Berzelius on Mr. Proufs theory of Atomic Weights. 371 



atomic weight of oxygen by 8. By new experiments upon oxy- 

 gen, nitrogen and atmospheric air, he went on to show other co- 

 incidences with the hypothesis of Proiit. 



These researches engaged another chemist, M. Marignac, to 

 undertake a series of experiments upon the atomic weights of 

 chlorine, bromine, iodine, nitrogen, potassium and silver. His 

 incomparable researches are a chef-Cfoeuvre, as much for consci- 

 entious exactness as for the judicious and varied methods em- 

 ployed. The results approach multiples of 125, but to make 

 them precisely so, requires larger corrections than the greatest va- 

 riations in the results of experiments. One alone, that of bro- 

 mine, is very nearly 80 times 12-5. The atomic weight of chlo- 

 rine, found to be 443-2, varies from 447*5, the nearest multiple, 

 by too large a quantity, to be an error in experiment. Marignac 

 was of the opinion that his experiments did not settle the ques- 

 tion at issue, as the approximation to the hypothesis of multiples 

 was still quite close ; but in the case of chlorine he admitted that 

 there was an undeniable exception. He endeavored, however, to 

 make an approximation to the hypothesis of Mr. Prout, by com- 

 paring the weight of two atoms of chlorine with one of hydro- 

 gen, in which he found the former to be almost exactly 73 times 

 that of the latter. 



But let us examine whether it is proper to view them in this 

 relation. You know that nearly all English chemists, and also 

 many others, consider water a compound of an atom of each ele- 

 ment, notwithstanding the fact that the amount of hydrogen gas 

 is in volume double that of oxygen ; and this view is adopted 

 because it is the most convenient, although mere convenience 

 cannot establish the truth in the case. It is well known that 

 several simple bodies combine with oxygen in such a manner 

 that the molecule of the oxyd sometimes contains one, some- 

 times two atoms of the radical, as an example of which I 

 mention here the series, Mn + 0, 2M+30, Mn+20, Mn + 30, 

 2Mn + 70. There is no doubt that the last, the hypermanganic 

 acid, contains 2 per cent, of the radical. It is besides known 

 that the hyper man ga^iate and hyperchlorate of potassa are iso- 

 morphous. From this it follows conclusively that the hyperchlo- 

 ric acid contains also 2 atoms of chlorine for 7 atoms of oxygen. 

 It must hence be an error to consider two volumes of chlorine as 



