486 JOURNAL OF THE ROYAL HORTICULTURAL SOCIETY. 



This shows the necessity for caution in ascribing irregularities of seed 

 shape to failure of dominance without first eliminating the influence of 

 the conditions of life. It is obvious that size and weight, when considered 

 as attributes in heredity, must necessarily be unreliable amid the 

 changing conditions of life, yet it may be noted, with all due reservation, 

 that in the parent seeds 6 of the largest ' British Queen ' just, balanced 

 16 of the largest ' Eclipse,' while 5 of the largest hybrid seeds balanced 

 the same 16 ' Eclipse.' As size and weight appear to correspond fairly in 

 seed Peas, this suggests that in hybrid Peas size is a purely maternal 

 character, unlike shape, which is a hybrid character. Professor Correns 

 has arrived at a similar conclusion in regard to the size of hybrid seeds of 

 Maize (1901). 



Second Generation (F 2 ). 

 DR x DR 



In 1903, 14 of the dominant round hybrids raised in Experiment 1 

 were duly sown : these produced 12 hybrid plants which, self -fertilised, 

 gave 1,755 seeds in the second generation. These seeds were distinctly 

 of two types, round and wrinkled, both types being often produced in the 

 same pod ; some pods contained all round, but none contained all wrinkled. 



The following is a list of the numbers of each type produced by the 

 12 individual plants : 



D R 

 Expt. Round. Wrinkled. 



2 27 6 



3 32 4 



4 24 11 



5 378 125 



6 15 5 



7 135 36 



8 62 12 



9 203 52 



10 287 107 



11 85 30 



12 72 25 



13 15 7 



1,335 420 



Actual ratio 3-1 D : 1 R 



Mendelian ratio 3D : IB 



Oat of the 1,755 seeds of the second generation 1,335 were round 

 dominants, and 420 were wrinkled recessives, giving a ratio of 31 D : 1 R, 

 which is close to the 3 : 1 expected by Mendel's law. It will be observed 

 that in individual plants where the numbers are few the ratios fluctuate 

 considerably, but where the numbers are 100 and over the ratios are on 

 the whole fairly constant. This shows the necessity of dealing with large 

 numbers in all quantitative experiments in heredity, otherwise the ratios will 

 be irregular. On the other hand, a few numbers will suffice for qualita- 

 tive experiments, as may be seen in the above experiments ; a dozen seeds 



