IN THE TROPICS. 
409 
coloured seeds were practically smooth in surface (or, on the other 
hand, quite wrinkled), though usually more or less square in outline. 
Even in the case of the largest seeds dimpled and wrinkled were 
usually quite distinguishable ; and the point could be settled by 
cutting sections, when dimpled seeds were found to have starch 
grains of the “round” type. In other cases where the eye alone 
could not distinguish dimpled from wrinkled with certainty, the 
microscopic test was found to be conclusive. 
Among the seeds with a white or nearly colourless testa, smooth 
and wrinkled cotyledons appeared in accordance with Mendelian 
expectation.* The above remarks apply without exception to F 2 
and to all the seeds in E 3 which could be dried. Many of the E 3 
plants were taken up in very wet weather, and in such cases it was 
impossible to discriminate between the shapes of the seeds. 
The colour of the testa appeared to follow perfectly definite rules, 
some account of which has already been given in a previous paper 
(page 333), but the description needs elaborating in some respects. 
Twenty-five plants were raised in E 2 , their seed-coat colours being as 
follows :— 
Eleven plants (m. p. f.) showed testas coloured like No. 2, except 
that in some cases the ground colour was yellowish like that of the E t 
plants, whilst in others it was greenish like the parental strain, f 
The purple spots and the 4 ‘ maple 5 5 marking was clearly visible in 
all these cases. 
Six plants (p. f.) resembled these in the ground colour of the testa 
and in the presence of purple spots, but the maple marks were quite 
wanting. 
In two others (m. f.) the purple spots were wanting, but the maple 
marking was obviously present. 
In two more (f.) both purple spots and maple marks were wanting, 
but the testas were obviously greenish and opaque. 
* In F 2 all 4 plants with white testas had seeds exclusively smooth— 
DD—when the expectation was 1 DD : 2 DR : 1 RR. This was probably 
accidental, since in F 3 , among the offspring of dimpled seeds, those with 
white coats showed the expected number of wrinkleds. 
t The relative numbers of the two kinds were not counted in this or the 
next generation, and I do not think that they were always distinguishable. 
