878 JOURNAL OF THE ROYAL HORTICULTURAL SOCIETY. 



" It has been stated by nearly all investigators that there is a tendency, 

 in the second and later generations, to revert to the parent forms. It 

 seems possible that there is a more accurate way of stating this. The 

 types that tend to occur in the second generation, as indicated by our 

 results, include all possible combinations of the characters of the two 

 parents. This, of course, includes the parent forms themselves, and we 

 find the parent forms repeated in the second generation, constituting, 

 apparently, certain definite portions of this generation. 



" Another important fact that is clearly revealed by the tables of 

 percentages is that the type that is most abundant in the second genera- 

 tion is the* same as the first generation type, whether the latter is of the 

 usual intermediate type or otherwise (see below). The exceptions to this 

 were so rare as to render them doubtful. 



" It was stated above that the first generation tends to be the same in 

 similarly bred hybrids, and is intermediate between the parents. Let us 

 consider the exceptions that occurred. In eleven of the fourteen crosses 

 there were no noticeable exceptions. In one case one of four hybrids 

 (first generation) resembled the male parent closely, and 66 per cent, of 

 the next generation was like it (Table III.) In another case, one out of 

 nine in the first generation differed from its fellows only in having no 

 velvet on the chaff. In the second generation there were, in general, 

 twelve types in these hybrids (Table XI.), six with velvet and six without. 

 The plant that was devoid of velvet in the first generation produced only 

 the six types without velvet in the second generation. In another cross, 

 which was the reciprocal of the last-mentioned (Table XII.), nine out of 

 twenty-seven hybrids were more or less irregular in one or both genera- 

 tions. Five of the first generation were more like the female parent, and 

 one more like the male parent than usual. In each of these cases only a 

 few of the possible types appeared in the second generation, and the type 

 most like the previous generation was in great excess. In two other cases 

 the only irregularity was the unusual preponderance of the first genera- 

 tion type in the second generation (Plats G19 and G22), and the absence 

 of several of the possible types. In five plats of this breeding bearded 

 forms appeared, though neither parent had beards. In every case the 

 beards appeared on heads having the form of the parent belonging to T. 

 vulgare. This probably indicates the presence of bearded parents in the 

 ancestry of the variety used [Farquahar]. 



" Three of the fourteen crosses were reciprocals of three of the others ; 

 e.g. in one case Farquahar pollen was used on Little Club (Table XL), and 

 in another, Little Club pollen on Farquahar (Table XII.) Other cases are 

 shown in Tables I. and IX., and Tables IV. and XIII. The same types 

 appeared whichever parent furnished the pollen, and in approximately the 

 same proportions. 



" While the results here reported are not sufficient to justify the positive 

 assertion that certain quantitative laws govern the transmission of parental 

 characters to hybrid offspring, yet they point so strongly in this direction 

 that we may state some of these laws provisionally, looking to future 

 investigation for their confirmation, modification, or rejection. 



" That similarly bred hybrids tend to be alike in the first generation, and 



