PREFACE. XV 



perhaps, venture to express a hope, that the reader will 

 not immediately judge them to be inaccurate, because they 

 do not exactly coincide with the analytical results of 

 other chemists. The processes of chemists have been so 

 much rectified and improved of late years, and so much 

 care is now generally bestowed upon experiments, that 

 we may rely upon the result (provided the data employed 

 be accurate) certainly within less than one per cent. I 

 believe indeed, that the error in many of Berzelius' analy- 

 ses does not exceed one thousandth part. But it is much 

 more difficult to obtain substances in a state of complete 

 purity than chemists in general are aware : it was in 

 reducing the different salts which I employed, to the 

 greatest possible degree of purity, that the greatest part 

 of my time was wasted. I have in all cases in which it 

 was in my power, deduced the atomic weights of bodies 

 from the rigid analysis of the neutral salts into which they 

 enter; because it is much easier to obtain neutral salts 

 pure, than any of the metallic bodies which constitute 

 their bases. Indeed, not a few of the metals have never 

 yet been exhibited in a state of absolute purity. Berzelius 

 has in general deduced the atomic weights of metals 

 from experiments on these bodies themselves. This I be- 

 lieve to be the principal reason why his atomic weights so 

 frequently differ a little from mine. For example, the atom 

 of zinc deduced from his experiments is 4*03225 ; while 

 the atomic weight obtained by me from the analysis of 

 sulphate of zinc is 4'25. The reason of this difference I 



16 



