CONTINUOUS ELECTKIC CALORIMETRY. 
139 
been criticised by Schuster, and also by Ames (‘ Paris Congress Reports,’ 1900) on 
the ground that the length of the lever by which the torque was measured was 
apparently taken as 28 centims. without verification. M. Guillaume replies in a 
footnote that the length would probably in anj r case be accurate to a tenth of a millim., 
or 1 in 2800 (which is about the limit of accuracy claimed for the best absolute 
measurements of the mechanical equivalent), but regrets that the original lever 
cannot be found for the purpose of verification. It appears, however, from the 
description given by Miculescu, and also from the wood-cut of the apparatus, that 
the weight was not supported on a knife-edge but suspended from a round liook 
clamped to the lever with a screw, so that it would be a matter of some difficulty to 
estimate the effective length. Moreover, it would be very difficult with a platinum- 
iron couple to measure so small a difference as 2° with a degree of accuracy higher 
than -01°, and Miculescu does not mention some of the necessary precautions. In 
addition, the correction for the external loss of heat from the calorimeter by radiation 
and conduction was regarded as being negligible in comparison with the heat-supply, 
and no correction was applied for it. It is easy to estimate, however, from the 
dimensions of the apparatus, that it could not have been much less than 0'5 per cent., 
and that it was probably larger on account of the smallness of the interval between 
the calorimeter and the jacket, which would considerably increase the heat-loss by 
conduction and convection through the air space. There are several other indications 
in the paper that the author did not really aim at a higher order of accuracy than 
1 per cent., which is about the limit usually reached in engineering experiments, and 
that the work should be regarded rather as an illustration of a method than as a 
serious absolute determination. 
(45.) The Work of Reynolds and Moorby. 
The work of Reynolds and Moorby (‘ Phil. Trans.,’ A, 1897), on the mean specific 
heat between 0° and 100° C. in absolute measure, stands in the same category as that 
of Rowland as an accurate determination of the mechanical equivalent. The two 
are not directly comparable, but if we assume the rate of variation of specific heat 
found in our experiments as the medium of comparison, the result of Reynolds and 
Moorby would give 4T79 for the number of joules in one calorie at 20° C., at which 
point Rowland’s corrected results give 4T81. If we took Ludin’s table, we should 
find 4T57. The formula of Winkelmann* gives 4T33, and that of Regnault 4TG7. 
The comparison of Reynolds and Moorby with Rowland is therefore a confirmation 
to some extent of the accuracy of the present experiments, as compared with 
Ludin’s, or with formulae proposed by other writers. On the strength of this 
comparison, I have generally adopted the mean value 4‘180 joules per calorie at 
20° C. as the most probable estimate of the “ Mechanical Equivalent,” assuming that 
the absolute value cannot be certain to a higher order of accuracy than 1 in 2000. 
* Winkelmann, ‘ Handbuch der Physik,’ Band 2, Abth. 2, p. 338. 
T 2 
