the boiling points of the two chlorides (TiCl, B. P. = 136o). Mendelejeff thinks that the discrepancies between the numbers obtained by Rose, Pierre and others as representing the atomic weight of titanium may possibly have been due to the unsuspected presence of EsCl, in the TiCl, analysed by them.

The discovery of eka-silicon is still in the future; yet looking to the history of gallium and scandium we may almost consider eka-silicon as one of the known elements.

III. The periodic law has also been successfully used as a guide in the comparative study of the properties of elements already known.

To which group of elements does beryllium belong? Is the formula of the oxide BeO or Be,O,, and of the chloride BeCl, or BeCl,? Is the atomic weight of beryllium 9 or 135?

The arrangement of the elements in accordance with the periodic law seems to necessitate the placing of beryllium in group II.; but recently amassed experimental evidence suggests, in the opinion of some chemists, necessitates in that of others, that this metal should be placed in the group characterised by the power of forming an oxide R2O. The atomic weight of beryllium = n. 91. The data regarding the specific heat of beryllium have been presented in Chapter I. par. 28, and it has there been shewn that, so far as specific heat data are concerned, the atomic weight of this element is probably represented by the number 91. The specific heat of beryllium, we found, increases rapidly as temperature rises, and in this respect shews an analogy with the specific heats of boron, carbon and silicon. If the mean values of the specific heats for the temperature-interval 0° to 100° of the metals in series 2 and 3 are multiplied into the atomic weights of these metals, it is found that the atomic heat decreases as the fusibilities of these metals decrease and as the atomic weights increase; thus,

The value to be assigned to the atomic weight of beryllium cannot however be regarded as finally settled by the determinations already made of the specific heat of this metal.

Nilson and Pettersson' insist that the molecular heats and specific volumes of the oxide and sulphate of beryllium, and also the atomic heat of oxygen in this oxide, establish the formulæ of these bodies to be Be2O, and Be. 3SO, respectively; and therefore, these chemists argue that the atomic weight of the metal is 13.65, and beryllium must be placed in the group which comprises aluminium, gallium, indium, &c. On the other hand, Brauner' gives data from which it would formula weight\

appear that the molecular volumes (i.e. specific gravity)

of beryllium oxide and sulphate, assuming the formulæ BeO and BeSO, as correct (Be = 91), are what might be expected from the position of beryllium (91) in group II. series 2.

Brauner gives in tabular form the data concerning the molecular heats (i.e. formula weight × specific heat) of metallic oxides, arranging the metals in groups and series. These data shew that the value of the 'molecular heat" of the oxides-calculated in each case for one atom of metal in the oxide-varies considerably: 'the oxides of the metals of a 'natural group have nearly the same molecular heats, but the ' value increases as the atomic weights of the metals increase.' If the atomic heat of the metal in each of these oxides is deducted from the molecular heat of the oxide-calculated as mentioned above the remainder represents the 'atomic heat of oxygen in the oxide.'.

Brauner gives the following table.

1 Ber. 13. 1459.

3 Loc. cit.

2 Ber. 14. 53.

These expressions must be taken as meaning, in one case, the product (formula weight spec. heat), and in the other, (formula weight of oxide × spec. heat) (atomic weight of metal in oxide x spec. heat).

'Atomic heat of oxygen in metallic oxides. (BRAUNER.)

The atomic heat of oxygen in the oxides increases as the atomic weights of the metals in each group increase; the value of the atomic heat of oxygen in beryllium oxide is smaller than the value for any other oxide in the group; hence beryllium should come in group II, series 2.

Carnelley's determination of the melting-point of beryllium chloride (see ante, par. 108) points to the beginning of group II as the proper position for beryllium, and hence to the number 9'1 as the atomic weight of this metal.

The general chemical characters of beryllium salts are summed up in the three statements' (Be = 91):

Hence we may conclude that there is a large probability in favour of the value 9'1 for the atomic weight of beryllium. This conclusion is supported by Hartley's observations on the spectrum of beryllium and his comparison of that spectrum with those of metals in group II. and III'.

Nilson and Pettersson have very recently succeeded in gasifying beryllium chloride; their determination of the density of this compound in the state of gas shews that the formula BeCl, (Be = 9′1) really represents the molecular weight of the substance".

The mean values for the atomic weights of the three metals, cerium, lanthanum, and didymium, deduced from the most trustworthy data are

Numbers somewhat smaller than the values in column (1) were formerly generally adopted. In the earlier work on these metals very varying results were obtained by different chemists. Mendelejeff proposed to multiply the generally accepted atomic weights of cerium and didymium by 15 and that of lanthanum by 2; he thus got the values Ce3 139; Di=138; La = 180. If this multiplication is performed on the more accurately determined values given in column (1) above, we have Ce 141; Di= 144; La = 1847.

=

Cerium forms two oxides, CeO and Ce,O, if Ce = 94, Ce,O, and CEO, if Ce 141. Mendelejeff placed cerium in group IV, the general formula for the highest oxide characteristic of this group being RO,; lanthanum he placed in the same group but in the series next after that which contained cerium; didymium found a place in group III (oxide = R2O1), coming after yttrium and indium and preceding erbium*.

1 C. S. Journal Trans. for 1883. 316.

2 Ber. 17. 987.

3 In a note (see Chem. News, 41. 49, note) Mendelejeff gives distinct reasons for thinking that this number, although that deduced from the best experimental evidence then available, is too small.

The atomic weight of erbium was not then determined with any accuracy.

The properties of cerium compounds were fairly well known at this time, and Mendelejeff's arguments were very strong'; the facts known concerning salts of lanthanum and didymium were however scanty, and Mendelejeff did not strongly press the arguments in favour of the new positions assigned them in his scheme of classification. More recent and trustworthy work has established the numbers in column (1) above as the equivalents of the three metals (the atomic weight of erbium has also been established as about = 166); hence the positions assigned by Mendelejeff to didymium and lanthanum must be altered. Cerium remains

in group IV (RO), series 8. Didymium occupies a position in series 8 of group V; the oxide Di,O, and the oxychloride DiOCl, which ought to exist if this position is correct, have lately been obtained by Brauner3; (Di̟O, is much more stable than Bi̟,O,, as would be expected from the position of didymium in the periodic arrangement). No place is however found for La= 1847; but if the equivalent of lanthanum (92.33) is multiplied by 15, then La (1385) will occupy a place also in series 8, but in group III (R,O,), being preceded by indium and succeeded by an element, as yet unknown, with atomic weight about 160.

The numbers obtained by Hillebrand for the specific heats of the three metals under consideration fully confirm the values assigned to the atomic weights of these metals by the application of the periodic law; thus,

CeF. H2O and 2CeF4.3KF. 2H2O. (Brauner, loc. cit.)

3 C. S. Journal, Trans. for 1882. 73.

Pogg. Ann. 158. 71.