if the solid which melts at 195° is raised to that temperature. and then slowly cooled, the product possesses the normal melting point, viz. 195'. When a substance crystallises in more than one system, one crystalline form approaches as nearly as possible to the other, one form appears to imitate the other both crystallographically and optically2; thus arsenious oxide crystallises in regular octahedra and also in rhombic prisms, the latter exhibiting an angle identical with the angle of the regular octahedron.

93. O. Lehmann3 has collected and discussed many instances of the exhibition of different physical properties by compounds possessing the same elementary composition*. The phenomenon, which may be called physical isomerism, presents analogies with allotropy (see ante, par. 67); in both, temperature is the most important condition affecting the change from one form to another, and this change is accompanied in both classes of phenomena by absorption or evolution of heat.

In the term allotropy are summed up similar phenomena. which appear to be best explained, in terms of the molecular theory, by the hypothesis that variations in the numbers of atoms, all of one kind, constituting a molecule, may be accompanied by variations in the physical, and to some extent chemical properties, of the substance which is composed of these molecules. But the term is also applied to phenomenae.g. the variation in melting points, &c. of solid sulphur-which are probably better explained by the hypothesis that what may be called the acting physical unit, or the physical molecule, of each allotropic form is constituted of a different number of chemical molecules. In this view of the matter,

1 See E. Lellmann, Ber. 15. 2835.

2 Pasteur, Ann. Chim. Phys. [3] 23. 267.

3 Zeitschr. für Krystallog. 1. 97. See also, in connection with the subject generally, the article ‘Isomerie, physikalische' in Neues Handwörterbuch der Chemie, Bd. 111. pp. 836-843.

On this subject see also Laubenheimer, Ber. 9. 760.

The term physical isomerism seems to have been first used by L. Carius, Annalen, 126, 214 (see also do. 130. 237).

the term physical isomerism would embrace phenomena common to elements and compounds.

94. Lehmann (loc. cit.) divides physically isomeric bodies into two classes: (1) those in which change from one form to another occurs at a definite temperature, the direction of the change being dependent on very small differences of temperature; (2) those which exhibit two forms, one more stable than the other, and in which change from one form to the other does not occur at a definite temperature, and is not reversible by heat alone.

Ammonium nitrate is an example of a substance belonging to the first class; the rhombic crystals of this salt, which separate at ordinary temperatures from an aqueous solution, melt at (about) 168°; as the molten mass cools crystals belonging to the regular system are formed, but at (about) 125° these change to rhombohedral forms, which at (about) 87° are converted into rhombic needles, from which, at 30° or so, the original rhombic crystals are produced. If the rhombic crystals are again slowly heated, the rhombic needle-shaped crystals form at (about) 30°; the rhombohedral forms appear at (about) 87°, the regular crystals at (about) 125°, and finally the solid melts at 198°. Again, if a little sulphur is melted on a microscopic slide (under a cover), and the slide is arranged so that temperature can be easily regulated', monoclinic crystals are produced, but, as temperature falls, these change into rhombic forms, and it is possible to regulate the temperature so that definite amounts of each form exist simultaneously, but on the slightest change of temperature the rhombic crystals grow at the expense of the monoclinic, or vice versa.

The behaviour of dibromopropionic acid when heated illustrates the nature of the changes which characterise substances belonging to Lehmann's second class of physical isomerides. This substance crystallises in rhombic forms. which melt at 64° (about); if the molten mass is heated a few degrees above this point rhombic crystals (M.P=64°)

1 Lehmann describes an apparatus for this purpose (loc. cit. pp. 102-3).

are again produced on cooling; but if the molten substance is heated many degrees above 64° and is then allowed to cool, small flat nearly right-angled tables are obtained which melt at 51° (about). If the less stable form (M.P. = 51°) is slowly heated under the microscope, growth of the other (more stable) crystals is noticed; the growth at first is rapid, then slower, but before the change has gone far the melting point of the less stable crystals is reached and the whole mass becomes liquid. If the more stable form is melted, heated some degrees above 64° and then brought into contact with crystals of both forms, growth of each modification proceeds until the crystals touch, after which the more stable (higher melting) crystals grow into the others until the latter are completely changed into the stabler forms. Another instance is furnished by paranitrophenol. This compound crystallises from hot aqueous solutions in monoclinic crystals, and from cold aqueous (or alcoholic) solutions in crystals belonging to the same system but differing in form and melting point from the others. By fusing either form and allowing the molten mass to cool, only the less stable (lower melting) crystals are produced; but if a little of the substance is melted on a microscopic slide, and a crystal of the second (stabler) form is placed in contact with the edge of the solidified mass, and heating is then again commenced, crystals of the stabler form begin to grow at the expense of the other crystals, at first rapidly, then more slowly, until both forms melt, the less stable at a lower temperature than the more stable.

