peated. The ether will expand and fill the space with its vapor, and the globe will hold just as much ether-vapor as if neither of the other two were present; and so we might go on, as far as we know, indefinitely. There is not here a chemical union between the several vapors, and we cannot in any sense regard the space as filled with a compound of the three. It contains all three at the same time, each acting as if it were the sole occupant of the space; and that this is the real condition of things we have the most unquestionable evidence.

You know, for example, that a vapor or gas exerts a certain very considerable pressure against the walls of the containing vessel. Now, each of these vapors exerts its own pressure, and just the same pressure as if it occupied the space alone, so that the total pressure is exactly the sum of the three partial pressures.

Evidently, then, no vapor completely fills the space which it occupies, although equally distributed through it; and we can give no satisfactory explanation of the phenomena of evaporation except on the assumption that each substance is an aggregate of particles, or units, which, by the action of heat, become widely separated from each other, leaving very large intermolecular spaces, within which the particles of an almost indefinite number of other vapors may find place. Pass now to another class of facts, illustrating the same point.

The three liquids, water, alcohol, and ether, are expanded by heat like other forms of matter, but there is a striking circumstance connected with these phenomena, to which I wish to direct your observation. I have, therefore, filled three perfectly similar thermometerbulb tubes, each with one of those liquids. The tubes are mounted in a glass cell standing before the con

UNEQUAL EXPANSION IN LIQUIDS.

11

denser of a magic lantern, and you see their images projected on the screen. You also notice that the liquids (which have been colored to make them visible) all stand at the same height; and, since both the bulbs and the tubes are of the same dimensions, the relative change in volume of the inclosed liquids will be indicated by the rise or fall of the liquid columns in the tubes. We will now fill the cell with warm water, and notice that, as soon as the heat begins to penetrate the liquids, the three columns begin to rise, indicating an increase of volume; but notice how unequal is the expansion. The ether in the right-hand tube expands more than the alcohol in the centre, and that again far more than the water on the left. What is true of these three liquids is true in general of all liquids. Each has its own rate of expansion, and the amount in any case does not appear to depend on any peculiar physical state or condition of the liquid, but is connected with the nature of the substance, although, in what way, we are as yet wholly ignorant.

But you may ask: What is there remarkable in this? Why should we not expect that the rate of expansion would differ with different substances? Certainly, there is no reason to be surprised at such a fact. But, then, the remarkable circumstance connected with this class of phenomena has yet to be stated.

Raise the temperature of these liquids to a point a little above that of boiling water, and we shall convert all three substances into vapor. We thus obtain three gases, and, on heating these aëriform bodies to a still higher temperature, we shall find that, in this new condition, they expand far more rapidly than in the liquid state. But we shall also find that the influence of the nature of the substance on the phenomenon has wholly

disappeared, and that, in the aëriform condition, these substances, and in general all substances, expand at the same rate under like conditions.

Why, now, this difference between the two states of matter? If the material fills space as completely in the aëriform as it does in the liquid condition, then we cannot conceive why the nature of the substance should not have the same influence on the phenomena of expansion in both cases. If, however, matter is an aggregate of definite small masses or molecules, which, while comparatively close together in the liquid state, become widely separated when the liquids are converted into vapor, then it is obvious that the action of the particles on each other, which might be considerable in the first state, would become less and less as the molecules were separated, until at last it was inappreciable; and if, further, as Avogadro's law assumes, the number of these particles in a given space is the same for all gases under the same conditions, then it is equally obvious that, there being no action between the particles, all vapors may be regarded as aggregates of the same number of isolated particles similarly placed, and we should expect that the action of heat on such similar masses would be the same.

Thus these phenomena of heat almost force upon us the conviction that the various forms of matter we see around us do not completely fill the spaces which they appear to occupy, but consist of isolated particles separated by comparatively wide intervals. There are many other facts which might be cited in support of the same conclusion; and among these two, which are more especially worthy of your attention, because they aid us in forming some conception of the size of the molecules themselves.

INTERSTICES IN SOLIDS.

13

If this mass of glass is perfectly homogeneous-if the vitreous substance completely fills its allotted space, and there is no break whatever in the continuity of the material then you would expect that its physical relations would not depend at all on the size of the surface affected. Suppose you wished to penetrate it with a fine wire. The point of this wire, however small, would not detect any difference at different points of the surface. Assume, however, that it consists of masses separated by spaces, like, for example, this sheet of wire netting. Then, although the surface would seem perfectly homogeneous to a bar large enough to cover a number of meshes, it would not be found to be by any means homogeneous to a wire which was small enough to penetrate the meshes. If, now, there are similar interstices in this mass of glass, we should expect that, if our wire were small enough (that is, of dimensions corresponding to the interstices), it would detect differences in the resistance at different points of this glass surface.

Make, now, a further supposition. Assume that we have a number of these wires of different sizes, the largest being twice as stout as the smallest. It is obvious that, if the interstices we have assumed were, say, several thousand times larger than the largest wire, all the wires would meet with essentially the same opposition when thrust at the glass. If, however, the interstices were only four or five times larger than the wires, then the larger would encounter much greater resistance from the edges of the meshes than the smaller.

It is unnecessary to say that no physical point can detect an inequality in the surface of a plate of glass, but we have, in what we call a beam of light, an agent which, in passing through its mass, does discover differences of the kind we have attempted to describe. Now, it

is perfectly true that we have no absolute knowledge of the nature of a beam of light. We have a very plausible theory that the phenomena of light are the effects of waves transmitted through a highly-elastic medium we call ether, and that, in the case of our plate of glass, the motion is transmitted through the ether, which fills the interstices between the molecules of this transparent solid; but we have no right to assume this theory in our present discussion.

Indeed, I cannot agree with those who regard the wave-theory of light as an established principle of science. That it is a theory of the very highest value I freely admit, and that it has been able to predict the phases of unknown phenomena, which experiment has subsequently brought to light, is a well-known fact. All this is true; but then, on the other side, the theory requires a combination of qualities in the ether of space, which I find it difficult to believe are actually realized. For instance, the rapidity with which wave-motion is transmitted depends, other things being equal, on the elasticity of the medium. Assuming that two media have the same density, their elasticities are proportional to the squares of the velocities with which a wave travels. The velocity of the sound-wave in air is about 1,100 feet a second or of a mile, that of the lightwave about 192,000 miles a second, or about one million times greater; and, if we take into account certain causes, which, though they tend to increase the velocity of sound, can have no effect on the luminiferous ether, the difference would be even greater than this.

Now, were the density of the ether as great as that of the atmosphere (say of a grain to the cubic inch), its elasticity or power of resisting pressure would be a million square, or a million million times that of the