286 AN INTRODUCTION TO MODERN GENETICS 



with a more or less average value of the measurement, and fewer with 

 either very high or very low values. The normal frequency curve is 

 sjmimetrical about the mean value, but we may get asymmetrical or 

 "skew" curves. The technique of comparing such curves with one 

 another is part of the subject of statistics ; full treatments on the matter 

 will be found in books on that subject. 



The variation expressed in such a frequency curve may not all be 

 hereditable, since different individuals in the population probably 

 developed in different environments which may have affected the 

 character measured. Before we can get an idea of the hereditable 

 variation, we must eUminate this environmental factor as far as pos- 

 sible. The variation which is left usually still shows a fairly normal 

 frequency distribution, but it may now have a much smaller spread. 

 It is, however, often still continuous over its whole range, and does not 

 usually consist of a few distinct groups of individuals, though this may 

 be true in a few populations which happen to be heterozygous for one 

 or two genes with marked effects. The common form of continuous 

 variation is nevertheless inherited by normal MendeHan genes, but it 

 is controlled by many genes each of which has only a slight effect. 

 Crosses between the extreme forms will therefore give intermediate 

 Fi's while in F2 segregation will cause very wide variation (cf. p. 159). 



A beginning has been made with identifying the variability of natural 

 populations due to mutant genes with clear-cut effects or to chromo- 

 some aberrations. The fullest analysis is of wild populations of Droso- 

 phila.^ Any given gene is, of course, pretty rare, and exists in 

 the heterozygous form (p. 289), but each individual fly may be hetero- 

 zygous for at least one mutant. Nearly all the mutants which have been 

 isolated from wild populations are deleterious; any useful ones would 

 be selected for and would comparatively soon spread through the 

 species. Autosomal recessives seem to be commoner than sex-linked 

 recessives or autosomal dominants. This may be explained partly 

 because there are more autosomal genes than sex-linked ones, and 

 partly because an autosomal gene more often occurs in a heterozygous 

 state in which it is hidden from the action of natural selection than do 

 either of the more exposed types such as sex-linked genes or dominants, 

 and will therefore be more widespread, for a given mutation rate and 

 selective disadvantage. 



There can be no doubt that many of the variations due either to 



1 Dobzhansky and Queal 1938, Dubinin et al., i934j 1936, Gordon 1936, 

 Sturtevant 1937, Timofeeff-Ressovsky and Timofeeff-Ressovsky 1927- 



