I9G Proceedings of the Royal Society of Victoria. 



The values of the coefficients from sets 3 and 4 are 

 OOS.Sl and 00391. The mean of these four gives, as 

 the coefficient at 20° C, -00392. This is larger than the 

 value for most metals other than iron. 



(4) Thermo Ei.ecthk; Height. 



As I had a piece of pure silver, and no other metal pure, 

 I resolved to find the thermo electric height of thallium 

 with regard to silver, and assume Professor Tait's result for 

 silver in order to obtain the absolute value for thallium. 

 Having done so, it was found that the thallium line thus 

 determined, crossed Professor Tait's copper line at about 

 70° C, and that copper was therefore an exceptionally 

 favourable metal with which to compare thallium. I there- 

 fore obtained ])ure copper and compared thallium with it, 

 and found that thallium was further below copper than 

 below silver; and on finall}" fiyi'^g coppei- and silver, I found 

 the lines should be very much closer together than the_y are 

 in Professor Tait's diagram, and that copper should be above 

 .silver and not below it. I therefore purified some lead, and 

 constructed a diagram of my own ibr the four metals — lead, 

 thallium, copper and silver. To obtain pure lead, I dissolved 

 some sheet lead in nitric acid, and precipitated it as sulphate 

 by adding dilute sul])huric acid. The sulphate thus obtained 

 was lieated with carbonate of soda and cream of tartar in a 

 Hessian crucible in an injector furnace, and lead obtained 

 which was assumed pure, tliough it contained a trace of 

 potassium. I used an astatic low resistance galvanometer 

 with a lamp and scale, at a distance of about four feet, the 

 .scale divisions being fortieths of an inch. The resistance of 

 the galvanometer was somewhat less than an ohm, but with 

 the leads and the wires of the thermo electric circuit, .the 

 resistance was a little over an ohm. So low an E.M.F. as 

 •00000 1 volt or U)0 absolute units gave a deflection of one 

 scale division. This appears to be about 30 times as 

 sensitive as the one used by Professor Tait twenty years 

 ago. To determine the exact value of a scale division, the 

 galvanometer was joined in series with an ordinary Daniell 

 cell and various high resistances, smd immediately after or 

 before its E.M.F. compared with a Latimer Clark cell, by 

 means of a condenser and balliatic galvanometer. In 



