376 



On the Laws of Inheritance in Man 



(ii). 



Nüw, it there be no immediate cross selectiou of other thau these three organs 

 and üo immediate direct selection we should expect to find : 



^u = PyJ'2i + Pm>'i3 +Pm'"i5'"« 



^52 ~ Pls'lS + Pn^'i'i^'u + Pm'm 



^m = Pnl'u ' Pw'"3i''j6 "t" Pi>n''i5 

 Ta« = /3is''i3»M + pw '■« + Pss'm 



Substituting the p's aiid the organic correlations in (ü) we find : 



TABLK III. 

 Coefficients of Cross Assortative Mating. 



We conclude trom this Table that : since the calculated values are all larger 

 than the observed, the hypothesis that only direct selection of these three 

 characters takes place is not valid. There nuist be direct selection of other 

 correlated organs, or in some manner, as yet inexplicable, also an immediate cross 

 assortative mating in man*. Generally, the results given in the present and 

 the previous section for assortative mating, and in section (iv) for preferential 

 mating, indicate that io future a greater degree of attention must be paid to 

 sexual selection. It can hardly be so significant in the case of man, where most 

 people would probably u, priori suppose it of no account, and yet fail to play an 

 important part in wild life. In particular, experimental enquiry on the relation 

 of homogamy to fertility, — the likeness not being due to in-breeding — would be 

 of very great value. It is clear, that " negative "i" natural selection accompanied 



* My own view, for whicli I have small evidence at present, is that tbo functions of sex are far inore 

 highly correlated witb the physical characters in man tban is generally supposed, and that the fertility 

 of any given pair is very delicately atuned to the relative proportions of their frames. Heucc wheu we 

 come to correlate the physical characters in the parents of adult cbildren, we tind not only high direct 

 but also high cross correlations, whicb seem inexplicable on any hypothesis of conscious assortiug at 

 mating. The only way to test this is to conipare the correlations of husbauds and wives at marriage 

 with tbose of parents of adult childrcn. We have already seen that these in certain cases sensibly 

 differ. (See p. 373 and p. 374 footnote.) 



t A selection for destruction not survival : see Phil. 7'ni»s. Vol. 200, A, p. 59. 



