The specific heats of both B and C rise rapidly with the temperature. + W. Preyer's value for Cr is 6-2, which is entirely without foundation though it will probably prove to be nearly correct. The atomic heat affords another proof of Lothar Meyer's classification of sodium with the alkali metals, rather than that of Mendeléeff, who, followed by W. Preyer, places it in the copper group. Its chemical affinities to potassium are very close; thus E. Augé (Comptes Rendus, 1890, cx. p. 1139) obtained a sodium alum, by regulating both concentration and temperature. To get comparable results, the specific heats must be observed at a constant temperature. With regard to meltingpoint, magnesium presents an irregularity in the zinc group, if the usual melting-point, 430°, is adopted; but the number has been corrected by more recent observation, restoring complete harmony to the series. The order of melting-points of Sn and Pb in Group IV., and of Sb and Bi in Group V., both apparent anomalies, will be seen to favour, rather than contradict, the laws of atomic heat and fusion; for in both cases the atomic heats are reversed to correspond with the melting-points. I have placed cobalt in IV.b, and nickel in VI., notwithstanding Zimmermann's values for their atomic weights; and in doing so, as before, I have been guided entirely by the atomic heats, leading to Mendeléeff's original classification, which placed Co before Ni on purely chemical grounds, from the analogies of the former to Rh and Ir, and of the latter to Pd and Pt (Ann. Supp. viii. p. 133 et seq.). Zimmermann (1886) finds Ni=58.71, Co=58.89. The above tables appear to establish the following laws :— 1. In groups I. and II. (which I will call the metallic groups), as the atomic weight increases, both the meltingpoint and the atomic heat decrease. 2. In groups III. and IV. and all the a groups (termed collectively intermediate groups) the atomic heat increases with the atomic weight, while the melting-point decreases; the former being characteristic of non-metals, the latter of metals. 3. In groups V., VI., and VII. (the non-metallic groups), as the atomic weight increases, so do both atomic heat and melting-point. 4. In the b groups the melting-point increases with the atomic weight, whilst the atomic heat decreases; the former being a non-metallic, the latter a metallic characteristic. The a may be called copper groups, and the b iron groups. It is to be noted that both groups IV. and V. become irregular in the same way when their last members are reached, these having more marked metallic characters. Thus both Sn and Pb in group IV., and Sb and Bi in V., have 2. The variation is greatest with elements of low atomic weights, becoming less and less as the atomic weight increases. I hope in the next place to show in a tabular form that the melting-points of the elements are intimately connected with their atomic heats, and that four general laws govern this connexion. Were sufficient data to hand with regard to latent heat of fusion &c., there is no doubt that many similar relationships would be established. There is evidently plenty of work before physical chemists in this direction. In the arrangement of the groups of elements I follow a plan used by Lothar Meyer, of using Roman numerals combined with lettering. Thus the Li group is I., the Cu is I.a.; Group II. is Ca &c. ; II. a Be &c.; II. Fe, Ru, Os. In placing the iron elements here, and not in a separate group, I am following W. Preyer, Das genetische System der chemischen Elemente, Berlin, 1893. The groups are entirely arranged by their atomic heats, but will be found to be practically identical with Lothar Meyer's classification. The melting-points taken are the most correct up to date, from (1) the late T. Carnelley's Physico-Chemical Constants, 1887; (2) H. Landolt and R. Börnstein, Phys.Chemische Tabellen, Berlin, 1894; (3) The Chemical Society's Journal. Several determinations are very rough, others are altogether wanting. The specific heats, which are much more complete, are from the last two sources and Watts' Dictionary of Chemistry.' They are all taken at as near as possible the constant t°, 15°. The atomic weights are from Landolt and Börnstein and the Chemical Society's Journal. From these and the specific heats I have calculated the atomic heats with much greater accuracy than has hitherto been thought Blanks denote that no reliable data are to hand. A query denotes that the observation has been made, but not accurately. necessary. 6 *Bunsen's value for Ca seems too high; the atomic heat of this element will probably prove to be less than that of K, |
