452 Prof. J. B. Farmer. On the [May 13, 



be regarded as the immediate characters themselves; they may, perhaps, 

 be roughly compared with that class of bodies known as ferments, setting up 

 in the extra-nuclear cytoplasm those changes which, in their sum-total, 

 constitute development, and producing those chemical re-arrangements of 

 which form, colour, etc., are the visible expression. 



It will be convenient, in the first place, to examine the results obtained 

 from experiments in cross-breeding, a field in which Bateson and others have 

 achieved so great a measure of success. We can afterwards try to relate these 

 results with the salient facts of nuclear structure which have been elucidated 

 within recent times. 



If two pure parents which differ in, say, one character from each other be 

 chosen and mated, the offspring almost always behaves in a definite and 

 uniform manner with regard to the character in question. Often it appears 

 as if they all inherited the character of one parent only, that of the other 

 having been entirely lost. But on allowing the hybrids to interbreed, it soon 

 becomes plain that the alternative character was only latent, and not lost at 

 all. It reappears in a fixed percentage of the offspring, usually to the extent 

 of 25 per cent. When these individuals, thus exhibiting the recessive 

 character, are mated, they continue to breed true, and the alternative 

 dominant character, which replaces the recessive one in 75 per cent, of the 

 individuals of that generation, never reappears amongst the recessive progeny 

 so long as they interbreed only with one another. 



As regards the individuals which make up 75 per cent, of the second 

 (the f2) generation, it is found that whilst all of them show the dominant 

 character exclusively, only one-third continues to breed true, whilst two-thirds 

 of them, when mated together, prove to be hybrids ; and in the next generation 

 these continue to throw out one-quarter pure dominants, one-quarter pure 

 recessives, and one-half hybrids. This is a remarkable result, and although it 

 is not always directly or obviously obtained, it occurs in so large a number of 

 instances as to prove sufficiently that the characters in question do behave as 

 independent units. It is not surprising that there should be apparent excep- 

 tions to the rule ; some of them have already been shown to be due to an 

 incomplete analysis of the real characters themselves, so that composite features 

 have been mistaken for units. But, at any rate, in the presence of so large a 

 number of cases which do conform to the Mendelian rule, they may be safely 

 left on one side so far as our present purpose is concerned, which is to find 

 an explanation of the numerous positive instances that are everywhere 

 recognised. 



It is possible to make certain statements, which almost partake of the 

 nature of inevitable conclusions, from a consideration of the mode of 



