32
B 3 65586 C
DEPARTMENT OF APPLIED STATISTICS
UNIVERSITY COLLEGE, LONDON.
Questions of the Day and of the Fray
No. VI.
EUGENICS AND PUBLIC HEALTH
[A LECTURE DELIVERED AT THE YORK CONGRESS OF THE
ROYAL SANITARY INSTITUTE, JULY 30TH, 1912]
BY
KARL PEARSON, F.R.S.
@ALTON PROFESSOR OF EUGENIOS
WITH SIX DIAGRAMS IN THE TEXT
CAMBRIDGE UNIVERSITY PRESS
0. F. CLAY, MANAGER
LONDON: FETTER LANE, E.C. 4 EDINBURGH: 100, PRINCES STREET
H. E. LEWIS, 136, GOWER STREET, LONDON, W.0.1
BOMBAY, CALCUTTA AND MADRAN:
WILLIAM WESLEY AND SON, 28, ESSEX STREET,
MACMILLAN AND CO., LIMITED
LONDON, W.c. 2
TORONTO: J. M. DENT AND SONS, LIMITED
CHTOAGO: UNIVERSITY OF CHICAGO PRESS
TOKYO: THE MARUZEN-KABUSHIKI-KAISHA
Price
1s. 6d.
Net
C.U.P.
Published by the Cambridge University Press, Fetter Lane, E.C. 4
UV
Ir
EUGENICS LABORATORY LECTURE SERIES.
1. The Scope and Importance to the State of the Science of National
Eugenics. By KARL PEARSON, F.R.S. Issued. Third Edition. Price
18. net.
II. The Groundwork of Eugenics. By KARL PEARSON, F.R.S. Issued.
Price 1s. net.
111. The Relative Strength of Nurture and Nature. Ly ETHEL M. EL-
DERTON. Issued. Price ls. net.
IV. On the Marriage of First Cousins. By ETHEL M. ELDERTON. Issued.
Price ls. net.
V. The Problem of Practical Eugenics. By KARL PEARSON, F.R.S. Issued.
Second Edition. Price 1s. net.
VI. Nature and Nurture, the Problem of the Future. By KARL PEARSON,
F.R.S. Issued. Price ls. net.
VII. The Academic Aspect of the Science of National Eugenics. By KARL
PEARSON, F.R.S. Issued. Price 18. net.
VIII. Tuberculosis, Heredity, and Environment. By Karl Prlar.SON, F.R.S.
Issued. Price 1s. net.
IX. Darwinism, Medical Progress, and Eugenics : The Cavendish Lecture,
1912. By KARL PEARSON, F.R.S. Issueil. Price 1s, net.
QUESTIONS OF THE DAY AND OF THE FRAY.
1. The Influence of Parental Alcoholism on the Physique and Ability of the
Offspring: a Reply to the Cambridge Economists. By Karl PEARSON,
F.R.S. Issued. Price ls. net.
II. Mental Defect, Mal-Nutrition, and the Teacher's appreciation of Intelli-
gence: a Reply to Criticisms of the Memoir on “ The Influence of
Defective Physique and Unfavourable Home Environment on the Intel-
ligence of School Children." By David HERON, D.Sc. Issued. Price
ls. net.
III. An Attempt to correct some of the Misstatements made by Sir VICTOR
HORSLEY, F.R.S., F.R.C.S., and Mary D. Sturge, M.D., in their
Criticisms of the Galton Laboratory Memoir : “ A First Study of the
Influence of Parental Alcoholism," etc. By KARL PEARSON, F.R.S.
Issued. Price 1s. net.
IV. The Fight against Tuberculosis and the Death-rate from Phthisis. By
KARL PEARSON, F.R.S. Issued. Price Is. net.
V. Social Problems : Their Treatment, Past, Present, and Future. By KARL
PEARSON, F.R.S. Issued. Price ls, net.
VI. Eugenics and Public Health : Lecture to the Congress at York. By KARL
PEARSON, F.R.S. Issued. Price ls. net.
Issued by the Cambridge University Press.
Biometrika : a Journal for the Statistical Study of Biological Problems. Founded
by W. F. R. WELDON, FRANCIS GALTON, and KARL PEARSON, Edited
by KARL PEARSON. Volumes I., II., III., IV., V., VI., VII., and
VIII, complete. Subscription price, 308. net per volume.
65
:
:
EUGENICS AND PUBLIC HEALTH
AN ADDRESS
TO PUBLIC HEALTH OFFICERS
[BEING A LECTURE DELIVERED AT THE YORK CONGRESS OF THE
ROYAL SANITARY INSTITUTE, JULY 30, 1912]
BY
KARL PEARSON,
F.R.S.
GALTON PROFESSOR OF EUGENICS
WITH SIX DIAGRAMS IN THE TEXT
LONDON
Published by the Cambridge University Press, Fetter Lane, E.C. 4
methicow
82.05
Psychology
3-28-1921
gen
(not to be neuse )
UNIVERSITY OF LONDON
THE FRANCIS GALTON EUGENICS LABORATORY,
UNIVERSITY COLLEGE, GOWER STREET, W.C.
This Laboratory was founded by Sir FRANCIS GALTOx, and is under
the direction of Professor KARL PEARSON, F.R.S.
Assistants: DAVID TIERON, M.A., D.Sc., ETHEL M. ELDERTOX,
AMY BARRINGTON; Hon. Sec.: H. GERTRUDE JONES.
National Eugenics is the study of agencies under social control that may
improve or impair the racial qualities of future generations, either physically
or mentally.
It was the intention of the Founder that the Laboratory should serve
(i) as a storehouse of statistical material bearing on the mental and
physical conditions in man, and the relations of these conditions to
inheritance and environment; (ii) as a centre for the publication or other
form of distribution of information concerning the National Eugenics;
(iii) as a school for training and assisting research-workers in the special
problems of Eugenics.
Short courses are provided for those who are engaged in social,
medical, or anthropometric work.
THE BIOMETRIC LABORATORY
This Laboratory is intended to forward the statistical study of
Biological Problems.
Assistants: Julia BELL, M.A., HERBERT E. SOPER, M.A., EVELINE
Y. THOMSON. Benington Student in Craniometry: KATHLEEN T.
RYLEY.
Until the phenomena of any branch of knowledge have been subjected
to measurement and number, it cannot assume the status and dignity of a
science.-FRANCIS GALTON.
The Laboratory is assisted by a grant from the Worshipful Company
of Drapers. It provides a complete training in statistical method, and
assists research-workers engaged on biometric problems.
Delivered at the Congress of The Royal Sanitary Institute at York,
July 29th to August 3rd, 1912.]
CONGRESS AT YORK.
LECTURE TO THE CONGRESS
By KARL PEARSON, F.R.S.,
Galton Professor of Eugenics, University of London.
EUGENICS AND PUBLIC HEALTH.
IT
a
(T is with very great hesitation that I stand on this platform to-night
before an audience, many of whom are experts in public health
matters, and several of whom have for years been first-hand workers
at social problems. It requires some courage, I assure you, to come
before you and say that having no training in public health, that having
no medical knowledge, and that having only second-hand acquaintance
with social problems, I still feel that I have a mission, a gospel to preach to
those who have. There are those (very likely represented in my audience
to-night) who feel, I know, strongly that a mere mathematician, a
figure-twisting statistician, has no business whatever in this field.
Well, I will frankly proclaim that I am no believer whatever in the
water-tight compartment theory of human knowledge. That theory is
opposed to all the history of scientific progress. Science has progressed
just as far as new methods and new ideas came into any branch of it, and
in nine cases out of ten those methods and ideas have come in from
outside, from what appeared to be at the time quite foreign divisions of
knowledge. Shall I give you illustrations ? What had the astronomers
made of their science before the advent of mathematics ? Mere astrology.
Then came Kepler, Newton and Laplace, and astronomy became the
most certain of all sciences. What part did electricity play in life,
municipally, socially or medically sixty years ago ? Then came the
Reprinted from tho Journal of The Royal Sanitary Institute, Vol. XXXIII, No. 8,
September, 1912.
4
Lecture to Congress.—Eugenics and Public Health.
mathematical physicists, Clerk-Maxwell, Lord Kelvin, Hertz, Röntgen,
and lo! our very habits of life and treatment were revolutionised.
And to take an example nearer home. What was the science of
medicine in 1840? Surely you will agree with me when I say, it was
an empirical science based upon clinical observation, and often largely
upon unjustified if not unjustifiable opinion. What has differentiated
modern medicine from early Victorian medicine? Is it wholly knowledge
gained at the bedside? Or is the advance due to the pressure of new
methods and new ideas breaking in from biology and from physies in their
broad senses? Is it not the laboratory work of the bacteriologist, of
the physiologist, of the parasitologist, and even of the physicist which
has revolutionised modern medicine? May it not well be that at least
one great branch of medicine, the public health service, has yet room for
further new methods and new ideas, and that some of these may come
from the mathematical statistician? Is there not room, nay need, for
the medical mathematician in the field of public health?
