478 THE AMERICAN NATURALIST [Vol. XLIV 



sort of a demand of them. The consideration is never- 

 theless, I think, worth consideration. 



The most serions, and probably fatal, objection to the 

 quantitative view outlined above is found in the later 

 possibilities of the mixed chromosomes. If the longitu- 

 dinal division is fortuitous in the synaptic pair it must 

 also be assumed to be fortuitous in the later splittings of 

 the same chromosomes in the embryo. The results 

 would give a mosaic of cells in some of which one and 

 in other cells another character will predominate. We 

 should expect therefore a sort of piebald or chimaera 

 type to result. The difficulty is not minimized by refer- 

 ring the results to all of the chromosomes instead of to a 

 single one. 



Unless we refer the problems of heredity to principles 

 apart from a material basis our only hope at present of 

 a scientific solution of the problem is to rely on such a 

 basis. There are three ways, however, in which we may 

 make use of such a physical material conception of ' ' seg- 

 regation." First, by postulating material particles in 

 the chromosomes of the germ cells qualitatively different 

 — particles that are sorted out at the reduction period. 

 Our analysis has shown that there are serious difficul- 

 ties for this interpretation. Second, by postulating a 

 quantitative factor as the basis of segregation; here also 

 difficulties are met with. Third, by assuming initial dif- 

 ferences in the germ cells of the hybrids due to the 

 same kind of differences that become patent in the 

 development of the embryonic organs where the re- 

 sults are not referable to segregation of chromatic ma- 

 terials, but due to regional differences or state of equi- 

 librium—the result of reactions between the cells. Here 

 it seems to me we find the most promising direction in 

 which to look for further light on the subject. For ex- 

 ample, the formation successively of brown, yellow and 

 black pigment by the follicle cells of a gray mouse sug- 

 gests that a similar process may take place in the germ 

 cells of hybrids. In the somatic cells no one supposes 

 that the differences are due to loss of chromatin, or to a 



