ON THE THERMAL CONDUCTIVITIES OF CERTAIN ROCKS. 137 



with them, of which only long-continued observations can by a final 

 average eliminate the inconstant actions. 



With all the loss of certainty which thence arises, the records of 

 earth-thermometers for one or two years only have frequently been 

 reduced, so as to afford approximate coefficients of amplitude-decrease 

 and retardation, from which corresponding conductivities in terms of the 

 heat capacity of a unit-volume of the rock- substance, and of the centi- 

 metre, gramme, and second, as fundamental units, are easily calculated, 

 when the fundamental units are known in which the concluded coefficients 

 were expressed. Coefficients of this kind were deduced by Kuppflfer 

 from Ott's extensive three-year observations at Ziirich about the years 

 1760-62, and from Dr. Leslie's, at Raith, near Edinburgh, in 1816 and 

 1817. The equivalents of these coefficients given in the table in the 



column of the ratio —, assume the units in which they were originally 



c 

 expressed to have been, the Paris-foot in the former, and the English- 

 foot in the latter reduction, those having been apparently (in the absence 

 of other intimations in Kuppffer's work), the scale-units of depth of 

 the Ziirich and Raith thermometers used and adopted by Kuppffer in 

 his calculations. The conductivity derived directly from the recorded 

 loss of range, which was nearly the same in one year as in the other, 

 between the two deepest (4 feet and 8 feet deep) thermometers at Raith 

 in 1816 and 1817, differs so much from that belonging to the higher 

 seated ones (at 1 foot, 2 feet, and 4 feet deep), and from Kuppffer's 

 mean value of the rate of logarithmic decrement for the whole of the 

 records at the different depths, that in view of the probability of the 

 lowest pair of thermometers being least of all affected in their ranges 

 by temperature-oscillations of short periods at the surface, a separate 



entry is given in the Table of the value of the ratio ■ — derived from 



c 



the two lowest thermometers only, on the supposition here adopted of its 

 being more certainly dependable than that arrived at from all the ther- 

 mometer indications together, without any choice or preference. 



Another logarithmic rate instanced by Kuppffer in the same paper as 

 the last two, is deduced from three' years observations at Strasburg, by 

 Hemmschneider, in 1821-23, of a thermometer fifteen feet below the sur- 

 face of the earth. This reduction seems to be as doubtful and uncertain 



as that last described, and the equivalent conductivity - is as high 



c 



(0'01267) as that denoting Kuppffer's mean rate for the Raith thermo- 

 meters is low (0*00444) among ordinary values. But from Rudberg's 

 one- to three- feet deep thermometer records in the Observatory plain of 

 coarse sand at Stockholm, three reductions for one year each, 1832-33, 

 1833-34, and 1834 alone (the last two by Quetelet and Angstrom) give 

 conductivity equivalences 0-0060, 0-0071, and 0-0093 very divergent but 

 generally resembling those found by Forbes and Thomson, -0067, -0079, 

 and -0087 for Calton Hill trap-rock and garden-sand. The Committee's 

 experiments gave -0036 for dry, and '0144 for thoroughly water-saturated 

 fine siliceous sand. But the source of divergence in the appajrent geo- 

 thermal conductivity at Stockholm was more probably, as Angstrom 

 recognised, in comparing Quetelet's with his own result, the irregular 

 course of temperature at the locality in two successive years, than possible 



