THE MECHANICAL EQUIVALENT OF HEAT. 493 
the accuracy of our determination of the water equivalent of the calorimeter, and oi 
the changes in it and in the capacity for heat of the water. If we reject Group B 
(and we have already shown that it has little value) the results are practically 
identical. 
Hence if we assume— 
(1.) The unit of resistance as defined in the ‘ B.A. Pmport,’ 1892. 
(2.) That the E.M.F. of the Cavendish standard Clark cell at 15° C. = 1‘4342 
volts. 
(3.) That the thermal unit = quantity of heat required to raise 1 grm. of water 
through 1° C. at 15° C. 
The most probable value of 
J = 4'1940 X lOht 
This, by reduction, gives the following :— 
J = 427'45 kilogrammetres in latitude of Greenwich {g = 981'17). 
J = 1402'2 ft.-lbs, per thermal unit C in latitude of Greenwich {g = 981'17). 
778*99 „ ,, F ,, „ „ „ 
Section XV. —Discussion of the Results. 
As stated in the Introduction, we proposed to determine the value of J in terms of 
the thermal unit there defined, viz., the quantity of heat required to raise unit mass 
of water through 1° C. at 15° C. Rowland has preferred to tabulate his results by 
giving the changes in tlie numerical value of J caused by changes in the capacity for 
heat of water. We can, however, deduce from his table the expression for the change 
in the specific heat of water over our range. 
Expressed in the same form as above, it becomes 
1 - '000400 - 15). 
The difference between our results on this point is marked, and it is evident that 
further investigation is required. Rowland finds that the minimum value lies 
between 30° and 35° C,; we hope to carry the investigation beyond that temperature, 
and some explanation of the difference in our results may then present itself. The 
whole question is probably one of thermometry, and possibly our revision of this part 
of the subject may bring our results into closer agreement. Rowland himself 
(p. 198) points out that the whole matter depends on a small difference which he 
* If we assume the E.M.F. of oui’ Clark cells to be the same as that of the Cavendish standard (and 
we are inclined to think we have over-estimated the difference), we get J = 4* 1930 X 10^. 
t The value obtained by us in 1891 = (4*192 -b) X 10^ ; supra, p. 365. 
