CHROMOSOME THEORY OF HEREDITY 257 



many normal variations are also inherited in the same way, 

 although their investigation, because of their slight deviation 

 from the normal, is attended with far greater technical difficulty. 

 As a working hypothesis, then, let me propose this — that the 

 hereditary constitution of an organism consists of a large number 

 of genes, or Mendelian unit-factors. 



Let us now turn to another field of work altogether, that of 

 cytology. If we examine the cells of organisms microscopically, 

 we find that they contain bodies known as chromosomes. 

 Their number is constant for a given species, and they exist 

 normally in homologous pairs, the members of which separate 

 from each other at some period prior to the formation of gametes. 

 A gamete thus gets one or other member of each pair of 

 chromosomes. Not only this, but, in the (very few) species where 

 the evidence permits a statement, the number of linkage-groups 

 is also the number of pairs of chromosomes. Thus, if Mendelian 

 genes were carried in the chromosomes, we should be given in 

 their behaviour the physical basis for ordinary Mendelian 

 ratios. P^urther, supposing that each chromosome carries a 

 great number of factors, the phenomena of linkage will be 

 explained if each chromosome represents a linkage-group, and 

 if the intensity of linkage varies inversely with the linear 

 distances of genes along the chromosomal framework. This is 

 Morgan's hypothesis, and the full evidence for it, with certain 

 special points of detail, may be found in his recent book. The 

 Physical Basis of Heredity. 



The evidence, direct and indirect, for the chromosomes 

 as the bearers of the genes is very varied. The strongest 

 support is to be found in the facts of sex-linkage, where the 

 observation of a different ratio from the normal Mendelian 

 one has been accompanied in many instances by a discovery of 

 a chromosomal difference which will account for the observed 

 divergence of ratio ; in the similar facts of non-disjunction, 

 which appear to constitute an experimentum crucis designed 

 by nature (see Morgan's summary in his Physical Basis of 

 Heredity), and by the very interesting cases in which organisms 

 have been found with double the ordinary number of chromo- 

 somes, whereupon the Mendelian ratios obtained are such as 

 might be expected from the random distribution of four instead 

 of two chromosomes to the gametes (Miiller, Amer. Nat. 48, 

 1914 ; Blakeslee, Belling, and Farnham, Science, 52, 1920, 

 p. 388). 



There are many other lines of evidence, but in this paper 

 we are deliberately taking on trust the bulk of the facts to be 

 found in standard books. 



Granted, then, for the sake of argument, that Mendelian 

 genes are represented by units in the chromosomes, have we 

 16 



