444 M, Berzelius on the [June, 



but tlie bulky state of the subarseniate of iron, and the lono- 

 washing which it requires, induced me once more to try the 

 acetate of lead, in the hope that the absence of iron would 

 render the composition of the precipitate less complicated. I, 

 nevertheless, deceived myself; and I had afterwards to sepa- 

 rate the lead by sulphuric acid, and the muriatic acid by 

 nitrate of silver. The acetate of lead gave a precipitate, which 

 weighed 85*8o parts. Treated with sulphuric acid, it gave 63*53 

 of oxide of lead, and with nitrate of silver 1*06 of muriatic 

 acid, which, both deducted from 85*85, leave 21*26 for the arse- 

 nic acid, equivalent to 13*88 of metallic arsenic, or 45*37 per 

 cent, of the weight of the ore. 

 The analysis therefore gave 



Arsenic 45*37 



Sulphur 19*34 



Nickel 29*94 



Cobalt, with a trace of copper 0*92 



Iron 4*1 1 7 



Silica -90 ^ 



loo758 



I shall not stay to inquire into the cause of the excess of weight 

 given by the analysis ; those who are accustomed to accurate 

 researches, know how easy it is to fall into such an error, 

 when every effort is made to lose nothing. I shall only add, 

 that if the silica of the ore was in the state of silicium, there 

 would scarcely be any excess. 



As to the chemical constitution of the ore analyzed, it is 

 evident that it is analogous to that of arsenical iron and 

 grey cobalt ; that is to say, that it contains an atom of quad- 

 risulphuret of the metal combined with an atom of biarse- 

 niuret, of the same metal ; calculating the composition on this 

 view, we have the following proportions : 



Arsenic 45*17 ^ 



Sulphur 19-32 



Nickel 35*51 



This is the composition of a combination containing nei- 

 ther iron nor cobalt. But the three metals in question may 

 occur, in a similar state of combination, mixed together in dif- 

 ferent proportions, without greatly influencing the proportions 

 of arsenic or sulphur, because the saturating capacity of nickel 

 and of cobalt are exactly similar, and that of iron very little^ 

 exceeds it. Consequently, when in the ore, which I have ana-, 

 lyzed, the weights of the cobalt and iron are added to that of 

 the nickel, we have 34*95, which agrees very nearly with the 

 calculated result. The variety analyzed on the two first expe- 

 riments, and that examined by M. Psaff, were then a mixture 

 of arse^iqal nickel with this last combination, which may be 



