66 Mr. H. H. Poole on Thermal Conductivity and Specific 



The distance between the P.D. leads on the central wire 

 being only 10*00 cm., a suitable modification was made in 

 the formula for K. The end spaces were on this and all sub- 

 sequent occasions filled with quartz sand, and the ends as 

 before covered with pulped asbestos. The next results 

 obtained are shown in the next three columns. The first 

 five results at approximately the same temperature show an 

 apparent variation of K with W ; the next three were so 

 inconsistent that it was evident that there must be some 

 source of error. This was traced to the potentiometer. In 

 this instrument the Thomson- Varley shunt is employed, and 

 if the double contacts forming the ends of the shunt become 

 at all foul, a small error in the reading is introduced. It 

 appeared that this error, which was measured by cutting out 

 the couple at the mercury key, might amount to about 15 or 

 20 microvolts. In all the previous basalt determinations W 

 and hence A were small, so that this source of error was 

 important. The error, if it occurred at all, would have been 

 far smaller in the determinations described in the previous 

 paper, as in them the values of A were much larger. In 

 all subsequent work particular attention was paid to the 

 double contacts, which were periodically cleaned with paraffin 

 oil. This had the effect of reducing the error to about 1 

 or 2 microvolts, the residue probably being due to thermal 

 effects at the terminals. After every reading of the E.M.F.s 

 of the couples this small error was measured and corrected 

 for. The effect of any remaining uncertainty was reduced 

 by using larger values of W. 



The subsequent results, shown in the last three columns of 

 the table, quite contrary to the previous ones, show a steady 

 fall of conductivity with rise of temperature. As this fall 

 was expected, it was concluded at the time that the previous 

 results w T ere unreliable owing to the error just mentioned. 

 However, another explanation of the discrepancy is possible 

 and will be mentioned later. The determinations were 

 brought to a conclusion by the breakage of the multiple 

 couple. 



Second Granite Experiment. 



In constructing new granite cylinders, of medium-grained 

 Newry granite, the cylinders were first roughly turned and 

 then the central hole, 4*3 mm. in diameter, bored. The 

 cylinders were then mounted on an axle through this hole 

 and turned down to the required size, so that the hole was 

 accurately centred when finished. Only two cylinders were 

 employed as in the original determinations, their lengths 



