EVOLUTION 407 



while in most races of man a long radius goes with a 

 long femur, a selection of radius might change the sign 

 of the correlation, so that, as in some of the anthropoid 

 apes, a long radius might be expected with a short femur. 



Now these remarks may suffice to show how difficult 

 is the problem of making a numerical determination of 

 the effect of a given environment. We may find a type 

 changing, but is this due to direct selection of a given 

 organ, or to the effect of this organ being correlated with 

 a directly selected organ ? It is possible indeed that the 

 change is partly direct and partly indirect. If the reader 

 will always bear in mind that the organism is a correlated 

 whole, he will appreciate how change of environment may 

 change a great variety of characters by only selecting one ; 

 further, that the fittest type of organ for one environment 

 might, owing to the principle of correlation, involve an 

 unfit type for a second organ, and that the best a selective 

 death-rate can do is to reach a balance of fitness in the 

 two. The selection of one organ may ultimately start the 

 selection of a second or even of a whole complex of organs. 

 That the environment is changing the whole organism we 

 may realise even quantitatively, but it will be very difficult 

 to assert that the environment does this by the selection of 

 certain special organs ; at the very best the determining of 

 the actually selected organs will only amount to a highly 

 probable guess.^ 



Now a very wide range of statistical measurements 

 shows us that whenever we consider the frequency distri- 

 bution of an organ, we are able, with but few exceptions, 

 to plot a continuous polygon of variation, and that the 

 mathematician can describe this polygon by a curve 

 defined by a few — three or four — numerical constants, 

 the most important of which are the mean, mode, and 

 standard deviation already defined. How are we to 

 measure the changes in these curves or their constants 



1 All causation as we have defined it is correlation, but the converse is 

 not necessarily true, i.e. where we find correlation we cannot always predict 

 causation. In a mixed African population of Kaffirs and Europeans, the former 

 may be more subject to smallpox, yet it would be useless to assert darkness 

 of skin (and not absence of vaccination) as a cause. 



