ON THE MECHANICAL EQUIVALENT OF HEAT. 
371 
sufficient to sav that a loss of 12 thermal units occurred per trial per 
unit fall of temperature along the shaft. 
Col. 1(5 gives, therefore, the product of this difference X 12, which represents 
the total loss by conduction. 
Col. 17. The difference of temperature between the brake and the surrounding 
air was taken as being equal to the difference of the mean discharge 
temperature of the w^ater and that of the air. The determination of 
the constant representing the loss of heat per unit difference of 
temperature is given later, and consequently, 
Col. 18 gives the product of this constant X the difference of temperature in 
Col. 17. 
Col. 19 gives the sum of the heat in Col. 9 added to all the corrections after¬ 
wards given. 
A further Table (p. 376) gives the work done, and the corrected values of the heat 
generated in these two trials, and the differences between them. 
The value of K in the last column is then found by dividing the difference of 
work in Col. 4 by the difference of heat in Col. 6. 
A slight inaccuracy has been pointed out to me by Professor Keynolds in the 
method of finding the mean temperatures of supply to and discharge from the brake. 
It was originally intended that the trials should, be of exactly one hour’s duration, 
and that the first series of readings should be taken one minute after the start. It 
was found impossible to do this, on account of the number of points requiring atten¬ 
tion in the first few minutes, and consequently I made all trials 62 minutes long, 
and took the first reading two minutes after starting. The mean used has not there¬ 
fore been obtained strictly in accordance with the middle breadth rule. Any error 
introduced would be of the occasional type, and should be eliminated in the mean of 
a number of trials. 
3 B 2 
