THE THREE CONDITIONS OF THE SEED ^5 



germination, whilst those of the leguminous Guilandina 

 bonducella treble their weight. But this cannot take us very- 

 far. A host of questions present themselves as we run 

 our eyes down the lists ; and when we endeavour to answer 

 them ofF-hand, a long vista of undetermined influences 

 opens up. It would be easy to discover differences and 

 to formulate distinctions ; but they would have little or no 

 meaning now ; and it would be futile at present to base 

 any general contrast, such as between families or between 

 genera, on these data. (As already remarked, indehiscent 

 fruits of the type found in palms and in genera like 

 Prunus and Sparganium do not come within the limits 

 of this discussion, but are referred to in Note 20 of the 

 Appendix.) 



We have yet to appreciate the value and to estimate the The diffi- 

 significance of such distinctions. Behind the varying behaviour subject°^*^^ 

 of the shrinking and the swelling seed lie the seed's life- 

 history and the cumulative effect of a multitude of inter- 

 relations as between the seed and the embryo, between the 

 kernel and its coverings, between the coverings and the 

 fruit, and, through the fruit, between the seed and the 

 parent and between the plant and its environment. It is, 

 therefore, obvious that we can only with some security ignore 

 the past when we have reason to believe that the seeds 

 are akin in their history ; and that is why it will be wise at 

 present to mainly confine our discussion of principles to the 

 Leguminosae. But even here it will soon be evident that 

 the risks increase as the affinities grow less, and the safest 

 road will often lie in the study of the varying behaviour of 

 seeds of the same species. 



Proceeding on these lines, we will inquire into the constancy The con- 

 of these ratios in the same species, a necessary preliminary shr"nkin^ *^^ 

 consideration, since' such estimates would lose much of their ^nd swelling 



, r . • r 1 1 . ratios in the 



value tor comparative purposes it the normal range is great, same species. 

 But few of the " shrinking " results admit of bein^ stated 

 in this fashion, as it was my wont in most cases to weigh 



