1913} SHULL—SEMIPERMEABILITY OF SEED COATS 181 
sible for this physical character of the membrane, which remains 
unchanged when the cell contents have been removed by dissolution. 
REICHARD (29) some years ago showed that the coats of Hordeum 
are full of tannin, and has now (30) raised the question whether the 
tannin present there has any effect on the semipermeability of its 
membranes. The colloidal condition of the tannin, and its known 
peculiarities chemically, suggested that it might be in a large 
measure responsible for the physical 
properties of the coat. He investigated 
the position of the tannin by micro- 
chemical methods, and found that in 
sections the barley grains showed a 
sharply defined layer of tannin which 
could easily form a continuous coat. 
Treatment of seeds with tannin solvents 
showed that with such treatment the Fic. 5.—Cellular arrange- 
layer of tannin becomes broader and _ ment of a portion of the inner 
more diffused.'- Alcohol (96 per cent) [aver of the testa.—Camera 
re : lucida, 4 mm. obj., 4 ocular, 
used alone does not dissolve tannin. 465 mm. tube 
Following the alcohol treatment with 
iron sulphate solution as a tannin stain shows that the tannin layer 
becomes if anything more sharply defined in alcohol of high grade. 
But if the seeds are soaked first in water, then the tannin shows 
marked solution effects when subsequently treated with alcohol. 
Alcohol seems to dissolve tannin only when it has been previously 
wet. 
According to REIcHARD’s view, the entrance of substances into 
the seed would depend upon their ability to dissolve the tannin 
layer, or to dissolve in it. Since strong alcohol does not dissolve 
tannin, seeds with such a tannin layer might lie in absolute alcohol 
indefinitely without loss of vitality. But the presence of sufficient 
moisture to dissolve or partially dissolve the tannin would allow 
rapid entrance of alcohol, and a consequent killing of the embryo. 
The bearing of these facts upon the resistance of dry seeds to organic 
solvents discussed in an earlier section of this paper is obvious. 
- Hydrates, especially NaOH, have a powerful solvent effect upon 
tannin; even in M/r1o solutions the solvent action is apparent, and 

