THE MORTALITY FROM ALCOHOL IN 
THE UNITED STATES 
EDWARD BONNELL PHELPS THE MORTALITY FROM ALCOHOL IN 
THE UNITED STATES 
THE RESULTS OF A RECENT INVESTIGATION OF THE CON- 
TRIBUTORY RELATION OF ALCOHOL WITH 
EACH OF THE ASSIGNED CAUSES 
OF ADULT MORTALITY 
EDWARD BUNNELL PHELPS, M. A., F. S. S. 
* 
Editor, The American Underwriter, of New York City, 
and Author of the Mortality of Alcohol, American Mortality Statistics for the 
Nine Years, 1900-1908, Neurotic Books and Newspapers as Factors 
in the Mortality of Suicide and Crime, A Statistical Study 
of Infant Mortality, Infant Mortality and Its Rela- 
tion to Woman's Employment, Etc. 
Reprinted from the Transactions of the Fifteenth International 
Congress on Hygiene and Demography, held at 
Washington, D. C., September 23-28, 1912 
7^ O THE MORTALITY FROM ALCOHOL IN THE 
UNITED STATES-THE RESULTS OF A RE- 
CENT INVESTIGATION OF THE CON- 
TRIBUTORY RELATION OF ALCOHOL 
WITH EACH OF THE ASSIGNED 
CAUSES OF ADULT 
MORTALITY 
INTRODUCTION 
ON February 2, 1911, Capt. Richmond P. Hobson, of 
Merrimac fame and of late years a Member of Congress 
from Alabama, announced on the floor of the House of Repre- 
sentatives his intention of printing in the proceedings a lecture 
of his on the subject of the effects of alcohol, and this lecture 
has since been reprinted from the Congressional Record under 
the title of "The Great Destroyer," has been widely circulated, 
and has been extensively quoted in public addresses by ministers, 
physicians, and prohibition advocates. In this speech Capt. 
Hobson cited alleged English statistics of mortality due to 
alcohol, undertook to apply them to this country, and thus 
arrived at the conclusion that "over 680,000 deaths per year 
in continental United States" are due to alcohol, and that 
"alcohol is killing our people at the rate of nearly 2,000 men a 
day every day in the year, as many Americans every year as 
all the wars of the world have killed in battle in 2,300 years." 
At the time of the delivery of this speech by Capt. Hobson 
the latest available mortality statistics of the United States were 
those for the calendar year 1909, which were presented in 
Bulletin 108, then recently issued by the Bureau of the Census. 
According to this bulletin, the total number of deaths of both 
sexes, at all ages, from all causes, in the registration area of the 
United States in 1909 was 732,538. Assuming that this death 
rate for the 55.3 per cent, of the population of continental 
United States included in the registration area was representa- 
tive of the country at large, this figure would predicate a total 4 
JOINT SESSION OF SECTIONS IV AND IX 
mortality of 1,324,662 for the entire country. With this tenta- 
tive figure as a basis of calculation, Capt. Hobson's assertion 
that alcohol was responsible for "over 680,000 deaths per year 
in continental United States" would indicate that 51.3 per 
cent, of all deaths, at all ages, from all causes, in the United 
States were due to the single factor of alcohol. Even were 
the presumptive total mortality of the country calculated by 
doubling the recorded mortality of the registration area-as some 
statisticians believe should be done, in order to allow for the 
higher negro mortality of the Southern States not included 
in the registration area-the total mortality of the United States 
in 1909 would have been but 1,465,076, and the alleged mor- 
tality due to alcohol would have amounted to 46.4 per cent, of 
this total. In short, Capt. Hobson's positive assertion that more 
than 680,000 deaths in the United States each year were due 
to alcohol could only be regarded as a declaration that alcohol 
was annually responsible for approximately one-half of the 
total mortality of the United States at all ages and from all 
causes. 
The preposterous nature of any such declaration, of course, 
is apparent on its very face, and I confidently venture the 
opinion that no man of unbiased mind and ordinary reasoning 
powers could regard such a statement as entitled to serious 
consideration. Mark Twain once grotesquely termed the rumor 
of his death "grossly exaggerated," and Capt. Hobson's state- 
ment was of like order. That fact was apparent, and the 
figure named literally fell by its own weight; but by way of 
natural sequence there followed the question, to what extent 
is this figure of 680,000 deaths per year from alcohol in the 
United States an exaggeration, or, in other words, about how 
many deaths in this country each year are presumably due to 
alcohol? No answer to this question was to be found in 
official figures, the only direct clues in the vital statistics of 
the Bureau of the Census and the various State and municipal 
mortality records being the meager returns for deaths directly 
due to alcoholism, delirium tremens, or the immediate effects 
of alcohol otherwise classified. 
In the last 60 years there have been at least three more or 
less exhaustive scientific investigations of English mortality THE MORTALITY FROM ALCOHOL 
5 
due to alcohol, but up to 1911 no investigation worthy of the 
name in this country, in so far as I was able to ascertain from 
this country's bibliography of alcohol or by correspondence with 
some of its principal authorities. In view of the importance and 
far-reaching ramifications of the subject, and the apparent lack 
of any credible figures, either official or unofficial, it seemed 
high time that a pioneer effort on these lines should be made. 
