RICHARDS AND BAXTER. ATOMIC WEIGHT OF COBALT. 125 



air to go backwards through tube 10 (stop-cocks 4 and 9 being closed), 

 and to bubble through the sulphuric acid kept in the vessel labelled 13. 

 The tube and vessel served as a safety valve and pressure gauge through- 

 out the whole operation. The ground joints were kept tight by means 

 of syrupy phosphoric acid. 



To the right hand end of the complex apparatus here described was 

 fused the automatic bottling apparatus, as indicated in P'igure 3. By 

 means of this arrangement the cobaltous bromide could be thoroughly 

 dried and enclosed in a suitable bottle for weighing without a moment's 

 exposure to the moist air of the room. In one case, where a platinum 

 boat containing several grams of the salt was reheated and rebottled 

 after weighing, the loss in weight was found to be only 0.11 milligram, 

 an amount not greater than the probable loss by sublimation at the high 

 temperature (400°) employed in the ignition. Hence the apparatus 

 evidently answers its purpose well. 



The preliminary analyses were so much easier to make than those 

 described in the preceding paper, — because of the ready solubility of the 

 salt used and its invariable freedom from the oxide whose presence had 

 caused trouble at first in the work upon nickel, — that only three were 

 de(!med necessary. 



The method of procedure may be easily gathered from previous de- 

 scriptions. The bromides of cobalt and silver were each weighed, as iu 

 the case of the nickel, and from the assumed molecular weight of argentic 

 bromide 187.885 that of cobaltous bromide was calculated. 



From the value for the atomic weight of cobalt obtained from this 

 preliminary series, the approximate proportion of silver necessary for the 

 complete decomposition of a given weight of cobaltous bromide was 

 calculated ; and iu the succeeding analyses the proper amount of silver 

 was weighed out in each case. After the silver had been dissolved with 

 precautions to prevent loss during the process, and after the solutions of 

 cobaltous bromide and argentic nitrate had been much diluted, the pre- 

 cipitation was effected in a roomy, glass-stoppered Erdenmever flask ; 

 and the mixture was very thoroughly shaken and allowed to stand. Tn 

 three experiments, Nos. 4, 6, and 7, the precipitation was completed by 

 titrating backward and forward with centinormal arp-entic nitrate and 

 hydrobromic acid solution, until the precise point was found ; * while in 

 Nos. 8, 9, 11, 12, and 13 the nepheloraeter was used to determine the 

 end-point. I Of course the work was done in yellow light or in darkness, 



* These Proceedinixs, XXVITI. 24 ; XXX. ."81. t Ibiil., XXX. 385. 