The name at least iatpo-pūOnpătukós, medical mathematician, is Greek
or rather Egyptian in origin, if to our generation belong the first attempts
to make mathematics a really serviceable handmaid of medicine. In
saying this I am perhaps overlooking an older link between medicine and
mathematics. I refer to the relation between medicine and the actuarial
profession: the actuary is only a specialised mathematician; but here our
positions are rather reversed, for the science of medicine is the handmaid
of insurance; it is the mathematician in that case who unties the purse
strings, and I am speaking rather of the day when the medical public
health service will directly seek the aid of the specialised mathematician.
That is the point I want to illustrate to you to-night, that is the
mission which leads me to this platform, and I think that must be the real
ground upon which the Council of The Royal Sanitary Institute requestel
me to lecture to this Congress. I have come that I may endeavour to show
you that in the field of public health, in the treatment of social welfare, an
entente cordiale of mathematics and medicine is of no less importance than
another partnership in the diplomatic sphere.
I must ask you to be patient with me while I endeavour to explain my
views, which, for the moment, may seem to touch only one side of the title
of my lecture. In the first place many of you are aware of an enormous
literature (if I may venture to call it such) which has sprung up during
the last ten years. I refer to the annual reports of the medical officers of
health and of the school medical officers. You are also aware of the still
more gigantic manuscript material, the hundreds of thousands of schedules
KARL PEARSON.
5
which lie behind this literature. I am convinced that some of you are
asking yourself the cui bono of this great national record, the very storage
of which is becoming a “burning" question in the smaller districts.
Well, those members of my audience who are medical officers of health
have certainly read their own reports, but how far have they had the
leisure to study those of their colleagues? How far again have they been
able to compare their schedules and records with those of other districts,
scattered by the hundred up and down this country? By comparison
of schedules, I do not mean the comparison of printed forms. It is quite
easy to draft a most all-embracing questionnaire ; I refer to the accuracy
and comprehensiveness with which the records themselves are kept, and
how far the questions asked are answerable, having regard to the education
of the questioned, the sympathy and tact of the questioner, and the practi-
cal limits to official cross-examination and to human life. I am, Ladies and
Gentlemen, a great reader of the literature I have referred to; I am by
no means certain that I do not know more of the contents of these medical
officers' reports than is known to any single medical officer. I welcome,
study and file every such report that the courtesy of the public health
officers (now in increasing numbers) sends to my laboratory. I have
spoken to you of the entente cordiale of medicine and mathematics as a
thing of the future, yet I venture to believe it is almost a fact of the
present. From the Midlands to Scotland, from the great manufacturing
towns of Yorkshire and Lancashire, thousands of the schedules to which
I have referred have reached my laboratory through the kindness of the
officers of the public health service. And in case any of those who have
loaned them are among my audience) I would appeal for patience and
time. I see work for a staff of fifty trained computers, where I have to
work with one of five. I would express my deep gratitude to these
medical officers and say that we are toiling to the conclusion of a mighty
task. But what we have learned from close examination of many schedules
from many counties is this:
First.-An elaborate questionnaire is of no service at all, if either
50 to 70 per cent. of the entries are not filled in, or filled in parrot-
fashion. Let me illustrate exactly what I mean. In some recent
inquiries as to infantile mortality there was a column for recording when
the father had “bad health.” In order that this entry should be really
useful some broad definitions ought to be given of health depending upon
weeks of sickness, absence from work, hospital attendance, and so forth.
For my own laboratory work we have issued the following scheme, but it
might be easily bettered or modified to suit special inquiries :-
6
Lecture to Congress.—Eugenics and Public Health.
a
HEALTH (ADULT, MALE OR FEMALE).
Health should be judged with reference to two standards based on the
answers to the questions:-
(a) How often has it been necessary for the subject to consult a doctor,
or seek advice or treatment at a hospital?
(6) How often has he been unable to work through illness?
SCALE.
V.R. Very robust.—He has never had to see a doctor, has never been in
hospital, nor been off work through illness.
R. Robust.—He has only seen a doctor about minor ailments, and has
only been off work for colds, etc.
N.H. Normally healthy.--He has not had more than one serious illness,
involving, say, a fortnight's absence from work during the last
ten years.
&
R.D. Rather delicate. He has had more than one serious illness, but
not more than one involving more than four weeks' absence
from work during the last ten years.
D. Delicate.-He is off work through illness at least four weeks in all
every year.
V.D. Very Delicate. He is in a chronic state of ill-health.
NOTE.-Serious accidents should be noted, and it might be asked how often he has received
money from a sick club, if a member.
What can possibly be learnt when the column of the health of the
father is left blank in 90 per cent. of cases, and there is no knowledge
whether the recorder was unable to ascertain it, or whether in all these
instances the father was in uniform good health ?
I think again of nearly 8,000 records from one town I have seen of
infantile mortality, where there is a column for the mother's health, and
in only 3 per cent. or 4 per cent. of these cases is she said to have had
other than good health! I wish indeed it were a fact, but I know from
similar towns where the health of the mother has been really carefully
investigated, that there is a fairly intimate relation between this health
and the survival of the infant. Indeed, if you look with the eye of a
statistician on these schedules, and apply tests well known in the profession,
you can put your finger at once on the men whose records are to be
trusted, and whose staff are enthusiastic in their work. Nay, more, one
can very often say within a given administrative district, which assistant-
officers, nurses, and visitors are capable and conscientious, and which are
filling in their schedules mechanically and superficially. I had quite
recently to return a long series of schedules dealing with all the physically
KARL PEARSON.
7
and mentally defective children of a large northern town, which had been
most kindly placed at my disposal, because, after a thorough examination
of them, it was clear from internal evidence that they had been filled in
without due inquiry and safe controls.
Secondly.--You may have the most conscientious district visitors, the
most energetic staff of school medical officers, but if nothing has been
done to standardise their judgments, the result will be hopeless confusion.
I know from the schedules how the most conscientious assistant school
medical officers, how the most enthusiastic and careful district visitors,
differ in what we term their personal equation. One visitor finds
8 per cent. of a certain characteristic, another working in an exactly
similar or the same district finds 25 per cent. Even more anomalous
results may be found in the school medical inspection returns. Yet in
nine cases out of ten records are pooled without regard to this vital
question of personal equation. Masses of material are indeed useless
because no real attention has been paid to this matter. There are
methods of correcting for personal equation, there is nothing new in the
problem, astronomers have felt the difficulty and met it for more than a
century. But it is the application of an old theory to a new field, and if
10 per cent. of public health officers have realised the difficulty, scarcely
1 per cent. have attempted to meet it.
With descriptive and qualitative characters, such as must be largely
used in work of this kind, absolute personal equation is of no importance,
relative personal equation is everything; and far more comprehensive
methods are needed for standardising the public liealth staff of each
district, and eventually for comparing the staff of one district with a
second. The standard man of the district ought to be compared with
each new comer, by letting the pair record independently the same 50 or
100 cases, and then from the results determining the exact nature of their
personal equation either with a view to correcting it on actual observations or
by an attempt to modify the new comer's methods of judgment. Probably,
as research in other fields has shown, the former is the safer plan, but
each local officer will have to consider as a matter of primary importance
this standardisation of his staff, and ultimately, if we are to draw con-
clusions as a whole for the relative conditions of different districts, there
will have to be a comparison of the standard men of these districts.
What value have the returns for defective teeth or eyesight for different dis-
tricts for comparative study of occupations and environments, when we know
that the mere introduction of a trained ophthalmologist or a dental surgeon
has sufficed in a given district or school to increase the reported defective-
8
Lecture to Congress.—Eugenics and Public Health.
ness by 100
per cent. to 200 per cent. of its value? What reason have we
for supposing that the prevalence among school children of mental defect
or of tuberculosis can really be estimated for the different districts by the
percentages returned to the Board of Education? Until some effective
system of interdistrict standardisation is called into existence, it is more
reasonable to believe that personal equation, which we know to vary
enormously, rather than local conditions may be the source of these
percentage differences. Unless the public health service proceeds in this
matter with great caution and in the true scientific spirit, we may have
the normal child of one parish segregated as feeble-minded or defective
when it crosses the borders into a second parish. We are almost certain
to have widely different standards for notification of the tuberculous in
different districts, and I anticipate most wonderful publications on the
influence of climate and environment on the prevalency of tuberculosis as
a result !
To those medical officers who have to direct a large staff of assistant
officers and district visitors, I would therefore emphasise, first, the idleness
of the records, if they are filled in without keen insight and enthusiasm.
All categories must be adequately defined and given, if feasible, quantitative
value, and, further, the staff must appreciate the importance of caution in
treating individual cases. This is our Scylla. But close alongside stands
our Charybdis, the wreck-bestrewn rock of unregarded personal equation.
I feel convinced that the public health service can only steer this course
under mathematical pilotage, and even then, were a veritable Ulysses at
the helm, there would still be a toll of failure. The tests for slovenly
record, the tests for differentiated records, and the means for correcting
the latter are essentially statistical and mathematical problems, and the
most valuable labour will be wasted until this is recognised.
But Scylla and Charybdis are not the only rocks in this newly-
discovered and largely uncharted sea. The value to be given to weight of
numbers is at present wholly unappreciated by many of the captains who
are seeking to traverse it. I should like to illustrate this point in a
variety of cases.
