296 



On the Mechanical Equivalent of Heat. 



correction might be applied for any residual differences of temperature 

 as observed, and in the same way with the slope of temperatures. 

 Further assurance was also obtained by making some trials with the 

 brake naked and others with it covered, fo as to reduce the loss of heat 

 to one-fifth; and. in the same way, with every circumstance which 

 <could affect the result trials, means were taken to vary the circum- 

 stances in different series of trials, so as to obtain an estimate of the 

 limits of error. All the appliances were most carefully standardised ; 

 and taking all the limits of error into account the limit of the sum 

 was less than 0003. 



In all 52 trials were included in the final result. Of these — 



25 with loads on the brake of 1200 ft.-lbs. at 70 H.P. 

 21 }, 600 „ 35 



f „ „ 400 „ 23 



From these, twenty-five separate determinations were made of the 

 •equivalent, subject to certain general corrections given below, which 

 gave a mean value 777 91. From this none of the separate deter- 

 minations differed by as much as 0*2 per cent., and when arranged in 

 -eight groups, according to the circumstances under which they were 

 made, the greatest divergence from the mean was 0*036 per cent. 



It was found impracticable to eliminate entirely from each deter- 

 mination the losses of heat due to radiation, conduction, leakage of 

 water, &c, and so it was found advisable to determine what these 

 losses were. This information was given by the trials themselves, 

 and the necessary corrections were applied to each separate determi- 

 nation. As illustrating the extent to which the method adopted did 

 eliminate these errors, it is interesting to remark that on the mean 

 value of the equivalent determined, without taking these errors into 

 account, the error was only 0*0192 per cent. 



Certain final corrections, which had the same values for all trials, 

 were made to the mean value of J, given above. These were intro- 

 duced on the following various counts : — 



I. Length of the brake lever (0-00042). 

 II. Salts dissolved in Manchester water (0*00003). 

 III. Air dissolved in water used in the trials (—0 00021). 

 IY. Reduction of weighings to vacuo ( — 0' 00120). 



V. Varying specific heat of water ( — 0*00006). 

 YI. Pressure on thermometer bulbs (—0*00037). 

 VII. Work done against gravity (0*00007). 

 VIII. Engagement of counter (0*00001). 



Their total effect was to introduce a correction factor of (1 — 0*00 J 25). 



The mean corrected value of the specific h?at of water between 

 freezing and boiling points, as measured in mechanical units at 

 Manchester, is found io be 776*94. 



