TAXONOMIC COMPARISONS 79 



(Droseraceae) , there is difference of size in the proportion of a 

 thousand to one, genotypic control of the difference is to be inferred 

 (Table lo). 



Structural changes in themselves have no necessary genetic effect, 

 but they may condition genetic isolation of stocks, and hence a 

 later genetic differentiation between species. But this differentia- 

 tion need not follow. We therefore find every relationship between 

 structural and numerical and genotypic differences in the chromo- 

 somes on the one hand and systematic differences on the other, as 

 follows : — 



(i) The absence of readily detectable differences within large 

 groups such as the gymnosperms — all of which, with a few excep- 

 tions, have twelve haploid chromosomes (cf. Sax, 1933) — and the 

 Acrididae (except three genera) ; or the absence of all differences 

 except polyploidy (e.g., the Pomoideae). 



(ii) The occurrence of differences between genera of an order 

 while the genera are themselves fairly uniform, e.g., Antirrhinum, 

 Orchis [cf. Tischler, 1928 h), Ribes (Meurman, 1929), Prunus (apart 

 from polyploidy, D., 1930 a). 



(iii) The occurrence of differences between sections of a large 

 genus, e.g., Primula (Bruun, 1930). 



(iv) The occurrence of differences between all the species of a 

 genus, e.g., Drosophila (Metz, 1916), Fritillaria, Tiilipa, Crepis, 

 Tradescantia, Muscari, Vicia (and polyploidy in nearly all large 

 angiosperm genera except Antirrhinum, Ribes and Orchis). 



(v) The occurrence of differences between geographical races or 

 sub-species (Nothoscord on species, Matsuura and Suto, 1935; Riimex 

 acetosa, Yamamoto, 1933 ; Ono, 1935 ; Drosophila pseudo-obscura, 

 Dobzhansky, 1934 (v. Tables 8 and 37). 



(vi) The occurrence of differences between forms not recognisably 

 distinct (v. Table 37, structural differences in some Orthoptera and 

 polyploidy in Silene, Blackburn, 1928 ; and numerical variation in 

 Viola canina, Clausen, 1931). 



A comparison of the mitotic chromosomes of different forms is 

 therefore of little value in placing them systematically unless we 

 know the type of variation that prevails in the groups in question. 

 If the forms have different chromosome numbers they are probably 



