Heat of Granite and Basalt at High Temperatures. 6?> 



scale. The circular grooves in which the junctions were- 

 situated, as shown, were about 2*5 mm. deep, the radial 

 orooves being onlv about OT> mm., so that the disturbance 

 of the heat-flow due to the latter was as small as possible. 

 Unfortunately, the deep circular grooves could not be kept 

 less than about 1*5 mm. wide at the top. As, however, the 

 grooves were V-shaped and the junctions, which were kept 

 as small as possible, were situated at the bottom of the V 

 which was filled with cement, the groove ought not to 

 introduce any large error. 



The circular grooves were cut with carborundum powder 

 by means of copper tubes of suitable diameter mounted on a 

 drilling machine. The straight grooves were cut with a 

 dentist's carborundum wheel. 



The single junction for measuring the mean temperature was r 

 as before, sunk in a straight groove cut in the opposite face. 



Very great difficulty was experienced in boring small 

 straight holes in the basalt, as the drill was inclined to bend 

 and so increase the size of the hole as well as rendering it 

 slightly crooked. The drilling was done by means of a soft- 

 nosed twist drill, run the opposite way to the twist, and 

 carborundum powder. The process was exceedingly slow, 

 and the resulting hole was about 4 mm. in diameter and not 

 as straight as it ought to have been. It was accordingly filled 

 with cement which was allowed to set and then a hole 

 1*5 mm. in diameter drilled centrally through the com- 

 paratively soft cement. 



An investigation on the error due to an unsymmetrical 

 distribution of the cement round the central wire, or to 

 eccentricity of the latter in the neighbourhood of the couples, 

 indicated that this was very small owing to the use of four 

 equally-spaced junctions. 



A more dangerous source of error is possible want of 

 uniformity of the cement layer at different points along the 

 axis. This would cause variations in the temperature, and 

 hence in the resistance per cm. of the central wire. There 

 is in fact a sort of inherent instability, as, if the thermal 

 insulation of the wire is greater at one point, there will be 

 more heat generated there owing to the positive resistance 

 temperature coefficient of platinum. This source of error 

 would cause an apparent variation of conductivity with 

 central current. A small variation of this kind was actually 

 noticed, but was apparently due to potentiometer errors as 

 will be explained later. For this reason, the same reliance 

 cannot be placed on these results as on those of later determi- 

 nations in which the chance of this error was greatly reduced. 



