1897] KARL PEARSON ON EVOLUTION 53 
to answer the problem: Is one sex closer to its mean, more con- 
servative to its type, than the other? and that the only scientific 
answer to this lies in the magnitude of the per centage variations of 
the two sexes for corresponding organs.” 
The meaning of this may be made clear by an example. Suppose 
a number of sticks, about a yard long, to be chosen by some rough 
process of measurement ; and suppose that more careful examination 
showed the average length of these sticks to be 3 feet, while half 
the sticks were between 2 feet 11 inches and 3 feet 1 inch in 
length. Suppose, further, a second group of sticks, whose average 
length is 6 feet, while half the sticks lie between 5 feet 10 inches 
and 6 feet 2 inches. Now, since one inch is the same fraction of 
three feet that two inches is of six feet, Professor Pearson asserts that 
the only scientific view of the variability of the two sets of sticks is 
that which treats the two bundles as equally variable; and he 
accordingly defines the “ coefficient of variation,’ or measure of 
variability, as the ratio of the “ Standard Deviation ” or “ Error of 
Mean Square” to the Mean. 
Now, it may at once be freely admitted that the coefficient of 
variation, as above defined, is an exact measure, and probably the 
best available measure, of the degree to which a group of animals is 
“close to its type”; that is to say, it is a measure of the extent 
and frequency of the mistakes a man would make, if he should 
simplify a discussion of these animals by using, instead of the 
individual animals, a series of perfectly average “types.” It is 
precisely the measure of accuracy of the customary morphological 
definition of a species or variety. But the student of evolution may 
have to concern himself with another measure of variability, when 
he asks not “ how close is the race to its type,” but “how much 
material for Selection is afforded by the variability of the race ?” 
The functional importance of a variation of known magnitude, 
and the effect of such a variation upon the selective death-rate, 
seem legitimate, if difficult, subjects of scientific inquiry; and if it 
can be shown that an organ in one sex gives more scope for the 
selective formation of varieties or races than does the corresponding 
organ in the other sex, it is surely legitimate (neglecting the possible 
complications due to peculiarities of heredity) to say that one sex 
is more variable than the other. For example, it is certainly 
possible, in the case of the common fowl, to produce races of cocks 
which differ more from each other in the length of their tail- 
feathers than do any hens yet produced; and the statement, that 
the tail of cocks in general is more variable than the tail of hens in 
general seems thereby justified, whether the “ coefficient of variation ” 
in the cocks of any one race be greater than that of the corre- 
sponding hens or not. 
