[Footnote 31: _Centralbl. für d. med. Wiss._, No. 55, 1874.] [Footnote 32: _Liebig's Annalen_, Bd. 182, S. 362.] [Footnote 33: _Ibid._, Bd. 209, S. 25.] [Footnote 34: _Zeitschrift für physiol. Chemie_, vi. 6.] [Footnote 35: "Beiträge zur Lehre von der Acetonurie und von verwandten Erscheinungen," _Deutsch. Archiv für klin. Med._, xxxiv., 2 Oct., 1883, S. 127.] Gerhardt early discovered a substance in the urine of diabetics and habitual drinkers which struck a deep-red reaction with chloride of iron. This he considered was the source of acetone, and was probably ethyl diacetate or diacetic ether, which by decomposition yields equal molecules of acetone and alcohol; thus: C_{4}H_{5}O_{3}C_{2}H_{5} + H_{2}O = C_{3}H_{6}O + CO_{2} + C_{2}H_{6}O. Ethyl diacetate. Water. Acetone. Alcohol. This view is still held by some, but others, in view of the recent discovery of Deichmüller and Tollens,[36] that diabetic urine when distilled yields decidedly more acetone than alcohol, have suggested that the substance is derived from aceto-acetic acid. [Footnote 36: _Loc. cit._] The first test suggested for acetone was Gerhardt's chloride-of-iron test. A solution of chloride of iron added to urine containing acetone strikes a burgundy-red color. But this reaction occurs with so many substances that it cannot be considered entirely reliable. Ralfe's modification of Lieben's iodoform test[37] is made as follows: About a fluidrachm (3.7 c.c.) of liquor potassæ, containing 20 grains (1.2 grams) of iodide of potassium, is placed in a test-tube and boiled; a drachm (3.7 c.c.) of the suspected urine is then carefully floated upon the surface. When the urine comes in contact with the hot alkaline solution a ring of phosphates is formed, and after a few minutes, if acetone or its allies are present, the ring will become yellow and studded with yellow dots of iodoform, which, in turn, will sink through the ring of phosphates and deposit itself at the bottom of the test-tube. A number of other substances {210} produce the iodoform reaction, but only one of these, lactic acid, is likely to be met in urine. [Footnote 37: _Clinical Chemistry_, Philadelphia, 1884, p. 100.] The perspiration, saliva, exudations, and effusions in diabetic cases have all been found, at times, to contain sugar.