some analogies with the processes of gaseous dissociation'. For these reasons Lehmann has summarised the phenomena characteristic of bodies of this class under the term physical polymerism, and the phenomena characteristic of bodies of the other class under the term physical metamerism. The former term implies that the physically different forms exhibited by a substance belonging to this class are to be regarded as associated with the existence of physical molecules, each formed by the grouping together of a different number of chemical molecules (as defined in Chap. I. par. 13, p. 25). The term physical metamerism on the other hand implies that the physical molecule of each different form of a substance belonging to this class is composed of the same number of chemical molecules, but that the arrangement of these is different in each case.

Lehmann's classification is certainly based on no fanciful analogies. Polymerism and metamerism are well marked phenomena among gaseous molecules; and the hypothesis of the existence of groups of molecules characterised by definite properties, but each of which groups is readily decomposed by heat, appears to be as simple as any other that can be proposed to explain the observed facts. Moreover this hypothesis is almost forced on our acceptance when we consider the numerous and varied phenomena summarised in the term 'molecular-compounds. But the analogy between the reactions of gascous molecules and the changes undergone by solid and liquid substances may be pushed too far; we ought to recognise how small and inexact our knowledge is of the molecular actions of the latter classes of bodies. Qualification of the terms molecule, polymerism, and metamerism by the adjective physical widens the meanings of these terms by making them applicable to a larger class of phenomena, but at the same time it makes the application less precise3.

1 See Book II. Chap. II.

2 See post, section 5.

3 Lehmann considers in considerable detail the phenomena attending the change of one form of a substance into another; he divides the changes into groups, according as both forms are solid, or one solid and one liquid, &c. As the

96. We have thus found that to trace the connections between the composition and the properties of changing material systems has always been regarded as the fundamental problem of chemistry. Attention has sometimes been almost confined to the composition of substances forming such systems, at other times the properties of the system and its components have been regarded as chiefly important. We found that as chemistry advanced it became necessary to know more than the mere elementary composition of bodies; having gained the atom and the molecule, chemists were soon convinced that the arrangement of the same atoms might vary, and that subject is important I give a brief resumé of some of Lehmann's results in this note, but the original paper cught to be studied by all who are interested in the subject.

A. Change of one, more complex, solid form of isomeride to another, less complex, solid form, attended with absorption of heat; physical molecules of both kinds are present simultaneously, but at a certain temperature change will occur. If one modification is heated alone, the normal temperature of change may be largely exceeded without a complete change to the second modification, but at such a high temperature contact with the second modification may determine sudden and complete change. B. Change of solid form to liquid form, occurring with heat absorption at a definite temperature dependent on the pressure; the change will not be complete, as molecules of both kinds will exist together. If the specific gravity of the solid form is greater than that of the liquid form, then on heating past the melting point there will be rapid expansion as the physical molecules of the solid form are separated into those of the liquid; this will be followed by a slower regular expansion. If the specific gravity of the solid is less than that of the liquid, expansion will be small, or even negative, until a point of maximum density is reached, after which expansion will proceed at the normal rate.

In some cases a solid form is changed, by the action of heat, into a liquid form, which, at a higher temperature, is again changed into a second solid form, e.g. when selenion is heated till it becomes viscous and is kept at this temperature for some time it changes into a crystalline form. So in the change of yellow to red phosphorus by the action of heat; in this case the molecules which form the liquid phosphorus are kept apart for some time, by the energy added as heat acting against cohesion, and so are allowed to re-arrange themselves in loose groups.

C. Change of liquid, to solid modification is complex: a few crystals form and determine the crystallisation of the whole mass; in some cases the liquid, especially if viscous, may be cooled below the temperature at which crystallisation normally begins, and may then pass into an amorphous solid form.

properties might therefore be correlated not only with atomic composition but also with atomic configuration. We traced this conception through the dualism of Berzelius and the unitary system of Dumas, Laurent, Gerhardt and others, through the hypothesis of compound radicles and that of types, to the time when Frankland and Kekulé gave it greater precision by arranging the elementary atoms in groups according to the maximum number of other atoms. with which each was found to combine.

