SPECIES CROSSES IN THE GENUS NICOTIAN A. 197 



inheritance before they are prepared to accept de Vries' 

 assertion.* 



Now the statistical behaviour of the offspring of species 

 crosses is a subject upon which practically nothing is known. 

 Gsertnsr and others have recorded the observation that in 

 a certain number of instances the hybrid form breeds true in 

 the second and third generations. But this proves nothing, 

 since it is well known that, even in cases where complete segre- 

 gation takes place, the new combinations which may arise 

 between the different characters concerned breed true in a 

 definite proportion of the plants which exhibit them, so that, 

 in the absence of large numbers of individuals, no definite 

 conclusion could be dra^vn from such an observation. It is 

 in fact impossible to prove the absence of segregation in cases 

 where the first hybrid generation is markedly sterile, with 

 the result that only a small proportion of the germs give rise 

 to mature offspring, because it is always possible to suppose 

 that the power of development is coupled with the particular 

 combination of characters shown by the original hybrid— a 

 combination which we already know to be capable of develop- 

 ment, and, in the conditions supposed, the only one of which 

 so much can be said. On the other hand, in the cases of less 

 sterile species hybrids, where a considerable number of indi- 

 viduals could be raised in F2 and onwards, it has been noticed 

 that whilst the majority of the individuals in the second 

 generation resemble the original hybrids more or less closely, 

 other individuals show considerable deviations in the direction 

 of one or the other parent form. This is precisely what would 

 be expected in the case of a cross between two forms which 

 differ in a considerable number of segregating characters, when 

 only a limited' number of individuals is available in the second 

 hybrid generation. 



One of the first objects of the present experiments there- 

 fore was to discover, if possible, a non-segregating hybrid 



* Since writing the above, I find that Prof. Bateson in hig inaugural 

 lecture on Genetics (Cambridge, 1908) accepts as a definite exception 

 Prof. Castle's discovery that the intermediates produced from a cross 

 between the long-eared lop rabbit and a short-eared variety breed 

 appro'cimatelv true. I have not had access to this work of Prof. 

 Castle's. 