For example, quite recently a distinguished investigator came to me
troubled in mind over contradictory results. The English workers had
tested twenty-eight cases of tuberculosis and found bacilli of the bovine
type in two. But the Germans had then tested one hundred cases and
found bacilli of bovine type in none. How many cases must be
investigated in order to settle the true percentage? That is a very nice
mathematical question, even if we can assume that the above-referred-to
KARL PEARSON.
9
Scylla and Charybdis were already cleared, which I, individually, am by
no means certain was the fact.
Again, to come nearer home, a distinguished assistant medical officer,
Dr. Ewart, of Middlesbrough, has apparently felt himself able on
seventy-nine cases of tuberculosis to determine the influence of mother's
age at birth of child on the liability of the child to phthisis. (Public
Health, May, 1912, p. 312. “Problems of Race.”)
“Problems of Race.") He tells us, on the
basis of his seventy-nine cases, that “this disease falls at the beginning
and with increasing incidence on those born later in life.” He suggests,
if I follow him, that the fall in the phthisis death-rate is due to the
limitation of families, so that the children born later in the life of
the mother are no longer born in the same numbers. He, however, gives
no comparative evidence of the distribution of children at age of birth of
mother now and thirty and more years ago. In the bulk of the population
there is little difference between the age of marriage now and fifty
years ago :
......
......
TABLE 1.-Average Age of Wives at Marriage in England.
1851
25.88
1890
26.03
1860
25.83
1900
26.24
1870
25.71
1907
26.45
1880
25.72
1909
26.65
The total change is under a year, and would not carry the mean wife
outside one of Dr. Ewart's groups into a second. Further, until the
present census returns are published, we have no real evidence that it is
the late children who are being dispensed with, and not an increasing
interval between births being allowed as the basis of limitation. Here,
however, are the seventy-nine cases on which Dr. Ewart has based his
theory. You will notice how the tuberculous are in excess among
the
earlier and later born :
TABLE 2.
Age of Mother.
Phthisis.
Per Mille Phthisical.
Per Mille Normal.
67
20 and under
21-25
26-30
31-35
36-40
41 and over
15
16
17
8bBT4919
101
190
203
215
178
113
256
289
203
147
38
79
1,000
1,000
10
Lecture to Congress.—Eugenics and Public Health.
In the comparison of the 3rd and 4th columns Dr. Ewart finds the
great objection to be that the tuberculous data were gathered from
South-East Lancashire and the normal population from the North
Riding of Yorkshire. I presume he means Middlesbrough, which is in
every respect differentiated from the dale population of the North Riding.
There is not the least doubt that the objection is a valid one. I have
data for the age of mother at birth of offspring from many districts,
agricultural and urban ; they vary widely, you may find any number from
3
per cent. to 9 per cent. of offspring born of mothers over 40 years
of
age, and any number from 2 per cent. to 7 per cent. of offspring born of
7
mothers under 21 years of age according to the district. Nevertheless,
this is not the great objection to Dr. Ewart's result; the real objection is
the paucity of cases he has ventured to use in drawing a sweeping
conclusion. How many cases ought we to use? At least a thousand and
better four thousand, and we ought to compare them against normal
material drawn from the same districts. I cannot make any such valid
investigation, for at present I have only 400 cases of tuberculosis with age
of mother at birth. Of these 100 are drawn from pedigree data in my
laboratory, 140 from a sanatorium in the north of England, 160 from the
Brompton Hospital in London by the kindness of Drs. Batty-Shaw and
Bernstein. Against these I set a normal population based upon some-
thing like 50,000 cases drawn from London, Northumberland, Lanca-
slire and Yorkshire in fairly appropriate proportions. The results are
given below:
TABLE 3.
a
Mothers' Age.
Phthisical Population
General Population.
20 and under
21-25
26-30
31-35
36-40
41 and over ...
I
56
224
308
211
158)
435
51
219
303
224
149
201
...
51 } 203
1,000
1,000
You will see that the younger mothers have a slight excess of tuber-
culous children, as Dr. Ewart found, but the older mothers a slight defici-
ency, contrary to what he found. The former result may be a secondary
effect of a rule I have elsewhere stated, that the elder or firstborn children
are more liable to tuberculosis, and such children come on the average from
younger mothers. But are such differences in this case really significant ?
KARL PEARSON.
11
You all know perfectly well that if you tossed a shilling a hundred times
and counted the heads and tails, you would be very unlikely to get 50 heads
and 50 tails; but if you got 35 heads and 65 tails, what would be the
significance of that? What is the probability that it is a mere chance
result, or is the coin actually loaded? Well, a mathematician would tell
you
that
you might expect that result once in about 1,000 trials, and
accordingly it is the more reasonable supposition that, having got it on a
single trial, the coin was biassed. Is that not exactly the same type of
problem as we have in our tuberculosis data ? We have taken 400 cases
out of a general population, and we must not expect to get them in exactly
the same proportions of mothers' ages at birth as they occur in the general
'
population. Now this can be illustrated quite simply. Put 1,000
counters into a bag, 51 yellow to represent the mothers of 20 and under,
219 blue to represent the mothers 21-25, 303 red to represent the
mothers 26–30, and so on; then proceed to draw a sample population
from this bag, one by one, returning each counter as drawn. Here are
some results :-
TABLE 4.-Per Mille.
Samples of 79.
Samples of 400.
Age of Mother.
Actual Pop.
1st Exp.
2nd Exp.
1st Exp.
2nd Exp.
60
20 and under
21—25
26-30
31-35
36-40
41 and over
::::::
51
219
303
224
149
54
26
329
253
177
139
76
51
215
278
228
190
38
235
283
230
115
77
87
255
253
215
150
40
...
From the first sample of 79 you would suppose a defect of young
mothers and an excess of old mothers; from the second a defect of old
mothers. Even from the 400 samples you might draw all sorts of startling
conclusions! The public health officer should toss coins, calculate the
probable number of trumps in his partner's hand at whist, and test generally
how far random samples deviate from the expected, before he ventures to
assert that his data indicate great scientific laws !
It is, I fear, only the trained statistician, however, who will be able to
tell us whether the deviations are the result of random sampling or repre-
sent true bias. Into that theory I cannot enter on this occasion, it is part
of the training which the lat po-uă nuărikós of the future will seek and obtain.
But one method which shows in rough and ready way the result is to plot
a certain quantity termed the probable error on either side of the expected,
and question to what extent the compared series falls outside this belt.
12
Lecture to Congress.--Eugenics and Public Health.
In the accompanying diagram this is done. Had the observed phthisical
series gone three times outside this belt we could not have asserted the
result to be significant. It goes once slightly outside, and that is no
AGE OF MOTHER AT BIRTH OF TUBERCULOUS AND NORMAL OPPSPRING.
Phthisical Population
General Population
300
2001
PREQUENC IES
MILLE
PER
100
20 & Under
21-25
26-30
31-35
36-40
41 & Over
AGE OP MOTHER
Diagram I.-To show there is no significant influence of age of mother at birth of child
on the chance of child becoming tuberculous.
significant deviation. Hence on the basis of five times Dr. Ewart's numbers
I am not able to assert that the age at birth of the mother has any influence
whatever on the chance of the child being tuberculous.
Now please remember that I do not state that no such relation exists.
I merely say that we can at present draw no inference at all, and that
KARL PEARSON.
13
a
there is grave danger when medical officers of health proceed on inadequate
data to draw very sweeping conclusions by wholly unscientific methods.
Until the complete ιατρο-μαθηματικός is trained to the public health service
most astonishing results will be reached unless medicine and mathematics
go hand in hand.
I must illustrate this neglect of numbers once more. I cite from the
statement of a medical officer of health published by the London Eugenics
Education Society, a society which professes to help forward the cause of
Eugenics.* He writes:
“ The influence, moreover, of one birth on the sex of the next, and the
relationship of the sex on the interval of time to the next impregnation,
show that the substance derived from the fætus persist after birth and
influence the succeeding one."
Now here is a direct statementſ that if parents have a boy at one birth,
it will influence the sex of the next child born to them. Now let us look
exactly at the data upon which this result is based. The author took 382
births and found :
TABLE 5.
Boy following
Boy.
Girl following
Boy.
Boy followlog
Girl.
Girl following
Girl.
90
102
90
2:9 years
2-6 years
100
3:0 years
2:8 years
Cases
Average Interval...
and again:
Sequence in Sex.
Change of Sex
180 times.
202 times.
...
...
Now no attempt is made to ascertain whether these differences are
statistically significant, but a sweeping scientific theory is based upon
them, that substances derived from the fætus persist after birth and
influence the sex of the next child!
What is the likelihood that in taking samples of 100 cases out of
material of which the average is 2.8, we shall get such deviations by pure
chance as 2.6 and 3.0? The author has given us no means of answering
it, but it involves the variability of the interval between two births of
which the average is 2-8 years, being something between a year and a year
and a half. This is so plausible that I have no hesitation in saying that
the observed differences carry no weight at all; they are due simply to
paucity of material. Look again at the second result. What should we
expect in 382 births ? Why, that there would be change of sex in 191
It occurred in 202 cases, or 11 times in excess of what would
cases.
* R. J. Ewart. “ The Aristocracy of Infancy and the Conditions of its Birth.” The
Eugenics Review. July 1911.
† This paper abounds in similar inadequately demonstrated theses.
