200 Mr. J. H. Gray. On a Method for determining the [June 14, 



from that on which the wire is. No heat can therefore be communi- 

 cated directly to the wire from the lamp. In the bottom of the box 

 above the lamp a number of thick copper pins is fixed, so as to catch 

 and distribute the heat. 3 mm. length of the other end of the wire 

 is soldered into a solid copper ball, diameter 5' 5 cm. In the ball & 

 hole 3 cm. deep is made, so as to admit the bulb and part of the stem 

 of a small and very sensitive thermometer. This thermometer is 

 graduated from 5 C. to 20 C., and can be easily read to within one- 

 fortieth of one degree. The bulb is surrounded by water. 



To prevent radiation from the surface of the wire, a tube of circular 

 section, diameter 1 cm., made of several layers of thin paper, sur- 

 rounds the wire all along its length. The air inside this tube soon 

 takes up the temperature of the part of the wire with which it is in 

 contact, and so practically eliminates radiation. 



A rough calculation gives for the maximum value of the loss due 

 to radiation, 5'5 per cent, when the' surface of the wire is exposed to 

 the air, the length being 4 cm. Unless the paper tube is effective, 

 the error due to radiation ought to be greater, the greater the length. 

 Exhaustive trials, however, proved that different lengths gave prac- 

 tically the same value for the conductivity. 



The other possible errors, besides radiation, to be tested for are : 



(1) The thermometer in the hot water may not indicate the tem- 



perature of the end of the wire. 



(2) The solder may cause some error. 



(3) The thermometer in the ball may not indicate the average 



temperature. 



(4) There may be a lag in the thermometer. 



(5) The temperature of the ball may not be the same throughout, 



and the thermometer may not indicate the temperature of 

 the wire where it enters the ball. 



All these errors are practically tested by using different lengths or 

 diameters of the wire, and the results obtained in the present investi- 

 gation indicate that the errors have been eliminated. 



To test whether the thermometer in the hot water indicated the 

 temperature at the end of the wire, a thermo-electric junction, made 

 of very thin platinoid and copper wires, was soldered to the wire just 

 where it entered the box. The other junction was tied to a thermo- 

 meter and immersed in water, which was heated till there was no 

 deflection in the sensitive mirror galvanometer which was used. The 

 temperature indicated by the thermometer was then found to be the 

 same as that of the thermometer in the hot water. 



An approximate calculation for the other end of the wire shows 

 that the temperature of that end is somewhat lower than that of the 

 ball, the greatest difference being 1'5 per cent. This difference was 



