102 L. B. WINTER AND W. SMITH.
which will reduco the alkaline copper solution. Owing to the careful
control of temperature, and the very small concentration of acid, we do
not consider that the method would tend to split off copper reduciiwg
substances from possible gluco-genetic bodies.
To determine how completely the sugar was extracted by the alcohol
employed, the residue left behind after the filtration of the alcohol was
shaken up and extracted with water. The filtrate now was perfectly clear
showing that the remaining protein was not readily redissolved. Copper
reducing determinations and polarimetric readings showed no measurable
amount of sugar present. To test whether the relation between the
polarimetric and copper reducing powers of glucose would be altered by
this treatment, a solution of pure glucose was concentrated in the
presence of tungstic acid employed for precipitation, extracted with
alcohol, and made up with water. The copper reducing power was then,
if anything, slightly lower than the polarimetric value. Possibly the
larger amount of free acid caused slight disaccharide formation(7);
normally all but a small amount of tungstic acid combines with the
protein and is so removed.
The filtrate is invariably slightly acid and therefore most likely to
stabilise the sugar originally present. It soon however became evident
that the instability of the original sugar made it essential that concen-
tration of the aqueous filtrate should be effected as quickly as possible.
This we have always done, and as soon as the final filtrate was obtained
we have made a quantitative comparison of the sugar by polarimeter
and copper reducing power. We have made no attempt to obtain in our
final filtrate all the sugar originally present in the blood. To have
attempted quantitative extraction would have lengthened the process
very considerably. Our sole concern was to obtain the sugar in a form
as nearly as possible approximating to that normally in the blood. Our
experiments from the beginning of protein precipitation to the polari-
metric reading usually occupied five hours for 100 c.c. blood; when less
blood was used the time taken was much less. Of this time only a com-
paratively short period is occupied in what we find to be the dangerous
stage, viz. the concentration of the aqueous filtrate.
The copper reduction method which has been adopted throughout is
that of Bertrand as modified by Wood and Berry(s); it had been
previously determined by us that this gives reliable and consistent
results; the large amount of fluid used and the convenience of a per-
manganate titration being points in its favour. A comparison of this
method with values obtained from the polarimeter was undertaken with