14
Lecture to Congress.-Eugenics and Public Health.
happen on a very long experience. Are those 11 cases significant or not?
Our medical officer of health assumes them to be so without any inquiry !
What is the problem? Why, the chance of a sequence is nearly one-half.
Therefore toss a coin 382 times and ask yourself whether 202 heads
instead of 191 would mean that the coin was biassed.
Why, the
mathematician will tell you that once in every three to four trials of such
an experiment the number of heads would differ from the theoretical value
DISTRIBUTION
07
ERALDS
AND
TAILS
IN
100
TRIALS OP
TOSSING
COIN
382 nos.
OKULIT LLO.
PERCENT
1
365
158
164
167
170
173
176
179
182
AB
200
203
206
209
212
215
218
221
304
185
NINO ER
101
OP
194 107
READS
Diagram II.-The black area relative to white shows the odds against a deviation of
Il heads or tails from the theoretical number 191 in throwing 382 coins.
191 by 11 or more units. In order to illustrate this point eight experi-
ments of tossing a coin 382 times were inade, and the following results
were recorded :
TABLE 6.
Heads.
Tails.
1st experiment....
2nd
3rd
4th
5th
6th
7th
8th
99
188
192
187
204
184
186
205
192
194
190
195
178
198
196
177
190
99
99
Experiment with sex of babies.......
Theory
202
191
180
191
KARL PEARSON.
15
a
No legitimate conclusion whatever can be drawn from such data.
Whether the theory that a previous birth influences a later birth be true
or not, it cannot be answered by such inadequate data and inadequate
statistical methods. They can only harm the science that Eugenists and
the Public Health Service alike wish to forward.
I could multiply such cases a hundredfold, they occur over and over
again when the medical officer of health endeavours to draw conclusions
from his data, without any training in the modern theory of statistics. I
do not cite such errors in a spirit of criticism or of anything but sadness
over such blunders. I feel that the science of eugenics—the study of
what improves or impairs racial efficiency-must go hand in hand
with the science of public health. Whatever makes for adequate
treatment of public health problems, makes for sure foundations for the
science of eugenics; whatever dedactions are drawn by superficial and
inadequate treatment from public health experience will damage the
science of eugenics as much as they damage the science of public health.
It is almost impossible to regret too keenly the hasty generalisations which
are being formed both in your science and in my own. No science can be
built up by popular discussions before its fundamental principles are firmly
based. To be a true science it must have developed appropriate methods,
have found its raw material and established a school of trained workers,
who are devoting their lives to its pursuit. Can that be said of eugenics
at the present time? I sadly fear it cannot.
If we sum up the situation, we cannot deny that the material is ample;
to a very large extent, although, of course, not wholly, the public health
service is producing it. That material is essentially statistical, and as
such must be dealt with by the methods of modern statistics. But these
methods cannot be learnt and applied, any more than the differential
calculus can be, without adequate training. I look forward to the time
when every great municipality in this country will have its trained statis-
tician, and, what is more important still, when there will be a Government
Statistical Bureau to supervise the reports of the chief state Statistical
Departments such as those of the Local Government Board, the Education
Department, the Registrar General's Department, and the Home Office.
Almost every continental government has now such a bureau; only Great
Britain, where practically modern statistics have been created, lacks such
an essential executive department. How can we expect municipal author-
ities to appreciate adequate treatment of their local data, when our chief
government officials are among the gravest offenders in this respect?
Please do not think I am exaggerating the state of affairs. I could
16
Lecture to Congress.-- Eugenics and Public Health.
a
*
illustrate the evil at every turn.
It will suffice to do so in one or two
cases. There are few questions of more importance from the public
health standpoint than the origin of cancer. You know that it is a widely
spread impression that cancer is associated with certain houses ; the idea
is that certain houses (the so-called “cancer houses ") receive more than
their due number of cancer cases. The conception is an important one,
and, if it were ever established, it would not only throw light on cancer,
but would open up a new field for the activities of the public health
officer. Well, how is such a problem to be answered ?
Let us suppose a
number of small compartments placed on a board, the number of these
compartments being equal to the number of houses in a given district, and
the size of these compartment proportional to the average number of occu-
pants of the houses.
Now let a number of marbles be taken equal to the
total number of cancer cases which have occurred during a given number
of years in the district. Let these marbles be thrown at random into the
compartinents. In some compartments will be found one, in some two,
and in some three or more marbles. If cancer cases have no relation to
houses, and we put aside the question of occupational mortality,* then
the compartments with several marbles will represent the distribution of
multiple cancer houses. The problem in a limited and simplified form can
be solved mathematically. If the actual number of houses with multiple
cases of cancer far exceeds those determined by our chance distribution,
then we should begin to believe that so-called “cancer houses” were a
reality. There will, of course, always be houses with multiple cancer
cases in any district; the real question is: Do they exceed those which we
should expect would arise from a random distribution of our marbles ?
To answer this even approximately we must know at least the total
number of cancer cases and the total number of houses in which they
occurred, classified according to their average size of households. Now
only a few years ago a Government report on cancer was issued by the
Registrar General for Ireland, and the author deals among other things
with the problem of “cancer houses." He had before him of necessity
the total number of cancer cases of the City of Dublin for ten years; he
inust have known the total number of inhabited houses in the City and
presumably the population. Yet all he tells us is that in twelve housest
* The multiple houses being determined in a district, it is of importance to ascertain
whether the cases in the same house belonged to (1) the same family; (2) followed the same
occupation.
† On any reasonable supposition as to number of cases, the multiple houses in Dublin
seem at least 50 per cent. less than the chance number, and we should be compelled to sup-
pose, if the data were worth anything, that the appearance of one case of cancer conferred
immunity on a house!
>
KARL PEARSON.
17
two cases of cancer occurred in those years. There is no statement as to
whether, considering the number of houses, the number of inhabitants,
and the number of cases of cancer, such a result would be likely to arise
from pure chance; he does not tell us how many houses had three or
more cases, only that there were twelve houses with two cases. From
this he solemnly draws the conclusion that:
“ In some instances more than one case of cancer has occurred amongst
different families living in the same house, or amongst successive occupants
of the same house." (Report p. 39.)
By the General Registry for Ireland I am told that it is now impossible
to ascertain the number of cancer cases that occurred in the City of
Dublin for those years, yet that number must have been before the anthor
of the Report!
In a similar manner the author illustrates the bearing of alcoho'i-m
and syphilis on cancer; he tells us that there were so many cases in which
cancer was associated in the same person with one or other of these
diseases, but there is not a word as to the frequency with which these
diseases occur in the City of Dublin! Now, Ladies and Gentlemen, I feel
sure you would simply smile if I told you that I had known 18 cases of
cancer associated with dark hair, and 10 cases of cancer associated with
Roman noses, without informing you of the total percentages of dark-
haired individuals and of Roman-nosed persons exposed to risk! Yet that
sort of information is precisely what this Government Report on cancer
provides us with. You may consider that this is an exceptional case, but
I assure you
it is not. I could cite equally fatuous argumentations from
the Local Government Board Reports, the Reports of the Prison Commis-
sioners, or the Reports of the Royal Commissions and Departmental
Committees of both this country and of Scotland; but time does not
permit, and my sole object is to emphasize how very important in these
matters a thorough statistical training must be. Throughout the country
immense masses of data are being collected by the public health and
school medical services, the conclusions which may be drawn from them,
if sound, will be of the highest national importance; they touch the very
problems with which eugenics has to deal, the consideration of those
agencies under social control which may improve or impair the racial
qualities of future generations. But if Government officials themselves
treat these problems superficially, who shall set a standard to the public
health service? If I can but make this audience realise the urgent need
for a Government Statistical Bureau to deal with problems of public
health, I shall have more than fulfilled my purpose to-night of aiding
both science and
my own.
your
18
Lecture to Congress.—Eugenics and Public Health.
Thus far I have tried to illustrate the dangers which flow from neglect
of personal equation, from careless observation, or from paucity of data.
But, given careful observation and ample data, a new series of difficulties
arises of an even more subtle kind. I shall indicate their nature best by
an example taken from the recent excellent report of Dr. Chalmers of
Glasgow, on the Administrative Treatment of Pulmonary Phthisis for
1910, the first year of compulsory notification in that city. By the
courtesy of Dr. Chalmers, I am able to add the notifications for 1911,
making 4,413 cases in all, and to use the results of the 1911 Census. We
then have the following table :
TABLE 7.-Housing of the Consumptive.
Number of rooms.
1
2
3
4 and over.
138
48:3
21:1
16.8
Percentage (Census, 1911) of population
occupying houses of
Percentage of 4,413 notifications of phthisis
occurring in houses of (1910 and 1911)......
22:3
53-6
15:5
8:6
+ 13.8%
- 13:8%
Now surely there is not the least doubt on the basis of these figures
of an association between phthisis and the size of the tenement? Dr.
Chalmers wisely draws no inference (at any rate at this part of his
Report), but leaves the figures to speak for themselves. I feel quite sure
that those figures will be interpreted as indicating that if we could only
Letter the housing, we should at once lower the phthisis death-rate. Now,
this is precisely one of the “difficulties of the subtle kind” which I want
to illustrate. We have found association, we must not without full
inquiry assert that it marks causation.
