10 Dr. Fletcher on the Determination of the B.A. Unit 



The coefficients of radiation were found to decrease with 

 decreasing difference of temperature between calorimeter and 

 jacket. When this decrease was regular the corresponding 

 value of c was used for each small interval of the main ex- 

 periment. When the decrease was small and irregular the 

 mean value of c for that day was used throughout. In the 

 revision of the calculations, stem- and stirrer-corrections were 

 neglected in the calculation of both c and A, it being obvious 

 that, both being small and quite regular, they are eliminated 

 in this way; and the value of c corresponding to the difference 

 between calorimeter and jacket for each small interval of the 

 main experiment was used in all cases. The mean results of 

 the two methods differ about 1 part in 1000, and the figures 

 in the table of results below are the means of both calcula- 

 tions. The mean values of c for the different experiments 

 vary between 0°*0035 and 0°*0046, the general mean being 

 0°-0040. 



The mean radiation-correction is about 5 per cent., and is 

 the most important source of variable error in the experiment, 

 as the temperature-differences are small, and errors of reading 

 have a large effect. But a 1 0-per-cent. error in the radiation 

 would only involve an error of 1 part in 200 in J; and as the 

 errors are irregular they are, in a great measure, eliminated 

 from the final result. 



The calorimeter was composed of 246 gr. copper, 45 gr. 

 brass, and 6 gr. solder. The specific heat of a mixture of 

 these proportions was measured with Regnault's apparatus. 

 Six determinations gave the value '0898 + '0005 for the mean 

 specific heat between 24° and 100°. Reduced by Bede's law 

 for copper to the mean temperature of my experiments, it 

 becomes '0877. The capacities of the coil and glass rods 

 were calculated from published tables. The whole capacity 

 is as follows : — 



Calorimeter .... 302-1 X -0877 = 26*49 



Coil 32-5 x -0324= 1-05 



Glass rods .... 9'0x -177= 1-59 

 Thermometer estimated at 1*25 





Total capacity .... 30*4 



The values of the deflections were obtained by a graphical 

 method. The galvanometer-readings fell into 4 groups, lying 

 about 26° and 3° 45' on each side of the zero-point. The 

 readings of each group were plotted separately as functions of 

 the time. From each curve the theoretical mean readings for 

 each interval between two temperatures used in the calcula- 



