SUMMARY. 399 



glass water-gauge indicated the progress of evaporation. Tlie whole aim 

 was to make the process a continuous one, and, had it not been for acci- 

 dents, a nearly constant source of heat and a nearly constant water-level 

 would have made it possible to keep u]) an ebullition of nearly constant 

 intensity for an indefinite period of time. 



The rock used was earlier diabase from the hanging wall of the Lodk 

 collected in the main Sittro Tm/ncl. It had undergone only a trifling amount 

 of decomposition 



Tlie ex])eriments were continued during a period of nearly five weeks, 

 unfortunately with an accident between the first and second, and another 

 between the second and third. On the average, three observations of the 

 difference of temperature of the ends of the thermopile, or, say, T — t, were 

 made during each twenty-four hours. 



Mathematical treatment and results. lu Ordcr tO obtain a COmprehoUSi VO vioW of 



the large number of data obtained it will be sufficient to assume the empirical 

 relation, 



T-t = a + ^X, 



where a and /? are constants to be calculated by the method of least squares, 

 X the time in hours corresponding to any particular T — (, and dated from 

 the commencement of the series of experiments to which the results belong. 

 Under variation of a, an apparent thermal effect not due to kaolinization 

 may be conveniently understood. 



For a a mean value of — 0.05" C. was found. The interior of the rock 

 was, therefore, invariably colder than the surrounding steam It follows, 

 also, that it is impossible, even after the lapse of a great interval of time, to 

 heat so large a mass of material to an equal temperature throughout. The 

 variation of a will add itself algebi'aically to /?; and unless the thermal 

 effect of kaolinization is comparatively large, will entirely vitiate the sig- 

 nificance of the latter constant. /3 gives nominally the rate of increase of 

 the temperatui'e of the interior of the rock per hour in consequence of a 

 thermal effect. Instead of reporting y5, however, it is more expedient to give 

 the corresponding rate B referred to a year as the unit, viz. : 



B = 8,165/3 



For reasons which appear in Chapter IX. the experimental data may 