With the kind cooperation of certain eminent medical authorities 
hereinafter named, I accordingly made the effort in question 
and published the results of my investigation in book form, 
under the title of "The Mortality of Alcohol-A Statistical 
Approximation of the Deaths in the United States in Which 
Alcohol May Figure as a Causative or Contributory Factor." 
THE COMPLEX NATURE OF THE PROBLEM 
The physiological and pathogenic influences of alcohol are 
a profound, complex, and preeminently debatable problem on 
which perhaps no two minds, either lay or medical, ever have 
been in entire accord. Even were every adult death immediately 
followed by an autopsy there doubtless would be a great many 
cases of difference of expert medical opinion as to whether in 
these specific instances alcohol had or had not played any 
appreciable part in causing death. And surely, in case of a large 
minority-if not, indeed, in a majority-of deaths nothing 
short of an autopsy could positively determine the actual im- 
portance of alcohol as a mortality factor. For these reasons 
it is apparent that no man who could authoritatively say precisely 
how many deaths are directly or indirectly due to alcohol now 
lives, ever has lived, or ever will live. At best the statistics 
of the subject, however credible and carefully compiled, can 
be but mere approximations. 
Ruskin made the profound observation that "the work of 
science is to substitute facts for appearances and demonstra- 
tions for impressions," and, measured by these standards, per- 
haps no approximation of the mortality of alcohol could be 
regarded as strictly scientific. The most carefully worked-out 
approximation is open to debate; and facts are not debatable. 
The subject of alcohol and its use and abuse being one of the 6 
JOINT SESSION OF SECTIONS IV AND IX 
most hotly debated topics of modern times, it is patent that no 
attempt at a serious approximation of the deaths due to alcohol 
would be received with unanimous approval, whatever the char- 
acter of its showings. On undertaking the work of trying 
to arrive at such an approximation by the most promising 
processes available, I clearly foresaw that any man on publish- 
ing to the world the results of his investigation would be con- 
fronted with the pleasant choice of the Scylla of prohibition 
disapproval, or the Charybdis of antiprohibition dissent, accord- 
ing as the resultant figures might prove lower than the former, 
or higher than the latter, would wish to have them. Some such 
option is generally one of the prices which must be paid for 
pioneer work in any field of wide public interest, however, 
and having secured the necessary expert cooperation I entered 
on the proposed work without either prejudice or suspicion 
as to its outcome in the way of high or low figures in the 
approximation of the mortality of alcohol thus deduced-and, 
I might add, with complete indifference as to the particular 
horn of the dilemma on which I was destined to be impaled. 
To the best of my knowledge, not even a semblance of 
official figures purporting to present the total number of deaths 
in which alcohol has played a causative or contributory part 
in the mortality of the United States now is, or ever has been, 
obtainable. The returns for deaths due to alcoholism, delirium 
tremens, etc., of course constitute but one item of that mortality. 
In Switzerland, as I understand, for several years the official 
death returns for the cities have been supplemented with con- 
fidential reports in cases of adult mortality in which alcohol 
was a contributory cause of death in the judgment of the attend- 
ing physician, thus affording at least some means of officially 
computing the adult urban mortality in which alcohol sup- 
posedly figures. As those familiar with "the Budapest system" 
are aware, the secondary causes of death are entered in the 
official mortality statistics of that city. With these two ex- 
ceptions, there are no mortality statistics in any country, in 
so far as I have been able to learn, which could be regarded 
as an official measure of the mortality directly or indirectly 
due to alcohol. It was therefore apparent at the very outset 
that any investigation of the subject in this country under THE. MORTALITY FROM ALCOHOL 
7 
present conditions must necessarily be entirely without the 
domain of official statistics. Whence, then, were any figures 
entitled to serious credence to be sought? That was the first 
and foremost of the many questions involved. 
The subject of the proposed inquiry being so pronouncedly 
pathologic, obviously the desired information could be obtained 
only from the medical profession. But from what branch 
or class of that profession could reliable data be expected? 
And taking into account the dearth of even unofficial medical 
records of deaths in the United States in which alcohol had 
played some part, how was the requisition for information to be 
framed so as to produce results of any real value? These 
questions, especially the last named, were of such serious mien 
as to seem at first thought almost unanswerable. Obviously 
it would be useless to look to any of the great medical asso- 
ciations, county medical societies, or medical bodies of lesser 
magnitude for the desired figures or estimates. It was no 
less clear that even the greatest and most experienced medical 
and surgical specialists would not be competent witnesses, their 
several experiences of course having been restricted to their 
respective lines of practice. To a certain extent this same 
objection apparently would apply to the house staffs of the 
great hospitals, for the reason that in their case there would 
certainly be what is known in the insurance world as "adverse 
selection," or an unquestionable preponderance of experience 
with the lower strata of humanity with exceptional alcoholic 
leanings. Were the inquiry to be confined to the ranks of 
"the general practitioner," that sturdy, old-fashioned body of 
physicians who have done so much to contribute to the health 
and happiness of the world, the question might arise as to 
whether these men had the up-to-date, specialized knowledge 
of the intricacies of the alcoholic problem necessary to qualify 
them as competent authorities on the subject at issue. 
