240 SCIENCE PROGRESS 



round again to the idea that it is by the combination of many 

 small (but definite and discontinuous) constitutional changes 

 that evolution proceeds. 



Further, the original very natural theory that, of two 

 contrasted Mendelian characters, one was due to the presence, 

 the other to the absence, of a whole factor is gradually being 

 abandoned. It is becoming necessary to think of the factors as 

 molecules, or groups of molecules, with the power of self- 

 reproduction, and the process of mutation as a change in one, 

 probably small, portion of the factor (see Morgan, loc. cit.). 



With this in mind, the function of a chromosome-gene 

 constitution, in conjunction with sexual reproduction, in pro- 

 moting variability is obvious. The independence of the genes 

 (an independence complete for those in different chromosomes, 

 partial for those in the same chromosome) permits of all possible 

 recombinations of any mutations which may crop up. If 

 two varieties of an asexually-reproducing organism exist which 

 differ from each other in respect of two Mendelian differences, 

 only these two varieties are possible until new mutation occurs. 

 But if the same factor-differences exist in a species with sexual 

 reproduction, they may be recombined to give a possible four 

 varieties. In general, in species without sexual reproduction, 

 if n mutations appear, n varieties are possible ; but in one 

 reproducing sexually, «* varieties are possible for the same 

 number of mutations. If one of these combinations should 

 happen to confer upon its possessor an advantage in the 

 struggle for existence, it will be able to establish itself in a very 

 short space of time. This is shown in the rapid spread of 

 such forms as the dark variety of the Peppered Moth, and its 

 speed can be calculated theoretically (see Bateson on Problems 

 of Genetics, 191 3, pp. 135-40 ; and Norton's appendix in 

 Punnett's Mimicry in Butterflies, 1915). Where the number of 

 factors is large, the possible recombinations are of enormous 

 number. This was beautifully shown in Baur's cross of two 

 species of Antirrhinum (Baur, Zeits. Ind. Abst. Vererb., 21, 

 1919) ; in the second generation, out of ten thousand plants, it 

 was hardly possible to find two alike ; and the process of segre- 

 gating new (and, be it noted, for the most part healthy and 

 *' normal ") types continued on the grand scale in later genera- 

 tions. It is important to notice that in this cross the first 

 generation hybrid was uniform. This uniformity of the first 

 generation followed by great diversity in later generations is 

 of course what is to be expected on Mendelian principles, and 

 on no others ; this experiment, and others like it, may be 

 regarded as establishing that the differences between species 

 are due, either wholly or in very large part, to Mendelian genes. 

 Further confirmation of this idea as regards local races and 



