On still further raising the temperature of pure clay (pelinite or clayite) no further reactions of importance occur, the material being practically infusible. If, however, any silica, lime, magnesia, alkalies, iron oxide or other material capable of combining with the alumina and silica is present as impurities in the clay substance, combination begins at temperatures above 900° C. This causes a reduction of the heat-resisting power of the material; the silicates and alumino-silicates produced fuse and begin to react on the remaining silica and alumina, first forming an impermeable mass in place of the porous one produced with pure clay substance, and gradually, as the material loses its shape, producing a molten slag if the 'clay' is sufficiently impure. As ordinary clays are never quite free from metallic compounds other than alumina, this formation of a fused portion--technically known as _vitrification_ (p. 37)--occurs at temperatures depending on the nature of the materials present, so that a wide range of products is obtained, the series commencing with the entirely unfused pure clay (china clay), passing through the slightly vitrified fireclays, the more completely vitrified ball clays to the vitrifiable stoneware clays and ending with materials so rich in easily fusible matter as scarcely to be worthy of the name of clays. The constitution of the clay molecule is a subject which has attracted the attention of many investigators and is being closely studied at the present time. It is a subject of peculiar difficulty owing to the inertness of clay substance at all but high temperatures, and to the complexity of reactions which take place as soon as any reagent is brought into active connection with it. Without entering into details regarding the various graphic formulae which have been suggested, it is sufficient to state that the one which is most probably correct, as far as present knowledge goes, is Mellor's and Holdcroft's re-arrangement of Groth's formula (30) HO\ /OSiO\ \Al2/ \O / \ / HO/ /\ \OSiO/ / \ HO OH which on removal of the hydroxyl groups might be expected to give the anhydride O\\ /OSiO\ \\Al2/ \ // \ / O// \OSiO/ though in practice this substance--if formed at all--appears to be instantly split up into Al2O3 and SiO2.