258 THE VARIATION OF ANIMALS IN NATURE 



are the same (i.e. really parallel), and, in Drosophila, mutations 

 with similar effects may occur in quite different loci. We 

 think, however, that parallel evolution may have played some 

 part in producing resemblances within restricted groups, 

 while, if it can be shown that two unrelated groups (such as 

 the Hymenoptera and Diptera) do in fact tend to throw parallel 

 variation, it is not necessary to know the locus in which the 

 mutation responsible occurred. 



(3) It has always been an important argument in favour 

 of the selective explanation of mimetic resemblances that no 

 other factor could be suggested which would account for the 

 phenomena. A very different view has been put forward by 

 Berg (1926, chapter vi). He advances the theory that the 

 ' geographical landscape ' profoundly influences the animals 

 subjected to it. By a geographical landscape he means ' a 

 region in which the character of the relief, climate, vegetation 

 and soils are united in one harmonious whole, which is typical 

 of a certain zone of the earth, recurring through its entire 

 area ' (I.e. p. 264). He supposes that ' the landscape does not 

 affect the organism by any one of its component agencies, 

 such as by its altitude above the sea-level, its temperature, 

 or the rocks forming its soil, but by the entire combination of 

 all the elements which constitute any given landscape ' (I.e. 

 pp. 264-5). Taken as a whole Berg's thesis appears to us a 

 very marked example of special pleading, but there may 

 nevertheless be some truth in his idea. It is well known (cf. 

 Zimmermann, 1930, 1931) that the relation between colour and 

 climate in the Hymenoptera is likely to lead to a certain degree 

 of convergence in the forms inhabiting one climatic region. In 

 some of the other cases where groups of species resemble one 

 another, it is possible that as yet undiscovered edaphic factors 

 determine the convergence, especially when the number of 

 species concerned is very large, as in some of the Lycid-coloured 

 groups. Sometimes there is great diversity in pattern as a 

 whole, whereas certain features are convergent in particular 

 regions. This may be seen in humble-bees (Bombus), which, 

 as shown by Vogt, usually have the pale hairs white in the 

 Caucasus and yellow in the Alps ; in the Pyrenees they 

 are also yellow, but the pale area is always more extensive ; 

 while England appears to form a region of melanism. In 

 some cases the colour alteration in the particular local direction 



