THERMO-ELECTRIC QUALITY UNDER PRESSURE. 289 



mechanical treatment, although they leave the general character of 

 the results unaltered, may nevertheless produce large changes in the 

 numerical values. For instance, a slight shift in the position of the 

 flat maximum would produce changes even of sign in the neighbor- 

 hood of 500 kg. 



The thermal e.m.f. of tin was measured at atmospheric pressure 

 against lead between 0° and 100°. Over this range its thermal e.m.f. 

 is given approximately by the formula: 



E = (0.230^-0.00067^2) ^ lO^^ volts. 



The Peltier heat against lead is 



P = (0.230-0.00134 t) {t -\- 273) X lO'^ volts, 



and the Thomson heat is 



(T = -0.00134 {I + 273) X lO"" volts/°C. 



This assumes the Thomson heat of lead to be zero. A positive e.m.f. 

 means that the current flows from lead to tin at the hot junction. 



Thallium. A different specimen from that of the resistance meas- 

 urements had to be used, the former piece having become very badly 

 oxidized. A fresh lot was obtained from the United States Smelting 

 Company. As provided by them it contained considerable impurity 

 of lead and cadmium. I purified it by electrolysis, by well known 

 methods. It was dissolved in excess c.p. H2SO4 and electrolyzed 

 from dilute aqueous solution onto a copper cathode with platinum 

 anode. The deposit is in the form of trees. These were repeatedly 

 washed in pure water, dried in a paraffine bath at 200° in a stream of 

 H2, and finally melted in a glass tube in a stream of H2. The H2, 

 however, was not perfectly pure, and a thin film of the yellow oxide 

 formed on the surface of the metal during melting. This was scraped 

 off, and the ingot formed into a wire by cold extrusion to 0.028 inch 

 diameter. The wire was kept under pure water while waiting for use, 

 which effectively prevented tarnishing. The outer wire was shellaced 

 immediately before assembling. There was no such trouble from 

 oxidation as that experienced during the resistance measurements. 



The purity of this sample was not so high as that of the previous 

 one, as shown by its temperature coefficient of resistance. The 

 average coefficient between 0° and 100° was 0.00475, which is con- 

 siderably less than that of the former sample, 0.00518. The several 

 crystalUzations in the form of various salts of the previous sample 

 doubtless accounts for its higher purity. The relation between 



