ters not how little oxygen is taken, provided only that the proper proportion of hydrogen is supplied. Then let us conceive the least possible portion of oxygen. Let the mind wrestle with the conception and reduce the volume of this gas until it is fixed at the smallest that can take part in a chemical action. Then conceive it combined with a volume of hydrogen twice as great. We may contemplate the infinitesimal droplet of water so produced, but to conceive a droplet any smaller is impossible, since this one contains no more than the least possible portions of its elements. To break this droplet of water would be to reduce it to fragments of hydrogen and oxygen. Again, then, do we find our minds in the presence of particles which can not be divided except at the sacrifice of their identity—in other words, in the presence of molecules.
But if molecules exist, the second question at once arises, Are they so closely packed as to constitute a continuous mass, or are they separated by intervening spaces?
A second class of phenomena directs our judgment here. We might detail the mathematical investigation by Cauchy, a third of a century ago, by which he demonstrated the impossibility of the dispersion of light in a substance whose minutest parts are absolutely homogeneous. It was proved that dispersion happens only on condition that "two contiguous portions of the medium, whose dimensions are moderately small fractions of a wave-length of light, are dissimilar." The molecules with intervening spaces would realize such dissimilarity.
But, confining ourselves to the experimental side of the problem, we find a variety of familiar phenomena ready to bear witness to this structure. Among them are porosity, expansion and change of physical condition, and the diffusion of vapors.
In regard to porosity, an old and homely experiment will give us a starting-point. We take a tall and narrow glass jar and fill it so completely with alcohol that the addition of a single drop will endanger an overflow. The jar appears to be full of a perfectly homogeneous liquor. But if a sheet of cotton wool, whose fibers have been previously well loosened, be at hand, fragment after fragment may, with care, be slowly introduced, without causing the overflow of a single drop, until the jar appears to be filled with moistened cotton instead of with alcohol. We have before us the surprising appearance of two bodies filling the same space at the same moment. Surely, however, we are not at liberty to adopt this explanation. For what should we call that which has no power to exclude another from the space which it occupies? To call it matter is to obliterate all distinction between matter and space. We are impelled to seek another explanation, and we find one more acceptable in the hypothesis that neither of the two bodies wholly fills the space which it appears to occupy—that spaces, too minute for even microscopic vision to detect, intervene between the ultimate particles of both, and to such an extent that these mate-
