In support of this arrangement, he quotes the fact that some of the elements appear to be able to form higher oxides in the presence of other oxides, which act as oxygen carriers (see pp. 174, 177-8), though these higher oxides are certainly not salt-forming. He also deduces, from the rates of hydrolysis of the sulphates, that the elements fall into two parallel series, according to the strengths of the hydroxides as bases, on which ground he justifies the distribution throughout series 8 and 9. There can be no doubt, however, that this disposition is far less in accordance with the behaviour and properties of the rare earth elements than is the first arrangement, which places them in a transition zone between barium and tantalum; it is impossible, for example, to reconcile the properties of praseodymium with those of columbium and tantalum, or to find the slightest analogy between neodymium and molybdenum or tungsten, as the second arrangement requires. The analogy of the rare earth group to the elements of Group VIII has been pointed out by many authors.[179] On the ground that the rare earth elements cannot be spread over the table in series 8-10, Steele[180] favours the early classification of Thomsen, according to which the elements are divided into three groups. The first, corresponding to Groups I and II of Mendelejeff’s table, consists of two sub-groups, each containing seven elements[181]; the second, corresponding to the first two long series of the periodic table, has two sub-groups, each of seventeen elements, of which the first and last seven are analogous--these elements fall into the same groups in the periodic table--whilst the middle three are interperiodic. These interperiodic elements are those which Mendelejeff places in Group VIII. The third division consists of one (or two) group(s) of thirty-one elements; here again, the first and last seven are analogous, whilst the interperiodic elements, which are seventeen in number, include the rare earth metals. [179] Compare Biltz, _Ber._ 1902, ~35~, 562. [180] _Chem. News_, 1901, ~84~, 345. [181] The inert gases are not included. Steele’s idea has been extended by Werner,[182] who has drawn up a table to illustrate it. In this classification, the elements are arranged in order of atomic weight, but arbitrary gaps are left in such a way that similar elements may fall into the same vertical columns, as in the periodic table. The arrangement has the advantage that the interperiodic elements, consisting of the rare earth elements and the elements placed in Group VIII of the periodic table, here do fall in the middle of their respective periods, but it has several drawbacks, and does not represent the transition of properties from element to element so well as the helical representation of the periodic table, which brings out most clearly the true relations between the elements, and the anomalous position of the rare earth metals. [182] _Ber._ 1905, ~38~, 914.