NUMERICAL RELATIONS OF THE ELEMENTS. .927 present regarded as elementary) or radicles of equal atomic weight, but endowed with distinct properties. b. A radicle intermediate, in properties and in its combining number, between two other radicles of the same group, might be produced by the union of half a molecule of the two extreme radicles. c. And, finally, the supposed constitution of these radicles (or bodies at present regarded as simple) might be assimilated to the compound radicles of organic chemistry of known constitution. There would be, however, this important distinction between the radicles of mineral chemistry and those of organic origin; viz., that the radicles of inorganic chemistry possess a stability indefinitely greater than those of the organic creation-a stability, indeed, of such an order, that the present resources of analytical chemistry are insufficient to effect their decomposition. The probability, on the other hand, of such views would obviously be negatived if the elements exhibited no such multiple . relation in their equivalents. Certain remarkable relations which exist between many of these numbers have been pointed out by various chemists. The whole subject of atomic weights has recently been submitted to a careful revision by Dumas (Ann. de Chimie, III. lv. 129). As the result of his investigations and calculations, Dumas concludes that, in a modified sense, Prout's law is true*; and he considers that the elementary bodies, the atomic weights of which he regards as accurately known, may be arranged in three groups, or rather two groups, if the duplication of the atomic weights adopted in this work be followed-viz., * Stas, however, arrives at an opposite conclusion. He has lately published a long and most laborious series of researches, which it appears almost impossible to surpass in precision. In this memoir he gives the following numbers for some of the most important elements (0=8) : These results have been obtained in each case by several different processes; and the differences between the various numbers thus arrived at for the same body are in all cases much smaller than the difference between the mean result and the whole number required by Prout's law. These numbers were obtained by operating with balances of unusual delicacy, and upon quantities much larger than is customary in researches of this kind-the quantities amounting in some cases to nearly a pound weight of the materials. The numbers thus obtained do not differ from those of Dumas, or indeed from those in general use, sufficiently to affect any calculations of analyses founded upon them; but from the variety of methods employed, and the extra 928 RATIO OF PROPORTIONAL NUMBERS. 1. Bodies which are represented by multiples of a whole number of the atomic weight of hydrogen. 2. Multiples by the number o'5 of that of hydrogen. 1. Bodies which are multiples by a whole number of the equivalent of hydrogen : 2. Multiples by 0'5 of the equivalent of hydrogen. The relations exhibited between the numbers of many of these bodies which are chemically allied are often very remarkable: 1. It has been observed that, in several instances where two elements are in close chemical relation to each other, they have atomic weights which are identical; this happens, for example, with the following pairs of bodies: Cobalt and nickel. 59 92 104°2 197'2 ordinary degree of care and precision with which the experiments were made, the results are of the highest value in relation to the hypothesis of Prout. M. Stas concludes his memoir with these words: "So long as we adhere to the results of experiment for the establishment of the laws by which matter is governed, we ought to consider the law of Prout as a pure illusion, and should regard the indecomposable substances of our globe as distinct bodies, having no simple relations between their atomic weights. The undeniable analogy in properties which is observed in certain of the elements must be sought for in other causes than those derivable from the relations in weight of their acting masses.' We cannot too jealously watch any twisting of experimental data to suit our theories; and those who are familiar with the speculations on the numerical relations of the atomic weights, cannot but feel that the severe method of induction from facts has in this case been more than usually departed from by the followers of a science which is pre-eminently one of experiment. RATIO OF PROPORTIONAL NUMBERS. 929 2. In other cases, the ratio of the atomic weights is as I to 2; for instance : 3. It has also been stated that, where three elements belong to the same natural group, the atomic weight of the intermediate element is frequently equal to the mean of those of the two exThis is true in the case of tremes. the number for sodium being the arithmetic mean of those for lithium and potassium: but this is the only case in which this relation is rigidly in accordance with the experimental numbers. Several groups agree very nearly with such a supposition, but the divergence is, notwithstanding, too great to admit of being attributed to errors of experiment. For example: In both these groups, the difference between the experimental and the calculated number of the intermediate elements amounts to 10; and it is probable that this difference is physically true. In cases like the lithium and calcium groups, it has been suggested, both by Pettenkofer and by Dumas, that the relation between the different members of the group may be analogous to that observed in bodies of organic origin which belong to the same homologous series. The reader who is desirous of pursuing this speculation will find it ably and clearly discussed by Dumas in his paper already cited (Ann. de Chimie, III. lv. 164). II. 3 0 |