116 



ROYAL SOCIETY OF CANADA 



plotted for ditterent dates, some of which are yhown in Fig. II. From 

 these curves the following values of the diffusivity have been deduced 

 between the dates and depths specified. 



Table op the Thermal Diffusivity at Different Depths, 

 Dates, and Seasons. 



With these we may include the results given last year 



The average value of k/r for the year 1896 as taken from the 

 probable curve is '0043, a result which is in fair accord with the value 

 •0036 obtained from the yearly curves. 



The results given in the preceding table exhibit several differences, 

 which require some elucidation. As explained last year, the rate of 

 diffusion of temperature varies considerably according to the condition of 

 the soil, and in particular is very greatly increased by percolation of rain. 

 It is our object to determine the nature and extent of these variations. 

 This we were unable to do last year owing to lack of time and of suffi- 

 cient observations. 



The value found for IX'C. 13, 14, 1894, is the highest which we have 

 yet obtained, l»eing about ten times as great as the average yearly value. 

 This exceptional result was entirely due to the rapid percolation of cold 

 rain and melted snow through the upper layers of the soil. In striking 



