question was debated for years, and our views on the subject have undergone frequent change. One remarkable fact is, that on further substitution the para- derivative only yields one trisubstitution product, but both of the other derivatives yield more than one tri-substitution product. As it is only in the third formula that the four remaining hydrogen atoms occupy similar positions, this formula has been assigned to the paracompounds 1. 4. In the first formula, 1. 2, the four remaining hydrogen atoms are in two different positions to the chlorine atoms, 3 and 6 are adjacent to the chlorine atoms, 4 and 5 are separated from them by a carbon atom. In the second. formula, 1. 3, the third chlorine atom may occupy three dif ferent positions-e.g. 2, between the two chlorine atoms; 4 and 6, adjacent to the chlorine atoms; and 5, separated by a carbon atom. As it has been proved by experiment that an ortho-di-substitution product can only yield two tri-substitution products, and the meta-compound yields three tri-substitution products, the formula 1 . 2 is assigned to the ortho- and 1.3 to the meta- derivatives.

Numerous comparisons have fully confirmed these hypotheses, and since the year 1874 the accuracy of these views has not been disputed.

These examples suffice to give an idea of the methods by which our intimate knowledge of the constitution of organic compounds has been acquired.

We may see in this a confirmation of the saying of Bacon:

Nec manus nuda, nec intellectus sibi permissus ad inveniendam veritatem multum valet. Instrumentis et auxiliis res perficitur, quibus opus est non minus ad intellectum quam ad manum.

§ 52. Physical Isomerism. Allotropy.-As the investigation of the constitution of organic compounds was extended, many instances were observed in which a larger number of isomeric bodies were discovered than could be accounted for by means of the formulæ derived according to the laws of atomic linking. In most of these cases the isomeric substances differed less in their chemical than in their physical properties, such as density, melting point, crystalline form, &c.

Such cases of isomerism which could not be explained by chemical formulæ were termed 'physical isomerides' to dis

H

6

6

tinguish them from the chemical isomerides,' which are caused by a difference in the mode of linking. This physical isomerism is very closely related to and is almost identical with allotropism.' The latter expression was introduced by Berzelius, and applied by him to describe the occurrence of elements in different forms or conditions, or allotropic modifications.' Before the molecules of elementary bodies were regarded as compounds of similar atoms, the existence of one and the same element in different modifications could not be explained in the same way as the isomerism of compounds. Hence the necessity of a special term to be applied to this class of phenomena. At the present time the expression 'allotropism' is also applied to compounds, and is synonymous with physical isomerism. There are several kinds of physical isomerism.

§ 53. Polymorphism.-Dimorphism and polymorphism are common forms of physical isomerism. When one and the same substance crystallises in two or more distinct forms, it is said to be dimorphous or polymorphous. Both elements and compounds exhibit this peculiar phenomenon. Wellknown examples of polymorphism are exhibited by carbon, which crystallises in the regular system as the diamond, and in the hexagonal system as graphite. Sulphur is deposited from fusion in monoclinic crystals and from solution in carbon bi-sulphide in rhombic crystals. Calcium carbonate (CaCO3) occurs in rhombohedral crystals as calcite and in rhombic crystals as arragonite. Silica is met with in two distinct hexagonal forms as quartz and tridymite. Titanium dioxide (TiO2) exists in three distinct forms as rutile, brookite, and anatase. Stannic oxide (SnO2), which is isomorphous with titanium dioxide, assumes the same forms as rutile and brookite, and perhaps anatase. These bodies are consequently 'isodimorphous' or 'isotrimorphous.'

6

There are many other examples of isodimorphism, such as that of the oxides of arsenic and antimony, As10, and Sb10, the sulphides of copper and silver, Cu,S and Ag2S. Many organic compounds are also dimorphous.

The form that a di- or poly-morphous body assumes on crystallisation depends chiefly on the temperature, and also

on certain other external conditions. If the crystallisation takes place from a solution, the nature of the solvent, the presence of other substances, especially of such as are isomorphous with one of the forms of the body in question, influence the form that body assumes.

The conditions under which many forms are produced are entirely unknown. We do not know under what conditions carbon crystallises as diamond, in spite of the numerous attempts which have been made to produce this valuable gem. In the case of many organic compounds one modification has been accidentally obtained, but the conditions under which it is formed still remain unknown.

The allotropic modifications of a substance differ considerably in their stability. Some modifications when once formed are very stable, but others can only exist within narrow limits of the conditions under which they are produced. As examples of the first class we have carbon as diamond or graphite; sulphur is an example of the second class.