If mere association were evidence of causation, there is ample proof
that back-to-back houses or one-apartment tenements produce high
infantile death-rates and high phthisis rates. The Galton Eugenics
Laboratory will show, in a forthcoming memoir, that in such houses more
alcohol is consumed, their tenants are more irregular and less cleanly in
their habits, their wages are lower and they pay less rent. I wish some
keen medical officer would also tabulate the cases of insanity, epilepsy
and imbecility in his district, with regard to the nature of the home in
which the patient had been living, and tell us the result. There is no
more important problem before both eugenist and medical officer than
а
KARL PEARSON.
19
whether man is the product of his environment, or the environment the
product of the man. Do the physically and mentally inferior get lower
wages and tend to drift towards the cheaper and dirtier type of house, or
does the house make the occupants physically and mentally inferior? If
neither statement is true, then what is the quantitative measure of their
relative shares in the joint result ?
Returning to our tuberculosis data, we have to ask three fundamental
questions :
(i.) We know that the age distribution is immensely important in the
case of tuberculosis. We ask are the age distributions of people living in
tenements of 1, 2, 3, etc., apartments the same ? Clearly, if persons of
the susceptible ages live more in 1 or 2 apartments, the relation between
smallness of house and prevalence of phthisis may be spurious. Now on the
basis of the 1911 Census for Glasgow, which gives the number of persons
of each age for each type of house, I have distributed 2,419 cases of
phthisis notified in 1910 with their ages among the people living in the
different types of houses in Glasgow.
I find as follows :
:
TABLE 8.- Distribution of Cases of Phthisis in Glasgou.
1
Number of rooms.
1
2
3
4 and over.
13.8%
0
48:3%
21.1%
16.8%
General population, 1911.
Expected phthisis distribution,
allowing for age
Actual phthisis distribution
18:5%
0
13.4%
22:3 %
46:3%
53-6%
21.8%
15.5%
8-6%
Clearly phthisis does not predominate in small houses because they are
more occupied by persons of susceptible ages. Allowance for age makes
little difference, although a priori this could not be assumed.
(ii.) Our second problem is this: Do the persons who by occupation
have greater liability to phthisis occupy in greater numbers smaller
houses?
Now, it is well known that certain trades have high mortality rates.
When I first came to investigate the matter I anticipated that trades of
all sorts with a high death-rate, the true dangerous trades, would show a
high rate of wages, but with few exceptions the exactly opposite is the
case ; there is quite a high correlation between the death-rate of a trade
* There is a possibility of fallacy even here, if the institutional cases are really the
old cases.
20
Lecture to Congress.- Eugenics and Public Health.
and the smallness of the wages paid in that trade. In other words, trades
with high mortality are trades of low wages, and their followers will
obviously be found in one and two-roomed houses. Here, again, we are
on the vicious circle with all its subtle difficulties. Do physically inferior
men follow “high mortality trades” because they can demand no higher
wage, and thus make the trade a high mortality trade? Or, does the
lower wage reduce the wage-receiver to a physically inferior state and thus
make him susceptible to disease? The main point to be borne in mind is
this : That in classes and families where there is no immediate question of
hunger, general health varies enormously and is hereditary. There is
no reason to suppose that general health and physique do not vary as
largely in the working classes, or that, in the main, commercial economics
do not regulate wages by fitness. In other words, we should expect to
find the physically weaker members of our society receiving lower wages
and occupying the poorer houses. That such persons are more susceptible
to phthisis goes almost without saying.
Dr. Leslie Mackenzie and Captain Foster, in their valuable Report on
the Physical Condition of Children attending the Public Schools of Glasgow,
1907, have divided the schools into four groups : the first, A, comprises
the poorer districts, the second, B, the poor districts, the third, C, those of
a better class, and the fourth, D, those of a still higher class. They state
that “these groups indicate a real social gradation,” and if we represent
their homes in the same manner we have just done those of the
phthisical we have the following result:-
TABLE 9.—Homes of School Children of Glasgow.
as
Number of roots.
1
2
3
4 and over.
Grade D
Grade C
Grade B
Grade A
0-6
30
74
14:8
22.0
48.0
63.9
72:6
398
36.8
24:1
11:3
3707
12:2
4:6
13
Phthisis........
22:3
53.6
15:5
86
Now Grade A contains thirty-four per cent. of all the school children
of Glasgow, and the phthisical have larger percentages of better homes
than these Grade A children. It seems to me that the social grade of
the phthisical patients deserves close consideration, and especially we
ought to know their employment and wage before and after onset. How
KARL PEARSON.
21
far did they originally spring from a social grade occupying the poorer
tenements? How far has that tenancy followed the onset of the disease ?
As suggestive in this matter we may note that Dr. Chalmers tells us that
sixty-five per cent. of the total notifications were ultimately obtained
through charitable or rate-aided sources, forty-four per cent. being from
poor law sources alone. Further, in the Glasgow special inquiry forty-
six per cent. of the adult phthisical males were unable to work. *
The instant we realise these points we see how urgent is the need for
further investigation. It may quite well be that the predominance of
one-roomed and two-roomed houses among the phthisical is the origin
of the disease; but, on the other hand, the predominance of such houses
may merely arise from the fact that sixty-five per cent. of the tuberculous
are in receipt of charitable relief, or that forty-six per cent. of the adult
males can do no work. These points will never be cleared up without far
fuller knowledge than we have got yet of the past history of phthisical
families. What was their original economic position and their trade?
Again this question of past history brings me to my third and last
point:
(iii.) What is the past physical history of the stocks of these phthisical
individuals ? Are they economically weak because they are phthisical, or
because they were originally of physically inferior stock ? In the
pedigrees of general degeneracy which we have collected in the Eugenics
Laboratory (i.e., pedigrees of epilepsy, insanity, mental defect, hare-lip,
cataract and albinism, etc.) there is a marked excess of phthisis, and such
families are as a rule economically as well as physically or mentally unfit.
To what extent do they contribute to the phthisical contingent in the
poorer homes ?
When you think over these points, I feel sure you must
agree with me that the matter is much more subtle than we can plomb by
merely comparing percentages of normal and phthisical homes of different
sizes. We have to ask ourselves what is the relation between phthisis
incidence and one or two-roomed homes for (i.) constant age distribution ;
(ii.) constant economic conditions antecedent to the onset; and (ii.)
constant physique in the stock. We have to reach a“ partial correlation.”
I can conceive no more hopeful or profitable inquiry than the attempt
to disentangle the multiple causes (trades or occupations, housing, heredity)
which bear on the incidence of phthisis in one of our large towns. It
would be relatively easy in relation to the system of notification-visits, but
the schedules I have so far seen do not appear to cover ailequately such a
fundamental inquiry.
* Report, pp. 7 and 26.
22
Lecture to Congress.-Eugenics and Public Health.
At the risk of wearying you I must illustrate further these subtle
difficulties of statistical interpretation. I take the following figures from
a report published by the Home Office, they deal with female inebriates :
TABLE 10.
Distribution of Fitness and Duration of Alcoholism in 885 Inebriates.
Years addicted to drunkenness.
Conditions for Hard Work
5 and under.
6-10
10/11
11-20
21-30
31-40
41-50
Totals.
214
7
Fit
Uofit
96
30
216
101
38
24
571
294
130
1
Totals
126
320
344
62
12
1
855
Percentages of Unfit
23.8
325
37.8
38.7
417
100
34:0
...
a
Now, can you conceive a better demonstration than this table that the
longer drunkenness lasts the more physically unfit becomes the drinker?
These numbers are so convincing that they will become a text wherever
the evil of alcohol is preached !
MANCHESTER UNITY FRIENDLY SOCIETY
AVERAGE WEEKS OF SICKNESS PER MEMOCR PER ANNUM
o
.
รรวมพวร 10 รพาพ
25
30
40
60
45
AGE
Diagram 111.-This diagram indicates how "unfitness" increases with
increasing age. At 65 the average number of weeks of sickness
is more than 8 times that at 30.
KARL PEARSON.
23
The author of the Report tells us that “the figures also demonstrate
the influence of continued drunkenness and the life associated with
drunkenness in the making of the unfit. The definite and regular increase
in the percentage of the unfit with every additional decade of habitual
drunkenness is unmistakable."
The percentage of unfitness has nearly doubled in 30 to 40 years of
drunkenness. Now please look at something of an entirely different
character. Here is a diagram giving the weeks of sickness of a great
Friendly Society for each year of age. You see at once that between
30 and 70 years of age the weeks of sickness are not merely doubled,
they are multiplied by ten! The person who has had thirty to fifty years
of drunkenness is, as a rule, thirty-five to forty years older than the person
who has had only five years of such a life, and on that very account we
must expect such persons to be less fit for hard work.
Clearly these statistics I have just shown you have no meaning at all
until we have allowed for the feebleness of the increasing age of those who
have had a longer duration of drinking. How are we to do this? Why,
there is one simple and direct method : we may inquire whether among
inebriate women of the same age, the unfit have had a longer duration of
alcoholism than the fit. This is the method adopted by Dr. David Heron
in his recently issued monograph on Extreme Alcoholism in Adults,
and the results are very remarkable :-
TABLE 11.
Mean duration of alcoholism.
