into nine great groups, all absolutely natural, each element in its proper place accurately fitted.

An examination of Table I will show,

1. Hydrogen stands at the head of the halogen group, with the members of which it makes countless substitutions. Hydrogen has always been an element difficult to class, but this is its proper place.*

2. Next follows the group of the metals of the alkalies, with sodium in its right place. With these, hydrogen is intimately

connected.

3. The calcium group begins with that metal, either because earlier members with lower atomic weights exist, but have not been discovered, or because the group naturally commences at a later stage than the preceding. The same thing is true with the scandium group next following.

4. The remaining groups need no particular description. They are eminently natural and very familiar.

5. In these groups following the horizontal lines, all members of each group have the same valency. To this there is no exception.

The foregoing applies to this table considered as groups corresponding to the horizontal lines. When it is intended to consider the serial relations according to columns they should be taken as one long serial without breaks.

Much importance has naturally been attached by Mendeléef to the progressive stages of oxidation in his groups, as for example, the series

Na,O, Mg,O,, Al,O,, Si,O,, PO̟„, SO̟, C10,.

39

69

And the corresponding hydroxides, hydrides, methides, etc. Of these it may be said that C1,O, is purely hypothetical. The rest of this series will be found in precisely the same order in the second column in Table I and all the corresponding elements in each horizontal line have the same valencies and tend to form similar compounds.

This table representing the first great division of elements includes all those whose ions function as anions with part of the

kathions.

*By Roscoe & Schorlemmer and in the Graham Otto treatise it is placed immediately before the halogens. By Ramsay at the head of the lithium sodium group.

Transitional Elements.

Intermediate between the two divisions is a small class of eleven transitionals. These elements have ions which at some valencies are colored and others colorless.

(As to tantalum and niobium, satisfactory information is deficient and they are placed here provisionally.)

An inspection of the table will show that the metals composing it have little relation with one another.

Most of them on the contrary are noticeable as having each some marked peculiarity. Titanium when heated combines with atmospheric nitrogen. Vanadium forms a compound radical simulating a metal so completely that all chemists were deceived until Roscoe detected the fact. Copper forms a solution which has the property of dissolving cellulose. Silver can assume a condition in which it is freely soluble in water; it is also remarkable for the high sensitiveness to light of some of its compounds. Tungsten is remarkable amongst inorganic substances for the unexampled complexity of its compounds, as first observed by Gibbs. Thallium is noted for uniting in a singular way the properties of an alkaline and those of a heavy metal.

The relations of this peculiar group are chiefly with the elements having colorless atoms only, as shown in Table II. Into this, they enter as it were in a block, at the center, increasing the number of groups to thirteen, or twelve if we take the first two as constituting a double group of univalent elements. In this way as we follow the groups downwards the valency increases successively from one to six, then begins at one again and increases regularly from one to six again. This will be better seen by referring to Table II. In every group each member has the same valency.

Whether it is probable that all the blanks represent so many elements remaining to be discovered seems doubtful. But it is certainly the opinion of many chemists that the number of existing elements as yet unknown, is large.*

* Cf. Ramsay, Inorganic Chemistry, p. 87.

TABLE II.-Elements with all Ions Colorless, and Transitional Elements (in italics) in numerical order, forming thirteen Natural Groups.

With elements having atoms always colored the relations of the transitionals are so slight that I hesitated about giving them a place in Table III and have done so chiefly to give emphasis to this fact, that in the entire series of elements there is not a single case in which an element having atoms always colorless appears in the regular numerical series between a transitional element and one with atoms always colored. Also that there is not a single case in which an element with atoms always colored appears in the numerical series between a transitional element and one having colorless atoms only.

This will be seen better by examining the diagram embracing the entire series. This perfect regularity seems to justify this new method of classification.

This group contains elements whose atoms function as kathions only.

In Table III will be found the series of elements with all colored ions and to these have been added the transitionals, distinguished by being printed in italics. The transitionals fit equally well into either of the two great divisions, that of the colorless and that of the all colored ions, with this difference that in the first division they fit into the center, in the second division they act as outliers to the respective series, connecting the colored with the colorless. This last function is however better shown by the diagram at the end. Their chemical relations are with the first division.

The colorless elements when arranged in vertical columns form groups according to the horizontal lines. Members of each group though closely connected in properties, differ widely in atomic weights.

With the elements having all colored ions the case is very different. They fall into four series, members of which have their atomic weights immediately following each other in unbroken succession.

The first of these series consists of the metals chromium 52, manganese 55, iron 58, cobalt 59, and nickel 59. This is a very well marked group, the chromates, manganeses and ferrates being isomorphous. Also the sesquisulphates of the three metals replacing each other in the alums.

Chromium and manganese were formerly always placed in the iron group until the exigencies of the Periodic Law required the transfer of chromium to the oxygen group and of manganese to the univalent halogen group, a translocation for which there seems no sufficient justification.