184 
Miscellanea 
a third, i.e. the intensity of a certain environmental influence and the intensity of a certain 
somatic character in the parent, say, on the intensity of the somatic character in the off- 
spring. Yet Major Darwin tells us we caiuiot do this because we cannot measure these 
things in the same unit ! — How suavely yet forcibly Sir Francis Galton himself would have 
ridiculed such ignorance in high places as is passed by the Editor of the Eugenics Journal ! — 
We can hear him now telling us how the intensity of each character could be measured by 
its grade, and how the problem turned on whether the same change in grade in the environ- 
ment and in the parental somatic character produced greater or less change in the grade of 
the filial somatic character. When we inquire whether inter-racially stature is more closely 
related to cephalic index or to eye colour, are we to be met by the statement that these 
characters cannot be compared because they cannot be measured in a 'common unit,' and 
then be told that it is not "wise to use words in scientific literature without endeavouring to 
attach a definite meaning to them ? " Every trained statistician knows that each character 
is measured in the unit of its own variability — in what he terms its standard deviation*, 
and that this standard deviation ))rovides him with a measure of the frequency of each value 
of the variate in question. It seems to me that the only correct sentence in this paragraph, 
is the author's statement that he himself has no idea what unit is 'common' to heredity and 
environment. 
But our author continues : 
"Take any quality, and we find that the human beings composing any community differ 
more or less considerably as regards that quality. Now we can measure the correlation 
between the differences shown in this quality and the differences of environment to which 
the members of the community in question had previously been exposed t. This is one 
correlation. Then we can also measure the correlation coefficient between, say, father and 
son, as regards the quality in question. Here is a second correlation ; and if we are told 
that the relative influence of environment and heredity is measured by the ratio between 
these two correlation coefficients, we cei'tainly do thus get a clear conception of what is 
meant j." 
But has the writer really obtained a clear conception of what such coefficients of correla- 
tion mean, when in the next j^aragrajjh he continues : 
" Imagine an ideal republic, in some respects similar to that designed by Plato, whei-e not 
only were all the children removed from their parents, but where they were all treated exactly 
alike. In these circumstances none of the differences between the adults could have anything 
to do with the differences of environments, and all must be due to some differences in inherent 
factors. In fact the environment correlation coefficient would be nil, whilst the hereditary 
correlation coefficient might be high§." 
Could any better evidence be adduced that the President of the Eugenics Education Society 
did not know what a coefficient of correlation meant at that date ? The coefficient of correlation 
for the environment might be anything from — 1 to -f- 1 ; the only obvious fact would be that you 
could not find its value, except in the form 0/0, from an environment which precluded any 
measure of variation. How again Sir Francis would have smiled at the notion that the 
coefficient of correlation for a constant environment must be nil. Why should we follow such 
* Of course he may or does need other constants to help iu the description of the frequency, 
t loc. cit. p. 153. 
X This seems to contradict the writer's previous assertion that two things are incomparable, if they 
have not a ^common unit' ! 
§ I wrote at once to Major Darwin pointing out the error of such a statement and he withdrew it in 
the next number. But the harm done by an article of this kind cannot be reversed by correcting a 
single misstatement. 