Age of inebriates.
Duration of At
less that of unit.
Among the ft.
1
Among the unit.
25-27
28-30
31-33
34-36
37-39
40-42
43-45
46-48
49-51
52 and over
8.53
10-49
11.15
11.92
12:74
13.45
14:72
13:45
16.87
14:31
8.46
10.35
11.08
13:36
15.71
13:35
13:43
14:25
13.90
12-60
07
+ 14
+ +07
- 1'44
- 2.97
+ •10
+ 1.29
.80
+ 2.97
+1.71
For 7 out of 10 age groups the tit have been longer alcoholic than the unfit.
Whereas we should have expected to find of two women of the same
age, one fit and the other unfit, the latter would have had the longer
duration of alcoholism, yet actually there is little difference and the fit
women, as shown by the last column, have on the whole rather the longer
duration of alcoholism.
24
Lecture to Congress.— Eugenics and Public Health.
Now this illustration is of immense importance because it leads us
directly to a most important principle : Two things like years of
drunkenness and increasing unfitness may be obviously associated together
and this association may lead us to sweeping conclusions, and yet the
association be not a true causation. We may have neglected an important
factor, in this case the increasing age of those who have been drinking
the longer. Here we reach the great rule of modern statistics : “When
investigating the relation of two characters which you find associated, test
whether they still remain related after you have given all other characters
likely to be influential constant values. Before you have done this you
certainly must not treat the relation as a causative one.”
The full theory of this method is what in modern statistics we term
the treatment by partial correlation. We say that for constant age the
partial correlation between years of drunkenness and physical fitness for
hard work appears on the basis of the Home Office statistics of the
inebriate reformatories to be zero.
Let us try to illustrate this important conception of partial correlation
from other data which have immense importance from the standpoints both
of eugenics and public health. My illustrations shall now be drawn from a
field which is attracting much attention at the present time, namely that of
infantile mortality. This field leads us at once to the questions of employ-
ment of mothers, of breast feeding, and of insanitary houses. These are
the very points on which men propose largely to legislate for unrepresented
women, and it certainly befits us men to be very sure of our premises.
Does employment of mothers increase the infantile death-rate? Is there
less artificial feeding where that death-rate is lower? Do back-to-back
houses show higher infantile mortality than through houses? Now first
let us look at breast feeding. I take data most kindly provided for me by
Dr. Pilkington, Medical Officer of Health for Preston. We have :-
TABLE 12.- Preston.
Mortality Rate 1st year of life.
Entirely breast-fed
13.2 per cent.
Partly or entirely artificially fed
15•7
Babies in bad health in first year.
Entirely breast-fed
5.5
Partly or entirely artificially fed
11.7
Dead and delicate babies.
Entirely breast-fed
18.7
Partly or entirely artificially fed
27.4
...
...
: :
...
: :
::
KARL PEARSON.
25
Could anything be more conclusive? We see that nearly 50 per cent.
more babies die or are delicate when the mother does not entirely breast-
feed. But Dr. Pilkington went further; he enquired into the reasons for
entirely or partly artificially feeding. The reasons may be classed into two
big groups, those cases for which it was needful either because the mother
could not provide the milk, or because the infant could not take it, and
those cases where the mothers gave up nursing optionally, because they
wished to return to the factory or because they disliked nursing. Now
let us examine the like rates when we distinguish these two classes.
We have :-
: : :
: : :
: :
...
...
: : :
: : :
...
TABLE 13.- Preston.
Mortality Rate, 1st year of life.
Entirely breast-fed
13.2
per
cent.
Optional artificial feeders
7.3
Necessary artificial feeders
20:5
Babies in bad health.
Entirely breast-fed
5.5
Optional artificial feeders
73
Necessary artificial feeders
14.2
Dead and delicate babies.
Entirely breast-fed
18.5
Optional artificial feeders
14.6
Necessary artificial feeders
34:7
It would appear from these results that in Preston, at any rate, it is
not the artificial feeding per se which is the source of the high death and
delicacy rates. The heavy rates arise solely in the case of those women
who necessarily adopt artificial feeding, either because they cannot provide
the food themselves, or because the baby is very delicate; the high rates
arise because the artificial feeding category covers largely mothers who
are themselves delicate or have delicate children.
It is well in such an important matter to give a control series. I am
able to do this by the courtesy of Dr. Greenwood of Blackburn. We
find from his data :
TABLE 14.- Blackburn.
Mortality Rate, 1st year of life.
Entirely breast-fed
Partly or entirely artificially fed
: :
6.3
per
cent.
15.8
...
26
Lecture to Congress.-Eugenics and Public Health.
...
.
i:
Babies in bad health.
Entirely breast-fed
24:0 per cent.
Partly or entirely artificially fed
19.0
Dead and delicate babies.
Entirely breast-fed
30•3
Partly or entirely artificially fed
34:8
But when we analyse these returns in the previous manner we find :
TABLE 15.— Blackburn.
Mortality Rate, 1st year of life.
Entirely breast-fed
6.3
per cent.
Optional artificial feeders
6.9
Necessary artificial feeders
22.0
Babies in bad health.
Entirely breast-fed
24.0
Optional artificial feeders
17.2
Necessary artificial feeders
20.0
Dead and delicate babies.
Entirely breast-fed
30•3
Optional artificial feeders
24.1
Necessary artificial feeders
42.0
These results absolutely confirm those from Preston, the death-rate
of artificially fed babies is swollen because they include mothers unable
to feed their children. These results suggest that it is not the artificial
feeding, but the health of the mother which is the dominating factor in
the mortality and delicacy of the infant.
I can bring home to you how very much more the health of the
mother means than the nature of the feeding by considering the per-
centage of dead and unhealthy babies in relation to nature of feeding and
health of mothers in Preston and Blackburn :
i :
...
...
29
Table 16.—-Percentage of Dead and Delicate Babies at end of 1st Yeur.
Preston,
Dil.
Blackburn.
Din.
30:3
Entirely breast-fed.
Entirely or partly artificially fed
1807
27.4 5
8.7
4.5
Mother's health good
Mother's health bad
23.2
38 6
15.4
28.4
45:3)
16:9
* Illustrations of the selective death-rate in infantile mortality, a death-rate recently
vigorously denied from this platform.
KARL PEARSON.
27
It will be seen that the difference of the rates in these two cases is of
a wholly different order.
Let us turn now to the question of mothers' employment.
Miss E. M. Elderton, of the Eugenics Laboratory, finds very small
relationships between infantile death-rate and the employment of mothers
in a great variety of districts; the influence of employment is not as
great as that of breast feeding, and sensibly less than that of mother's
health.
For example taking Preston :-
TABLE 17.-Employment of Mothers.
PRESTOX.
Neiiber before nor after.
Before and after.
Before only.
Death rate
1 ..icacy rate
Death and delicacy rate
17.7
9:1
26:8
17.2
99
27.1
185
5:3
23.8
Here we have employment making a difference of only about 3 in the death
and delicacy rate, whereas we have already seen that in the same town the
mother's health makes a difference of 15 in the same rate!
I ask you to look at the same thing from Blackburn :-
Table 18.- Employment of Mothers.
BLACKBIRY.
Neither before por after
Before and after.
Before only.
Death rate
Delicacy rate
Death and delicacy rate
10-6
22:0
32:6
13.9
20-0
33.9
18.9
21.1
40-0
Here we have the paradox that the mothers employed before and after
have better habies than those employed before but not after. The real
solution no doubt lies again in the health of the mothers, the mothers not
returning after their confinement being on the average in weaker health.
If we take all employed as against unemployed, we find the death and
delicacy rates of the two classes are 32.6 and 35.6, and even this difference
is incomparable with that of health which gives 28.4 and 45.3 for healthy
and unhealthy mothers! How about unhealthy and healthy fathers, some
of you may ask? Well, we have both reports on the father's health and
on his occupation. The latter may be classified into skilled, factory and
general labour groups. It is well known that the mortality rate and
presumably, therefore, the health of general labourers is far below that
of skilled workers and factory hands. The results for Blackburn are as
follows:
28
Eecture to Congress.-Eugenics and Public Health.
...
Table 19.- Death and Delicacy Rates of Babies for Occupation of Father.
Skilled Workers
31.1
Factory Hands
30.4
General Labourers
42:3
Employed Mothers 32-6 Unemployed Mothers 35.6
The occupation of the father is thus more influential than the employment
of the mother, and if we legislate for the one, we must legislate for the
other. Indeed, if we legislate for either of them we ought to legislate for
the “baby-pacifier," which gives still more marked differences in the rates.
.
TABLE 20.--Infantile Death-rate-Rochdale.
General rate 1909
9.0
Rate after dummy teat stage
7.5
Mother not employed
8.0
Mother employed 12.3
Dummy teat not used ... 4.0 Dummy teat used ... 8.6
Percentage range in death-rate associated with mothers' employment ... 44
Percentage range in death-rate associated with dummy teat
.61
Now turn to the health of the father. The accompanying table shows
us the relative influence on death and delicacy rate of babies in the first
year of life for Preston :-
...
...
*
. . .
..
• • •
TABLE 21.- Death and Delicacy Rate for Infants.
General Rate 1908, 25.6.
Range.
