In the table on p. 225 iodine and tellurium are placed in series 7, but I > Te. The older determinations of the atomic weights of these elements made Te> I; nor is this result contradicted by the recent work of Wills'. Nevertheless as the numbers obtained by Wills range from 12607 to 1280, we are, I think, justified in provisionally placing tellurium in group VI and iodine in group VII. To reverse the positions of these elements would be entirely to obscure the analogies of both with other elements2.

Uranium is another element the comparative study of the properties of which has been much advanced by the application of the periodic law. The atomic weight of this element has been established as = n. 120. If n = 1, the three oxides of uranium must be formulated UO, UO̟, and U„O̟; but there is no place for an element with this atomic weight and forming these oxides in the periodic arrangement. If however n=2, then (U=240) the oxides become UO,, UO,, and UO, and uranium finds a place in VI-12. The preceding members of this group-Cr, Mo, W-yield oxides (RO) which are acid-forming. But a comparative study of the relations between the properties of oxides and the atomic weights of the elements in these oxides shews, that as the atomic weights of the elements in a group increase, the acid character of the higher oxides formed by these elements becomes less marked (e.g. CrO, is more markedly an acid oxide than MoO, or WO,). Now the highest oxide of uranium is an acid-forming oxide, but its acid functions are less marked than those of CrO,, MOO,, and WO,; salts corresponding to K,CrO, and K,Cr,O, in which Cr is replaced by U are known. Uranic chloride, UCl, if U = 240, resembles MoCl in being volatile and decomposable by water.

The atomic volume (i.e.

1 C. S. Journal Trans. for 1879. 704.

atomic weight) of the four spec. gravity

2 Brauner has recently obtained values for the atomic weight of tellurium varyfrom 124'94 to 125°4 (mean =125); he has shewn that the process employed by Wills gives too high results unless great precautions are taken. (See abstract in Ber. 16. 3055.)

metals, Cr, Mo, W, U, increases as atomic weight increases, the values being Cr=76; Mo= 11; W = 11; U=12′5.

Hence from the comparative study of uranium compounds guided by the periodic law, we appear to be justified in adopting 240 as the atomic weight of this metal.

Recent determinations of the densities of gaseous uranium bromide and chloride, and of the specific heat of pure uranium, have fully confirmed this number (see ante, Chap. I. pars. 19, and 25).

112. The facts enumerated in the preceding pages undoubtedly establish the periodic law on a firm basis, and justify the employment of this law as one of the main guides in a general scheme of chemical classification'.

The following arrangement of the elements (the table is taken, with a few alterations, from a paper by Mendelejeff in Ber. 13. 1804) is in the opinion of Mendelejeff himself the best for clearly setting forth the general teaching of the periodic law. (See next page.)

Each group-except group VIII-contains members belonging to odd and to even series; or it may be said that each vertical column, or large series, is subdivided into two parts having seven elements in each. The entire column, comprising an odd and an even series, forms a 'long period '; the seven members in the even or in the odd series form a 'short period.' The members of group VIII form ‘transition 'periods' from series 4 to 5, 6 to 7 (probably 8 to 9), and I to II. Including the 'transition periods,' each 'long period' theoretically contains 17 elements.

Because of its peculiar properties, and also because of the anomalous relations between the values of its atomic weight and those of succeeding elements, hydrogen is regarded as the sole representative of group I, series I.

Comparing series, we find closer analogies between corresponding members of odd or of even series, than between those of odd and even series: thus, comparing series 4 and 6,

1 It is very unfortunate that Mendelejeff's Treatise on Chemistry, in which, as I understand, this law is made the basis of a general system, should not be published in some one of the languages of Western Europe.

and 4 and 7, potassium and rubidium are seen to be more closely related than potassium and silver; calcium and strontium, than calcium and cadmium; vanadium and niobium, than vanadium and antimony. Again, comparing series 5 and 7, and also 5 and 6, it is seen that the relations between zinc and cadmium, or between arsenic and antimony, are closer than those between zinc and strontium, or arsenic and niobium.

Omitting the typical elements, it may be said that, as a rule, the most markedly nonmetallic elements are placed in odd series. Also, that the passage from an even to an odd series is accompanied by a gradual change, but that from an odd to an even series by a more sudden change in the properties of the elements; thus chromium and manganese resemble copper and zinc much more than selenion and bromine resemble rubidium and strontium, or than tellurium and iodine resemble cæsium and barium. It may also be laid down as a general proposition that volatile organometallic compounds are formed only by metals which occur in odd series; should such compounds be hereafter formed containing metals which belong to even series, the properties of the compounds in question will probably differ much from those of the volatile organo-metallic compounds at present known. (Mendelejeff.)

The elements which form the 'transition periods' (group VIII) possess many characteristic properties. They are very infusible, have small atomic volumes, and occlude oxygen and other gases; oxides of the form RO, are met with in this group only; the highest oxides are basic or very feebly acid; these metals form stable alkaline double cyanides KRCу, K,RCy,, or K,RCу,, and also stable ammoniacal compounds'.

The elements in series 2 (from lithium to fluorine), and perhaps the first member of series 3, viz. sodium, are grouped together as 'typical' elements. There is no 'transition 'period' coming between the even series 2 and the odd

M. C.

1 See Mendelejeff, Chem. News, 40. 267.

16

series 3 as there is between series 4 and 5, 6 and 7, and 10 and II. The mean difference between the atomic weights of two elements in successive even series and in the same group (e.g. between potassium and rubidium, or between rubidium and cæsium) is 45; but the mean difference between the atomic weight of an element in series 4 and the corresponding element (i.e. the element in the same group) in series 2 is 35: hence we should expect to find the relations of series 2 to other series different from the general mutual relations exhibited by these other series. As the lower members of an homologous series of carbon compounds are sometimes characterised by the possession of properties which do not belong to the higher members, so the elements with atomic weights ranging from 1 to 19 (23) are characterised by special properties; they are 'typical' elements.

113. As the atomic weight increases in each group, the basic character of the higher oxides formed by the members of the group becomes more marked, and at the same time these oxides become more easily reduced. It is also to be noted that the composition of the more stable haloid and oxyhaloid salts (and in some cases of the more stable salts as a whole) tends, as atomic weight increases, to correspond in form with an oxide containing less oxygen than the highest oxide. These statements hold good more especially for those members of a group which occupy the odd series. Group V presents a good example. Sb,O, is more basic than P,O,, and Bi̟O, is marked by an almost complete absence of acid properties. The highest oxides of this group belong to the form RX, (see p. 243); the stable haloid and oxyhaloid salts of phosphorus, vanadium, niobium (PF, VOCI,, NbCl,), belong to the same form, but the bismuth haloid and oxyhaloid salts are BiCl,, BiBr,, BiOC1, BiOBr, &c., which belong to the form RX,, characteristic of the lower oxide Bi̟O ̧.

5

The first and last members of a series, and more especially of a 'long period,' present marked differences in their general chemical behaviour; thus lithium, potassium, and rubidium, the first members of the long periods 1, 2, and 3, are strongly positive, whereas the last members of the same periods,