sublimed,
melted at 115 deg. to 117 deg. C., and contained sulphur. The crystals
were insoluble in water, almost insoluble in alcohol, but readily soluble
in ether.

By distilling roasted coffee with superheated steam, Erdmann[148]
obtained an oil consisting of an indifferent portion of 58 percent and
an acid portion of 42 percent, consisting mainly of a valeric acid,
probably alphamethylbutyric acid. The indifferent portion was found to
contain about 50 percent furfuryl alcohol, together with a number of
phenols. The fraction containing the characteristic odorous constituent
of coffee boiled at 93 deg. C. under 13 mm. pressure. The yield of this
latter principle was extremely small, only about 0.89 gram being
procured from 65 kilos of coffee.

Pyridin was also shown to be present in coffee by Betrand and
Weisweiller[149] and by Sayre.[150] As high as 200 to 500 milligrams of
this toxic compound have been obtained from 1 kilogram of freshly
roasted coffee.

As stated above, the empyreumatic volatile aromatic constituents of the
coffee are without question formed during and by the roasting process.
According to Thorpe,[151] the most favorable temperature for development
of coffee odor and flavor is about 200 deg. C. Erdmann claimed to have
produced caffeol by gently heating together caffetannic acid, caffein,
and cane sugar. Other investigators have been unable to duplicate this
work. Another authority,[152] giving it the empirical formula
C_8_H_10_O_2, states that it is produced during roasting, probably
at the expense of a portion of the caffein. These conceptions are in the
main incomplete and inaccurate.

By means of careful work, Grafe[153] came closer to ascertaining the
origin of the fugacious aromatic materials. His work with normal,
caffein-free coffee and with Thum's purified coffee led him to state
that a part of these substances was derived from the crude fiber,
probably from the hemi-cellulose of the thick endosperm cells.
Sayre[154] makes the most plausible proposal regarding the origin of
caffeol. He considers the roasting of coffee as a destructive
distillation process, summarizing the results, briefly, as the
production of furfuraldehyde from the carbohydrates, acrolein from the
fats, catechol and pyrogallol from the tannins, and ammonia, amins, and
pyrrols from the proteins. The products of roasting inter-react to
produce many compounds of varying degrees of complexity and toxicity.

The great d