Oct., 1875.] ELLIOTT SOCIETY, 87 



itely settled. It is true, some of the weights ultimately adopted may coincide 

 more nearly with the results of the above processes, but it is, at least equally 

 possible that they deviate farther, and if chemical analogies or relations require 

 large deviations, such as the halving or doubling of some of these weights, our 

 arithmetic and geometric structures will give way at once. The possibility of 

 finding for the numerical series of atomic weights such expressions as above 

 given is not to be overlooked, since these expressions may lead to the percep- 

 tion of new chemical relations, but they must hold a subordinate place until 

 such relations have been brought out. A certain succession of numbers 

 being given, with tolerable regularity of sequence, some latitude permit- 

 ted in choice of constants, and at same time a moderate deviation being 

 allowed for errors of computed results, it generally would not be difficult 

 to find one or more algebraical expressions for the terms of the sequence, 

 fulfilling certain conditions of approximation. It is not impossible that 

 formulae yet more complex, or of forms more peculiar, might be found yield- 

 ing results stiU more closely agreeing with the atomic weights of the table, even 

 when the order of sequence should sometimes be inverted, as would be the case 

 if V, Qb, Ta were removed to line — 5, and Cr, Mo, W to hne — 6. 



In that portion of our Table that we have had under consideration is contain- 

 ed all the perissads, save H, and many artiads ; but there still remain a large 

 number of the latter to be considered, for which can be found no suitable place 

 in the preceding scheme. These I arrange in groups in lines below the for- 

 mer, but separated from them by a horizontal line drawn through the table. 

 They lie therefore on the positive side of line whose index is 0, and are thus 

 properly placed except Cr, for all the rest lie in the positive third of table II, 

 p. 16, except the Platinum group, which lie in in the middle third and may be 

 placed here without violating the principles adopted for the construction of the 

 Table. In the first of these lines is the Iron Group, well recognized, and re- 

 markable for the small range of the atomic weights of its five chief members, 

 Cr, Mn, Fe, Co, Ni. To the addition of Cu to this group some may demur ; 

 none will object to prefixing Al, separated and yet connected by the dash, 

 nor probably to annexing, with similar dash, the element U, supported by the 

 analogy of Uraninite with Magnetite, Frankhnite, &c. It is remarkable that 

 the atomic weights of two of the members of this group are identical, and the 

 half of that of the extreme member nearest to them would differ from their 

 own by 1 ; also that the weights of two other members differ by 1, while the 

 double of that of extreme member nearest to them will coincide with the weight 

 of one of them. The elements on the next line below I can scarcely call a 

 group, but write them here to shew the remarkable " parallelism " of five of 

 their number, Y, Zr, Ce, La and D, with those members of the Iron Group 

 under which they are written. The other two, Er, Th, are written here for a 

 reason, very unscientific indeed, butnot without influence I imagine, in other 

 schemes than the present, — I know not where else to put them. I am not suf- 

 ficiently famihar with the elements on this line to make a better arrangement. 

 In the next line. I place the Platinum group, prefixing In, and arranging them 

 so as to exhibit the ' ' parallelism " of In, Ku, Ko, Pd, with those elements in 



