We have therefore a number of systems, each capable of being equated to a unit of energy, but of which the terms are in some as yet unknown way related to each other, and often more directly than they can be related to mass and weight. This relation is only an approximate one in the case of specific heats; it appears to be an absolute one in the conveyance of electrical energy. The arrangement of the elements in the periodic table must therefore be considered by the light of such general views.
I would venture to suggest, as a tentative method of solving this problem, that it be considered whether mass or weight are such invariable properties of matter as have generally been taken for granted. That the relative weights or masses in which elements combine always retain their invariable proportion is true, so far as we can determine; but it must be remembered that we cannot cause them to combine except under a very limited series of conditions. For example, the act of combination equalises the temperature of two combining atoms; it also, in all probability, equalises their electrical charges. It is a legitimate speculation whether, could we maintain a difference in their temperature or in their electric potential, their atomic weights might not also change. Indeed, we are ignorant whether mass is changed by alteration of temperature. Experiments made by Sir John Airy, and interpreted by Professor Hicks, appear to show that variation of temperature is not without some influence on gravitational attraction. Others by Professor Landolt point in the same direction.
It therefore appears to me not impossible that the mass of the atoms may be affected by the various and different properties which they possess, some to a greater, some to a lesser extent. It must be admitted that atoms differ from each other in the readiness with which they combine with those of the same kind to form molecules; and that molecules of different elements differ from each other in their capacity to form molecular aggregates. Take, for example, such cases as caesium and fluorine, each intensely active, but towards different objects: caesium the most electro-positive of the metals, and fluorine the most electro-negative of elements. Surely their activity must be due to some cause which cannot but exert influence on their other properties, such as their mass and their gravitational attraction, as it doubtless has influence on their specific heats, and on many of their other physical properties. And contrast these instances with helium and with argon, the most indifferent of substances, the atoms of which are unwilling, and apparently unable, to pair even with themselves; it is hardly conceivable that these peculiarities should leave their other, and, as we are in the habit of thinking, invariable, properties unaffected. I venture to suggest that these powers of combination, due to some configuration or to some attractive force, tend to lessen the gravitational attraction by which we measure their atomic weights; that helium and argon, which possess little, if any, of such power to combine, show what may be termed the normal atomic weights, inasmuch as their gravitational attraction is subject to no deduction attributable to their reacting powers.