200 ANNUAL OF SCIENTIFIC DISCOVERT. 



sificatlon of chemical elements into families of the kind I allude to, in a sys- 

 tem of grouping, in which the first idea, or rather the first germ of the idea, 

 may be traced to a remark made by M. Dumas, in one of his reports to this 

 Association, and which is founded on the principle of arranging- them in a 

 series, in each of which the atomic weight of the elements it comprises are 

 found among the terms of an arithmetical progression, the common differ- 

 ence of which in the several series are 3, 4, 5, 6, 8, and 9 times the atomic 

 weight of hydrogen respectively. So arranged, they form six groups, which 

 are fairly entitled to be considered natural families, each group having com- 

 mon properties in the highest degree characteristic; and what is more remark- 

 aide, the initial member in each group possessing in every case the character- 

 istic property of the group in its most eminent degree, while the others ex- 

 hibit that property in a less and less degree, according to their rank in the 

 progression, or according to the increased numerical value of the atomic 

 equivalent. Generally speaking, I am a little slow to give full credence to 

 numerical generalizations of this sort, because we are apt to find their authors 

 cither taking some liberties with the numbers themselves, or demanding a 

 wider margin of error in the application of their principles, than the precision 

 of the experimental data renders it possible to accord ; so that the result is more 

 or less wanting in that close appliance to nature which makes all the differ- 

 ence between a loose analogy and a physical law. But in this instance it cer- 

 tainly does appear that the groups so arising not only do correspond remark- 

 ably well in their theoretical numbers with those which the best authorities 

 assign to their elements, but that it really would be difficult to distinguish the 

 Clements themselves into more distinctly characteristic classes by a consider- 

 ation of their qualities alone, without reference to their atomic numbers. 

 When we find, for instance, that the principle affords us such family groups 

 as oxygen, fluorine, chlorine, bromine, and iodine, self-arranged in that very 

 order; or again, nitrogen, phosphorus, arsenic, antimony, or bismuth; when 

 we find that it packs together in one group all the more active and soluble 

 electro-positive elements, hydrogen, lithium, sodium, and potassium, and in 

 another the more inert and less soluble ones, calcium, strontium, barium, and 

 lead, and that without outraging any other system of relations, it certainly 

 does seem that we have here something very like a valid generalization: and 

 I shall be very glad to learn, in the course of any discussions which may arise 

 on such matters as may be brought before us in the regular conduct of our 

 business, from those more competent to judge than myself, whether I have 

 been forming an overweening estimate of the value and importance of such 

 generalizations. I will only add on this point, in reference to what fell from 

 our excellent President in his address to the assembled Association last night, 

 that this kind of speculation, followed out, would seem to me likely to termi- 

 nate in a point very far from that which would regard all the members of 

 each of these family groups as allotropes of one fundamental one ; inasmuch 

 as the common difference of the several progressions which their atomic 

 weights go to make up, are neither equal to, nor in all cases commensurate 

 with, the first terms of these progressions. For instance, in the chlorine 

 group, the first term being 8, the common difference is 9. Something very 

 different from allotropism is surely suggested by such a relation. It would 

 rather seem to point to a dilution of energy of one primary element by the 

 super-addition of dose after dose of some other modifying element; and this 

 the more strikingly, since we find oxygen standing at the head of very dis-- 

 tinct groups having very striking correspondences in some respects, and 