The elimination of these several groups of pathologic and 
pathogenic authorities had seemed all but to exhaust the entire 
panel of medical witnesses when it occurred to me that some 
of my distinguished medical friends occupying the positions 
of medical directors of great insurance companies might not 
only be capable of furnishing the desired information, but, as 8 
JOINT SESSION OF SECTIONS IV AND IX 
a matter of fact, perhaps the best possible authorities on the 
subject owing to their extensive experience on wholesale lines, 
so to speak. I take it that all intelligent men, even without 
the slightest insurance training or experience, by this time are 
aware of the fact that all applications for life insurance are 
finally passed upon by the medical directors of the companies 
in question, these officials constituting the medical court of 
last resort in so far as the approval or rejection of the applica- 
tion for insurance is concerned. In a lesser degree this same 
rule applies to the larger casualty or personal accident insurance 
companies, and in all applications for life or personal-accident 
insurance there are certain questions regarding the drinking 
habits of the applicants whose answers to these questions are 
made part and parcel of the proposed insurance contracts. 
Necessarily, therefore, the medical directors of these com- 
panies are compelled to keep posted, at least in a general way, 
on the latest knowledge as to the relations of alcohol and 
human morbidity and mortality, and hence are constantly in close 
touch with this phase of the alcoholic problem. 
Furthermore, the medical director of a large life or accident 
company annually passes on many thousands of applications, 
and, the alcoholic habits or experience of the applicant in 
each case being an important factor in determining whether 
the application is to be approved or rejected, the medical 
director of a well-run insurance company might be said to 
maintain a closer relation with the actual alcoholic habits of the 
community at large than does any other class of physicians, 
with the possible exception of those on the staffs of "gold- 
cure" establishments. For these reasons I concluded that the 
medical directors of some of the larger insurance companies 
might prove exceptionally competent witnesses in the proposed 
quest for an approximation of the mortality of the United 
States in which alcohol appreciably figured. The source of 
the desired information being determined, the next and all- 
important question was just how could the proposed inquiry 
be so formulated as to bring it within the limitations of the 
medical directors' field of work and yield tangible results on 
the basis of their actual experience. THE. MORTALITY PROM ALCOHOL 
9 
THS BASIC PLAN OF THE INVESTIGATION 
The so-called law of average being the bedrock foundation 
of all sound forms of insurance, and the literature of alcohol 
and alcoholism apparently indicating that there are almost as 
many distinct positions as commentators on these subjects, the 
possibility-not to say the eminent desirability-of in some 
way averaging the data collected from the medical directors 
who might be pressed into service in the proposed investigation 
naturally suggested itself. In the home offices of all well-run 
life and accident insurance companies carefully compiled records 
of the causes of policy holders' deaths are kept and interesting 
tabulations of these records for long stretches of years have been 
published by some of the larger companies, but in no insurance 
office, so far as I am aware, are the contributory causes of death 
systematically recorded. In default of actual records of deaths 
in the case of which alcohol figured only as a minor factor, 
of course it would have been impossible for medical directors 
to furnish the number of deaths directly or indirectly caused 
by alcohol in the case of their respective companies, and evi- 
dently the nearest approach to exact figures at their hands 
would be in the nature of estimates based on their singularly 
broad experience in passing on applications and death claims. 
It was instantly apparent that the more numerous and the 
more sharply differentiated the classes on which the estimates 
were based the more accurate would probably be the final 
results of these estimates-and the more numerous the available 
checks on those results. I therefore conceived the idea of 
calling upon each medical director who might participate in 
the investigation for his personal, independent estimate of 
the percentage of male deaths at adult ages directly or in- 
directly due to alcohol in the case of each of the nominal 
causes of adult male deaths in which alcohol could possibly have 
figured to any appreciable extent. The plan so mapped out 
provided for applying to the total number of male deaths at 
adult ages in the case of each of these causes the average of 
the percentage estimates obtained from the medical directors 
for that cause of death, thus working out the presumptive 
number of adult male deaths from that nominal cause in which 10 
JOINT SESSION OF SECTIONS IV AND IX 
alcohol really figured, and then by the addition of these totals 
arriving at a tangible approximation of the total adult male 
mortality in which alcohol was a factor in the registration 
area of the United States for which specific mortality figures 
for recent years were available. With this total as a rationally 
established basis, it seemed possible that at least a fairly safe 
approximation of the total alcoholic mortality of continental 
United States might be calculated. 
Having thus tentatively formulated a plan for a pioneer 
effort at an approximation of the alcoholic mortality of the 
United States, I consulted Dr. Brandreth Symonds, president 
of the Association of Life Insurance Medical Directors of 
America, chief medical director of the Mutual Life Insurance 
Co. of New York, and for more than 20 years identified with 
that great company as medical examiner, assistant medical 
director, medical director, and chief medical director, and re- 
quested his frank opinion as to whether in his judgment the 
proposed plan was practicable and workable, and might be ex- 
pected to produce results of real value. Dr. Symonds replied 
that the plan was not only new and novel, but entirely work- 
able and well worth the trying, and that I could count on his 
hearty cooperation with the proposed investigation. The re- 
sults of this encouraging decision on the part of so eminent 
a medical authority of the life insurance world were the 
investigation and the publication of its results under the title 
of "The Mortality of Alcohol" in the latter months of 1911. 
