ON THE ELECTROLYTIC METHODS OF QUANTITATIVE ANALYSIS. 309 



deposits are secured. The quantity stated above is greater than that 

 formerly mentioned, because there is a greater loss with warm solutions 

 than at the ordinary temperature. 



The temperature should not be allowed to rise too high, otherwise 

 Ti-regular deposits may result, and there may be excessive loss of ammonia. 

 A very small Hame is sufficient to keep the liquid warm enough. 



In dealing with small quantities the remarks made with reference to 

 nickel apply with greater force here ; for accurate results the quantity of 

 ammonia and strength of current should both be moderated, or the whole 

 experiment carried out on a reduced scale. Thus, it was found that a 

 certain volume of cobalt solution which gave 0-0612 grme. of cobalt when 

 electrolysed in a large basin in the usual way, gave only 0-0601 grme. in 

 a simultaneous determination carried out on a reduced scale in a platinum 

 crucible ; although deposition was quite complete in the latter case, the result 

 is decidedly lower, and this is explained by the fact that the deposit was 

 entirely soluble in hydrochloric acid while the former left a slight black 

 residue. A similar pair of experiments at another time gave 0-0603 and 

 0-0595 grme. respectively, again showing an advantage in favour of the 

 crucible experiment. 



The influence of other substances present in the solution is in many 

 cases similar on the whole to what was found to be the case with nickel. 

 There are, however, some exceptions to be noted in this respect. 



In the early cobalt experiments without nitrate, ammonium chloride 

 was occasionally tried in place of sulphate, and the results thus obtained 

 seemed generally superior. One of the best results obtained was got by 

 employing 10 grme. of ammonium chloride and 2''grme. of ammonia. It 

 was found, however, that solutions containing both chloride and nitrate 

 were rather erratic in their behaviour, sometimes giving vei-y poor and 

 irregular deposits. It is also more difficult to regulate the quantity of 

 ammonia in such cases. If it is desired to determine cobalt in a solution 

 containing ammonium chloride, it would appear preferable to conduct the 

 ■electrolysis without the addition of ammonium nitrate. 



The most striking diflTerence from nickel is shown when cobalt solutions 

 are electrolysed in presence of arseniate. We have seen that the presence 

 of ammonium arseniate in large quantity has no efl"ect on the nickel 

 deposit. This is far from being the case with cobalt. Deposition is re- 

 tarded, the deposit is very dark and rough, and the numerical result is 

 far too high. The deposit contains a large quantity of arsenic, but the 

 solution is apparently free from arsenite. Metallic cobalt would appear 

 to be readily attacked by ammoniacal arseniate solution, which is pro- 

 bably reduced to arsenite. But any arsenite formed would be promptly 

 reduced electrolytically to arsenic, which would contaminate the 

 deposit. 



Another point which shows how unsafe it is to assume that what holds 

 for nickel does so also for cobalt is the behaviour in presence of zi7ic salts. 

 This comes into the domain of separations, but it may be mentioned here. 

 With small quantities of zinc in presence of nickel, the deposit obtained 

 by the usual method contains the whole of both metals, as already noted. 

 In the case of cobalt a solution containing zinc gives on electrolysis by the 

 ordinary method a deposit which is not very good in appearance, but is 

 practically normal as regards numerical result. The decanted liquid gives 

 a pure white precipitate of zinc sulphide on the addition of hydrogen 

 sulphide. To what extent, if at all, this diflference depends on the some- 



