingenuity to 
suggest an error of adjustment, which will modify one and not 
the other. 
For the sake of clearness, I have assumed that we start with 
all the lines continuous in both spectra ; in practice this condi- 
tion is not easily assured: commonly some lurking error, will, 
without especial pains, cause them to appear broken upon a 
fixed source of light; but we disregard this, and consider, as 
we bring the instrument into new positions, only the difference 
of displacement of the solar and telluric lines. The simul- 
taneous observations of this difference, in each of two spectra, 
is the essential condition relied on, not only in theory but in 
actual work. 
It will be remembered that many lines in the spectrum are 
only seen when the sun is low. These are clearly due to ab- 
sorption in our atmosphere. Many thousands, as we know, are 
due to absorption in the sun’s atmosphere. There remains a 
large number of lines not coincident with any we produce at 
the electrodes of our battery, and always present in the spec- 
trum. Of these we do know that they are either caused by 
the sun’s atmosphere or ours, without always knowing which, 
for these can only be inferred to be telluric from their growing 
stronger as the sun sets,* and this, though easily determined in 
the case of a single line, becomes a task of great labor where 
we deal with thousands. It is evident, however, that after 
having used known telluric lines to determine the fact that the 
refrangibility of solar lines only is altered, we can reverse the 
process, and classify unhesitatingly hereafter, all lines as tel- 
luric which are unaffected by the changes that compel others 
to betray their solar origin. To merely see these two spectra 
with clearness, then, is to be enabled to pick out the telluric 
lines from the others, as though they were mapped before us. 
They are mapped in fact, and it becomes, under the proper 
conditions, a matter of simple inspection to determine them. 
I hope then, it will not seem too assuming a title, if I speak 
of this as a new method in solar spectral analysis. 
I have only to add, that in all my trials of this method, I have 
* Or by experiment on artificial lights viewed through intervening atmospheres. S. P. Langley—New method in Solar Spectrum Analysis. 
146 
constantly so arranged each experiment, that I remained in in- 
tentional ignorance as to which spectrum came from the eastern, 
and which from the western limb, until I had determined the 
point by the different behavior of these solar and terrestrial lines, 
which I have been able to do thus far correctly in every instance. 
I believe, in fact, that the effect under proper conditions is so 
marked that the observer hereafter need not take pains to 
guard against the unconscious bias of his measurement toward 
a desired result. I have, in these preliminary experiments, 
however, thought it necessary to take precautions that no such 
prepossession, if it existed, might influence me unawares. 
I ought to add that the proposed method is considerably 
limited, not only by the need of very powerful instruments 
(for these can be supplied), but by the need of very good ob- 
serving weather, which unfortunately we have not at command 
when we want it. 
Useful tests of the desired condition are the duplication of 
such lines as 1529 of Kirchoffs scale, and the more celebrated 
“ 1474,” known as a close double since Professor Young’s 
duplication of it last year. 
I have not yet made any measurements of the displacement 
by this method, which appears to me as thus far described to 
be less adapted to quantitative results than that already em- 
ployed by Professor Young, but capable by the use of the 
micrometer of giving exact numerical data. I ought to add, 
however, that Professor Rood, of Columbia College, has offered 
the valuable suggestion, that by the use of a double refracting 
prism, we can cause the spectra to overlap, thus not only over- 
coming some mechanical difficulties which I have not enlarged 
on, but obtaining at little trouble all the advantage of micro- 
metric measurement. I regret that the instrument reached me 
so late that I have not yet been able to give a trial to this, 
from which I hope excellent results. Nor have I been able to 
introduce photography with other contemplated improvements, 
such as the simultaneous use of both right and left spectra of 
the same order, thus doubling the displacement; and the use 
of three reflections instead of two, on each ray before it reaches 
the slit, thus inverting the portion of each limb under examina- 
tion, and bringing the very edges of the sun into juxtaposition, 
instead of a portion somewhat within the edge, as at present. 
I have only to add in conclusion, that while the instruments 
used have been principally those of the Allegheny Observatory, 
to which institution my services belong, I should have been 
unable to carry out this research without certain indispensable 
adjuncts, the expense of which has been defrayed from an ap- 
propriation placed at my disposal by the trustees of the Bache 
fund. 
Allegheny Observatory, April 14, 1877.