54 PROCEEDINGS OP THE AMERICAN ACADEMY. 



On a and /3. On guard-ring. 



'c hil t>l 'c ha 'ft 



12.6° 49.0° 87.3° 14.0° 48.0° 84.6° 



Accordingly we get 0.8° as an approximate value for the mean excess 

 of temperature of the hars a and /?, for the significant 16 cm. between 

 mid points of hot and cold couples, above the temperature of the corre- 

 sponding part of the guard-ring bars. Now from p. 35 it appears that 

 the lateral flow from the same 16 cm. of each bar, when it is 4.2° warmer 

 than the guard-ring, is 0.0242 calorie per second. With a difference of 

 0.8° we may take the lateral How from each bar as 0.0046 calorie per 

 second, about 0.9 per cent of the amount which is conducted along the 

 hot end of the bar by a gradient of 4.6° per centimeter. 



Making correction accordingly, and taking account also of the slight 

 effect produced by the expansion of the main bars, we get from the value 

 0.9392, given above for g c -f- g h , the ratio k h ~ k c = 0.947 ; and hence 



/= f-^-lU (t h - t c ) = (0.947 - 1) -f- (87.3 - 12.6) = -0.0007. 



This study of the value of f raised again the question of the accuracy 

 of calibration of the thermo-electric couples ; and a good deal of addi- 

 tional attention was given to this matter in February, as the following 

 summary will show : 



Combination ijives 



2055 x 10- 8 at 89.4°, 

 1703 x 10~ 8 at 17.4 . 



Mr. Churchill found values from which we get 



2070 x 10~ 8 at 89.4°, 

 1722 X 10" H at 17.4°. 



Ratios of Churchill values to February values 



