October 13, 1892 



NATURE 



571 



in such a way as to bring out the relation that there is a 

 constant difference of 4 between each halogen and its 

 positive analogue : — 



\'a, 23 - 4 -= 19, F ; K, 39 - 4 = 35, CI ; Rb, 85 - 4 = 81 

 Br; Cs, 131 - 4 = 127, I. 



In a similar way the oxygen group is made into a 

 negative column having positive analogues in the H2n 

 -roup and showing a constant difference of 8 : — 



Mg, 24 - 8 = 16, O ; Ca, 40 - 8 = 32, S ; Sr, 88 - 8 = 80, 

 Se; Ba, 136-8 = 128, Ta. 



Of course chemists have long been familiar with various 

 numerical relationships between groups of allied elements, 

 but this does not appear as sufficient evidence for altering 

 the atomic weights of Br, CI and Se, unless these re- 

 lationships can be conclusively shown to be the necessary 

 result of a general law. 



But apart from such defects as have been pointed out, 

 it will be seen that the proposed grouping breaks down 

 altogether after the third group. The author is hardly 

 fair when he says (Preface, p. iv.) : — " While the multiple 

 relations subsisting among the atomic weights of the 

 other series of elements are highly interesting, they do 

 not possess, in the present state of our knowledge, that 

 degree of precision which is the distinguishing feature of 

 the series Hn and H2n. An exception might, however, 

 be made in favour of the series H3n, &c." 



Asa matter of fact it is not a question of " degree of 

 precision " at all, for, as far as we can see, the other 

 groups do not lend themselves to the Bodeian method ; 

 at any rate, not in the form applied to the groups Hn, 

 H2n, and H3n. We give the author's results as com- 

 pared with those obtained by the application of his own 

 method :— :• 



Group H^n. 



j; = 16 ; a = 32 ; Ti =48 ; Ge = 72 ; Zr = 92 ; Sn = 116 ; 

 La = 140 ; X ~ 164 ; D = 18S ; U — 240. 



The numbers obtained by the rule (i, 2, 3, 4, &c. X 32, and 

 16 subtracted from each product) are 32, 48, 80, 112, 144, 176, 

 :o8, 240, &c., which, after Titanium, do not represent any 



jtni^ weights in the group till Uranium is reached. 



Group H^n. 



B = 10 ; P = 30 ; V = 50 ; As = 75 ; Nb = 95 ; Sb = 120 ; 

 X - 140 ; jr = 165 ; Ta = 185 ; Bi = 210. 

 Calculated (I, 2, 3, 4, &c. x 30 and 10 subtracted from each 

 product) the numbers are :— 30, 50, 80, no, 140^ 170, &c. 



Group Hdn. 



X - \Z; jr - 36 ; Cr = 54 ; Mo = 96 ; x - 144 ; W = 186. 

 Calculated (i, 2, 3, 4, &c. x 36 and 18 subtracted from each 

 product) the numbers are : — 36, 54, 90, 198, cSrc. 



Group Hth. 



N = 14 ; Si :== 35 ; [Fe = 56 ; Mn = 56 ; Ni = 56 ; Co = 56] ; 

 [Pd = 105 ; Rh - 105 ; Ru zz 105 ; Da = 105] ; [Au = 196 ; 

 Pt = 196 ; Ir = 195 ; Os = 196]. 



Calculated (l, 2, 3, 4, &c,, X 35 and I4subtracted from each 

 product) the results are : — 35, 56, 91, 126, 161, and 196. 



