VARIATION IN THE GENUS EREBIA. 9 



establish itself as the typical number. (A further discussion of 

 this -^vill be given.) 



It is plain also that if the M of -ij is not preserved from 

 generation to generation owing to the inequality of the sexes, 

 then that M has no very much better chance of establishing 

 itself than any other. Hence a state of fluctuation. 



The cases typified by (5) and (6) are, I believe, of more 

 general application than (1) and (4) ; and also they demand less 

 theoretical hypothesis for their explanation. But I found the 

 latter more convenient to deal with at first owing to the simplicity 

 of the figures when given in full. 



I think it will be found, w^ien dealing with actual data in 

 reference to some particular character or characters, that the 

 species under consideration cannot be very definitely referred to 

 any of the six categories given above, but possesses some of the 

 qualities of several in a not very marked degree. Also many 

 instances will occur to the reader's mind of animals with sexes 

 disproportionate in numbers, and yet with no marked degree of 

 variability, &c. ; but it must be remembered that onlj^ one factor 

 in a very complex mechanism, that of heredity, is l3eing taken 

 into account. When the rule is stated — that equality in the 

 number of the sexes tends towards equilibrium, i.e. fixation of 

 one character or degree of a character, and that inequality wdien 

 coupled with variability in the preponderating sex tends tow'ards 

 continued fluctuation — it must be borne in mind that this rule 

 would be strictly true, only on the condition of all other factors 

 being equal ; but in nature, where interaction is universal, its 

 particular effect may be entirely obliterated or at least modified. 

 Nevertheless, if it is shown to hold good in theory, its possible 

 effect must not be ignored in practice. 



In the foregoing Part, a factor has, as it were, been isolated 

 from all the other factors with which it is naturally compounded, 

 and has, so to speak, been examined iti vacuo. We will, in the 

 next Part, turn to actual data, and consider them in the light of 

 the theoretical considerations already discussed. 



To recapitulate so far : equality in the numerical proportions 

 of the sexes tends towards constancy of characters. If one sex 

 is already constant, the variability of the other will tend to con- 

 form to the constancy of the one, owing to the prepotency of 

 reversion. If both sexes are variable and there are no marked 

 reversionary tendencies, then a mean will be struck between the 

 two variabilities, which will become constant, according to the 

 laws of chance. If, however, the sexes are very unequal in 

 numerical proportions, and the preponderating sex is variable, 

 whether one sex is constant or not, the variability of one or both 

 sexes is preserved, owing to the impossibility of the laws of 

 chance working regularly. 



(To be continued.) 



