225 



on tlie nature of the soil. If A^^ denote the range in Centigrade 

 degrees at a depth p, and A and B two constants, we have 



Log. A^, = A + B/) 



/specific heat 

 And B depends upon V conductivity ^^ the soil. Now this con- 

 stant B is found to have the following negative values : * 



Observatory —0.0547 ; Garden — 0.0440 ; Craigk-ith _ 0.0317. 

 Consequently the conducting power of the strata is in the order 

 just written, the first being the lowest, the last the greatest (sup- 

 posing the difference of specific heat immaterial). By extending 

 the Logarithmic curve, the depth at^'which the range has any 

 amount maybe found.t Thus the range will be reduced to ^J^ of 

 a Centigrade degree, or may be reckoned insensible, at the follow- 

 ing depths : — 

 Obser^'atory, 58 feet ; Experimental Garden, 72 feet ; Craigleith, 97 feet. 



Now it is,remarkable that the above variations in the value of B 

 exceed those contained in M. Quetelet's table, which includes all 

 the observations made in different parts of Europe ; shewing that 

 the increase of the value of B with the Latitude, which was thought 

 to have been observed (Quetelet, Mhnoire sur les Variations Diurne 

 €t Annuelle de la Temperature Terrestre, p. 61), was quite acciden- 

 tal ; and that the value of B must depend solely on the constitution 

 of the soil in which the experiments are conducted. 



IV. The epochs of the winter minimum of 1(337 are ill deter- 

 mined for the upper thermometers, owing to the great irregularity 

 of the winter cuive, and also because the observations were com- 

 menced too late to obtain them correctly. The epochs of maxi- 

 mum temperature give, however, a complete confirmation of theory, 

 and of the preceding deductions. We find that, with a single ex- 

 ception (the shortest thermometer at Craigleith), and that excep- 

 tion is doubtful, all the thermometers at Craigleith attained their 

 maximum first, then those at the Experimental Garden, and lastly, 

 at the Observatory, as the following table shews : — 



* In order to render the results directly comparable with those in M. 

 Quetelet's excellent paper in the Brusscl's Transactions, the French foot 

 and Centigi-adc degree are employed as unities. 



t The values of A are l.l(i4, 1.176, 1.076 in the same order as before. 



