56 



NATURE 



\Nov. 21, 1889 



tains brings the whole succession of past geological changes 

 within the scope of a general theory of geographical evolution. 



A. J. Jukes-Browne. 



The Composition of the Chemical Elemen:s. 



My excuse for troubling your readers with this well-worn theme 

 is that a definite hypothesis is possible, which, should it be fully 

 borne out by the facts, appears to afford a remarkably complete 

 explanation of the periodic law, as set forth in Prof. Mendeleeff's 

 table. 



The periodicity exhibited by this table is double, alternate 

 series presenting members which have high or low atomic 

 volumes, are fusible or infusible, &c. 



Should the elements be really simple atoms, it- would be im- 

 possible to account for this fact without introducing occult differ- 

 ences of quality, from which it has been all along the aim of 

 chemical science to free itself. Undoubtedly periodical variations 

 in the size and shape of the atoms might account for the dual 

 periodicity of their properties, but nothing satisfactory can be 

 gleaned from such an explanati m. Besides, we are accustomed 

 to regard differences of properties in compounds as dependent on 

 composition, even should their molecular weights be similar. It 

 may also be urged that, if the elements are supposed single, their 

 properties should vary with increase of weight in some con- 

 tinuous manner, and not sway to and fro so remarkably, 1 am 

 aware that Prof. Mendeleeff himself does not take this view (cf. 

 Chem. Soc. Journ., October 1889), but it is one that is widely 

 spread, and is held by other eminent chemists. 



It is, however, possible to push too far such analogies as that 

 of a series of organic compounds. Important differences exist 

 between such a series and that of a natural family of the ele- 

 ments : for example, the specific refraction equivalents are not 

 at all analogous in the two cases. Specific heat determinations 

 show that, as a rule, an element moves as a single .'■olid mass. 

 But these considerations need prove nothing more than that we 

 must be prepared to deal, in the case of the elements, with 

 affinities of a different order — perhaps brought into play by vastly 

 different conditions — from those found in ordinary compounds. 



If the elements are assumed to be composite radicles, then, in 

 stating their hypothetical composition, there is material ready to 

 hand. The famous principle known as "Occam's razor" 

 applies here as elsewhere. Hypothetical elements should only 

 be introduced where other considerations are plainly in favour of 

 the suppositions involved. 



The elements form natural families of two groups each, six of 

 them having for their types the following : Li, Be, B, C, N, 

 and O. 



Since the properties of the typical element run all through the 

 members of a family, then (on the hypothesis that properties 

 depend upon composition) we should expect it to be found in the 

 formulae of the remainder. 



The hypothesis here advanced is, that the periodicity of the 

 properties of the elements is due to the dependence of the pro- 

 perties of each element upon those of the typical element of the 

 family to which it belongs, together with the mode of its c -m- 

 bination with oxygen. In other words, that the elements, with 

 the exception of the first six, are, in a qualified sense, compound 

 oxygen radicles. 



The reasons for the adoption of oxygen are : (i) the remark- 

 able coincidence of the figures for each family upon this hypo- 

 thesis ; (2) that the atomic weights of the oxygen family of 

 elements are whole multiples of that of oxygen ; ^3) the relations 

 disclosed between the numbers of atoms composing the ele- 

 ments, which cannot be other than the result of law ; and (4) 

 the fact that all the elements combine with oxygen, which is also 

 the most plentiful element in Nature. 



Supposing any natural family complete, its two groups are 

 given by the following formulas, R being its typical element : — 



P f Group {a) -. R0„ ROg, ROg, RO^, RO14. 

 \ Group {b) : KO, R.Oj, R^O^, R.O;,, R^Oi^. 



The seventh and eighth families are very incomplete, but may 

 be represented in the same way. 



It will be noted that the numbers of atoms in these formulae 

 are as follow : — 



fS, 6, 9, 12, 15. 

 [2, 5, 8, II, 14. 



The common difference in each group being 3, and the numbers 

 4, 7, 10, and 13 being absent. 



The resemblance of these figures to the atomic weights of the 

 ten typical elements (including four hypothetical ones) is very 

 close. One is almost tempted to regard them as the primitive 

 forms of the combination of matter, and to return to Prout's 

 hypothesis. 



