same time they show that as the value for CH, between the methides and ethides is not far from the truth, the values for the methides may be taken as nearly correct, and where a value for the methide is wanting an approximate one can be got by subtracting from the value for the ethide 75 for each CH2. Thus we get the following Table for the methides only : TABLE XXVI. (M21)*. ... (M2)3. ... 5.1 ZnMe, HgMe. SiMe. SnMe,. PbMe. BMe,. NMe,. PMe,. 5.7 6.7 7.5 8.2 5.6 5.5 6.1 ...... Here again values are given for the methides of N, O, S, Cl, Br, and I, only as checks. The first point to ascertain is whether in the true metallic methides (M2) follows the laws proper to organic compounds or inorganic metallic compounds. In the organic compounds CH, has a value 9, and the other H of CH a value 1.1, so that CH, is 2, and thus the values for SiMe, and SnMe4 are less than for Me: therefore the organometallic compounds do not behave as ordinary organic compounds; but as we have seen the valency of the metallic atoms play an important part in the values of (M27) for their inorganic compounds, it will be best to try the effect of treating the methides as we did the metallic chlorides and similar compounds. That is, we must divide the values of (M2) in Table XXVI. by n, where n is the valency of the metallic atom; then, subtracting 2 for CH, from each of the results, we get the values of F/n for the metallic atoms. These are given in the first row of the following table, the second row containing the values from Table XVIII. TABLE XXVII. Zn. Hg. Si. Sn. Pb. B. N. P. As. Sb. 1.6 2.1 1.35 1.75 21 1.2 1.2 1.5 1.55 1.9 .8 85 1.05 F/n F/n from Ta. XVIII. ... .6 .95 .6 It appears that the values of F for the metallic atoms in the organometallic compounds are double the values in the inorganic compounds, whether the atoms are dyad, triad, or tetrad; and this seems to correspond closely with what we found in the stud, of the ratio B/F for the compound acid radicals, because for basicities higher than one this ratio came out only half the value for basicity one, which corresponds to a value of F double what would be expected from other considerations. As these facts show a characteristic influence belonging to compound radicals, it will be instructive to return to the case of the nitrates of the dyad metals, in which, according to what we have just seen, the values of (M2) in Table XXII. when divided by 21' and reduced by 3-6, the value for NO3, ought to leave values for F/2 for Ca, Sr, and Ba which are double the values obtained from the halogen compounds. Performing these operations we get for The values of F/2 from the nitrates are only 1.5 and not 2 times the values from the chlorides; and thus the nitrates of these metals do not come satisfactorily under this principle, and there are too many steps in the process of determining the value of F/2 from the nitrates to enable us to fix upon a probable cause of the discrepancies in these nitrates. With the organometallic compounds we have finished the data for compounds to be discussed in the present paper, and have now an opportunity to make a comparison between the values contributed by the atoms of the metals to MB and to (M2) in a molecule, though to keep it clear that in the compounds of the metals we have had to take account of valency n we will give the values of B/n and F/n in the following table. There is considerable difficulty in getting the atomic volumes of such atoms as P and As, because the relations of the chlorides, bromides, and iodides are not consistent with those of the chlorides, bromides, and iodides of the monad metals; the volumes of the lower members of the monad and dyad series are also difficult to assign, being small. In the case of P, As, and Sb the volumes adopted are those given by Thorpe (Journ. Chem. Soc. 1880) as the domains of these atoms in liquids at their boiling-points; they are therefore probably too large, but not necessarily much too large, because the domains of the halogens in organic compounds at their boiling-points are almost identical with their limiting volumes in these compounds, and also in metallic compounds. Accordingly the values of the limiting volumes of the atoms given in the next table are only such general approximate values as represent best the facts of the simplest compounds. A glance at the numbers for the Li and Be families of metals shows that the ratio of B to F is not constant, as we found to be approximately the case with non-metallic atoms and radicals; but when we get to the families which are half non-metallic and half metallic, as in the P family and the C family, we see that the ratio, except in the case of C itself, is not far from 10, the value which we have hitherto found for non-metallic atoms and radicals. For further assurance as to the reality of this distinction between metals and nonmetals, the best course will be to determine values of M2l for the uncombined metals. 3. Determination of M2l for the Uncombined Elements. Data are available for both methods of calculating M2l, and that of the Kinetic Theory of Solids will be taken first as being applicable to a greater number of substances. At first the molecular mass will be regarded as unknown, and for M the usual atomic weight will be used in equation (9). The following table contains the values of T the melting-point and of M/p from L. Meyer's 'Modern Theories of Chemistry,' and the values of M2/ [not (M2/)*] calculated from them by equation (9), as well as the values of the ratio (M/p)/M2l [not (M/p)/(M2)]. Equation (9) applies strictly only to the metals, but in a formal manner it is here extended to some non-metals. The first point of importance brought out by this table is that in the main families the ratio (M/p)/M2l has a constant value characteristic of each family, and in the subfamilies a value simply related to that of the main family: thus in the first family the range in the value of the ratio is only from 2.9 for Li to 2-6 for Cs, with a mean value 27, of which the value for Cu and Ag in the subfamily is a quarter. In the second family the ratio is constant from Be to Ba with a value 10, but the mean value 1.3 for Zn and Cd in the subfamily is greater than that of the main family, which is in strong contrast to the behaviour of the copper subfamily, and may have some significance, as will be seen presently; the distinct manner in which Hg separates itself with a value 3 should be noticed. In the third family there are only two members, Al and La, which both have the same value for the ratio, 9, and in the subfamily Ga has 3 times this value, in which respect it is similar to Hg, and In has double the value of the main family, Tl has perhaps 1.5 times that of the main family. In the fourth family there is only one main representative Ce with a value 75, of which that for Sn is exactly double, as was the case with In and the third family; the value for Pb appears, like that of Tl, to be 1.5 times the value of the main family. In the fifth family a certain arbitrariness in our separation into main and sub-families becomes apparent, because P, As, Sb, and Bi have the close relationships of a main family; Di, which is the only metallic representative of the fifth family, has a value 62 for the ratio, of which those for As, Sb, and Bi are double, and that for P is again double those for As, Sb, and Bi. In the sixth and seventh families, where the metallic character disappears, regularities in the ratio (M/p)/M2 disappear, but in these nonmetallic elements the tendency is towards M/p being proportional to (M27) and not to M27. In the eighth family the ratio (M/p)/M2l has a mean value 41. The second point of importance brought out by Table XXIX. is that the ratio (M/p)/M2 for each main family exhibits a remarkable relationship to the valency of the family when the main family is metallic. This is shown in the next table, where the first row contains the valency of the family, the second the value of the ratio (M/p)/M2 for the family, and the third the product of these two. This table shows that if n is the valency, or perhaps it is safer to say the order of the main family, then `n (M/p)/M2/` |