To a practiced operator the indications yielded by the use of this test are of great value; but beginners are exceedingly liable to mistake its various reactions, and to report the urine as saccharine when normal traces only of sugar are present. The bismuth test of Bottger, as greatly improved by Nylander, is fairly delicate, and not so easily misread as Fehling's. A large volume of reagent being used with a comparatively small quantity of urine, the precipitate of earthy phosphates does not interfere in the least with the reaction. On boiling about 3 drachms of Nylander's solution and 20 minims of urine for a minute or two, the liquid darkens with a trace of sugar, and becomes opaque and black if the latter is present in quantity. The reagent is prepared by dissolving 494 grains of caustic soda, 247 grains of Rochelle salt, and 154 grains of subnitrate of bismuth (free from silver) in 13 fluid oz. of distilled water. It should be decanted for use from any sediment. [Illustration: DR. PAVY'S APPARATUS.] In those cases where the amount of glucose present is required to be determined, Dr. Pavy's ammonia cupric process distances all compeers for ease of application and delicacy of end-reaction, combined with considerable accuracy. His solution differs from that of Fehling in containing ammonia, which dissolves the cuprous oxide as soon as it is formed, yielding a colorless solution. It is only necessary, therefore, to note the moment that the blue color of the liquid is exactly discharged, in order to tell when all the copper present has been reduced. Pavy's solution is prepared as follows: Dissolve 356 grains of Rochelle salt and the same weight of caustic potash in distilled water; dissolve separately 73 grains of recrystallized cupric sulphate in more water with heat. Add the copper solution to that first prepared, and when cold add 12 fluid oz. of strong ammonia (sp. gr. 0.880), and distilled water to 40 fluid oz. The estimation is thus conducted: Dilute 10 c.c. of the ammoniated cupric solution--equivalent to 5 milligrammes of glucose--with 20 c.c. of distilled water, and place in a 6 or 8 oz. flask. Attach this by means of a cork to the nozzle of an ordinary Mohr's burette, _b_, preferably fitted with a glass stopcock, and filled previously with the diluted urine. The small tube, _c_, which traverses the cork is intended to permit the escape of steam. Now raise the blue liquid in the flask to active ebullition--not too violent--by the aid of a spirit lamp or small Bunsen flame. Turn the stopcock in order to allow the urine to flow into the boiling solution at the rate of about 100 drops per minute (not more or much less) until the azure tint is exactly discharged. Then stop the flow, and note the number of cubic centimeters used. That amount of dilute urine will contain 5 milligrammes of glucose. To render the determination as accurate as possible, the urine should be diluted to such an extent that not less than 4 or more than 7 c.c.