DETERMINATION OF THE THERMAL CONDUCTIVITY OF MARBLE. 265 



The wires used tor the junction were .5 ni.in. in diameter; coarser 

 wires having been found to produce thermal ettects by direct conduc- 

 tion from parts outside Uic sphere. Even the glass-tube itself was 

 found to be a similar source of disturbance, so that it was necessary 

 to protect the upper part of the tidjc from direct contact with the 

 water. This was etfected by means of a slightly larger metal tube. 

 This metal tube was provided with three metallic strips of a cpuidran- 

 tal form, wliich, fitting close over the upper half of the sphere, were 

 screwed to three similar strips from l)eIow. Where the tidoe met the 

 three quadrantal strips, it expanded into a disc, which iitted well on 

 the sphere near the junction-hole. After the insertion of the glass- 

 tube containing the junction, the space between it and the wall of the 

 hole was filled with a paste of zinc sulphate, minium, and linseed oil. 

 A thin coating of the same paste was spread over tlie under surface 

 of the disc, which when the strips were screwed tight, prevented any 

 water from [)assing into the hole from the outside. By these arrange- 

 ments, the heat conducted directly through wires or their connections 

 from the part outside the sphere, was diminished to such a degree as 

 to l)e inappreciable. These precautions were f jund to be absolutely 

 essential.* 



Another thing to be l<:)oked to carefully was the stirring of the 

 water in the baths. The more vigorous the stirring of the water in 

 the hot and cold baths, the greater the range between the maximum 

 and minimum temperatures. The reason was that when the agitation 

 was not sufficient, the surface of the sphere was not at the same 

 temperature as the bath itself. With increased agitation, however, 



•Although Professors Ayrton and Perry speak of very fiue wire, the wires I found in 

 their balls could hardly be so designated; the copper was 1.3 m.m. in diameter; the iron 

 .7 m.m. Further the copper wire, simply coated with gutta-percha and cotton, seemed to 

 have been directly exposed to the bath, and the iron does not seem to have had any cover- 

 ing at all. As Professors Ayrton and Perry spoakof careful insulation of wires from water 

 and from one another, probibly the wires I found wereinsertedby somebody else afterward. 