Dr. Symonds having kindly consented to serve on the medical 
jury to which the case was to be submitted, I then enlisted 
the cooperation of two other medical directors of long and 
broad experience, namely, Dr. Eugene L. Fisk, medical director 
of the Postal Life Insurance Co., and Dr. William L. Gahagan, 
medical director of the United States Casualty Co., both of 
New York City, and secured from each of these three physicians 
his independent estimates of the percentage of male deaths at 
adult ages in which alcohol had probably figured as a causative 
or contributory factor in the case of each of 106 causes of 
death listed in the mortality statistics for 1908 of the Bureau 
of the Census, the statistics in question being the latest avail- 
able when the investigation was commenced. All told, 187 THE MORTALITY FROM ALCOHOL 
11 
causes and classes of causes of death were listed in the census 
report in question, but as the preliminary investigation was 
to be restricted to male deaths at adult ages, 81 of these causes 
or classes of causes were stricken out as without the scope 
of the investigation in the judgment of several physicians called 
upon to pass on the list. For similar reasons the three medical 
directors of whom percentage estimates were requested were 
asked to confine their estimates to the probability in the case 
of male deaths between ages 20 and 74, inclusive, their official 
experience not including infant or child mortality, and the 
consensus of medical opinion being that male deaths at age 
75 or over in which alcohol appreciably figured practically 
constituted a negligible element. 
the; results of the investigation 
In the lists of primary causes of male deaths at adult ages 
submitted to the three medical directors for their independent 
percentage estimates in each case, the 106 causes in question 
were broadly classified in three groups, as follows: 
Group A. Twenty-eight causes of adult male deaths in which 
alcohol may have been an important contributory factor and 
sometimes the principal causative factor. 
Group B. Forty-eight causes of adult male deaths in which 
alcohol may have been a minor contributory cause, or at least 
a distinctly disturbing factor. 
Group C. Thirty causes of adult male deaths in which 
alcohol was not primary or secondary cause, but may have been 
a harmful contributory factor. 
Without going into needless details regarding the findings 
in the case of each of these groups, the results of the ap- 
plication of the average of the percentage estimates for each 
separate cause of death to the total number of male deaths 
at adult years charged to that cause in each of the three groups 
may be summarized as follows: 12 
JOINT SESSION OF SECTIONS IV AND IX 
Groups 
Total number of 
male deaths from 
these causes 
(ages 20 to 74, 
inclusive) 
Number of male 
deaths from these 
causes in which 
alcohol figured, 
according to 
average of 
physicians' 
estimates 
Group A 
73,747 
107,422 
17,689 
i8,337 
13,664 
852 
Group B 
Group C 
Totals 
198,858 
32,853 
The above figures merely state on the broadest possible lines 
the immediate results of the entirely independent percentage 
estimates of the three medical directors in the case of each of 
106 causes of male deaths at adult ages in the registration area 
of the United States in 1908, on the basis of which was calcu- 
lated a statistical approximation of the total alcoholic mortality 
of both sexes at adult ages in continental United States. The 
above group totals, of course, make no showing of the indi- 
vidual estimates or of the pronounced differences between the 
three physicians' estimates in many cases which, in my judg- 
ment, materially contribute to the credibility and reliability 
of the final showings as thoroughly representative of American 
medical opinion as to the mortality directly or indirectly due 
to alcohol. The opinions of equally competent physicians as 
to the part which alcohol plays in modern mortality materially 
differing, according to their several viewpoints, and the esti- 
mates on the averages of which the findings of this investiga- 
tion are based having been made without so much as the 
slightest semblance of consultation in any case, it was inevitable 
that in some cases the estimates would widely vary. Had 
they not so varied they would not have been representative of 
American medical opinion on the subject. 
It was, therefore, fortunate rather than unfortunate that 
one of the medical directors whose estimates figured in this 
investigation proved to be a physician of pronouncedly anti- 
alcoholic leanings-that is to say, one of the class of physicians 
who believe alcohol to be a particularly potent mortality factor. THE. MORTALITY FROM ALCOHOL 
13 
As an inevitable result of this view of alcohol's pathogenic 
effects, this physician's estimates were almost invariably higher 
than those of the two other medical directors, so that the 
averages of the three estimates throughout the tabulation of 
deaths from 106 different causes probably were fairly represen- 
tative of ultraconservative, conservative, and pronouncedly anti- 
alcoholic American medical opinion. In fact, some physicians 
have expressed the opinion since the publication of the results 
of the investigation that even were a poll of two or three hun- 
dred American physicians to be made and the averages of 
their percentage estimates applied to the total number of 
deaths at adult ages in the United States, the final showings 
of that poll probably would not materially differ from the 
findings based on the averages of the three medical directors 
who figured in this investigation. Further proofs of the prob- 
able approximate correctness of these findings are hereinafter 
briefly summarized in this paper. 
