5ss 



THE AMERICAN NATURALIST [Vol.XLVI 



proposed series of ratios seems to be taken for granted 

 and results are viewed only from this standpoint. 



It is hoped that the previous discussion may operate to 

 cheek this practise of referring examples of coherence 

 to preconceived gametic ratios without considering the 

 possibility that an apparent agreement may be only a 

 chance approximation. 



So long as the preconceived series is taken for granted 

 and intermediates are not considered, the results of all 

 experiments will seem to give additional evidence in sup- 

 port of the series. 



We have seen that the results of previously reported 

 experiments do not correspond to the ratio in the ex- 

 pected series more closely than they do to others that are 

 intermediate, and furthermore that in some cases, at 

 least, the associations are not due to relations of the 

 determinants inside the gametes at all, but occur after 

 the stage of karyapsis, or nuclear fusion, has been 

 reached (Cook and Swingle, 1905). 



Summary 



The theory of gametic coupling assumes that correla- 

 tions between two Mendelian character pairs are caused 

 by attractions or repulsions between character-units or 

 determinants, previous to the formation of the germ cells. 

 These attractions or repulsions are supposed to increase 

 the number of gametes bearing certain combinations of 

 determinants. 



The further assumption that the various degrees of 

 association observed between different character-pairs 

 will fall into a regular series represented by powers of 2, 

 as in simple Mendelian hybrids, appears to have been 

 accepted without adequate analysis of the data on which 

 it was based. 



An examination of the early examples shows thai it 

 was only by neglecting the possibility of intermediate 

 ratios, and thus begging the question, that the observed 