Father's health good
24.4 Father's health bad 54.2 29.8
Mother's health good 23.2 Mother's health bad 38.6 15.4
Father's labour high class 21.1 Father's labour low class 31.8 10.7
Entirely breast fed 18.7 Entirely or partly art. fed 2704 8:7
Mother unemployedt
26.8
Mother employedt
27.11 0:3
You will I think agree, on examining this table, that parental health
is fifty per cent. more important than parental occupation or breast
feeding, and that all four are immensely more significant than the
employment of women.
But every medical officer of health is surely bound to tell us that if he
had money enough he could give us health? I hope it may be so, but
,
the question is rather the measure of the gift he can provide us with.
* Excessive owing to smallness of numbers, but high at Blackburn also, where we have
death and delicacy rate 34:3, but for father's health good 30:7, father's health bad.
48:5, range 17.8.
+ Before and after.
KARL PEARSON.
29
Has he anything like the same balance at his bank that Nature has to
draw upon at hers?
Let me illustrate my point, and the pitfalls we may fall into by
superficial reasoning. By the kindness of Dr. Anderson, of Rochdale, I
have been able to see much material as to the effect of back-to-back or
non-through houses. In something like 2,000 houses in Rochdale the infant
mortality in the first year of life was 9.2 in the through-houses, and 12:7
in the non-through houses. Could any evidence better support that of
.
Dr. L. W. D. Mair's Report to the Local Government Board on the evil
of back-to-back houses ? Why, the back-to-back house increases the
death-rate by thirty-eight per cent.! Shall we, however, after what we
have seen to-night, rest content with such a crude result? Shall not
we ask whether the type of people living in through and in back-to-back
houses are the same ? Whether they start physically, mentally and
economically on the same planes ? Dr. Mair tried to make the people in
the two types of houses of the same class, but he gives us no reliable
estimate of their economic means or their habits, and he is obliged to
confess that even after selecting his back-to-back houses he found a
difference of 1s., or twenty per cent., in the average weekly rent. Without
any selection of houses at all, we find in Rochdale every bad quality
associated with the back-to-back house in most intimate degree : over-
crowding, uncleanliness, poor means, irregularity of father's work, habits
of parents. The difference of death-rate has no significance until these
quantities have been equalised in back-to-back and through houses !
Examine the following table :-
TABLE 22.-Through and Not-through Houses.—Rochdale.
Through houses. Not-through houses.
Infant mortality
9.2 per cent. 12:7
per
cent.
Poor means of parents 66-8
99.1
Bad habits of parents
17.0
57.1
Till we have at least allowed for the poor means and the bad habits of
parents in back-to-back houses, it is really idle to discuss the infantile
mortality rate in such houses, But you may ask, how do I know that bad
habits are so influential ? That has also been provided by Dr. Anderson,
and here it is :-
TABLE 23.—Infant Mortality in Rochdale.
General Rate
104 per cent.
و
>>
Range.
per cent.
Through house 9.2 Not-through house 12:7 3.5
Parents of good habits 7.6 Parents of bad habits 16:3 8.9
per cent.
per cent.
30
Lecture to Congress.- Eugenics and Public Health.
We see that even if the back-to-back house be partly the source of the
higher infantile death-rate, the medical officer of health who could alter
parental habits would achieve more than twice as much as he who carried
a bye-law against back-to-back houses. But what if the higher death-
rate in back-to-back houses is really due to the predominance of parents
of bad habits and bad physique in the cheaper houses ? Is not this
cheaper house exactly where we should expect to find them, and shall we
gain anything socially by transferring them to through houses ? The
test question is: What infantile mortality have parents of good habits
in back-to-back houses? This is answered in the next table.
TABLE 24.-Influence of Back-to-back Houses and Purental Habits on
Infantile Death-rate. Rochdale, 1910. .
Through houses. All parents.
Not-through houses.
All parents.
At six months.
At a year
6.9
9.2
9.9
12:7
Parents with:
Parents with :
Good bahits. I
Bad habits
Good babits.
Bad habits.
5.8
At six months
At a year
124
168
6:3
7.7
12:7
16:1
7.6
Percentage of parents with bad habits: Through houses, 17.0; not-through houses, 57·1.
This shows us that in Rochdale at any rate it is the character of the parent
and not of the house wbich kills the child.
By the courtesy of the medical officer of health for Sheffield, I have
received data from which I am able fully to confirm these Rochdale results.
TABLE 25.---Sheffield: Death and Delicacy rates for Female Babies
(circa 4,000).
Character.
Rates.
Range.
Health of Mother
Health of Father
Habits of Father
Habits of Mother
Good
Good
Good
Good
6.4
708
7.5
7.4
Bad
Bad
Bad
Bad
14.6
13.0
12.5
109
8:2
5:2
5.0
3.5
Employment of Mother
Type of House
Nnt employed 7.9
Through
77
Employed
Back-to-back
8.5
8.3
0-6
0:6
Thus the health and habits of both parents are far more influential than
employment of mother or type of house.
KARL PEARSON.
31
Such results force me to believe that as the Public Health Service
places more and more data at the service of social investigators, we shall
realise more and more how man moulds his circumstances and not mere
environment moulds the human herd. Take this very question of health.
I have tried to investigate its distribution and variation in the pro-
fessional classes, where on the whole the cruder forms of environment
have little influence, and here how markedly, considering the many ex-
traneous sources of bad health, do we find it hereditary! I would ask
you to look at the diagrams which demonstrate this. The one exception
appears to be the offspring of very delicate parents, and the probable
GBWBRA
RBALTI.
175
PATARR
AND DAOG ATBR.
(1937)
150
V. ROBUST
12
100
75
50
25
o
ROBUST
DADOPTER
XORMALLY HEALTHY
-25
CAN HEALTH
DAUOAT KRS.
-50H
-100
PATHERS.
-125-
DELICATB
- 150
MEAN HEALTH
-175
100
75
175 150 125
V. ROBUST
50 25 0 -25 -50 -75 -100 -185 - 1501-175
ROBUST NORMALLY HEALTHY
DBLICATE
PATHER
Diagram IV.-Showing the heredity of General Health in the Professional
Classes. 1937 cases of father and daughter. Increasing delicacy of
father marks increasing delicacy of daughter. Note the anomaly in
the case of delicate fathers.
32
Lecture to Congress.—Eugenics and Public Health.
reason for this is shewn in my last diagram which indicates why the
delicate children of delicate parents are not recorded in their due
proportions : they die before their health can be recorded in larger pro-
portions than the children of healthier parents. I once remarked to a
coastguardsman: “What a healthy spot for your bairns !” “Well," he
said in reply, “I says of children some is un'ealthy wherever you puts
’em, and the others ag'en is ’ealthy no matter where they lives.”
But if, as I have endeavoured to show you to-night, health is a real
hereditary character and the health of the parents is far more important
than the question of back-to-back houses, one-apartment tenements, the
>
GENPRAL
HEALTH
tos
MOTHER
AND
DAUGHTER.
150
V. ROBUST
(1878)
125
100
75
50
ROBOST
25
o
- 25
MBAN HEALTH
DAUGHTERS
DAJOHT2??
NORMALLY HEALTHY
-50
15
-100-
MOTHERS.
-125
DBLICATB
-150-
MBAN HEALTH
-175
175
75
0
150 125 100
V. ROBUST
50 25
ROBUST
- 25 -50 - 75
NORMALLY HEALTHY
-125 - 150 - 175
DBLICATE
MOTHER
Diagram V.-Showing the heredity of General Health in the Professional
Classes. 1878 cases of mother and daughter. Increasing delicacy of
mother marks increasing delicacy of daughter. Note the anomaly in
the case of delicate mothers.
KARL PEARSON.
33
employment of mothers or breast feeding, what function, some of you may
ask, am I leaving to the Public Health Service? Well, Ladies and
Gentlemen, when I say that you cannot produce brains by multiplying
universities and technical colleges, am I arguing that such institutions are
of no service at all? If brains are bred, not manufactured by colleges,
are we to neglect the fact that, although a keen razor can never be made
of bad steel, a good steel requires setting and tempering before it can fitly
perform its functions? Because I believe that health in the main is the
product of sound stock, and that you cannot produce it by changing the
habitations of unsound stock, is it not possible for me also to hold that
PATHER AND SON.
.
PERCENTAGR OP SONS DBAU.
10+
16+
20
75
175 150 185 100
V.ROBUST.
50 25 O - 25 -50 -75 -100 - 125 - 150 - 175
ROBOST NORDLALLY HEALTHY.
DELICATE
DATHBR.
Diagram VI.-To explain anomaly in offspring of delicate parents.
Percentage of dead sons to various classes of fathers. The
delicate fathers have more dead sons and so a larger propor-
tion of their sons escape the record. Only the healthy offspring
of delicate parents survive to be recorded.
31
Lecture to Congress.- Eugenics and Public Health.
+
increased sanitary efficiency will render the work and life of the healthy
still fuller and more effective?
