220 Experimental Zoology 
A variety has the same number of unit-characters as has the 
parent species, and when the two are crossed the unit-characters 
pair, each with its kind, subsequently in the second generation, 
separating again to give the Mendelian ratio. On the other hand, 
when a new species arises, a new unit-character is formed that 
is assumed to have no pair in the parent form, hence de Vries 
assumes that it must divide and go to all the cells when the Men- 
delian characters simply separate. It is not evident why a new 
unit-character should be assumed to divide and an old one that 
has lost its pair should not divide. It seems simpler to assume 
that both kinds always divide, but that in the variety the con- 
dition of latency has appeared, and that this condition alter- 
nates in the hybrid’s germ-cells with the old character, while in 
the “progressive” mutant the new character always dominates.’ 
Linnean species are, according to de Vries, large groups com- 
posed often of many elementary species, some of which at least 
— if we exclude local races due to environment — take the rank 
of varieties of ordinary systematic writers. Linnean species are 
separated from each other partly by progressive, partly by retro- 
gressive and degressive, characters,” i.e. they may differ both by 
true specific differences (progressive) and also by varietal char- 
acters (retrogressive and degressive). 
In this connection it is worth while to call attention to a point 
that de Vries has considered in regard to the origin of the muta- 
tion itself. The new type originates by the union of a male 
and a female cell — spermatozoén and egg. In which of these 
did the mutation first appear? If only in one, the mutation is a 
1 As stated above, de Vries ascribes the behavior in inheritance of a progres- 
sive character to the absence of a fellow-unit with which to pair. Is it not more 
probable that the new character is a modification of some original one which is 
its pair? From this point of view the behavior ofa progressive character is not 
due to its pairing or lack of pairing, but to some inherent quality that makes it 
dominate in the pair. There is some evidence that all new characters do not 
behave in this way, and that some of them are recessive to the original characters 
as seen in the long hair of the guinea pigs. 
? Progressive characters are mew characters. Retrogressive characters are 
those that have disappeared, z.e. become latent. Degressive characters are la- 
tent characters that have become active. 
