Proceedings of the British Association. 



91 



The difference of these numbers, such as it is, is evidently not attri- 

 butable either to errors of observation or to the inadequacy of the 

 experiment to afford consistent results ; a comparison of the curves for 

 two years, in which every observation is projected, shows the most 

 minute general conformity in their flexures and intersections. It is 

 rather probably to be attributed to the observations having been com- 

 menced too late in the winter of 1836-7 to obtain the true minimum for 

 that period, from which circumstance the superficial range would come 

 out too small, and B would therefore be diminished, as it is in each of 

 the three cases. 



From the observations of 1838 we deduce farther this important 

 result, that the oscillations of annual temperature would be reduced to 

 -— of a Centigrade degree (or virtually extinguished) at a depth of 

 49 feet in trap tufa. 

 62 feet in incoherent sand. 

 91 feet in compact sandstone. 



The differences in the value of B determined by various observers 

 (varying in the results quoted by M. Quetelet in his excellent memoir 

 on this subject, from — .0526 to — .0384) do not depend upon difference 

 of geographical position, but on the various constitutions of the soil 

 operated upon, a circumstance hitherto wholly neglected. 



II. — Epochs of Maximum and Minimum Temperature. 



Minimum : 



1837 



1838 



Maximum : 



1837 



1838 



3 Feet (French). 



Trap. 



Mar. 31 

 Feb. 26 



Aug. 6 

 Aug. 8 



Sand. 



Mar. 23 

 Mar. 3 



July 31 

 Aug. 6 



Sand- 

 stone. 



Mar. 20 

 Feb. 23 



Aug. 5 

 Aug. 16 



6 Feet. 



Trap. 



Apr. 9 

 Mar. 14 



Sept. 

 Sept. 



Sand. 



Apr. 5 

 Mar. 19 



Aug. 24 

 Aug. 31 



Sand- 

 stone. 



Mar. 26 

 Mar. 3 



Aug. 19 

 Aug. 23 



12 Feet. 



Trap. 



May 6 

 Apr. 20 



Oct. 17 

 Oct. 19 



Sand. 



Apr. 30 

 Apr. 22 



Oct. 6 

 Oct. 14 



Sand- 

 stone. 



Apr. 10 

 Mar. 20 



Sept. 11 

 Sept. 19 



24 Feet. 



Trap. 



July 26 

 July 18 



Jan. 8 

 Jan. 5 



Sand. 



Sand- 

 stone. 



July 12 

 July 8 



May 18 

 May 12 



Dee. 30 



Jan. 4 



Nov. 1 1 



Nov. 2 



These dates, derived by graphical interpolation, are only approximate. 



The progressive retardation of epochs as we descend is too evident to 

 require to be pointed out. The maximum occurs 5{ months after that 

 of the air in the two first-named strata, whilst the conducting power of 

 the sandstone is so superior as to accelerate this epoch by seven or 

 eight weeks, compared with the trap or loose sand. Were this result 

 deduced from thermometers placed at one depth only, its exactness 

 might be doubted. It is derived, however, also from the intermediate 

 ones. 



By a simple graphical method it is easy to deduce approximately the 

 rate of propagation of heat downwards in each of these soils, resulting 

 from the whole observations taken together. The observations at dif- 

 ferent depths confirm one another; but the minimum in 1837 was, as 



