160 



ROYAL SOCIETY OF CANADA 



The deviation is considerable in November, but we must remember 

 that an agreement of better than a degree by an ordinary thermograph 

 is not to be expected. The difference for October in the case of the 

 differential recorder was computed from only eight charts at different 

 dates throughout the month, so the agreement here is accidental. 

 The agreement for December is more satisfactory. 



In the following table we reproduce the table of differences for 

 the several years, including the results during the past year, 1907-08. 



July 



August 



September 



October 



November 



December 



January 



February 



March 



April 



May 



June 



TABLE II. 

 1903-04 1904-05 1905-06 190&-07 



-5°.5 Fah — 2°.6 Fah. 



-4°.7 " — 2°.5 " 



-5°.8 " — 4°.4 " 



^TO Q « .... 



-4°.6 " — 5°.7 " 



-8°.2 " — 3°.0 Fah. — 8°.2 " 



-110.2 " — 2°.2 " — 6°.3 " 



-8.°1 " — 2°.7 Fah. — 1°.6 " — 4°.3 " 



,509 « _2°.0 " — 2°.5 " — 2°.9 " 



-6°.8 " — 3°.l " — 3°.7 " — 4°.6 " 



.30 J « _2°.6 " ' — 2°.6 " 



— 2°.2 " — 2°.5 " 



Temperature Forecasts. 



1907-08 

 -1°.9 Fah. 

 -4°. 4 " 

 -1°.6 " 

 -3°.0 " 

 -2°. 9 " 

 -3°.6 " 



-1°.5 " 



-r.8 " 



-2°.8 " 



As an example of the way our records afford a means of forecast- 

 ing the temperature conditions of the next day, we include the fol- 

 lowing table of predictions, taken from the April records. It is from 

 a careful study of the general variation of the differences for each 

 daily chart, as well as from the actual magnitude of these differences. 

 No single observation at one time can give anything like as accurate 

 a prediction as a general examination extending over several hours. 

 We always have to consider what has happened before, going back 

 over as much as 24 or even 48 hours by inspection of the continuous 

 variations in the temperature gradient in the lower strata of our 

 atmosphere. 



