892 JOURNAL OF THE ROYAL HORTICULTURAL SOCIETY. 



but that the second generation tends to repeat in Mendelian proportions 

 whatever is shown by the first generation, whether *' Dominant ," Inter- 

 mediate, or " Becessive." 



That is to say, if the individuals of the first generation are of the 

 " Dominant " type, they will reproduce that type in the second generation 

 in the ratio of 3 D : 1 R ; but if they happen to be of the " Recessive " type 

 they will reproduce that same type in the second generation in the ratio of 

 3 R : 1 D ; while if the individuals of the first generation are of the Inter- 

 mediate type, they will reproduce that type in the second generation in the 

 ratio of 1 D : 2 DR : 1 R. 



If future experiments confirm this, it will be an important advance in 

 our knowledge of heredity, and to the practical hybridist the information 

 will be invaluable, as he will be able to select whatever he wants in the 

 first generation, whether it happens to be " Dominant," Intermediate, or 

 " Recessive," and by the aid of the Mendelian Principles he will be able 

 to calculate the result beforehand. 



Having dealt hitherto simply with single characters, and found them 

 to be strictly Mendelian, we now proceed to the much more complicated 

 question of two and three characters taken together. For this purpose 

 Prof. Spillman's tables are admirable and require no analysis ; we can 

 take them as they stand in his original paper. 



Firstly we will take those tables in which two characters are considered 

 together. 



In Tables II., VII., and VIII., varieties with " long heads " and " velvet 

 chaff" were originally crossed with varieties with "short heads" and 

 " glabrous chaff','' the result being thirty hybrids with " Intermediate heads " 

 and " velvet chaff." These thirty hybrids, self-fertilised, produced thirty 

 plats, each containing six types, on the average in the following proportions, 

 viz. :— 19-9 (1 4- v) + 12 (1 + g) + 36-2 (si + v) + 12-7 (si + g) + 17"2 

 (s + v) + 6-8 (s + g). 



Now, if we take the Mendelian formulae for the two single characters, 

 we get for the one 1 1 + 2 si 4- Is, and for the other 3 v + 1 g. 



Now the possible combinations between them would be six, in the 

 following proportions, viz. : — 3 (1 + v) + 1 (1 + g) + 6 (si + v) + 2 (si +g) 

 4- 3 (s + v) -i- 1 (s + g), the approximate percentage of which would be 

 respectively 18 + 6 + 86 + 12 + 18 + G. 



This Mendelian expectation agrees well with the actual percentages 

 respectively of Prof. Spillman's experiments given above, viz. : — 19*9 4- 7*2 

 4- 36-2 + 12-7 + 17-2 + 6*8. 



In Tables IV., XIII., and XIX., varieties with "long" and "bearded 

 heads " are crossed with varieties of " short " and "bald heads," the result 

 being twenty-one hybrids with "intermediate bald heads." These twenty- 

 one hybrids, self-fertilised, produced 21 plats, each containing six types, 

 on the average in the following proportions, viz. : — G'G (1 4- be) + 19'1 

 (1 4- ba) 4- 14-0 (si + be) + 35-4 (si + ba) + G'5 (s + be) + 17*7 (s + ba). 

 The Mendelian formuhe for the two single characters are respectively, 

 1 1 4- 2 si + 1 s and 8 ba + 1 be, and the possible combinations between 

 them would be six, in the following proportions, viz. : — 1 (1 + be) 4- 8 

 (1 + ba) + 2 (fll + be) + G (si + ba) + 1 (s 4- be) 4- 3 (s + ba), the ap- 



