THE SHRINKING AND SWELLING SEED 193 



Anona muricata, number of seeds tested 1 2, range of seed-coat ratios 

 31 to 38. 



Canna Indica^ number of seeds tested 10, range of seed-coat ratios 

 18 to 25. 



With regard to the influence of variation in size as inter- 

 preted by weight on the seed-coat ratio in the same species, I 

 am not able to give many data, as my mode of work rarely 

 admitted such a comparison. The following results for 22 The influence 

 seeds of Guilandina bonducella indicate that there is no great 

 difference. The range of the variation in weight was from 

 33 to 45 grains, the proportion of the seed-coats being slightly 

 greater in seeds above than below 40 grains. Thus : 



Above 40 grains in weight, average seed-coat ratio 58-5. 



Below 57'5- 



However, carefully guarded observations on a large 

 number of seeds of the same age and from the same plant are 

 necessary for the elucidation of this point. 



As respecting the influence of appendages on the proper- The influence 

 tional weight of the seed-coverings in resting seeds, we will deal 

 first with hairs and then with wings. In the table subjoined 

 there are given the data for eight kinds of hairy seeds belonging (A) Hairs, 

 to three families, Asclepiadeae, Malvaceae, and Convolvulaceas, of 

 which the two first are especially notable for the hairiness of the 

 seeds in some genera. The results are arranged in order accord- 

 ing to the proportional weight of the hairs, commencing with the 

 seeds where the relative weight is smallest. It is evident that 

 ordinary pubescence as illustrated in the case of Ipomaea pes-capr<e 

 adds but little to the weight of a seed. Nor does it make much 

 difference if a pubescent or puberulous seed, as in the case of 

 Ipomasa tuba, is bordered by longer hairs at the angles ; but when 

 these hairs are abundant and woolly, as in the seed of If omasa 

 peltata, the hairy covering may make up 9 per cent, of the total 

 weight. The proportions for these convolvulaceous seeds are 

 probably typical of a good many seeds of other families. The 

 great development of hairs that we find in some Asclepiads and 

 in some malvaceous genera is not common in the plant-world. 



13 



