THK PHENOMENA OF REVERSION 317 



while such stripes are very seldom present in the horse and ass, 

 and are even then only very faint ; but we must nevertheless 

 suppose them to be derived from the ancestral form of the two 

 species. And, again, when certain races of pigeons are crossed, 

 offspring are produced in which the plumage has the slaty-blue 

 colour of the wild rock pigeon, although the races used for 

 crossing were of quite another colour ; in this case, moreover, the 

 descent from the wild rock pigeon is certain. Similar instances 

 also occur in plants. The hybrids of Datura ferox and D. 

 Icpvis, in both of which the flowers are white, regularly bear 

 blue (purple ?) flowers, and Darwin * has shown that this is to 

 be looked upon as a reversion to ancestors which possessed blue 

 flowers, as, in fact, is the case at the present day in an entire 

 group or species oi Datura. 



I will now attempt to explain these three instances in accord- 

 ance with my theory. In them, as is universally the case, rever- 

 sion must be attributed to the presence of old unmodified 

 determinants in the germ-plasm, which take the place of the 

 younger homologous determinants as regards obtaining the 

 control of the cell or region of cells in question. Similar 

 assumptions must be made in every theory of heredity. In his 

 theory of pangenesis, Darwin makes use of old gemmules for 

 this purpose, while de Vries assumes that reversion is due to the 

 presence of old pangenes. Some unmodified portion or other 

 of the hereditary substance must always form the starting-point 

 in attempting to explain the problem ; and the only question is, 

 whether we are to remain satisfied with such a statement, and 

 leave everything else in obscurity, or whether it is possible to 

 obtain a certain insight, in principle at any rate, into the question 

 as to why these minute parts can remain unmodified, and why 

 and under what circumstances they can suddenly obtain control 

 precisely in the region of transformed parts which are homol- 

 ogous to them. 



A solution of the first of these problems has already been 

 given above. It has been shown that, according to the principle 

 of selection which controls the whole, a transformation such as is 

 required by the vital conditions of a species never necessitates 

 the transformation of all the determinants relating to the parts 

 to be transformed, but that this process is only necessary in the 



* Darwin, ' Animals and Plants under Domestication,' Vol. II, p. 254. 



