1821.] Atomic Weight of various Metals and Acids. 123 



of sulphate of bismuth. This experiment, in conjunction with 

 the two others, seems to me to leave no doubt that bismuth 

 weighs 9, and oxide of bismuth 10. 



Berzelius's number is only 0-131 less than 9. It nearly coin- 

 cides with the atomic weight of bismuth which I deduced from 

 some experiments made by me on bismuth many years ago. 

 But I employed the bismuth of commerce in these experiments, 

 which i? not quite pure. This I consider as the cause of the 

 deficiency in these old experiments. 



II. Tin. 



The only salt of tin which exhibits a regular crystalline form, 

 at least as far as my experiments extend, is the muriate. But 

 this salt contains a considerable quantity of water, from which it 

 cannot be deprived without converting it into an insoluble chlo- 

 ride. The muriate is soluble in water ; but we do not readily 

 succeed in throwing down its muriatic acid by means of nitrate 

 of silver ; for when we add that salt to muriate of tin, the solu- 

 tion becomes muddy, and the liquid, even though set aside for 

 months, never becomes transparent. We may analyze this 

 muriate of tin by first throwing down the tin by means of an 

 alkali, neutralising the liquid by nitric acid, and then precipitat- 

 ing the muriatic acid by means of nitrate of silver. But this 

 mode of analysis is too complicated to give results to be depended 

 on, when an error of one-eighth of a grain would render the 

 whole useless. 



I next tried whether the fuming liquor of Libavius, which has 

 been shown to contain no water, would not afford an easy 

 method of analysis. When it is poured into distilled water, the 

 tin is precipitated in the state of a white hydrate, and the chlo- 

 rine is converted into muriatic acid. I separated the tin, dried, 

 and weighed it; threw down the muriatic acid by means of 

 nitrate of silver ; the chloride was washed, dried, fused, and 

 then weighed. Two successive analyses made in this way did 

 not exactly agree with each other ; but the mean of both did 

 not dirier materially from the previous analysis of this bichloride, 

 for which we are indebted to Dr. John Davy. 



Foiled in these attempts to discover the exact weight of an 

 atom of tin, I attempted to reduce the peroxide by heating it in 

 a glass tube, and passing hydrogen gas over it ; but I was not 

 able in this way to reduce the whole tin to the metallic state : 

 part was reduced, while a portion obstinately remained in the 

 state of protoxide. 



My e uts, though unsuccessful, led to the notion that 



tin weight of an atom of tin was iu all probability 7-25. I, 

 therefore, weighed 7*25 grains of pure tin in a platinum crucible,. 

 dissohed il in weak nitric acid, evaporated the solution to dry- 

 MM very slowly, and then gradually exposed the crucible to a 

 red heat. The peroxide of tin formed in this way weighed 



