168 PROCEEDINGS OF THE AMERICAN ACADEMY 



2. Radiation. 



The correction for radiation varies, of course, with the difference of 

 temperature between the calorimeter and jacket ; but, owing to the 

 rapid generation of heat, the correction is generally small in propor- 

 tion. The temperature generated was generally about 0°.G per 

 minute. The loss of temperature per minute by radiation was approxi- 

 mately .0014 0° per minute, where 6 is the difference of the tempera- 

 ture. This is one per cent for 10°. 7, and four per cent for 14°. 2. 

 Generally, the calorimeter was cooler than the jacket to start with, 

 and so a rise of about 20° could be accomplished without a rate of 

 correction at any jioint of more than four per cent, and an average 

 correction of less than two per cent. An error of ten per cent is thus 

 required in the estimation of the radiation to produce an average 

 error of 1 in 500, or 1 in 250 at a single point. The coefficients 

 never differ from the mean more than about two per cent. The 

 observations on the equivalent, being at a great variety of tempera- 

 tures, check each other as to any error in the radiation. 



The losses of heat which I place under the head of radiation include 

 conduction and convection as well. I divide the losses of heat into the 

 following parts: 1st. Conduction down the shaft; 2d. Conduction by 

 means of the suspending wires or vulcanite points to the wheel above ; 

 3d. True radiation ; 4th. Convection by the air. To get some idea 

 of the relative amounts lost in this way, we can calculate the loss by 

 conduction from the known coefficients of conduction, and we can get 

 some idea of the relative loss from a polished surface from the experi- 

 ments of Mr. Nichol. In this way I suppose the total coefficient of 

 radiation to be made up approximately as follows: — 



Conduction along shaft . . .00011 



" " suspending wires .00006 



True radiation 00017 



Convection 00 100 



Total . . . .00140 



The conduction through the vulcanite only amounts to .0000002. 



From this it would seem that three fourths of the loss is dm: to 

 radiation and convection combined. 



The last two losses depend upon the difference of temperature 

 between the calorimeter and the jacket, but the first two upon the 

 difference between the calorimeter and frame of the machine and the 

 wheel respectively. The frame was always of very nearly the same 



