78 PETER GUTHRIE TAIT 



represented by a succession of three parabolas with quite different parameters. 

 In the case of iron these peculiarities occur at high temperatures, which Tait 

 was able to measure by means of two alloys of platinum and iridium whose 

 thermoelectromotive force was very approximately proportional to the tem- 

 perature difference. These were known as M and N. Tait hoped to get a 

 series of such alloys having the same properties ; but though many specimens 

 of various percentage compositions were supplied him by Johnson and 

 Matthey, never again did he obtain a pair possessing the same simple 

 proportionality. The final experiments on iron at high temperatures were 

 entrusted to C. Michie Smith and myself in the winter of 1873. The three 

 wires M, N, and the particular specimen of iron under investigation had their 

 ends bound together to form one triple junction, while the other ends were 

 arranged so that the circuit M-N or the circuit N-Iron could be alternately 

 thrown into the galvanometer circuit. The triple junction was then inserted 

 within the hollow of a white-hot iron cylinder ; and as this cylinder cooled 

 to lower temperatures, the two circuits were thrown in rapid alternation into 

 the galvanometer circuit, and practically simultaneous measurements were 

 obtained of the N-Iron and M-N currents. 



Nickel and cobalt were not easily obtained in the early seventies ; and 

 the first piece of nickel experimented with was a narrow ribbon not more than 

 two feet long, supplied by F. Lecoq de Boisbaudran. The following letter 

 to Andrews touches upon the work with these magnetic metals. 



38 GEORGE SQUARE, 

 EDINBURGH, 13/12/75. 



My dear Andrews, 



Many thanks for your letter. I have been extremely remiss in not long 

 ago thanking you for the Nickel and Cobalt you kindly sent me. I know you will 

 be glad to learn what they have told me. Here it is : 



1. The new specimen of nickel gives almost exactly the same results as those 

 in my Thermoelectric Diagram. So that very curious result is verified. 



2. The Cobalt specimen was not coherent enough for any but qualitative 

 results: but it has shown me that cobalt lies (in the diagram) between Iron and 

 Nickel (at moderate temperatures), cutting copper, platinum, lead, zinc, cadmium, &c., 

 so that the observations of a few neutral points will tell me all about it except 

 (of course) the sinuosities which I have reason to think its line will show somewhere 

 about a white heat But I may be altogether wrong in this. Meanwhile with Crum 

 Brown's assistance I am preparing to deposit electrolytically films or foil of pure 

 cobalt. 



