﻿928 Mr. A. H. Davis on Natural 



bridge was balanced with a small current (0*01 amp.), thus 

 allowing for slight temperature changes of the fluid. An 

 additional resistance dH now introduced into the third arm 

 destroyed the balance, which was then restored by increasing 

 the current through the bridge and thus heating the wire 

 TT until its resistance had sufficiently increased. Correction 

 for the cooling at the ends of the test wire was automatically 

 effected by having in the other arm of the bridge a shorter 

 piece OC of the same wire * also immersed in the liquid 

 under examination. To be effective, this compensating wire 

 must be greater than a certain minimum length. Ayrton 

 and Kilgour f have given a calculation showing that for a 

 6-mil wire at about 300° C. temperature excess in still air, 

 the effect of the heavy leads extends about 1 cm. from each 

 end. For finer wires it would be less, and also it would be 

 less if the cooling medium were a liquid, for these carry off 

 more heat than air, so that the end effects become less 

 important. In the present experiments CC was never less 

 than 2 cm., and so the end effect should be entirely 

 eliminated. 



The energy dissipated in the uncompensated length (/) of 

 TT was calculated from the current (i) through the wire 

 and its (hot) resistance R. The temperature excess (0) 

 of the wire above its surroundings is given by dR/R a, where 

 R is the resistance of the wire at 0° C. and ' "a' the temperature 

 coefficient of resistance. The heat loss H (in calories) per 

 cm. length of wire per ° C. temperature excess is given by 



H =S^> w 



where 



<9=dR/R «, 



4*18 being the factor required to convert watts to calories 

 per second. 



Details of the wires used are given in Table I. 



The rectangular vessel containing the liquids under 

 examination had a height of 12 cm. and a base of 17 x 10 

 cm., and the wires (3 cm. apart) were 3*5 cm. below the 

 surface of the liquid. This vessel required rather larger 

 quantities of some of the liquids than were available. A 

 smaller one was found with carbon tetrachloride to give the 

 same result as the larger, and so it was generally used 



* Callendar introduced this method of eliminating end effects in 

 resistance measurements, using it for many similar purposes. Proc. 

 Phys. Soc. Lond. xxxiii, p. 137 (1921). 



t Ayrton and Kikrour, Phil. Trans. A, clxxxiii. p. 371 (1892). 