The diamond and graphite can exist unaltered side by side, and it is only at a very high temperature that the diamond is converted into graphite. On the other hand the rhombic form of sulphur is only stable below, and the monoclinic form above, a temperature of 95°6 C.; both modifications can exist unchanged for some time outside these limits. But they are in a state of unstable equilibrium, which is easily upset by heating, or shaking, or more particularly by contact with a crystal of that modification which is stable at the prevailing temperature, when the whole mass is converted into this form.

Many dimorphous organic compounds behave like sulphur in this respect, and as a rule only one modification is stable, and the other unstable above and below a certain definite temperature.

This kind of physical isomerism is supposed to be due to a difference in the arrangement of the particles or molecules, which are in themselves identical. The accuracy of this hypothesis cannot be proved, as we do not possess any method by which the nature, or even the size, of the molecules of solid

1 Moissan obtains diamonds by dissolving carbon in molten iron and cooling under high pressure.

bodies can be ascertained. But when we see that under suitable conditions crystals of both modifications can be obtained from one and the same liquid, it seems probable that these modifications are composed of similar molecules, just as different kinds of buildings can be constructed from the same kinds of bricks. This class of isomerism may be termed 'isomerism of aggregation.'

§ 54. Physical Isomerism of the Molecules.-There are also cases of physical isomerism caused by a difference in the molecules. The examples of real polymerism belong to this class, e.g. when a body has different molecular weights in the gaseous and liquid states. In the case of sulphur the molecules at temperatures near the boiling point consist of six atoms, So, which are split up at higher temperatures into molecules consisting of two atoms, S. Many organic and inorganic compounds such as certain aldehydes, acetic acid, nitrogen peroxide, &c.-exhibit analogous behaviour.

The allotropic modifications of phosphorus are probably due to differences in the number of atoms composing the molecules. If phosphorus be heated above 210° in a closed vessel too small to permit the element being completely converted into vapour, it passes from the gaseous state into a red solid modification, from which the colourless variety is regenerated, if sufficient space be offered for complete volatilisation. The red modification is produced from the compressed and the colourless from the expanded vapour at the same temperature (210-300°). It is therefore probable that both modifications already existed in the state of vapour as isolated molecules. A difference in the vapour can only be due to a difference in the molecules. It is not yet known whether this difference is to be ascribed to polymerism.

§ 55. Optical Isomerism. The most remarkable form of isomerism is that in which the isomeric bodies crystallise in forms which are identical in all their individual parts, such as angles and faces, and are symmetrical but not superposable, and bear the same relation to each other that an object bears to its reflected image in a mirror, or that a right-hand glove bears to a glove for the left hand. This peculiar behaviour is generally associated with another remarkable property, viz.

the bodies are optically active. One turns the plane of polarised light to the right, to the same extent that the other does to the left. The bodies thus acting on polarised light are divided into two classes. Some substances are optically active only when they are in a solid and crystalline state; others are optically active as liquids, either in solution or in a molten state; and a few gases or vapours are optically active. The members of the first class either crystallise in the regular form or are uniaxial and crystallise in the quadratic or hexagonal systems. If the two kinds of crystals are placed in parallel lines it is noticed that certain hemihedral faces which occur on the right side of the one set of crystals are found on the left side of the other crystals.

Cinnabar, quartz in the form of rock crystal, chlorates, bromates, periodates, thiosulphates, sodium sulphantimoniate, and some organic bodies belong to this class.

As the rotation of light by these substances depends on their crystalline form and ceases when the substances are brought into the liquid state by fusion or by solution, it is evident that the rotation is not due to the nature of the molecules, but is caused by a peculiarity in their arrangement. It is assumed that the molecules are arranged in a spiral form, and that in one form of crystal the spiral turns to the right and in the other to the left.

The second class of optically active compounds exhibits this property in the liquid state. In these cases the molecules are free to move about and do not take up fixed positions. Hence it appears that the rotation of light is not due to the relative position of the molecules, but to their peculiar nature.

Of course this does not exclude the possibility of these substances (if they are capable of crystallising) exhibiting a peculiar arrangement of the molecules. This is indeed the case with many compounds; e.g. tartaric acid (CHO) crystallises in two different forms, which are non-superposable and bear the same relation to each other that an object does to its reflected image. Only a few of these compounds crystallise in the regular system (amylamine alum) or are optically uniaxial (strychnine sulphate): these bodies rotate the plane of polarised light in the crystalline state.