( vi ) 



According to the known method of investigation the effects 

 due to individual differences (sex, differences due to nourish- 

 ment, &c.) could be eliminated, but it appeared to him 

 (Prof. Meldola) somewhat doubtful whether the "probable 

 error " could be properly estimated and allowed for in a 

 species in which the alternate generations differed so con- 

 siderably in size. In every individual of one generation 

 there are present in a latent form the hereditary tendencies 

 of the other generation, and the experiments of Weismann 

 and Edwards have shown how slight are the influences which 

 determine the production of one or the other form in a 

 seasonally dimorphic species. In other words, the measure- 

 ment of the rate of divergence would in such cases be com- 

 plicated by the continued tendency of some of the individuals 

 of one generation to approach those of the other in character ; 

 so that, although the extreme forms might be selected for 

 pairing in one generation, it was at present doubtful whether 

 the particular character selected (say size) would necessarily 

 correspond in the two broods. There might be such a 

 correspondence, but no experiments had hitherto been made 

 in this direction, and the point therefore remained doubtful. 

 Supposing, by way of example, that half a dozen of the 

 largest pairs of the spring brood were selected and mated, 

 and a similar number of the smallest pairs. The effects 

 upon the size of their descendants could only be measured 

 in the next spring brood ; but in the meantime an autumn 

 brood would have been interpolated, and the question as to 

 whether this intermediate smaller brood influences the trans- 

 mission of the selected character (large or small size), and if 

 so, to what extent, and further, whether the superior or 

 inferior size of the spring parents is associated with a corre- 

 sponding superiority or inferiority in the size of their 

 autumnal progeny, must be, as it appeared to him (Professor 

 Meldola), in the first place settled by experiment before the 

 "rate of divergence" required by Mr. Galton could be 

 estimated in a manner that would place the results beyond 

 the region of doubt. That periodic heredity does tend to 

 produce instability of character is shown by the frequency 

 with which intermediate forms have been bred in all 



