Southerden (_Nature_, March 26, 1903), the olfactory cells being
directly stimulated, not by the ordinary vibrations of the
molecules, but by the agitations accompanying chemical changes.

The vibratory hypothesis of the action of odors has had some
influence on the recent physiologists who have chiefly occupied
themselves with olfaction. "It is probable," Zwaardemaker writes
(_L'Année Psychologique_, 1898), "that aroma is a
physico-chemical attribute of the molecules"; he points out that
there is an intimate analogy between color and odor, and remarks
that this analogy leads us to suppose in an aroma ether
vibrations of which the period is determined by the structure of
the molecule.

Since the physiology of olfaction is yet so obscure it is not
surprising that we have no thoroughly scientific classification
of smells, notwithstanding various ambitious attempts to reach a
classification. The classification adopted by Zwaardemaker is
founded on the ancient scheme of Linnæus, and may here be
reproduced:--

I. Ethereal odors (chiefly esters; Rimmel's fruity series).

II. Aromatic odors (terpenes, camphors, and the spicy,
herbaceous, rosaceous, and almond series; the chemical types are
well determined: cineol, eugenol, anethol, geraniol,
benzaldehyde).

III. The balsamic odors (chiefly aldehydes, Rimmel's jasmin,
violet, and balsamic series, with the chemical types: terpineol,