EK. V. McCollum and N. Simmonds 61 
an animal in a serious pathological state, but, when furnished 
from the beginning of the feeding experiment, may extend the 
period of growth or maintenance and delay the appearance of 
signs of failure. 
In conducting the work reported in this paper, we followed the 
plan of feeding a diet of purified food substances together with 5 
per cent of butter fat to supply an abundance of the fat-soluble 
A. This diet was complete except that it was free from the water- 
soluble B. The rats were confined to this food mixture until 
they either had become stationary in weight or were declining. 
By the 5th week nearly all were either stationary in weight or 
were failing and they almost invariably showed signs of paralysis 
at about this time. When the rats were thus prepared, the 
material to be tested for the water-soluble B was put into the 
diet. The animals then either continued to decline or responded 
with growth. This method served to show within 2 weeks 
whether the substance B in significant amount was in the prepa- 
ration under investigation. This procedyre makes the test de- 
cisive. It also greatly shortens the time required to make the 
test. 
The scheme of applying successively to raw navy beans the 
solvents ether, benzene, and 95 per cent alcohol, in the order 
named, and of feeding in separate experiments the extract and 
residue in the case of benzene and alcohol showed that benzene 
does not remove the water-soluble B from ether-extracted beans 
while hot alcohol does. The extraction with hot alcohol is 
not complete in a Soxhlet apparatus in 18 hours. 
A surprising result was observed when the material which was 
dissolved from beans by hot alcohol was deposited on dextrin and 
the latter then extracted with hot benzene. The benzene-soluble 
material obtained by this procedure was very effective in inducing 
growth and in causing the prompt recovery of rats in the poly- 
neuritic state. A benzene extract prepared directly from ether- 
extracted beans, however, does not contain an appreciable amount 
of the substance B. Benzene, therefore, does not remove the 
water-soluble B from beans directly, but once this substance is 
extracted by alcohol it is soluble in benzene. 
Acetone does not extract the water-soluble B when the solvent 
is applied directly to the ether-extracted beans, and extracts only 
traces when applied to the alcohol-soluble matter from beans. 