The association of nitrogen with silicon and the metals 

 of the iron and platinum groups is, to say the least of it, 

 incomprehensible. We have thought it desirable to give 

 this analysis, for no reason is given in the paper for this 

 particular grouping after the third series, beyond the well- 

 known chemical relationships of the elements which, as 

 we have seen, is sometimes violated in a most unaccount- 

 able way. The groups are obviously not constructed by 

 the Bodeian method ; the atomic weights are modified in 

 many cases by one or two units, and the result is a 

 classification which differs only from the received classi- 

 fication on points which cannot possibly be conceded by I 



NO. I 198, VOL. 46] 



chemists. The reason why silicon is separated from its 

 analogues is as follows : — 



Now, if silicon were the true analogue oF titanium, the 

 oxides of these elements should be isomorphous, whereas the 

 crystalline form of quartz is hexagonal, while rutile, anattise, 

 brookite, zirconia, and tinstone (similar oxides of members of 

 the series H4n) are tetragonal ; consequently, silicon does not 

 belong to the series H4n. 



This is a point, and out of justice to the author we give 

 it for what it is worth,^ but the atomic weight of silicon 

 has been determined by the vapour density of its chloride, 

 and the result is fatal to Mr. Wilde's classification. His 

 attempt to justify the atomic weight 35 by an appeal to 

 the specific heat is unfortunate, because he takes the old 

 determination by Regnault (o'i76) instead of the more 

 recent determination by Weber (0*203 at 230^ C). More- 

 over, he is inconsistent in not allowing the same correction 

 for boron and the other elements which deviate from 

 Dulong and Petit's law. 



We cannot go much further into the details of this 

 paper. Enough has been written to justify the dis- 

 appointment which we expressed at the outset, and it is 

 only the intrinsic importance of all questions bearing 

 upon the origin of the elements that has warranted such 

 extended treatment. It appears that the numerical rela- 

 tions which are brought out by the author's method have 

 either long been known or else — as in his application 

 of the Bodeian method — they do not exist beyond a 

 limited number of groups. The results do not take us 

 beyond the point at which chemists were left by 

 Dobereiner, Pettenkofer, Dumas, and numerous other 

 chemists who, for three-quarters of a century, have 

 directed attention to such numerical relp-tionships. In 

 some respects — such, for example, as in the exactness 

 with which the atomic weight of an element is the arith- 

 metical mean of the elements above and below it in the 

 same series — Mr. Wilde's numbers express the relation- 

 ship more closely than those of any other author ; but 

 this agreement is simply brought about by forcing the 

 atomic weights into the requirements of the case. The 

 increase in density as the odd and even seriesare ascended, 

 is nothing more than an imperfect way of stating the 

 well-known relationship between atomic weight and 

 atomic volume, which is so much better shown by 

 Lothar Meyer's curves. The table of elements presented 

 by Mr. Wilde ignores that fundamental principle of 

 periodicity or recurrence of properties which is the key- 

 note of Mendeldeff's system, and which has led to the 

 general adoption of that system by chemists. We do not 

 pretend that Mendel^eff's classification is faultless; the 

 illustrious founder of the Periodic Law would be the firs|» 

 to admit that his system has certain imperfections. M«:'v 

 Wilde has emphasized a few of these in his preface,, an^ 

 he somewhat summarily dismisses the whole schenie in 

 the following words : — ^ 



From the numerous discrepancies which present themselves 

 in the classification of the elements when, arranged in the 

 regular order of their atomic weights, it will be obvious that 

 the idea of recurring properties or periodic functions, in terms 

 of the vertical series of Newlands or the horizontal series of 

 Mendeleeff, has no more relation to chemical science than the 

 law of the increase of population, or th : laws of variation and 

 inheritance in organic species. 



This paragraph, penned in the present year, will, per- 

 haps better than any other statement that could be repro- 

 duced from the paper, enable chemists to form a correct 

 estimate of the value of the work and of the author's 

 qualifications for deaUng with the question of the origin 

 of the chemical elements. R. Meldola. 



1 " Stannic and titanic oxideg resemble silica both physically and chemi- 

 cally they might be expected to form analogous compounds, and 



be isomorphous with slica. as Marignac (18S9) found actually to be the 

 case." MendeWclTs " Principles of Chemistry," v»L ii. p. 95. 