The existence of y^?^r elements between H and Li is indicated 

 as well by the gap which exists between them as by this hypo- 

 thesis. That Fe, Co, Ni, &c., have formulae commencing 

 with Rj, is shown by the fact that they recur regularly in the 

 series having these formulas, their comparative infusibility and 

 low atomic volume indicating also this composition, as well as 

 the fact that, if it were otherwise, the rule observable in the first 

 six families would be broken through. It is, again, hardly pos- 

 sible to suppose that the seventh family, the halogens, should' 

 contain the electropositive hydrogen, although the latter would 

 then lose its unique position, and in this case the difference be- 

 tween the calculated values of Ag and I (i8"9) agrees very nearly 

 with that between those observed (18 '87), the ratio of these 

 latter being very exactly determined by Stas. This, however, is 

 a matter which may well be left undecided for the present. 

 Should fluorine be a fundamental element, the halogen series 

 will break the rule which holds for at least six out of the 

 remaining seven families. 



The following table is constructed on the lines of Mendeleeff's. 

 The seventh and eighth families are placed first in order, and 

 the calculated and observed atomic weights are placed under- 

 neath their respective formulae. Want of data is indicated by 

 blanks, but the rarer metals are omitted, although they mostly 

 correspond to the formulae R20y. It will be no ed that the 

 arrangement gives Mn, Fe, Co, Ni, and Cu an intelligible 

 position in the series. 



It is not to be expected that the calculated and observed 

 figures will perfectly agree, although in some thirty cases the 

 average variation is o'5 of a unit. The chief variations occur ia 

 two series, in which, however, the natural order is preserved, 

 viz. Ti, V, and Cr, with an average error of 4 "5, and all the 

 elements containing Ojj, from tungsten to bismuth, in which the 

 mean difference is 9. It will be noted that this difference holds 

 even in the case of the eighth family, in which the f irmulas con- 

 tain the hypothetical R", R"', and K'''', showing that the errors 

 arise from a common cause. The atomic weights, since the dis- 

 covery of the periodic law, have not been decided upon without 

 reference to one another. This whole series is separated by a 

 huge gap from the rest of the atomic weights, which is only 

 filled in at intervals by the less common metals of the earths, &c., 

 and consequently an error in one of them would certainly affect 

 the whole. Similarly, the differences of 4 between the observed 

 atomic weights of Ca and Sc, and Sc and Ti, are anomalous. 



On the other hand, the coincidences exhibited by the table 

 cannot be the work of chance, and, considering the inexactitude 

 of the determinations of many of the atomic weights, the fact 

 that the average of the differences between the observed and 

 calculated numbers in the large majority of the elements is only 

 one unit, and that the remainder appear to arise from a single 

 cause, is remarkable, especially when we consider the facts 

 which are brought to light by this mode of representation. The 

 law that elements essentially similar differ only by an atomic 

 weight of O3, or its multiple, surely deserves attention. When, 

 again, the difference between the two groups of any natural family, 

 and the periodicity of the properties of the elements, are ex- 

 hibited as the result of composition, the conclusion becomes 

 apparent that we have in the hypothesis at least a guide for 

 future research. 



The atomic volumes of the groups commencing with RO are 

 smaller than in those commencing with ROj. These correspond 

 to the " even " and "odd " series of Mendeleeff, Other proper- 

 ties follow, thus affording a possible clue as to hoxv the charac- 

 teristics of the elements depend upon their composition. 



Without trespassing further upon your valuable space, I will 

 conclude by quoting Dr. Gladstone (Pres. Address, Chemical 

 Section, Brit. Assoc, Soulhport, 1883): — 



" The remarkable relations between the atomic weights of the 

 elements and many peculiarities of their grouping, force upon 

 us the conviction that they are not separate bodies created with- 

 out reference to one another, but that they have been originally 

 fashioned, or built up from one another, upon some general 

 plan. This plan we may hope to understand better ; but if we 

 are ever to transform one of these supposed elements into an- 

 other, or to split up one of them into two or three dissimilar 

 forms of matter, it will probably be by the application of some 

 method of analysis hitherto unknown." 