95. I do not think that a rigid classification of physical isomerides into two groups can be carried out. The examples given shew that there are certain broad differences between the two classes; but a detailed consideration of these examples, and of others to be found in Lehmann's paper, seems to me to lead to the conclusion that there exists no firmly drawn line of separation between the phenomena exhibited by substances placed in different classes. This will, I think, be more apparent if some of the facts enumerated are represented in a roughly graphic manner. Let us compare the

action of heat on ammonium nitrate, dibromopropionic acid, and paranitrophenol.

Let that form of ammonium nitrate which crystallises in regular crystals be called A, that which crystallises in rhombohedral forms B, that in rhombic needles C, and that in ordinary rhombic forms D. Let the line ab represent the b 168° interval of temperature through which ordinary ammonium nitrate must be heated until it melts; now 125° let the molten substance cool, that part of the line ab between 168° and 125° represents the temperature87° interval through which the salt exists in form A, 36° that part between 125° and 87° represents the tem

perature-interval through which form B is stable, that part between 87° and 36° the interval through which form C is stable, and lastly the portion below 36° represents the interval of stability for form D. The operation represented by the line ab is reversible; whether we begin at a or b, the salt goes through the several stages roughly indicated in the diagram.

(80°)

64°

51°

Now turn to dibromopropionic acid. Let the rhombic form melting at 64° be called A, and the small tables melting at 51° be called B. The line ab represents the temperatureinterval through which A is stable; on heating A to 64° it melts, but on cooling, it so to speak runs back along the same line: now let A be raised from a to c, say 16° above its melting point, and then allowed to cool; a new substance, B, has been produced, and this substance is stable throughout the intere val ce (between d and e it is solid, between c and d it is liquid). Although B cannot be changed into A by heat alone, yet when B is somewhere between c and d (i.e. when it is molten) it may be partially converted into A by contact with small portions of A. If the symbols A and B are used to represent the two forms of paranitrophenol, then B may be almost wholly converted into A by contact with A. Because of its lower melting point, B has been called the less stable form of dibromopropionic acid,

but if we consider that it cannot be changed into A by the action of heat we should say that, thermally looked at, B is the more stable form of this substance.

If we

b 195°

130°

now treat the facts concerning the diphenyl derivative mentioned on p. 184, par. 92, diagrammatically we shall have this result: (let the form melting at 195° = A, and that melting at 99° = B). The line ab represents the temperature-interval through which A must be raised in order to melt it: let the molten substance cool slowly, it runs back along the same line ab; [? do any crystallographic changes occur along this line]; but let A cool a quickly, it seems to get, so to speak, shunted

off the line of normal stability on to the line bd: cd represents the interval through which it exists in solid form and ce the interval (roughly) through which it is stable in liquid form, for when B has again been heated to 125°130° it solidifies, and is found to have come back to form A, i.e. to the line of normal stability1 (ab).

This diphenyl derivative appears to belong to both of Lehmann's classes: some of the changes which it undergoes are to a great extent reversible and occur at definite temperatures, others are not reversible and occur gradually throughout a considerable interval of temperature.

Substances of which ammonium nitrate is the type appear to be less profoundly modified by the action of heat than substances belonging to the class represented by dibromopropionic acid. Substances belonging to the first of these classes shew analogies with the so-called molecular compounds (e.g. compare the action of heat on crystallised sodium phosphate, or on hydrated cobalt salts, with that on dibromopropionic acid or on paranitrophenol); and the course of the change brought about by the action of heat on these bodies shews

1 It seems probable that if A were heated to 195° and cooled quickly to 130°, and then very rapidly, it would solidify at that point, and afterwards be found to melt at 195°.

2 See post, section 5, par. 102.