EXPERIMENTS ON KAOLINIZATION. 



301 



Table IV. finally gives tlie data obtained for the calculation of a after 

 the experiments in Table III. had been completed, together with the results 

 of calculation. The nomenclature is the same as before. 



Table IV. 



If the values of a in Tables I. and II are compared with that in Table 

 IV., a difi"erence of about 3 per cent, will be found. This may be due partly 

 to a change in the internal structure of the bismuth bars, partly to the fact 

 that both bismuth and silver were attacked by the sulphur fumes generated 

 in consequence of the presence of iron pyrites in the rock. In the case of 

 bismuth this action merely produced a thin, colored coating of sulphide on 

 the exterior. The silver, however, was so deeply corroded that its use had 

 to be abandoned, and in subsequent experiments this metal was replaced by 

 platinum. 



The data for ^t show a difference of temperature between the interior 

 and exterior of the rock-chamber, which is much greater than was antici- 

 pated. Moreover, the consecutive values of this quantity gradually de- 

 crease, indicating thereby an apparent increase of the temperature of the 

 rock itself 



Tables V. and VII. contain the data obtained in the determination of a 

 resi^ectively before and after the measurements of ^t made during the inter- 

 mediate week. In Table VI. these measurements are given, together with 

 the date, barometric height, and water-level, I (in inches from the bottom as 

 zero), corresponding to each Jt. The figures for barometric height were 

 obtained from a small aneroid. No reliance can therefore be placed on the 

 values as absolute, though the fluctuations are probably represented with 



