420 Scientific Proceedings, Royal Dublin Society. 



is partially eliminated from the wings, and they are light ; where y appears 

 it is not : only in tlie case of the third group, which is dilute, this elimination 

 is not clearly visible. 



In the whole case there are thus three pairs of factors concerned, viz. : 



P, Purple and its recessive r, red. 



D, Densing and its recessive d, diluting. 



E, Eliminating colour from the wings, and its recessive e, non-eliminating. 



Their second crosses will be as follows : — 



P D E = Dense purple with light wings 27 



P D e = Dense purple with dark wings 9 



P dE = Dilute purple with light wings 9 



r BE = Dense red with light wings 9 



r d E = Dilute red with light wings 3 



r B c = Dense red with dark wings 3 



P d c = Dilute purple with dark wings 3 



r d e = Dilute red with dark wings 1 



Thus there ought to be eight groups ; and it is very probable that only 

 six were found because the dilute purples and reds with dark wings were not 

 distinguishable by the eye from those with light wings. 



Several minor suggestions might be made : first, that factors may not drop 

 out and leave none in their place, or — which is the same thing — come in 

 without displacing others. In connection with sweet peas it has been 

 suggested that the numerous cultivated varieties have arisen from the wild 

 " by a process of continuous loss." In the table above, the variety at the top 

 (^FBE) is the same in colour as the wild Sicilian form. It is suggested that 

 one or more of the factors PB and E have dropped out and given us our 

 cultivated varieties : for instance, that E dropped out of the wild Sicilian 

 variety and gave us a purple variety with dark wings. That may be in part : 

 it may be that E dropped out, but when it did so, another factor, in this case 

 its recessive, e, took its place. 



A tendency has been manifested on the part of some workers, particularly 

 in America, to take it for granted that every observable character must have 

 an " absence." It may be suggested now that there may be danger even in 

 the narrower assumption that every character has an alternative. It is 

 certainly wrong to assume that every dominant character can have only one 

 recessive and every recessive only one dominant. 



It might also be suggested that the use of the words epistatic and hypoHtatic 

 might be revised. They are used to indicate the relative positions of the 



