Although furfuraldehyde is the natural decomposition product of pentosans, furfuryl alcohol is the main furane body of coffee aroma. This would indicate that active reducing conditions prevail within the bean during roasting; and the further fact that carbon monoxid is given off during roasting makes this seem quite probable. If one admits that caffetannic acid exists in the green bean; that upon oxidation it gives viridic acid; and that it is concentrated in the outer layers of the bean, as certain investigators have claimed, then there is chemical proof of the existence of oxidizing conditions about the exterior of the bean. In any event, however, the fact that oxidizing conditions predominate on the external portion of the bean is obvious. Accordingly, our meager knowledge of the chemistry of roasting indicates that while the external layers of the roasting beans are subjected to oxidizing conditions, reducing ones exist in the interior. Future experimentation will, no doubt, prove this to be the case. Attempts have been made to retain in the beans the volatile products, which normally escape, both by coating previous to roasting[163] and by conducting the process under pressure.[164] However, the results so obtained were not practical, since the cup values were decreased in the majority of cases, and the physiological effects produced were undesirable. In cases where the quality was improved, the gain was not sufficient to recompense the roaster for the additional expense and difficulty of operation. Various persons have essayed to control the roasting process automatically; but the extreme variance in composition of different coffees, the effect of changing atmospheric conditions, and the lack of constancy in the calorific power of fuels have conspired to defeat the automatic roasting machine.[165] It is even doubtful whether De Mattia's[166] process for roasting until the vapors evolved produce a violet color when passed into a solution of fuchsin decolorized with sulphur dioxid is commercially reliable. Many patents have been granted for the treatment of coffees immediately prior to or during roasting with the object of thus improving the product. The majority of these depend upon adding solutions of sugar,[167] calcium saccharate,[168] or other carbohydrates,[169] and in the case of Eckhardt,[170] of small percentages of tannic acid and fat. In direct opposition to this latter practise, Jurgens and Westphal[171] apply alkali, ostensibly to lessen the "tannic acid" content. Brougier[172] sprays a solution containing caffein upon the roasting berries; and Potter[173] roasts the coffee together with chicory, effecting a separation at the end. [Illustration: GROUND COFFEE UNDER THE MICROSCOPE]