SUMMARY 167 



at one time or another, which is very improbable. Again, since 

 the inheritance of polydactylism in man and the lower animals is 

 alternative, Professor de Vries' theory involves the corollary that 

 mammals have descended from ancestors which possessed six 

 digits, one of which has become latent, which also, to say the least, 

 is very improbable. 



283. Taking all the facts into consideration, I think there can 

 be no doubt that reproduction of at least many mutations is alter- 

 native from the first. The only matter open to doubt (and even 

 that, as we see, is not open) is whether the alternation involves 

 alternative inheritance (unit segregation and gametic purity) or 

 merely alternative development (alternative patency and latency). 

 Great stress is often laid on the frequency with which alternative 

 characters are reproduced in the second mongrel generation in what 

 are known as the Mendelian proportions. Thus, if round and 

 wrinkled peas are crossed, the average proportions in the second 

 mongrel generation are three round peas to one wrinkled ; if rose 

 comb is crossed with pea comb, the proportions are nine walnut 

 combs, three rose, three pea, and one single. These proportions 

 are certainly very interesting and remarkable. Beyond doubt 

 they suggest segregation rather than patency and latency. But 

 suggestion is not the same thing as demonstration. The latter 

 must depend on surer evidence, such as that furnished by pure- 

 bred varieties. Here, as I say, we have positive testimony that the 

 Mendelian phenomena do not indicate segregation. 



284. To sum up : we have a choice between two possible 

 explanations of Mendelian phenomena. On the one hand is the 

 orthodox Mendelian doctrine of dominance (patency and latency) in 

 the first hybrid generation, and segregation and gametic purity in 

 the pure dominants and recessives of succeeding generations ; on 

 the other hand is the theory of temporary patency and latency in 

 the first generation and more perfect and permanent patency and 

 latency in the pure dominants and recessives of succeeding genera- 

 tions. I am not aware of any fact in the whole range of Mendelian 

 literature that contradicts the second theory. But the first is 

 contradicted by the facts (i) that pure dominants and recessives 

 may occur in the first hybrid generation ; (2) that apparently pure 

 dominant races may produce recessives, and recessive races domi- 

 nants ; (3) that dominants and recessives may occur in altogether 

 disproportionate numbers in any generation succeeding the first ; 

 (4) that extracted* dominants and recessives nearly always 

 bear traces of their hybrid origin (i.e. display some degree of 



