THERMAL CONDUCTIVITIES Of SOME ELKCTftlCAL INSULATOKS. 435 



difference of temperature ' t',. If for example the cylinder be placed in a gas, it 

 will lose heat from its surface at a rate which may, for a small range of temperature, 

 be taken proixrtional t> excess of the temj>erature of the surface over that of the 

 gas, and at the surface we shall have 



where k is the conductivity of the medium of the cylinder, H the normal to its 

 surface, h the loss per second from 1 sq. centim. of surface when one degree hotter 

 than the gas, V the temperature of the gas. 



The expression for the temperature difference r, r, then involves a term in lijk 

 which, although not large, will vary with the material used and with the variations 

 of h as the experiment is carried out at different temperature* 



In order to reduce as far as possible the effect of li on the temperature difference 

 (' u r,, the cylinder of material to tie tested was cast in and kept surrounded by a 

 brass cylinder with walls about '5 millim. thick, which has the effect of making the 

 surface of the material in contact with the brass nearly an isothermal surface. The 

 value of h has then a negligible influence on the temperature difference observed, and 

 its value need be known only approximately. 



The measurement of the temperature difference v, may be made in several 

 ways, e.g., by two thermometers with long thin bulbs arranged with their axes 

 parallel to that of the wire, or by two similar bulbs of a small gas thermometer 

 arranged to read differentially, or by two thermojunctions placed at the two points 

 and connected in opposition, or by two straight wires of a metal of known coefficient 

 of increase of electrical resistance with temperature, arranged parallel to the heating 

 wire, and forming two sides of a resistance bridge. 



Of these methods the latter seemed to promise best and was first tried. So long 

 as the material in which the wires were eml>edded was not subjected to great changes 

 of temperature, it gave excellent results, but when the apparatus was cooled down to 

 the temperature of liquid air, the wires were found to change their resistance in an 

 irregular manner, which was traced to the contraction and occasional cracking of the 

 material during the cooling process. 



Increasing the time of cooling through a given temperature interval did not get 

 rid of this trouble, and after some time the wires were replaced by the two bullis of a 

 small differential air, and eventually hydrogen, thermometer.* This arrangement 

 again answered extremely well at ordinary temperatures, but at the temperature of 

 liquid air irregularities were met with which seemed due mainly to the capillary 

 forces acting on the thread of strong sulphuric acid which acted as an index, being 

 large compared to the forces due to the pressure of the hydrogen at these low 

 temperature* 



* LEE*, ' Pmc. Miinclietiter Lit. and Phil. Sue.,' xlvii., p. viii. (15)02). 



3 K 2 



