XIV] Abseiice (^f Segregation 251 



organs of the fertile stems, and the mixed pollen of those 

 stems. Janczewski concludes that in this way a new specific 

 type could arise which, but for its mixed pollen-grains, 

 would be indistino'uishable from a natural and oricrinal 

 species. 



This extreme irregularity and sportiveness in the degree 

 of fertility manifested is a frequent concomitant of partial 

 sterility in hybrids. The fact that the sterility once lost 

 does not return, obviously suggests that it is due to the 

 presence of some dominant factor or factors, and that they, 

 or the compound they make, can be eliminated by bud- 

 variation. Other reflections will occur to the reader which 

 we cannot now consider, but I feel it difficult to accept 

 such an example without further comment as a proof that 

 in ordinary sexual reproduction segregation may fail as a 

 normal event. 



It is most desirable that search should be made for 

 genuine cases of failure of segregation amongst animals. 

 The example of the human mulatto is as yet one of the 

 best which ordinary experience suggests"^. In experimental 

 breeding perhaps the single clear instance is provided by 

 Castle's experiment (48) on the lengths of ears in rabbits. 

 He crossed the long-eared lop rabbit with ordinary short- 

 eared individuals. F^ had ears of intermediate length, and 

 the subsequent generations retained this character very 

 definitely. The later details have not yet been published, 

 but Professor Castle very kindly showed me many of his 

 records which plainly established the fact that no return to 

 the parental types took place. 



In some experiments made with the butterfly Paj^ar^e 

 egeria I obtained a comparable result. In Spain and the 

 plains of Southern France ege^na has a bright fulvous brown 

 colour (intersected wdth darker brown), while in England 

 our representative form egeriades has the bright brown re- 

 placed by primrose yellow, In the Loire valley and in 

 Brittany a race almost exactly intermediate between cgeria 

 and egeriades is tound. By crossing the two extremes the 

 intermediate type was produced with great exactitude. /% 

 was not raised in any adequate numbers, but those that 



* See p. 208. 



