BISHOP. — THERMAL CONDUCTIVITY OP LEAD. 687 



nickel at this point varied little from that at any point. The wires were 

 0.3 mm. in diameter. 



10. Change in temperature in the oil surrounding the shell. — It was 

 impossible to maintain this oil at exactly a certain temperature. The 

 temperature was therefore allowed to vary, part of the time increasing 

 and part of the time decreasing, and by the same amount both times. 

 This total change did not exceed ^a few hundredths of a degree. See 

 Readings. 



1 1 . Variation in the agitation. — The fan for agitating the oil on the 

 outside of the shell was always driven in the same direction, as it was 

 evident from the motion of the oil that this direction gave much better 

 agitation close to the shell. 



On the inside of the shell the fans were usually driven in the same 

 direction, and always at the same speed, but a change in the direction 

 did not appear to produce any change in the deflection. It would 

 appear from a consideration of the construction of the heating coil that 

 a change in the direction of rotation inside the cavity should produce 

 a change in the galvanometer deflection. That it did not may be 

 due to the position selected at which to attach the nickel wires on the 

 inside. 



12. Amount of oil in the cavity. — The fire-clay plug had two holes 

 by which oil could enter or air leave when the shell was immersed in oil. 

 At first the shell was immersed in oil and the air exhausted from over it 

 by means of a pump, but it was found that this was an unnecessary pre- 

 caution. A small wedge which could be easily withdrawn was placed 

 between the two hemispheres, thus allowing the air and oil to enter the 

 cavity at the separation of the spheres, and also through the fire-clay 

 plug. The oil was agitated in the cavity and at the same time heated to 

 expel the air. The wedge was withdrawn and the holes in the fire-clay 

 plug sealed with shellac. 



13. Calibration of the ammeter. — The ammeter read from zero to 

 5 amperes, the smallest division being 0.05 of an ampere. The resist- 

 ance of the tenth of an ohm in series with the ammeter was accurate 

 to g 1 ^ of 1 per cent. The drop of potential across its terminals could be 

 balanced against the Carhart-Clark cell to an accuracy of 0.1 per cent. 

 Consequently it was possible to determine the current to within 0.1 per 

 cent. 



14. Determination of voltage. — This was balanced directly against 

 the Carhart-Clark cell, and the galvanometer was sensitive to better 

 than 0.1 of 1 per cent. 



