their Determination in general. 273 



considered as important preliminaries for the establishment of a 

 natural system. It seems to me, however, that in trying to 

 establish such a law, we should proceed more cautiously than 

 has been in some instances the case hitherto. Laws must have 

 established facts to rest upon ; as long as the latter are wanting, 

 laws are out of the question. In this point I agree entirely with 

 Schafarik*, and consider it very fortunate that chemistry at pre- 

 sent is not a mere aggregate of facts. Such an aggregate, how- 

 ever, were better than one of unproved hypotheses. 



According to M. Dumas, there is a certain law according to 

 which several elements have the same equivalent. In the first 

 in stance t, this law was sought to be established with regard to 

 manganese and chromium, which were said both to have the equi- 

 valent 26. As has been remarked (p. 271), Dumas has been 

 already obliged to retract the number 26 for manganese as being 

 inaccurate, although advanced with such positiveness. Granted, 

 however, that manganese were actually 26, still the equality of 

 the equivalents of manganese and chromium would not have fol- 

 lowed. For according to the investigations of Peligotf, the 

 equivalent of manganese is not 26, as Dumas gives it, but 26-24; 

 and until experimental proofs are furnished, it seems unjustifi- 

 able to convert this into the round number 26 ; at least this 

 ought not to be done for the sake of an imaginary law. The 

 equivalents of manganese and of chromium are therefore decidedly 

 not equal. 



Lately M. Dumas § has endeavoured to prove the equality of 

 the equivalents of nickel and cobalt. According to his determi- 

 nations, both have the equivalent 29*5, According to what was 

 before said, and accoi'ding to my own determinations, I believe 

 that I have sufficient ground for considering this regularity as 

 being as yet doubtful. The specific heat of the two metals in 

 no ways supports the equality of their equivalents. The spe- 

 cific heat of nickel is, according to the determination of Regnault, 

 0"1109, and that of cobalt =0'1069. These numbers are not 

 equally great, as they should be if nickel and cobalt have the 

 same equivalent. On the other hand, they are very favourable 

 to the equivalent numbers given by me (Ni = 29 and Co = 30), 

 for they arc almost exactly inversely proportional to these num- 

 bers; for 



0-1109 : 0- 1069 =30 : 28-99. 



Although the relation of these numbers cannot by itself prove 



* Ann. der Chem. und Pharm. vol. cix. p. 100. 

 t Comptes Rendus, vol. xlv. p. 7 IB. 

 J Ann. de Chim. et de. Phys. (3) vol. xii. p. 628. 

 § Ibid. vol. Iv. p. 150. 



Phil. Mag. S. 4. Vol. 18. No. 120. Oct. 1859. T 



