448 



Mr. R. S. Willows on the Variation of the 



peraturejust greater than that corresponding to P (fig. 3) 

 after it had been standing for a month. The experiments, 

 therefore, showed that above 36° the volume-coefficient of 

 expansion was greater than that at lower temperatures, but 

 no sudden increase in volume was detected at the temperature 

 at which the sudden fall in resistance took place. 



Fig. 9. 



40 



.30 



:eo- 



•'5 20 25 30 35 



The amount of heat given out in cooling through various 

 ranges was next tried. The ordinary calorimetric method 

 was first tried, but was abandoned for the null method of 

 Hesehus [Journ. de Physique, 1888, p. 489) and Watermann 

 ( Phys. Rev. vol. iv. p. 961). In order to detect a small change 

 in the amount of heat given out, a fairly large amount of the 

 solid must be used, and this must be kept at a constant tem- 

 perature for some time before putting in the calorimeter in 

 order to insure a uniform temperature throughout. This is 

 difficult to accomplish imless a series of liquids with different 

 boiling-points is used, or the same liquid boiling under dif- 

 ferent pressures. In addition, if a small variation in thermal 

 capacity does occur, we should expect it to be very small ; 

 probably for these reasons this method did not give definite 

 results. 



When a glass tube containing amalgam forms part of a 

 thermo-electric circuit, if there is a change in the amalgam 

 arising from the formation of a different compound of zinc 

 and mercury, we might expect a change in the thermo-E.M.F. 

 due to a couple formed of the amalgam and another metal. 

 To test this some zinc amalgam was introduced into the inner 

 branches of a W-shaped tube, and various other metals were 

 used in turn to complete the circuit through a delicate low- 

 resistance galvanometer. One junction was kept in ice and 

 the other heated, the thermo-E.M.F. being measured by 

 balancing it against the difference of potential of two points 

 on a wire in the circuit of a Daniell cell running through 



