124 
PROFESSOR HUGH L. CALLEXDAR OX 
Hence, the complete equation of the method, when corrected for the effect of the 
variation of the temperature gradient in the fine flow-tube, becomes 
EC = J.s-Q <19 + h 0 (1 + hjBJsQ - a cld/6) <16 .(9). 
If we divide through the equation by cI6, and write s = s 0 (1 -f- cl), since our 
object is to determine the variations of s, we obtain 
EC jcl6 — Js 0 Q = = J.s 0 Q d -j- /? 0 (l — a c 16/6) + h 0 2 /BJsQ . . ( 10 ), 
in which D is employed as an abbreviation for the expression on the left-hand side, 
which it is most convenient to calculate as the first stage in the reduction of the 
observations. 
By combining the observations for two different flows, Q' and Q", for which the 
current is adjusted to give approximately the same rise of temperature, cW, if D' and 
D ' are the corresponding values of the difference observed, we obtain 
(IT - D")/(Q' - Q") = J s,d - V/3J,Q' Q".(11). 
Since the last term, which represents the effect of the correction sought on the 
variations of the specific heat, is generally very small, the equation may be readily 
solved by approximation, employing the value of h found in the first instance bv 
neglecting the term involving Jr. 
As an example of the order of magnitude of the correction, and of the method of 
application, we may take the following experimental results, which were given as an 
illustration of the method at the meeting of the British Association in 1897, and 
were quoted in the ‘ Electrician ’ of that date. In all the earlier series of observa¬ 
tions, three independent flows were taken, with the same rise of temperature at each 
point, with the object of verifying the theory of the method, and detecting possible 
sources of error. If the values calculated from the largest and smallest flow 
disagreed with the observation on the intermediate flow, it was a sure sign of some 
error or defect in the work. I thought at first that it might be possible to determine 
the conduction error experimentally in this manner, but the effect proved to be 
too small. 
Table VIII.—Example of Calculation of Specific Heat of Mercury. # 
Flow, Q. 
Rise, <16. 
Watts, EC. 
EC/cW. 
•1400Q. 
I). 
A 2 /3J*Q. 
Results, 
corrected. 
(1) 
8-753 
11-764 
14-862 
1-2632 
1•2255 
•0377 
•0008 
/ i= -0546 
(2) 
6-740 
11-8720 
11-696 
•9851 
•9435 
•0416 
•0010 
Jsd= - -00202 
(3) 
4-594 
12-301 
8 • 488 
•6901 
•6433 
•0468 
•0015 
J.s= -13798 
* These results are given merely as an illustration of the method: they were not corrected for the 
absolute values of the resistances and other minor sources of error. 
