Nevertheless, the occurrence of such numerical relationships continued, as already stated, to excite speculation. Döbereiner, in 1829, pointed out that in certain groups of correlated elements, consisting each of three members, the middle member had an atomic weight practically identical with the arithmetic mean of the atomic weights of the others; and similar observations were made by Gmelin, Dumas, Gladstone, and Strecker. An approach to the recognition of the general law underlying these facts was made by Newlands in England, and independently by De Chancourtois in France, who were the first to indicate that the properties of the elements are related to their atomic weights. This conception was developed by the Russian chemist, Mendeléeff. In Mendeléeff’s arrangement, first published in 1869, the elements are so grouped that their properties are periodic functions of their atomic weights. The general statement of what is now known as the Periodic Law may be put in this form: If the elements are arranged in order of increasing atomic weight, the properties of these elements vary from member to member of the series, but return more or less nearly to the same value at certain fixed points in the series. This is observed to occur in the atomic value, or valency, of the several members; also in their specific volumes, melting-points, ductility, hardness, volatility, crystalline form, thermal expansion, refraction equivalents, and conductivities for heat and electricity, in their magnetic properties and electro-chemical behaviour, and in their heats of chemical combination, etc. The first chemist of note to grasp the significance of Mendeléeff’s generalisation was Lothar Meyer, who, dealing at the outset with one of the characteristic properties of the elements—viz., their specific or atomic volumes (that is, the values obtained by dividing their specific gravities into their respective atomic weights)—greatly developed the principle of periodicity, representing it graphically in a most striking and suggestive manner, leading up to a classification almost identical with that of Mendeléeff.