360 APPENDIX B 



quently. The dominant character is usually the ancestral or 

 normal type from which the recessive variety has been derived. 

 Thus the normal agouti coat in guinea pigs is dominant over 

 all the derived varieties. There are exceptions, however ; for 

 example, in guinea pigs the derived rough coat is dominant over 

 the normal smooth coat. Professor De Vries maintains 1 that 

 Mendel's law obtains only when a variety in which a character 

 has become latent is crossed with a variety in which that 

 character is visible that is, when a regressive variety is crossed 

 with its own ancestral type. If the character is absent, not 

 latent, in one of the conjugating pairs, as when a progressive 

 variety mates with its ancestral type, he believes that the union 

 tends to be more or less sterile ; or, if offspring arise, that they 

 tend to blend in varying degrees the characters of both varieties 

 and to reproduce the blended type in descendants ; that is, they 

 breed true, there being no Mendelian segregation. 2 Apparently 

 he is of opinion that characters which have once appeared in a 

 species, but which are seemingly absent in descendants, are never 

 entirely lost, but become latent. Thus, according to him, a 

 white flower, derived from a blue or red variety, has the colour 

 dormant. It is probable that, in the example given, he is right. 

 Large regressions occurring in a single generation usually indicate 

 latency, not complete loss. But in the immense majority of 

 instances that is, when small and newly-acquired characters 

 disappear in a single generation, or when, in the absence of 

 selection, larger and more ancient characters slowly disappear 

 in the course of many generations there is every reason to 

 believe that the loss is complete. We are driven to this con- 

 clusion by our knowledge of the extreme parsimony of nature. 

 All progressive evolution consists of an increase in the complexity 

 of the germ-plasm. It is reasonable to suppose that that which 

 can be added can also be subtracted. De Vries' hypothesis 

 unnecessarily endows the germ-plasm with an almost unimagin- 

 able complexity. It supposes, for example, that all the characters 

 of man's remotest ancestors are latent in him all the traits of the 

 long and varied line of lower animals from which he is descended. 

 Perusing Mendelian literature, the reader is apt to gather 

 the impression that segregation of parental characters in 

 descendants is the normal, and blending quite the exception. 

 As a fact, however, segregation is rarely seen except when 

 varieties are crossed, a thing which of course seldom happens in 

 nature. In normal pure or intra -varietal breeding blended 

 inheritance is the rule. Thus, when two human beings mate, the 

 inheritance is rarely Mendelian. Professor Castle states, "In 

 man a condition of hypophylangia (two-jointed instead of three- 

 jointed digits) is dominant over the normal condition. . . . 



1 Species and Varieties, p. 276, et sea. : Open Court Publishing Co. 



2 Ibid., Lecture IX. 