But we saw that the expression equivalency (or valency) of atoms gradually came to be used in a loose and inexact manner. We found that the comparison of monovalent with divalent, &c. atoms, when unchecked by accurate dynamical knowledge, led to the belief that the term in question expressed in some vague way quantitative measurements of interatomic forces, and to the conclusion that, inasmuch as one divalent atom could directly bind to itself two other atoms, while one monovalent atom could act directly on only a single other atom in a molecule, therefore the divalent atom was capable of exerting twice as much force as the monovalent atom. The latter part of the foregoing sentence may I think be taken as fairly representative of the loose and slipshod way in which dynamical language has too often been used in chemistry.

We found that attempts were made to build a general theory of valency on a shifting quasi-dynamical foundation; but the account given in this section of Lossen's criticisms of the expressions 'a bond,' 'a valency,' 'a unit of affinity,' &c. has I think been sufficient to shew how inexact, while apparently precise, and how narrow, while apparently far-reaching, the theory in question really is.

The objections raised against the atomic theory in recent years by some chemists, who nevertheless made free use of the essentially atomic conceptions of modern chemistry, led, it seems to me, to a looseness of thinking about atoms, molecules and equivalents, which has done no little harm. Parts by weight were spoken of as if the expression were synonymous with atom; equivalents were regarded as acting and

reacting one on the other; there appeared to be a possibility of chemistry retracing her steps to the time when no precise meaning was attached to any of the terms atom, molecule, combining weight, equivalent, but each was used as nearly synonymous with the others. From the possibility of such retrogression we have been saved by the general advance of physical science. As the molecular theory of matter became more precise and its applications more far-reaching, it was impossible for chemists to employ conceptions essentially molecular and atomic and at the same time to express chemical changes in a notation based on the notions of a pre-molecular era. It became necessary to choose definitely between the atom and the equivalent, and the great body of chemists has certainly chosen the former.

But as soon as attempts to found a theory of chemical actions on the basis of equivalents was abandoned, it was seen that the conception of equivalency might be retained and applied to the elementary atoms. To keep distinct the conceptions implied in the terms equivalent and atom, and at the same time to arrange the atoms in equivalent groups, is one of the problems of modern chemistry. On this distinction and on this resemblance is based the theory of isomerism. The study of isomerism has done much, we found, to render precise the conception of the molecule as a structure with properties dependent on the nature, the number, and the arrangement of the constituent atoms.

We endeavoured to subdivide the conception expressed in the words 'arrangement of atoms in a molecule' into parts, and to demonstrate by illustrations the existence of a connection between each of these parts and the properties of the molecule. These illustrations led to clearer notions concerning valency of atoms, and the meaning of structural formulæ ; these formulæ we regarded as expressing the actual valencies of the atoms in the molecule,-i.e. the number of atoms directly acting on and acted on by each atom, and as expressing also the distribution of the atomic interactions, i. e. the nature of the atoms in direct mutual connection; but we tried to attach no quantitative meaning to the symbols used

M. C.

13

for expressing atomic valencies and distributions of atomic interactions.

The theory of valency, as thus used, leads to dynamical conceptions, but regards these as outside its sphere: it points the way along which progress will be made. Attempts must be made to apply thermal, optical, and other physical methods of research to the investigation of chemical problems; thus we may hope to gain clear and precise knowledge regarding the connection between structure and stability of molecules, in so far as the latter is measured by variations in the quantities of energy associated with each molecule.

APPENDIX TO SECTION IV.

97. To have given a detailed account of Lossen's criticisms of the generally accepted views regarding 'valencies' or 'units 'of affinity' in the text of the section on isomerism, would have involved too great an interruption of the main argument of that section. But as Lossen's criticisms seem to me of great importance I propose to give some account of them here.

The many and varied hypotheses concerning valency set forth by chemists of acknowledged authority may be divided, says Lossen, into three groups :—

I. Those hypotheses which regard ‘an affinity' as a definite quantity of matter, or as an action of some kind proceeding from a definite quantity of matter.

II. Those which regard 'an affinity' as a part of an atom, or at least as something connected with a part of an atom. III. Those which regard the 'affinities' of an atom as definite forms of motion of the atom.

I. Erlenmeyer1 has developed the conception of ‘Affini'valencies.' He states, as a rule without exceptions, that 'in all 'chemical combinations a constant quantity of one element

1 For references to the work of the various chemists mentioned, see Lossen, loc. cit.