THE SHRINKING AND SWELLING SEED 223 



that of leguminous seeds is brought out for purposes of illustration in 

 the case of the seeds of the two species of Ipomcea^ which represent the 

 convolvulaceous regime (p. 207). 



(16) The special difficulty presented by the seeds of Hura 

 crepitans is next referred to. These seeds follow the fashion of some 

 indehiscent fruits, like those of the Oak (Quercus) and of the Coco-nut, 

 the seed-coverings losing considerably in weight before the kernel 

 attains its mature' size (p. 208). 



(17) The author then utilises a large amount of data obtained for 

 seeds where the swelling phenomena of the coats and kernel were alone 

 observed. The results for forty-four species are thus employed, of which 

 all but ten are leguminous ; and it is remarked that although in about 

 two-thirds the kernel has a larger swelling ratio than the coats, there is 

 no very determinate result in this direction, genera behaving sometimes 

 consistently and at other times being divided in this respect (p. 209). 



(18) In order to wrest their story, another arrangement of these 

 "swelling" data is adopted, from which it appears that whilst with 

 the coats the swelling ratios, as respects impermeable, variable, and 

 permeable seeds, are not far apart, with the kernels the ratios for all 

 three types of seeds difter markedly from each other, the impermeable 

 seed showing the largest, the permeable seed the smallest, and the 

 variable seed a ratio intermediate in amount (p. 210). 



(19) It is then elicited that this behaviour of the coats and the 

 kernel in these three seed-types is to be connected with the water- 

 percentage of the resting seed, but less in the case of the coats, where 

 the differences are small, than with the kernels, where the differences 

 are large, the kernel of an impermeable seed holding on the average 

 not much more than half the water held by the kernel of a permeable 

 seed and possessing a much greater swelling ratio. With the kernels 

 of variable seeds, where the swelling ratio is intermediate in extent, the 

 water-percentage is also intermediate in amount (p. 212). 



(20) Two other points are then referred to, the first being that in 

 seeds with oily kernels, where the unusually low water-percentage 

 indicates that the oil supplies the deficiency, the swelling ratio, as in 

 the case of the seeds of Ricinus^ is also unusually small. The second 

 is the decrease in the relative weight of the coats as the seed ripens ; 

 and in this connection it is remarked that just as with fruits the growth 

 of the pericarp is always in advance of the seed-growth, so with seeds 

 the growth of the coats is always in advance of the kernel (p. 214). 



(21) Having dealt with the shrinking and swelling processes of the 

 kernel, we now differentiate in the case of dicotyledonous albuminous 

 seeds between the two components of the kernel, the embryo and the 

 albumen. The role of the embryo in this connection is now studied. 

 For this purpose only seeds with embryos possessing large, flat cotyledons, 



