properties repeat themselves periodically, and that the order of the faculty of the elements to combine with other elements also corresponds with the order of their atomic weights. This is seen in the following simple example.

All the elements having an atomic weight of not less than 7 and not more than 35·5 fall into two series:

Li 70 Be=9.1 B=110 C=12·0 N=14·0 0=16 F=19.0 Lithium Beryllium Boron Carbon Nitrogen Oxygen Fluorine Na=23'0 Mg = 24.3 Al=27.0 Si = 284 P-310 S=32·1 Cl = 35.5 Sodium Magnesium Aluminium Silicon Phosphorus Sulphur Chlorine

Each pair of elements present a great similarity in their chief properties; this is especially marked in the higher saline oxides, which in the lower series are :

or

Na2O, MgO, Al2O3, SiO2, P2O5, SO3, C1207,

Na,O, Mg,O,, Al2O3, Si2O4, P2O5, S206, Cl2O7.

Thus the atomic order of the elements exactly corresponds to the arithmetical order from 1 to 7. So that the groups of the analogous elements may be designated by the Roman ciphers I to VII: and when it is said that phosphorus belongs to group V, it signifies that it forms a higher saline oxide P2O. And if the analogues of argon do not form any compounds of any kind, it is evident that they cannot be included

in any of the groups of the previously known elements, but should form a special zero group which at once expresses the fact of their chemical indifference. Moreover, their atomic weight should necessarily be less than those of group I: Li, Na, K, Rd, and Cs, but greater than those of the halogens, F, Cl, Br, and I, and this a priori conclusion was subsequently confirmed by fact, thus:

The five well-known alkali metals correspond to the newly discovered argon analogues, and the atomic weights of both exhibit the same common law of periodicity. But the halogens and alkali metals are the most chemically active among the elements, and are, moreover, of opposite chemical character, the first being particularly prone to react with metals and the others with metalloids, the former appearing at the anode and the latter at the cathode. They must therefore stand at the two extremes of the periodic system, as in the scheme on page 24.

Although this arrangement best expresses

the periodic law, the distribution of the elements according to groups and series in the table on page 26 is perhaps clearer.

Here x and y stand for two unknown elements having atomic weights less than that of hydrogen, whose discovery I now look for.

A reference to the above remarks on the argon group of elements shows first of all that such a zero group as they correspond to could not possibly have been foreseen under the conditions of chemical knowledge at the time of the discovery of the periodic law in 1869; and, although I had a vague notion that hydrogen might be preceded by some elements of less atomic weights, I dared not put forward such a proposal, because it was purely conjectural, and I feared to injure the first impression of the periodic law by its introduction. Moreover, in those days the question of the ether did not awaken much interest, for electrical phenomena were not then ascribed to its agency, and it is this that now gives such importance to the ether. But at the present time, when there can be no doubt that the hydrogen group is preceded by the zero group composed of elements of less atomic weights, it seems to me impossible to deny the existence of elements lighter than hydrogen.

Let us first consider the element in the first