RESEARCHES ON THE DEVELOPMENT OF STARCH-GRAINS. 299 
spindle shaped, form on their surface one starch-grain, or 
sometimes two or three, which is at first conical; when it 
has attained the thickness of the corpuscle, which has in the 
meantime become rod-shaped, it increases in size almost 
solely in a plane which is parallel to that of the corpuscle. 
It appears from this that not only does the unequal growth, 
which produces the excentric structure, depend upon the 
mode of nutrition of the grain, but also the unequal in- 
crease of its diameters which produces the flattened form. 
The layer of the corpuscle which immediately invests the 
grain is delicate, and is more or less swollen (fig. 41). The 
further behaviour of the corpuscles is similar to that de- 
scribed in other cases; they become less dense and stable, 
are reduced to a small swollen gelatinous residue, and 
finally disappear. 
The formation of starch in the parenchyma of the young 
tubers (figs. 37—40) takes place in essentially the same way 
asin the root. I have not succeeded in observing the cor- 
puscles in this case before the appearance of the starch- 
grains: the apices of the minute spindles already contained 
them. Both corpuscles and grains increase in size and 
become much larger than in the root.!. Their further beha- 
viour will be described in the following section: 
The starch-grains in the rhizome of Canna gigantea (figs. 
46—49) are very large, triangular, and flattened; they are 
excentric, and are either simple or partially or entirely com- 
pound, consisting of a few, seldom more than ten, small 
grains which are usually arranged in a row. 
The starch-forming-corpuscles resemble those of Amo- 
mum Cardamomum in their development, and at first in 
their form also; they differ from them only in that they 
usually contain a tabular crystalloid, which is either octa- 
hedral or cubical, and which only becomes apparent on treat- 
ment with water. The first stages in the development of the 
starch-grains are the same as in Amomum; they are formed 
excentrically or even superficially in the corpuscle, and there 
may be one, two, or three; they have at first a rounded form 
which is somewhat flattened at the point of attachment. 
The corpuscles now behave very differently from those 
of Amomum ; they grow in one direction only and acquire 
an elongated form. ‘The starch-grain grows, as in Phajus, 
in a plane which is parallel to that of the corpuscle which 
formed it, and the hilum lies towards the free end. The 
crystalloid lies in a projecting portion of the corpuscle. 
The further behaviour of the corpuscles is the same as it 
! This is weil seen in sections which have been hardened in alcohol, 
