262 Maternal Characters in Seeds [ch. 



coloured and uncolpured in the proportion 3:1, the coloured 

 all have indent seeds and the white-flowered all have round 

 seeds. Segregation is therefore normal, and the fact that 

 all the F^ seeds are indent is in some way brought about by 

 the nature of the maternal envelopes in which the seeds 

 develop. How this influence is exerted we cannot suggest, 

 but perhaps there is some quality in these seed-coats which 

 causes the loss of water on ripening to take place irregularly 

 and so induces an irregular shrinking of the cotyledons. 

 Such an account is difficult to apply, for the seed-coats seem 

 uniform and homogeneous, and as the next case proves, the 

 influence of the seed-coat, whatever it be, operates in an 

 extraordinarily capricious and specific manner. 



Until recently the account given above was supposed to 

 apply to all crosses of round with indent. Going over 

 seeds which were harvested in 1904 and 1905 without 

 examination at the time I find that one case is altogether 

 different from the rest. In this the round parent was 

 Nain de Bretagne, a small, very round, white-flowered 

 variety. Indent fertilised by this variety gave F^ seeds 

 indent. These when sown became coloured F^ plants, but 

 their /% seeds, instead of being all indent as in the examples 

 described, are quite definitely indents and rounds, in the 

 usual 3 : I ratio'*. Moreover among the few F^_ plants of 

 which the seeds were harvested, was one which had only 

 round seeds (in coloured coats) and it is evident that this 

 plant came from a round F^ seed which was sown without 

 note being taken of its character. 



We are thus presented with the exceedingly definite and 

 specific fact that one round-seeded variety, by virtue of its 

 intrinsic nature, behaves quite differently from the others 

 that have been tried. Presumably its reserve-materials 

 (and by inference the ferments which lead to their formation) 

 have some distinctive property such that though ripening 

 in a seed-coat which would make the other round peas 

 shrink as indents, they are still able to retain their own 

 characteristics. Such a fact may well be remembered in 

 any discussion of the nature of specificity. 



* Several were intermediate in appearance, as in such cases must be 

 expected. Three plants, for instance, had 339 indent, 119 round, and 

 39 uncertain. 