The tabulation of the preliminary showings above presented 
indicates that 32,853 of the total of 198,858 adult male deaths 
from the 106 causes in question were directly or indirectly 
due to alcohol, or, in other words, that 16.5 per cent, of the 
total number of deaths in question were deaths in which alcohol 
presumably played some part. A careful comparison of the 
number of male and female deaths directly due to alcoholism 
in the registration area of the United States, and also in New 
York City, apparently indicating that the male mortality in 
which alcohol figures in this country must be at least five 
times as large as the female mortality of this class, and this 
conclusion being confirmed by every one of the physicians and 
registration officials to whom the matter was submitted, it 
seemed entirely safe to assume that the percentage of female 
deaths at adult ages directly or indirectly due to alcohol would 
not exceed one-fifth, or 20 per cent., of the male ratio. The 
16.5 per cent, of male deaths between ages 20 and 74, inclusive, 
directly or indirectly due to alcohol in the case of the 106 causes 
considered being found to constitute about 13.2 per cent, of the 
total male deaths from all causes at ages 20 and upward, the 
female mortality in which alcohol presumably figured was 
consequently calculated on the basis of one-fifth of this per- 14 
JOINT SESSION OF SECTIONS IV AND IX 
centage rate, or 2.6 per cent, of the total female mortality from 
all causes at ages 20 and upward, and so computed the total 
adult alcoholic mortality of the registration area of the United 
States in 1908 apparently reached the total of 38,288 deaths, 
or 8.4 per cent, of the total mortality from all causes at ages 
20 and upward. 
The presumptive adult mortality of the registration area 
in which alcohol figured as a factor in 1908 having been ap- 
proximately computed, it only remained to apply the percentage 
figures thus worked out to the estimated population of the 
remainder of the country, with the adjustment necessary for 
adapting them to the much larger rural population of the non- 
registration area, in order to arrive at a statistical approxima- 
tion of the total mortality of continental United States in which 
alcohol presumably figured as a causative or contributory factor. 
Making the necessary allowances for the estimated rural and 
urban populations of the nonregistration area and the supposedly 
resultant mortality rates of that area, the computation of the 
alcoholic mortality of the entire country, worked out on the 
lines hereinbefore summarized with the averages of the three 
medical directors' estimates as a basis, fixed the mortality of the 
United States in 1908 in which alcohol appreciably figured as a 
factor at 65,897, or in round figures 66,000, this total being about 
5.1 per cent, of the supposed total mortality of the country 
from all causes at all ages, or 7.7 per cent, of the total mortality 
from all causes at ages 20 and upward. Otherwise stated, this 
figure indicates that alcohol might be supposed to play some part 
in about 1 in every 20 deaths at all ages, or in about 1 in every 
13 deaths at adult ages (ages 20 and upward). As is stated 
in the conclusions presented in my book (p. 73), "it should 
be clearly understood that this figure by no means signifies 
that alcohol was the direct cause of 66,000 deaths, the num- 
ber in question presumably including all of the deaths in which 
alcohol played any appreciable contributory part. Consequently 
the number of deaths thus computed is not properly com- 
parable with the number of deaths accredited to any particular 
cause in the annual mortality statistics of the registration area, 
as in every case those figures deal with deaths immediately due 
to the cause named." THE MORTALITY FROM ALCOHOL 
15 
THE RESULTS MEASURED BY OTHER STANDARDS 
According to Capt. Richmond P. Hobson's widely advertised 
congressional speech and subsequent booklet on "The Great 
Destroyer," alcohol is responsible for more than 680,000 deaths, 
or approximately one-half of all deaths from all causes at all 
ages in continental United States. According to the results 
of this pioneer investigation of the subject in the United States, 
alcohol is responsible, directly or indirectly, for not more than 
66,000 deaths, or approximately one-twentieth of all deaths 
from all causes at all ages in continental United States. Both 
figures are but approximations, but the discrepancy between 
them is somewhat pronounced. Which of them is apparently 
more firmly grounded and entitled to more serious considera- 
tion? The discussion of the results of my investigation in 
medical, temperance, and lay periodicals stretching out over 
very nearly 12 months has failed to bring to light any semblance 
of a previous investigation of the probable mortality from 
alcohol in the United States, and it would seem safe to assume 
that there had been no such previous investigation. There 
apparently being no prior data on these lines available, the 
presumptive accuracy or inaccuracy of the conclusions of this 
investigation, as contrasted with Capt. Hobson's tenfold assump- 
tions, can only be measured by a comparison of the methods 
by which the two estimates-almost as far apart as the poles- 
were arrived at, by a checking up with the best foreign data 
obtainable, and in the light of the most competent, unbiased, 
American opinion. 
The processes of my approximation of 66,000 deaths in 
the United States in 1908 in which alcohol may have figured as 
a causative or contributory factor, based on the average of 
the independent estimates for each of 106 causes of death by 
the medical directors of three large American insurance com- 
panies, have already been sufficiently outlined. Now, as to 
Capt. Hobson's own statement of the methods by which he 
arrived at his conclusion that the number of deaths in this 
class should be fixed at "over 680,000 deaths per year in 
continental United States." On pages 4 and 5 of the reprint 
of his lecture on "The Great Destroyer" I find this explanation 
of the faith that is in him: 16 
JOINT SESSION OF SECTIONS IV AND IX 
It is difficult to say in any particular case whether having alcohol 
in the system caused a patient to take a disease or caused a patient 
to die, and alcoholism attributed to men who die in delirium tremens 
is the only record of death ordinarily kept against alcohol. But the 
British Government, in conjunction with English life insurance com- 
panies, from the records of millions of cases, has been able to deter- 
mine the death rate of total abstainers and of those who drink. 
