The Fanycncs. 04') 



of the hybrid to hve, and in the second place the fcriiliiy 

 of the hybrid itself. If, however, they are fertile, the 

 unpaired characters probably sinii)ly divide in the primary 

 hybrids, at the moment of sexual reproduction, in a vej^e- 

 tative way, and this would explain the constancy of such 

 hybrid races. 



Progressive mutations are due to the formation of 

 new pangenes. Dissimilar entities arise in the idioplasm 

 instead of only similar ones, and this is the process 

 which we have called pre-mutation. The pre-mulalcd 

 pangenes tend to be inactive at first, either because they 

 do not exist in sufficient numbers or for other reasons. 

 Obviously it is very probable that in different si)ecies 

 similar pangenes may lead to the origin of the same new 

 pangenes ; and this might perhaps explain many phencjni- 

 ena of parallel progressive mutability. 



Lastly we must suppose that the pangenes, or grou])s 

 of them in each of the conditions referred to, may l)e 

 more closely or more loosely associated with the remain- 

 ing ones. If the association is a close one it will remain 

 the same through all generations, and the species or 

 variety is immutable with regard to the character in 

 question. If the equilibrium is an unstable one, the 

 character in question is mutable; and slight external in- 

 fluences may turn it into a stable c(Mi(lition and thus in- 

 duce the visible mutations such as those of the Ocuothcras. 

 Unfortunately, however, the nature of these inlhuMu-i's 

 is still unknown. The stable condition, which in this way 

 arises out of the mutable one, can be either active or 

 latent. 



It would be easy to extend this discussion further: 

 suffice it, however, to say that the relation of pangen- 

 esis to new discoveries is evervwhere more or less ol>- 