Meanwhile, every medical officer in this country, who realises the
enormous power of gaining knowledge which his position gives him, can aid
in building up a science upon which a sure foundation for race fitness can be
constructed. Only let us not be hasty, let us not rush upon the basis
of most superficial statistics to conclusions from which still more super-
ficial legislation will flow. We are at the parting of the ways now. The
public health service of this country is bound to develop marvellously in
the next fifty years. It may, overlooking the infinite subtlety of these
social problems; proceed experimentally before it has fully inquired and
attained to adequate knowledge. In that case it is bound sooner or later
to lose the confidence of the nation, and it will only reach results by
wasting endless public money in experiments. But, on the other hand, it
may content itself by collecting material, by observing and recording; in
short, by building up a science of human society, before it converts that
science into an applied art. This is the only safe way; the public con-
fidence will only be won when we laymen find that in nine cases out of ten
(ten in ten if that be possible to any group of human beings) the public
health service is in the right. But if this safe course is to be the course
of the future, I feel sure it must take its origin from a new concordat:
the union of medicine and mathematical statistics for a common end, the
end indeed of my own science of eugenics, namely, the discovery of what
agencies under social control will improve or impair the racial qualities of
future generations. Under that definition every public health officer is a
eugenist; but if he is to be an interpreter as well as a recorder, the day is
with us when he must seek aid from mathematics as the handmaid of
public health.
Without that aid neither eugenics nor public health can,
to use the words of Friar Bacon, discover its own ignorance nor find its
proper remedies.
1

UNIVERSITY OF MICHIGAN
3 9015 07487 0794
Bia etric Laboratory Publications.
shed by the Cambridge University Press, Fetter Lane, E.C. 4
1
T
-
DRAPERS' COMPANY RESEARCH MEMOIRS.
BIOMETRIC SERIES.
1. Mathematical Contributions to the Theory of Evolution.—XIII. On the
Theory of Contingency and its Relation to Association and Normal
Correlation. By KARL PEARSON, F.R.S. Issued. Price 48. net.
II. Mathematical Contributions to the Theory of Evolution.—XIV. On the
Theory of Skew Correlation and Non-linear Regression. By KARL
PEARSON, F.R.8. Issued. · Price 58. net.
III. Mathematical Contribations to the Theory of Evolution.—XV. On the
Mathematical Theory of Random Migration. By Karl PEARSON, F.R.S.,
with the assistance of JOAN BLAKEMAN, M.Sc.Issued. Price 58. net.
IV. Mathematical Contributions to the Theory of Evolution.-XVI. On Far-
ther Methods of Measuring Correlation. By Kael PEARSON, F.R.S.
Issued. Price 4s. net.
V. Mathematical Contributions to the Theory of Evolution.-XVII. On
Homoty posis in the Animal Kingdom. By ERNEST WARREN, D.Sc.,
Alice LEE, D.Sc., EDNA LEA-SMITH, MARION RADFORD, and Karl
PEARSON, F.R.S.
(Shortly.
VI. Albinism in Man. By KARL PEARSON, E. NETTLESHIP, and C. H. USHER.
Text, Part I., and Atlas, Part I. Issued. Price 358, net.
VII. Mathematical Contributions to the Theory of Evolution.-XVIII. On a
Novel Method of Regarding the Association of two Variates classed
solely in Alternative Categories. By KARL PEARSON, F.R.S. Issued.
Price 4s, net.
STUDIES IN NATIONAL DETERIORATION. Price 38. net each,
I. On the Relation of Fertility in Man to Social Status, and on the changes
in this Relation that have taken place in the last 50 years. By DAVID
HERON, M.A., D.Sc. Issued.
II. A First Study of the Statistics of Pulmonary Tuberculosis (Inheritance).
By KARL PEARSON, F.R.S. Issued.
III. A Second Study of the Statistics of Pulmonary Tuberculosis. Marital
Infection. By ERNEST G. POPE, revised by KARL PEARSON, F.R.S.
With an Appendix on Assortative Mating, by ETHEL M. ELDERTON.
Issued.
IV. The Health of the School-Child in relation to its Mental Characters. By
KARL PEARSON, F.R.S.
[Shortly.
V. On the Inheritance of the Diatheses of Phthisis and Insanity. A Statis-
tical Study based upon the Family History of 1,500 Criminals. By
CHARLES GORING, M.D., B.Sc. Issued.
VI. A Third Study of the Statistics of Pulmonary Tuberculosis. The Mortality
of the Tuberculous and Sanatorium Treatment. By W. P. ELDERTON,
F.I.A., and S. J. PERRY, A.I.A. Issued.
VII. On the Intensity of Natural Selection in Man. (On the relation of
Darwinism to the Infantile Death-rate.) By E. C. Snow, D.Sc. Issueil.

EUGENICS LABORATORY PUBLICATIONS.
SOLD SQUARE, LONDON, W.
MEMOIR SERIES.
1. The Inheritance of Ability. Being a statisticnl Examination of the Oxford
Class Lists from the year 1800 onwards, and of the School Lists of Harrow
and Charterhouse. By EDOAR SCHUSTER, M.A., Formerly Galton Research
Fellow in National Eugenies, and E. M. ELDERTON, Galton Research Scholar in
National Eugenics Issued. Price 48. net.
II. A First Study of the Statistics of Insanity and the Inheritance of the
Insane Diathesis. By DAVID HERON, D.Sc., Galton Research Fellow. Issued.
Price 38 net.
III. The Promise of Youth and the Performanee of Manhood. Being a
statistical Examination into the Relation existing between success in the
Examinations for the B.A. Degree at Oxford and subsequent Success in profes-
sional Life. (The professions considered are the Bar and the Church.) By
EDGAR SCHUSTER, MA, D.Sc., Formerly Galton Research Fellow in National
Eugenios. Issued. Price 28. 6d. net.
IV. On the Measure of the Resemblance of First Cousins. By ETHЕТ. М.
ELDERTON, Galton Research Scholar, assisted by KARL PEARSON, F.RS.
Issued. Price 38. 6d. net.
V. A First Study of the Inheritance of Vision and of the Relative
Influence of Heredity and Environment on Bight. By Adry BARRINGTON and
KARL PEARSON, T.R.S. Issued. Price 4s. net.
VI. Treasury of Human Inheritanee (Pedigrees of physical, psychical, and pntho-
logical Characters in Man). Part I and II (double part). Diabetes insipidus,
Split-Foot, Polydactylism, Brachydactylism, Tuberculosis, Deaf-mutism, and
Legal Ability.) Issued by the Galton Laboratory, Price 14s. net.
VII. The Influence of Parental Oceupation and Home Conditions on the
Physique of the Offspring. By ETHEL M. ELDERTON, Galton Besearch Scholar.
Shortly.
VIII. The Influence of Unfavourable Home Environment and Defective
Physique on the Intelligence of School Children. By DAVID HERON, D.Sc.,
Galton Research Fellow. Issued. Price 4s, net.
IX. The Treasury of Human Intelligence (Pedigrees of physical, psychical, and
pathological Characters in Man). Part III. (Angioneurotic Oedema, Hermaph-
roditism, Denf-mutism, Insanity, Commercial Ability.) 188ued. Price 6s. net.
X. A First Study of the Influence of Parental Alcoholism on the Physique
and Intelligence of the Offspring. By ETHEL M. ELDERTON, Galton Research
Scholar, assisted by KARL PEARSON, F.R.S. Issued. Second Edition. Price 4s.net.
XI. The Treasury of Human Inheritance (Pedigrees of physical, psychical, and
pathological Characters in Man). Part iv. (Cleft Palate, Hare-Lip, Deat-
mutism, and Congenital Cataract.) Issued. Price 10s. net.
XII. The Treasury of Human Inheritance (Pedigrees of physical, psychical, and
pathological Characters in Man). Parts V and vi. (Haemophilia.) Issued.
Price los. net.
XIII. A Second Study of the Influence of Parental Alcoholism on the Physique
and Intelligenee of the Offspring. A Reply to certain Medical Critics and an
Examination of the rebutting Evidence cited by them. By KARL PEARSON,
F.R.S., and ETHEL M, ELDERTON. Issued. Price 48. nat.
XIV. A Preliminary Study of Extreme Alcoholism in Adults. By AMY
BARRINGTON and KARL PEARSON, F.R.S., assisted by DAVID HERON, D.Sc.
Issued. Price 4s. net.
XV. The Treasury of Human Inheritance (Pedigrees of physical, psychical, and
pathological Characters in Man). Parts VII and VIII. (Dwarfism.) With 49
Plates of Illustrations and 8 Plates of Pedigrees. Issued. Price 158, net.
XVI. The Treasury of Human Inheritance. Prefatory Matter and complete
Name and Subject Indices to Vol. 1. With Frontispiece Portraits of Sir Francis
Galton and Ancestry. Issued. Price 38, net.
XVII A Second Study of Extreme Alcoholism in Adults. With special
reference to the Home-Office Inebrinte Reformatory data. By DAVID HERON,
D.Sc. Isstied. Price 58. net.
Buckram covers for binding Volume 1, of the Treasury of Human Inheritance with
impress of the bust of Sir FRANCIS GALTON by Sir GEORGE FRAMPTON can be obtained
from the Eugenics Laboratory by sending a postal order for 25. 9d. to the Hon. Secretary.
A large photograph (11" by 13") of SIR FRANCIS GALTON by the late Mr. Dew SMITH
can also be obtained from the Laboratory by sending a postal order for 103, 6d. to the
Hon. Secretary