is seriously impeded. Most of those alkaloids which might possibly be
present do not yield the gas when treated in this manner, and
therefore may be disregarded.


SUGAR.

Glucose, so characteristic of _diabetes mellitus_, is not difficult of
detection or estimation. The facility with which it reduces alkaline
cupric, argentic, bismuthous, ferric, mercuric salts, indigo and
potassic picrate and chromate solutions has been utilized for the
preparation of several ready methods for its determination. Trommer's
test consists in adding enough cupric sulphate to color green, then
excess of alkali, and boiling. Yellow to brick-red cuprous oxide forms
as a heavy precipitate if glucose is present. The organic matter of
the urine prevents the precipitation of cupric hydrate on the addition
of the alkali. This test is delicate and deservedly popular. Fehling's
well-known solution contains sodio-potassic tartrate, which serves the
purpose chiefly of retaining the copper in solution. Unfortunately,
Fehling's original solution has a tendency to become hyper-sensitive
if kept long, a proneness to change that is much increased on
dilution. When so altered, the solution will yield a more or less
copious precipitate of cuprous oxide on merely boiling, and quite
independent of the presence of glucose. This decomposition is obviated
by preserving the copper salt in a separate solution from the tartrate
and alkali, and mixing before use. Schmiedeberg substitutes mannite
and Cresswell glycerin for the Rochelle salt, in order to render the
solution stable. Some prepared by the writer over twelve months ago,
according to the suggestion of the latter physician, has since shown
no signs of decomposition, and is now as good as it was then. For