JUNE 30, 1923] 
NATURE 
877 

Letters to the Editor. 
(The Editor does not hold himself responsible for 
opinions expressea by his correspondents. Neither 
can he undertake to return, nor to correspond with 
the writers of, rejected manuscripts intended for’ 
this or any other part of NATURE. No notice is 
taken of anonymous communications.) 
The Mechanical Equivalent of Heat. 
Wit the assistance first of Dr. J. K. Roberts, 
now of the National Physical Laboratory, and later 
of Mr. E. O. Hercus, I have been engaged for some 
years upon a determination of the mechanical equi- 
valent of heat. It is believed that an indication of 
the lines upon which the experiment is being made 
may be of use to other workers in this branch of 
physics. 
A number of determinations of what may be called 
‘the electrical equivalent of heat have been made, 
including the very thorough work of Jaeger and 
Steinwehr at the Reichsanstalt, but since the time of 
Joule the only direct measurements of the mechanical 
ent of heat are those of Rowland published in 
1880 and of Reynolds and Moorby. The work of the 
former has for long been regarded as of high accuracy. 
Reynolds and Moorby’s result is in terms of the mean 
calorie, and there is considerable room for doubt as to 
the value of that calorie in terms of the 15° or the 
20° calorie. This doubt arises from the conflicting 
values found for the specific heat of water from, say, 
60° to roo° C. It appeared, then, to be desirable to 
have a direct determination of the mechanical 
equivalent of sufficient accuracy as to be available 
for comparison with the electrical equivalent of heat. 
Such a comparison may throw light on the absolute 
values of the electrical units. It must be admitted, 
however, that to obtain the necessary accuracy in the 
value of the mechanical equivalent for that purpose 
will be a problem of some difficulty. But there appears 
to be no reason, if the same attention is given to the 
question as has been given to the realisation of the 
electrical units, why it should not be attained. 
In our experiment, work is indirectly converted into 
heat; the work done and the heat developed are 
_ directly measured. The work is found, asin Rowland’s 
experiment, in terms of a couple and a number of 
revolutions ; the heat is measured by a continuous 
flow calorimeter in terms of a quantity of water and 
its rise of temperature. A correction is made for 
‘the heat lost during an experiment. The relation 
-between these quantities is 
2rnmgd = Jw(t, -t,) +L 
where J ergs per calorie is the mechanical equivalent 
of heat. The apparatus is designed so that the heat 
lost can either be directly determined, or be eliminated 
by taking the difference between the equations for a 
heavy and a light run. 
The efficiency of an apparatus for finding the 
electrical or the mechanical equivalent of heat, which 
may be briefly called a J apparatus, is expressed by 
two characteristics, namely, (1) the percentage of 
the heat developed which is lost, and (2) the accuracy 
with which the lost heat can be determined, or 
eliminated from the expression for J. 
We have gradually developed, after many failures, 
an apparatus which, measured by this test, is an 
efficient one. We set out in the following table 
‘average figures for the power absorbed, and the 
percentage of heat lost in experiments by the observers 
named : 
NO. 2800, VOL. 111 | 

Percentage of 
Observer. Power. 
Heat lost. 
Rowland : 3 o-4 H.P 3 
Callendar and Barnes 0°03. 4; 2 
Reynolds and Moorby _. _| 70 D 0-8 
Laby and Hercus 0-2 4 about 0-2 

Any apparatus for the direct determination of J 
is a brake dynamometer. Reynolds and Moorby, for 
example, used the Froude hydraulic brake, which is 
the same in principle as the devices used by Joule and 
by Rowland, but the design is more efficient. We 

Fic. § 
decided to enclose the brake in a vacuum flask, in 
order to obtain high thermal insulation, and to use 
continuous flow calorimetry. The brake we are 
using is an electro-magnetic induction brake, which 
is closely analogous to an induction motor. The 
construction of the brake is shown in Figs. 1 and 2. 
An electromagnet (see Fig. 2) rotates about a vertical 
axis; in the rotating magnetic field so produced a 
copper cylinder (Fig. 2) and an iron core are placed. 
The copper and iron cylinders are attached by means 
of a glass tube to the inner sleeve of a bearing. This 
part of the apparatus is called the stator. The 
rotating magnetic field induces eddy currents in the 
copper cylinder, which is thereby heated, and the 
Teaction between these currents and the rotating 
field causes a couple to act on the stator. 
The couple acting on the stator is balanced by the 
tensions in two wires attached to the torsion wheel 
carrying two weights, one of which is shown at the left 
of Fig.1. The only details which need be mentioned of 
this part of the apparatus are the devices used to 
