Page:Proceedings of the Royal Society of London Vol 60.djvu/157

142 The most notable change from previous tables is the increase in the value of hydrogen and the decrease in that of carbou, but the necessity of this has been gradually recognised by the principal workers on the refraction of organic bodies. This in no way affects the well-determined value CH2 = 7*6.

It should be borne in mind that the specific refraction cannot claim a constancy equal to that of the atomic weight. The latter is generally believed to be identical under all circumstances, though the element may be capable of combining with another in two or more multiple proportions. On the other hand, several of the elements, as oxygen and iron, exhibit two or more specific refractions, which are not in multiple proportion, but depend upon the manner of combination. The best recognised of these are given in the third column, and the existence of others is indicated by an “ &c.” Beside these well-marked differences, there are many smaller variations, scarcely, if at all, beyond the limits of experimental error, which depend upon differences of physical condition or chemical structure. The numbers given in column 3 are therefore subject to an uncertainty, which may in some instances amount to 5 per cent. Where there is a greater divergence among the values observed, or where the deductions have been made from only one specimen, it is indicated by a query.

In the paper “ On the refraction equivalents of the elements ” previously referred to, it was shown that if the metallic elements be arranged in the order of their specific refractions, they are roughly in the inverse order of their combining proportions.

In the lecture at the Royal Institution, I showed that this inverse order followed an approximate law, namely, that the “ specific refractive energy of a metal is inversely as the square root of its combining proportion.” This generalisation was proved for univalent metals, the figures showing (with the exception of sodium) a practically constant value for the product of the specific refraction and the square root of the combining proportion.

By the aid of the table in the first part of this communication, the generalisation can now be tested throughout the whole range of the metallic elements.