Observations of Underground Temperature. 131 



The true value of t / — must, as we have seen, be '116, to a 



very close degree of approximation. 



33. When we consider the character of the reduction we have 

 made, and remember that the data were such as to give no sem- 

 blance of a theoretical agreement when the first five years' term 

 of observations was taken separately, we may be well satisfied 

 with the approach to agreement presented by these results, de- 

 pending as they do on only eighteen years in all, and we may 

 expect that, when the average is of a still larger term of observa- 

 tion, the discrepancies will be much diminished. In the mean 

 time we may regard the semiannual term we have found for 

 the three-feet deep thermometer as representing a true feature 

 of the yearly vicissitude ; and it will surely be interesting to find 

 whether it is a constant feature for the locality of Edinburgh, to 

 be reproduced on averages of subsequent terms of observation. 



34. It may be remarked that the nearer to the equator is the 

 locality, the greater relatively will be the semiannual term ; 

 that within the tropics the semiannual term may predominate, 

 except at great depths; and that at the equator the tendency 

 is for the annual term to disappear altogether, and to leave a 

 semiannual term as the first in a harmonic expression of the 

 yearly vicissitude of temperature. The facilities which under- 

 ground observation affords for the analysis of periodic variations 

 of temperature when the method of reduction which I have 

 adopted is followed, will, it is to be hoped, induce those who 

 have made similar observations in other localities to apply the 

 same kind of analysis to their results; and it is much to be 

 desired that the system of observing temperatures at two, if not 

 more depths below the surface may be generally adopted at all 

 meteorological stations, as it will be a most valuable means for 

 investigating the harmonic composition of the annual vicissitudes. 



III. Deduction of Conductivities. 



35. Notwithstanding the difficulty we have seen must attend 

 any attempt to investigate all the circumstances which must be 

 understood in order to reconcile perfectly the observed results 

 with theory, the general agreement which we have found is quite 

 sufficient to allow us to form a very close estimate of the ratio of 

 the conductivity of the rock to its specific heat per unit of bulk. 

 Thus, according to the means deduced from the whole period of 

 eighteen years' observation, the average rate of variation of the 

 logarithmic amplitude of the annual term through the whole 

 space of twenty-one feet is "1157, and of the epoch of the same 

 term, -1154. The mean of these, or -1156, can differ but very little 



K2 



