Broadly speaking, three types of crystal structure may be recognised: ( a ) interpenetrating arrangements of atomic and electric lattices, the electrons being symmetrically disposed with reference to the atoms; ( b ) neutral lattices, of which sulphur and phosphorus among the elements are examples; ( c ) ionic lattices, to which class most salts belong, in which the electrons are unsymmetrically arranged so as to be in close relationship with particular atoms. Sir J. J. Thomson pointed out that by a decrease of the closeness of this association between atoms and the linkage electrons, this type merges into the first. Franck similarly distinguishes between three kinds of linkage: (i) where the electrons remain in unchanged quantum orbits, i. e . homopolar linkage due to van der Waal’s forces owing to mutual polarisation of the atoms; (ii) where the linkage electrons occupy new quantum orbits which may be binuclear, i. e ., homopolar linkage such as occurs in molecules of many gases; and (iii) true heteropolar linking where electrons have been bodily transferred from one atom to the other, and occupy new quantum orbits around a single nucleus. Linkages of type (i) clearly produce crystal lattices of class ( a ), usually possessing electric conductivity, while those of class (ii) are concerned in the formation of insulators having neutral lattices of type ( b ). True homopolar linkage should give no active frequencies of vibration, or, if any exist, they should be, according to Schaefer, of extremely weak resonant amplitude, and should lie in the region where the wave-length of the radiation is comparable to the atomic spacing. Nernst suggests that all crystals are built up by homopolar linkages, but this would appear to be contrary to experience, crystals generally showing sharply defined absorption maxima in the infra-red. Kossel is quoted by Bragg as supposing the CO 3 group in carbonates to be held together by heteropolar attraction between the ionised atoms of the group. De, however, considers the binding to be homopolar, while Bragg and Chapman, Chapman, Topping and Morral, Chapman and Ludlam, Kornfeld, and Lennard-Jones all agree in supposing the group to consist of ions having effective charges modified by deformation of the electronic shells. Fajans and Joos, on the other hand, state that, contrary to Kossel’s assumption, there seems to be little reason for supposing the internal binding of the CO 3 ion to be by heteropolar linkages, they themselves being inclined to think that the setting up of an induced electric moment in the oxygen atoms, by reason of the deformability, provides a sufficient explanation of the stability of the group.