Statistics compiled by insurance companies show that the death rate 
for the population at large is 1,000 death per year out of every 61,215 
of the population, and that the death rate of total abstainers is 560 
per year out of the same number, and for liquor dealers 1,642 death 
per year out of the same number. These figures, resulting from many 
millions of cases, can be taken as accurate. They show that 440 deaths 
out of every 1,000 deaths, nearly one-half of the deaths that occur, 
are due to alcohol. Applied to this country, over 680,000 deaths per 
year in continental United States or over 725,000 per year in the United 
States and its possessions. In other words, alcohol is killing our 
people at the rate of nearly 2,000 men a day every day in the year. 
On the strength of nearly 20 years' experience in the con- 
duct of an insurance magazine, and a fairly comprehensive 
acquaintance with the records and literature of life insurance, 
I confidently challenge this statement that "the British Govern- 
ment, in conjunction with English life insurance companies, 
from the records of millions of cases, has been able to determine 
the death rate of total abstainers and of those who drink." 
If the British Government ever has undertaken any such task, 
it has completely escaped my attention, and I should be very 
glad to have the date and title of the official record of any 
such investigation, and the page on which the alleged results 
as stated by Capt. Hobson are presented. That the bare figures 
quoted by him were presented in a British "Bluebook" I am 
well aware, now having before me the three ponderous volumes 
of the Report of the Inter-Departmental Committee on Physical 
Deterioration (Cd. 2175, Cd. 2186, and Cd. 2210), published 
in 1904, in which they appeared, but I therein find positive 
evidence that Capt. Hobson has entirely misunderstood, and 
misquoted, the figures in question, which he characterizes as 
"statistics compiled by insurance companies." In the first 
place, life insurance companies do not bother with "the death 
rate for the population at large," their mortality compilations 
practically being confined to their own respective mortality THE MORTALITY FROM ALCOHOL 
17 
experiences. In the second place, the figures cited by Capt. 
Hobson, and incorrectly cited, were not compiled or presented 
by the British Government, but were presented as an appendix 
to the evidence of a Mr. W. McAdam Eccles and a Dr. Robert 
Jones before the inter-departmental committee. In the third 
place, as will be seen by reference to page 64 of Volume III 
of the committee's report, the figures as presented in this ap- 
pendix actually conveyed a very different meaning from that 
attributed to them by Capt. Hobson, reading in the report as 
follows: 
comparative mortality op adult males, PUBLICANS, and 
abstainers 
The registrar general has ascertained that of 61,215 men between 25 
and 65 in the community, 1,000 die in one year; but of 61,215 publicans, 
1,642 die in one year; but of 61,215 Rechabites (abstainers), 560 die 
in one year. 
The figures for the death rates of "men between 25 and 65 
in the community" and for publicans between those ages, I 
find on pages ix-x of Dr. John Tatham's letter to the registrar 
general in the Supplement to the Fifty-fifth Annual Report 
of that official, published in 1897 (C. 8503), but the death rate 
for Rechabites (abstainers) alleged as above apparently is 
not included in that voluminous report, and presumably was 
cited from other sources. Not only are the figures quoted from 
Dr. Tatham's letter to the registrar general those for the 
years 1890, 1891, and 1892-or, in other words, 20 years old- 
but they afford no measure whatsoever of the mortality of 
alcohol, merely tracing the comparative mortality of publicans 
between ages 25 and 65 and the general male community 
between those ages. Even should the death rate cited for 
Rechabites (abstainers)-the authority for which is not men- 
tioned-have rested on sufficient experience and been so care- 
fully worked out as to entitle it to serious consideration, the 
comparison of such a death rate among a carefully selected 
class of men between ages 25 and 65 and the death rate among 
the general male community at those ages would be equally 
valueless as a means of measuring the relative mortality of 
abstainers and nonabstainers, the one presumably being a select 18 
JOINT SESSION OF SECTIONS IE AND IX 
class of insurable risks-and, incidently, total abstainers-and 
the other, the general run of the male community, including 
men of all classes, both drinkers and nondrinkers. 
As a matter of course, the death rate of a selected, insurable, 
body of men in any age group might be expected to be materially 
lower than that of the entire male community in that age 
group, and consequently Capt. Hobson's novel method of under- 
taking to measure the mortality properly chargeable to alcohol 
by subtracting from the death rate of the adult male com- 
munity at large the alleged death rate of a select, insurable, 
body of abstainers of like age, and then announcing the dif- 
ference as the alcoholic death rate, is not only unique but 
utterly valueless. So computed, his assertion that alcohol is 
responsible for "over 680,000 deaths per year in continental 
United States" obviously falls of its own weight, as previously 
stated. Were it not for the fact that the figure in question was 
the only specific figure at which the mortality from alcohol in 
the United States had been fixed in any quarter prior to 
my recent investigation, and that it had been widely quoted, 
it would scarcely have been worthy of the actual demonstration 
of its absurdity on the strength of official records with which 
I have dignified it. 
