606 



PROCEEDTNOS OF TTIE AJIERfCAN ACADEMY. 



It appears that both the mean gradient at the spirals and the mean 

 difterence of temperature between the main bars and the guard-ring 

 bars increased a few per cent during the course of the observations, 

 keeping pace with each other pretty closely in this change. This is a 

 sufficient approach to equilibrium for our purposes. 



A summary of the results from the observations of February 9, 

 which were in a general way similar to those of January 26 and 27, 

 is given below. The c temperature of the main bars was found by 

 direct thermo-electric observation ; the other main bar temperatures 

 were estimated from this one by the use of data furnished by the 

 gradient measurements and the M against G measurements. The 

 guard-ring temperatures were found by means of the subtraction here 

 shown : 



Plotting an M-G curve, we get, as the excess of the mean tempera- 

 ture of the main bars, between the cross-sections a and e, over the 

 mean temperature of the corresponding part of the inner guard-ring, 

 6.32°. 



Taking the thermal conductivity of iron to be 0.134 at 216°, an 

 estimate which is in accord with our measui-ements of the temperature 

 coefficient of thermal conductivity, we get, as the total How of heat 

 per second in each bar at the junctions a and e, 



TT 



2 X - X 0.403 X 0.134 = 0.848 calorie. 

 4 



Accordingly we get, as the lateral outflow from each bar per second, in 

 the stable condition, per degree excess of mean temperature of the bar 



