Many conflicting data are given by various authorities respecting the composition of coffee. Many attempts have been made to ascribe the properties which are valuable in coffee to particular chemical characteristics. In only one particular are these statements reliable. The properties of caffein, the alkaloid which gives coffee its principal stimulating value, are well known, and the effects which coffee and tea produce upon the nerve centers are correctly ascribed to this ingredient. Caffein, however, has so little taste or flavor that it plays no important role whatever in giving to coffee the characteristics of aroma, taste and flavor, which commend it to the palates of so many people.

Spencer, in his studies in the Bureau of Chemistry under my direction (Bulletin No. 13, Part Seventh, pages 902-903), gives the following data for the average composition of coffee:

Unfortunately the samples representing the roasted coffees were not the same as those representing the green coffees, so the data are not strictly comparable. The principal changes, however, which are produced in roasting coffee are: first, reduction of the moisture; second, the practical elimination of sugar. The percentage of fat and the percentage of caffein do not seem to be materially influenced by the roasting process. The percentage of tannin as reported by Spencer is not nearly so great as that of other analyses which he quotes in the same bulletin. For instance, in the analyses of 7 samples the caffetannic and tannic acid combined, which represent tannin, varied from a maximum of 23.10 percent to a minimum of 19.50 percent, and in these same analyses the caffein varied from a maximum of 1.53 percent to a minimum of 0.64 percent.

One of the important constituents of green coffee is sugar. Approximately 8 percent of sugar are found in green coffees. In the work conducted by Spencer considerable quantities of this sugar were extracted and obtained in a crystallized state and identified as pure sucrose. The presence of sugar in green coffees has not been given sufficient emphasis in the explanations which have been made concerning the development of the aroma of roasted coffee. The analytical data show that the sugar is practically eliminated, that is, converted largely into caramel. Caramel is a flavoring substance of high value, and there is no question of the fact that much of the prized aroma and flavor of roasted coffee are due to the presence of caramel.

Although it may reflect upon the skill and ability of the chemist, the truth should be told respecting the nature of the changes produced on roasting. The delicacy of flavor, taste and aroma, which is produced, while doubtless correlated with chemical changes, has not yet been accurately connected with any particular change that occurs. It is quite customary to state that the pronounced flavor of roasted coffee is due to certain changes in the oils. The data show very little change of any kind in the oils, and the greater quantity of them is fixed and non-volatile. Of all the problems of chemistry there is none more difficult than the determination by chemical means of a flavoring or odoriferous substance. Minute quantities of these substances are sufficient to give character to a product, and yet they elude the most delicate balance of the chemist.

For historical purposes rather than for any expression of definite composition references may be made to the other constituents of coffees which some analysts have claimed to discover. The data on which these discoveries rest are not of sufficient accuracy to warrant any definite statements.