4 RESEARCHES ON THE 



ism, we have in most cases examined the light reflected from layers of crystals, by 

 Haidinger's dichroscopic lens, and have given the colors of the ordinary and extra- 

 ordinary images as obtained in this way. As a curious physical result, we may 

 here mention that, in general, the cobalt color predominates in the ordinary image. 

 We are indebted to Prof. Dana for the determination of the systems to which 

 many of our crystals belong, and of their principal forms, as well as for our figures, 

 and embrace this opportunity of expressing our grateful acknowledgment of his 

 valuable assistance. 



METHODS OF ANALYSIS. 



The accurate quantitative determination of the different elements which enter 

 into the constitution of the ammonia-cobalt bases and their salts, is attended with 

 great difficulties. We have in general found it necessary to study out with much 

 labor the methods of analysis proper to be used in each particular case ; and it has 

 been only after many trials that we have at length been able to obtain accurate 

 results. Before proceeding to the description of the compounds in question, it may 

 therefore be proper to state the analytical methods employed. 



Cobalt. The determination of the cobalt in these salts may, in most cases, be 

 very easily and accurately effected by the following process. A weighed portion 

 of the salt is gently heated in a deep platinum crucible, with a quantity of pure 

 and strong sulphuric acid sufficient to moisten the whole mass. Some effervescence 

 is generally produced by the addition of the acid, but there is no danger of loss if 

 the crucible be sufficiently large, and if the heat be applied only after the first 

 action of the acid is over. The mixture is to be gently heated over a spirit lamp, 

 until the excess of the acid, sulphate of ammonia, and other volatile matters have 

 been expelled. During the whole time of heating, the cover of the crucible must 

 be so placed as to prevent the possibility of loss by spattering, and at the same 

 time to permit the escape of volatile matters. When, however, the quantity of 

 acid has not been too great, the whole process goes on very quietly to the end, when 

 the mass becomes dry. The heat is finally to be raised, for an instant, to low red- 

 ness, the cover of the crucible being quickly lifted off and then replaced. The 

 crucible is then to be allowed to cool and weighed, when the quantity of cobalt may 

 easily be calculated from the weight of the dry and pure sulphate. After the 

 weighing, the mass in the crucible must be carefully examined. It should have a 

 fine rose color, and be perfectly soluble in warm water, leaving no black residue. 

 In case this is observed, which happens only when the heat has been too high, a 

 drop of sulphuric acid and a few drops of oxalic acid may be added, and the whole 

 evaporated to dryness, and again ignited. When, however, there is much oxide of 

 cobalt present, it is better to reject the analysis at once. With a little care and 

 practice the operation succeeds almost invariably, and the result, as we shall here- 

 after show, leaves nothing to be desired in point of accuracy. When chlorine is 

 present in the salt to be analyzed, a little free chlorine is sometimes found among 

 the products of the action of the sulphuric acid, and the platinum crucible is 

 slightly acted upon. In such cases we usually add a little oxalate of ammonia to 



