134 Proceedings of the Royal Irish Academy. 



absorbing oxygen from the air to form the double compounds of cobalt 

 and ammonia. A small quantity of a sulphite destroys the reaction, 

 as it changes the tint to a deep brown. Thiosulphates and some other 

 reducing agents do not act in this way. 



These experiments lead to the conclusion that the colour coefficient 

 of nickel with regard to cobalt is 3"16, in all cases, or, in other words, 

 that the tint of nickel and cobalt solutions is independent of the acid 

 radical in combination with the metals, and depends only upon the 

 metal in solution. It is evident that nickel and cobalt may be esti- 

 mated by means of this reaction. As an example of its application to 

 this purpose, I give the following account of the manner in which small 

 quantities of nickel may be estimated. 



The nickel must be dissolved in an acid, and the solution diluted 

 to any convenient quantity, e.g., 50 or 100 cubic centimetres. Into 

 each of three cylinders -0078125 grm. of Co as Co CI2 is placed. This 

 amount of cobalt is afforded by 6-25 c. c. of the standard Co CI2 solu- 

 tion. Calling the cylinders No. 1, I^o. 2, and !N^o. 3, we place in 

 ISTo. 1, -024531 grm. of nickel in solution, and in No. 3, -0248458 grm. 

 To the three cylinders we then add 25 c. c. of the standard ammonium 

 carbonate. Cylinder No. 2, which contains only cobalt solution and 

 ammonium carbonate, is then made up nearly to 150 c. c, and No. 1 

 and No. 3 are filled up to that quantity. Cylinder No. 1 has then a 

 purple tinge, while cylinder No. 3 has a blue tinge. By adding from 

 a burette the solution whose strength we wish to determine to No. 2, 

 until its tint is intermediate between No. 1 and No. 3, we make with 

 great accuracy the required determination. In all cases the cylinders 

 should be held, whilst under comparison, with their lower extremities 

 at some inches distance above a sheet of white paper. Three experi- 

 ments, that by no means reached the highest limit of accuracy, gave the 

 following results : — 



Ni in sohition. Ni found. 



•02469 grm. (1) -02425 



(2) -02475 



(3) -02500 



•02466 = mean. 



It is evident that a similar plan of estimating cobalt would be still 

 more accurate on account of the higher colour efficiency of that metal. 



The partially opaque brown solution obtained by mixing strong 

 solutions of nickel and cobalt might, I think, be used for making 

 standards for the purposes of colorimetrical analysis. For instance, the 

 brown solution mixed with a few drops of potassic bichromate cannot 

 be distinguished from Nesslerised ammonia. Probably the tests used 

 to compare the solutions of steel, in Eggertz's process for the estimation 

 of carbon, might be made in a similar manner. They would have the 

 advantage of being permanent. 