The ridiculous figure of 680,000 deaths from alcohol per 
annum in the United States disposed of, the approximation of 
about 66,000 deaths arrived at as the result of the recent in- 
vestigation would seem to be entitled to serious consideration 
as a starting point for further investigations, if fairly com- 
parable with the results of the English investigations of many 
years since, and the official figures for the towns of Switzerland 
previously mentioned in this paper, no other standards of com- 
parison apparently being available. At least, such would seem 
to be the case, unless some tangible, unanswerable arguments 
against the tentative acceptance of this figure in default of 
more exact data were to be brought forward. Up to date, to 
the best of my knowledge, no such arguments have been 
advanced, the leading medical journals of this country in their 
review of the published results of the investigation having 
practically concurred in the conclusion that the final approxima- 
tion of 66,000 deaths directly or indirectly due to alcohol was THE MORTALITY FROM ALCOHOL 
19 
presumably not far away from the actual figure. In so far 
as the three English investigations of the subject are con- 
cerned, but one of them, to wit, that conducted by the Harveian 
Society of London in 1879-1882, was laid out on such lines as 
to permit of any comparison of its results with those of my 
investigation. 
As read before the Harveian Society November 16, 1882, 
and published in the British Medical Journal of January 20, 
1883 (Vol. I for 1883, p. 97 et seq.), the report of the English 
investigation in question dealt with returns for 10,000 deaths 
in the city of London collected from London physicians, and 
showed that 1,402 deaths, or almost exactly 14 per cent., of the 
total of 10,000 deaths apparently had been accelerated or 
partly caused by the abuse of alcohol, or were wholly due to it. 
About 75 per cent, of the returns in question had come from 
private practitioners, and the remaining 25 per cent, from 
workhouse infirmaries, lunatic asylums, hospitals, and inquest 
reports, and all, it will be remembered, dealt with London 
mortality. My investigation, on entirely different lines, showed 
an apparent mortality of about 7.7 per cent, of the total mortality 
of the United States at adult ages in 1908, as contrasted with 
the Harveian Society's showing of a mortality of 14 per cent, 
at adult ages in the city of London in or about 1880. Taking 
into account the pronounced differences in the periods and 
fields of the two investigations, how do their findings com- 
pare, or, in other words, is the discrepancy in the findings 
greater than might have naturally been expected? I venture 
to believe that it is not, and for these reasons: 
The English investigation not only dealt with the mortality 
conditions of a generation ago, but solely with urban mortality, 
and with the urban mortality of the greatest city in the world. 
About one-quarter of its figures came from infirmaries, asylums, 
hospitals, etc. 
In the 28 years intervening between 1880 and 1908 the 
corrected death rate of England and Wales decreased from 19.5 
to 14.7, or by all but 25 per cent., and even assuming that the 
Harveian Society's figure of 14 per cent, of the total adult 
mortality of London was also a fair measure of the alcoholic 
mortality of England and Wales in 1880, it would obviously 20 
JOINT SESSION OF SECTIONS IV AND IX 
be unfair to compare the supposed alcoholic mortality of the 
United States in 1908 with that of England and Wales 28 
years earlier. As a well-known matter of fact, in both countries 
the rural mortality is always much lower than the mortality 
of the cities, and it would seem fairly assumable that what- 
ever the alcoholic mortality rate of the great city of London 
might be, the alcoholic mortality rate of the entire United 
States, with approximately two-thirds rural population, would 
be materially lower. 
Furthermore, Mulhall's statistics show that the per capita 
consumption of alcohol in the United Kingdom in 1871-1880 
was 2.10 gallons as against a per capita consumption of 1.14 
gallons in the United States in 1880, and the figures of the 
Board of Trade of Great Britain and Ireland for 1900 charge 
the United Kingdom with a per capita consumption of absolute 
alcohol of 2.08 gallons, as compared with one of 1 gallon 
for the United States. All these facts would forecast a much 
lower alcoholic death rate in the United States, as a whole, 
in 1908 than in London in 1880, and the materially greater 
amount of female drinking in London than in this country, 
and the materially greater percentage of known female mortality 
directly chargeable to alcohol would be certain to make the 
London alcoholic death rate much higher. In fact, were 
the ratio of 55.26 female deaths for each 100 male deaths in 
which alcohol figured, as shown by the Harveian Society's 
London investigation, applied to the known figures for the 
registration area's total adult mortality considered in my in- 
vestigation, the deaths of both sexes at adult ages directly or 
indirectly due to alcohol would have amounted to 11.2 per 
cent, of the total adult mortality, as compared with the Harveian 
Society's figure of 14 per cent, for London in 1880. 
In short, taking into account only the more important dif- 
ferences in the conditions governing the two investigations, 
it would seem that the difference between the Harveian Society's 
measure of the alcoholic mortality of the city of London in 
1880 as 14 per cent, of the total adult mortality, and my ap- 
proximation of the alcoholic mortality of this entire country in 
1908 as 7.7 per cent, of the total adult mortality was no greater 
than might have been expected. The records for the leading THE MORTALITY FROM ALCOHOL 
21 
towns of Switzerland in 1900-1903 suggest that 10.3 per cent, 
of the deaths among adult males were in whole or part due to 
alcohol, and that 1.9 per cent, of the deaths among adult 
females were so due, and taken either separately or jointly 
these figures do not materially differ from the showings of 
this investigation for the United States-the Swiss ratio of 
female to male deaths, or 1.9 to 10.3, being almost identical 
with my assumption of a ratio of 1 to 5. In conclusion it 
may be truthfully said, I think, that there is nothing in the 
English figures of 30 years ago or the Swiss figures of recent 
years to raise any serious question as to the probable approxi- 
mate accuracy of the results of this recent investigation in 
this country. 
