686 Wisconsin Academy of Sciences, Arts, and Letters. 
corrosion channels have penetrated but a few layers, there ap¬ 
pear irregularities along the walls of the channels; but no good 
evidence could be secured as to whether the light or the dark 
layers were more susceptible to diastase action. 
A noticeable fact in connection with these corroded grains 
is that the portions of the layers bordering directly on the 
corrosion channels show a margin of orange-stained material 
which blends gradually into the violet of the unaffected por¬ 
tions. 
Hrabbe’s observations led him to believe that the sub¬ 
stance of the starch grain is removed, molecule by molecule, and 
that there is no general penetration of the grain. He used 
iodine as a stain and found that the parts of the grain remain¬ 
ing showed no difference in staining properties from the intact 
grain. His results, when compared with the conditions ob¬ 
served in Figure 28 ? show how little reliance can be placed on 
observations of corroded grains in water, even when stained 
with iodine. 
Flemming’s triple stain shows the borders of the corrosion 
channels plainly differentiated in corroded Canna grains; the 
material bordering the canal takes the orange stain. Iodine 
is not a good differential stain, and does not show slight dif¬ 
ferences either in the composition or the structure of the 
starch grains. "Whether or not there is a penetration of dias¬ 
tase in all cases beneath the surface of the corrosion canals, 
there is plainly, in the case of Canna starch, a transition layer 
in all surfaces which are being corroded. The presence of 
orange-stained material over the entire corrosion surfaces of 
the grains suggests very strongly that a substance is found at 
the time of solution of the grain similar to that present on 
the surface of grains which are being formed. 
The experiment was tried of crushing large Canna grains 
which had previously been stained by the triple stain. The 
crushing was effected by pressing on the cover of a freshly 
made slide with an eraser before the balsam had hardened. A 
number of deep radial cracks running from the surface in¬ 
ward are formed in this way. This fact probably has no sig¬ 
nificance in determining the finer structure of the starch 
