92 STUDIES IN SEEDS AND FRUITS 



tions before proceeding to describe in detail the results of my 

 investigation of these seeds from this particular standpoint. 



One of the first noticeable things in these groups is the 

 Thepre- predominance of leguminous seeds in the two groups of 

 leguminous s P ec i es possessing impermeable seeds exclusively or in part, 

 plants in the This has already been established by Dr Gola and others, but 



groups J J 



characterised principally by Professor Ewart from a more extensive experi- 

 ableeeds. e ence ; and one may be content with showing here how this pre- 

 dominance is suggestively emphasised in my own less numerous 

 results. Of the 105 species of seeds named in these groups, 44, 

 or about two-fifths, are leguminous ; and of these three-fourths 

 are claimed by the two groups containing plants with impermeable 

 seeds. If the quality of impermeability was equally distributed 

 amongst the natural orders, we should expect to find leguminous 

 plants and plants of other orders about equally represented in 

 the total of the impermeable and variable groups. As a matter 

 of fact, however, this is far from being the case, since as many 

 as 82 per cent, in the groups just named, and as few as 19 per 

 cent, in the permeable group, are leguminous. 



Various points in this classification of seeds will be eluci- 

 dated as this work proceeds. 1 will, however, notice that the 

 distinction between the impermeable group, where all the seeds 

 are impermeable, and the variable group, where both kinds of 

 seeds are found, is often rather arbitrary. In truth, it would 

 almost appear from Professor Ewart's tables that very few plants 

 have seeds that are impermeable without exception. But careful 

 consideration has convinced me that the real distinction between 

 the two groups ought to lie not so much in the total absence or 

 in the presence of permeable seeds, but in the degree of constancy 

 of the characters within a species. Without some familiarity 

 with the seeds in question one would be often likely to in- 

 clude amongst impermeable seeds some that have lost their 

 imperviousness through a slight defect developed during the 

 shrinkage process. With typical impermeable seeds there always 

 will be a very high percentage of seeds normal in this respect ; 

 whilst with typically variable seeds there will be, as a rule, con- 



