whether the experiments which have so far been made on this point are sufficiently accurate to decide with certainty whether or not the loss of sugar is exactly equal to the sum of the carbon dioxide and alcohol produced. It has been shown by Buchner and Meisenheimer [1906] that glycerol is a constant product of alcoholic fermentation by yeast-juice (p. 95), and no other source for this than the sugar has yet been found, so that it is not improbable that a small amount of sugar is converted into non-carbohydrate substances other than carbon dioxide and alcohol. It has also been shown [Harden and Young, 1913] that the deficit of sugar is not due to the formation of hexosephosphate (p. 47), which has a lower reduction than glucose, and that the solution from which the sugar (either glucose or fructose) has disappeared actually contains some substance of relatively high dextrorotation and of low reducing power. [p032] However this may be, it may be considered as established that during alcoholic fermentation sugar is converted by an enzyme into some compound of less reducing power, which again yields sugar on hydrolysis with acids. The exact nature of this substance has not been ascertained, but it appears likely that the process is a synthetical one resulting in the formation of some polysaccharide, possibly intermediate between the hexoses and glycogen. A similar phenomenon has been observed with living yeast by Euler and Johansson [1912, 1], and Euler and Berggren [1912], whose interpretation of the observation is discussed later on (p. 57). (/d/) /Fermentation of Different Carbohydrates. Autofermentation./ Yeast-juice and zymin ferment all the sugars which are fermented by the yeast from which they are prepared, and, in addition, a number of colloidal substances which cannot pass through the membrane of the living yeast cell, but which are hydrolysed by enzymes in the juice and thus converted into simpler sugars capable of fermentation [Buchner and Rapp, 1898, 3; 1899, 2]. Of the simple sugars which have been examined, glucose, fructose, and mannose are freely fermented, l-arabinose not at all, whilst the case of galactose is doubtful. Galactose is, however, fermented by juice prepared from a yeast which has been "trained" to ferment galactose [Harden and Norris, 1910]. As regards both the rate of fermentation and the total amount of carbon dioxide evolved from glucose and fructose by the action of a definite amount of yeast-juice, Buchner and Rapp obtained practically identical numbers. Harden and Young [1909], using juice from top yeast, found that fructose was slightly more rapidly fermented and gave a somewhat larger total than glucose, whilst mannose was initially fermented at almost the same rate as glucose, but gave a decidedly lower total, the following being the average result:--