the Heat-conductivity of Stone. 257 



the special object of these earlier experiments was to familiarize 

 us with the method, we have not employed the results so ob- 

 tained for the calculation of the conductivity. 



The considerable difference between the value of E obtained 

 for the ball of 5J centims. radius and for that of 6^ centims. 

 radius is due, to a certain extent, to the difference in the stone 

 and difference in the surface, the smaller ball being of finer grain 

 and having a smoother surface. The rock from which the balls 

 were turned is a fine-grained variety of stone, which is largely 

 used for building-purposes in many parts of Japan. Our 

 colleague, Mr. J. Milne, Professor of Geology at this College, 

 informs us that, with its varieties, it is a typical rock of the 

 country, forming in many districts large mountain-ranges. 

 It is essentially felspathic ; and the particular specimens em- 

 ployed were, in the original state, probably a porphyritic tra- 

 chyte, and, from the traces of hornblendic crystals which are 

 apparently to be found in it, was also rhyolitic. Both the fel- 

 spathic base and the enclosed crystals have been decomposed, 

 especially the latter, which have been altogether kaolinized. 

 Although the rock is generally light-coloured, it carries with 

 it a slight greenish tinge, due to the chloritization of a portion 

 of the hornblende. The crystals of hornblende and the felspar 

 are all of small dimensions, probably never more than two or 

 three millimetres long; so that the specimens have as a mass 

 a fine-grained homogeneous character. They also do not ap- 

 pear to be at all calcareous. Owing to the nature of the pro- 

 ducts of decomposition, the rock is soft and probably porous, 

 and therefore to a certain extent permeable to water. 



VIII. We now pass to the consideration of the possible 

 errors made in the determination of K and E by the method 

 of experimenting that we have adopted. In the actual ther- 

 mometer-readings there need not have been an error of more 

 than o, 05 C. ; but a single measurement of the actual tempe- 

 rature of the centre of the stone ball may possibly have erred 

 by as much as o, 2 C, due to want of perfect equalization of 

 temperature in the compensating-bath &c. As, however, the 

 observations were all corrected by drawing time-curves, such 

 errors were eliminated to an indefinite extent : and here it may 

 be stated that regular curves were found to pass through 

 nearly every point, as plotted from the observations on a large 

 sheet of squared paper. 



Our methods of reduction being based on Fourier's mathe- 

 matical calculations, really consist in finding exisothermal 

 curves from the given observations. Suppose such an exiso- 

 thermal curve found, then the value of v for any time t will 

 be subject to an error which we may call Sv. Now it may be 



Phil Mag. S. 5. Yol. 5. No. 31. April 1878. S 