a supplemental confirmation of the findings 
An entirely unexpected and exceptionally valuable confirma- 
tion of the general showings of my investigation was forth- 
coming about three months ago in the form of a personal 
letter and attached tabulation sent to me by my esteemed friend, 
Dr. Cressy L. Wilbur, chief statistician of the Division of 
Vital Statistics, Bureau of the Census. Under date of June 26 
last Dr. Wilbur wrote me that he had just finished reading 
my book on "The Mortality of Alcohol," and added: 
It may interest you to see some estimates that I made myself be- 
fore examining the averages or individual opinions (of the three 
medical directors on whose percentage estimates the findings of the 
investigations were based). Subsequently I added in red ink the aver- 
ages, so that you can conveniently compare them with my offhand 
estimates. Of course, I look at it from a somewhat different point 
of view, regarding not merely the form of the term and its medical 
significance, but the kind of cases that were included under such term 
in the course of the practical compilation of causes of death. Never- 
theless, on the whole, I think there is a very remarkable correspondence, 
and I should, of course, be disposed to modify certain percentages 
of my own to correspond with the general consensus of opinion. 
The tabulation attached to Dr. Wilbur's letter bore at its 
head a note to the effect that "the estimates below were made 
before examining the tables (before even turning over the 
pages on which they appear or reading the headings or any part 
thereof) and are therefore entirely uninfluenced by the ratings 22 
JOINT SESSION OF SECTIONS IV AND IX 
given therein." Then followed his entirely independent per- 
centage estimates for each cause of death, and, as stated in 
his letter, the average percentage in each case as presented in 
the tables in my book was subsequently added in red ink. A 
bird's-eye view of the surprising concurrence, in the main, 
of Dr. Wilbur's estimates, with the averages of the three medical 
directors' estimates, is presented in the following tabulation 
of the figures for all causes of death to which my investigation 
charged as many as 500 male deaths between ages 20 and 74, 
inclusive, in the registration area of the United States in 1908. 
Causes of Death 
Total male 
deaths 
between ages 
20 to 74, inclu- 
sive, from 
these causes, 
in registration 
area of the 
United States 
in 1908 
Male deaths from these causes, between 
ages 20 to 74, inclusive, apparently due 
in whole or part to alcohol 
According to 
averages of medical 
directors' estimates 
According to 
the estimates of 
Dr. Wilbur 
Per Cent 
Number 
Per Cent 
Number 
Bright's disease 
17,513 
30 
5,254 
40 
7,005 
Tuberculosis of lungs. 
33,9°° 
12 
4,068 
IO 
3,390 
Heart disease 
Pneumonia (lobar and 
22,887 
l6 
3,662 
20 
4,577 
unqualified) 
14,044 
22 
3,090 
IO 
1,404 
Cirrhosis of liver. ... 
4,123 
67 
2,762 
75 
3,092 
Apoplexy 
u,748 
22 
2,585 
30 
3,524 
Alcoholism 
2,025 
IOO 
2,025 
100 
2,025 
Suicide 
6,035 
23 
1,388 
20 
1,207 
Diseases of arteries... 
Other accidental in- 
2,803 
23 
645 
40 
1,121 
juries 
5,163 
IO 
5l6 
5 
258 
Paralysis 
2,329 
22 
512 
25 
582 
Totals 
122,570 
21.6 
26,507 
23 
28,185 
The difference of 1,678 deaths between the book totals and 
Dr. Wilbur's totals for the 11 more important causes of death 
above listed amounts to 6.33 per cent., and, making allowance 
for certain minor causes of death which Dr. Wilbur has in- 
cluded in his percentage estimates but which were not included 
in the medical directors' estimates, the total number of adult 
deaths in continental United States in 1908 in which alcohol 
presumably played some part according to Dr. Wilbur's esti- THE MORTALITY FROM ALCOHOL 
23 
mates, would have been about 71,600, as compared with the 
approximation of about 66,000, suggested by the report of 
my investigation. This difference, in other words, would have 
been one of less than 9 per cent, of the total tentatively fixed 
on the basis of the three medical directors' estimates, or con- 
siderably less than 1 per cent, of the presumptive total mortality 
at age 20 and upward in the United States in 1908. It would 
seem that Dr. Wilbur's statement in his letter to the effect 
that "on the whole, I think there is a very remarkable cor- 
respondence," was an entirely justifiable summary of the com- 
parison of the two sets of estimates so entirely independent 
and made on such radically different bases. Of course, Dr. 
Wilbur's estimates were in no sense official, but were made 
in his purely personal capacity, but coming from such an ex- 
pert source they seem to me to shed a flood of light on the 
probable approximate accuracy of the results of my investiga- 
tion, and I have been more than glad to present them in con- 
nection with this paper, with Dr. Wilbur's express permission. 
As stated in the introduction to my book, "At best, however, 
the results of any such inquiry can be but an approximation, 
and, as a matter of course, any one of the estimates or numerical 
conclusions presented in the tables accompanying this paper 
is fairly open to competent analysis and discussion. If it can 
be proven incorrect, the sooner it is disproved the better." 
[7787Q]