of the Elements. 425 



a general correspondence ; but iron manifests a tendency to form 

 persalts, copper to form subsalts, and hence the two metals are 

 usually widely separated in our classifications. I hope to show 

 that they form with nickel a most characteristic natural family. 

 Again, arsenic produces acid oxides, and bismuth basic oxides ; 

 but the antagonism disappears on the introduction of antimony 

 as an intermediate term. I am inclined, indeed, to place great 

 confidence in gradational differences, not as a means for sever- 

 ance, but for association. 



That certain natural families have not long ago been recog- 

 nized and made available in practice, has arisen, I believe, from 

 the accordance of an undue consideration to certain unimportant 

 dififerenees. Our attention has been arrested by the most su- 

 perficial rather than by the most fundamental characters. For 

 instance, the sulphydrate of magnesium is soluble in water, 

 that of zinc in acidulated water, that of cadmium in moderately 

 strong acids. Hence the three metals, despite their great ana- 

 logies, have been referred to three artificial classes. The point 

 I wish to establish is, that, in our attempts at classification, we 

 must rely upon natural characters, and not upon some arbitraiy 

 rule, such as the degree of temperature at which a metal decom- 

 poses water, or the sti'cngth of acid in which a metallic sulphide 

 is soluble. Doubts may ever arise as to the proper distribution 

 of some particular element, but the existence of natural families 

 of elements, to which all artificial arrangements must give place, 

 is I conceive incUsputable. 



Group I. Fluorine — Chlorine, Bromine, Iodine. 



These four substances have one marked property in common, 

 a property not pertaining to any other element witli which we 

 are acquainted, namely, that of combining with hydrogen in the 

 proportion of atom to atom, gaseous atomic volume to gaseous 

 atomic volume. The combinations moreover take place without 

 any condensation ; the I'csulting compounds are powerful acids, 

 and exhibit a general resemblance in their characters and beha- 

 viour. Chlorine, bromine, and iodine present a marked simi- 

 larity and gradation of properties. Fluorine is separated from 

 the other members of the group by certain specialities. 



The atomic weights, which also express the vapour densities 

 referred to hydrogen, form a remarkable sequence* : — 



CI . . 3.5-5 242-5 



Br . . 80-0 -3— = »0 8 



I . . 127-0 



Sum = 212-5 Mean diff"erencc = 44. ? 



* The atomic numbers made use of in this paper .ire those employed by 

 Phil. Mag. S. 4. Vol. 13. No. 88. June 1857. 2 G 



