878 Report oF THE HorTICULTURAL DEPARTMENT OF THE 
of the seed as a whole. Observations on this point showed, as 
has already been stated, that the specific gravity of the shell-like 
covering of the grape seed is about 1.35. 
Now as to the relation of the mass of the testa to the mass of 
the inclosed contents in seeds of different sizes: A number of 
observations were made, from which it appears that the thickness: 
of the testa of a small seed varies very little or none at all from 
that of a large seed. From this observation a conclusion of 
practical importance is deduced, namely, that the higher range 
of specific gravity observed uniformly to obtain in the case of 
small grape seeds than of large ones is due simply to the larger 
proportion of testa in the small seeds as compared with the large 
ones. For in the case of the small seeds the volumes of the 
kernel decreases more rapidly than does the volume of the seed 
coat, Since the reduction in diameter falls mostly on the kernel, 
this leaves an increasingly larger proportion of testa. 
Air-dry grape seeds of the highest specific gravity, say from 
1.13 to 1.16, reveal no air spaces between the kernel and the seed 
coat, but beginning at about 1.12 and from there down a slight 
shrinking away from the seed coat is noticeable. This shrink- 
ing uniformly occurs first on the chalazal side of the seed. Ata 
specific gravity of one or two one-hundredths less a separation 
appears on one side of the seed also. At a somewhat lower spe- 
cific gravity still, a separation appears on both sides. As the 
specific gravity of the seed decreases more and more, the volume 
of unoccupied space within the seed coat increases. It appears 
then, from these structural differences that, even if the kernel 
of the seed had in every case the same composition, nevertheless 
these differences in specific gravity of the seed would appear. 
Further, given seeds all of the same volume, it appears that the 
greater the amount of unoccupied space within the seed, the less 
must be the volume of the kernel and presumably therewith the 
less the mass of reserve material. Seed selection then, under 
these or similar conditions, would not in reality be selection 
according to specific gravity at all in the common understanding 
of that word, but would simply be selection according to size, 
and differences in vigor of the resulting plant would have to be 
attributed to differences, not in specific gravity, but in size. Un- 
