144 



ROYAL SOCIETY OF CANADA 



judge by the records of the inhabitants of that district, as the winter of 

 1903-4. For instance, on one occasion when we recorded our greatest 

 difference, colder on flie mountain, the mercury was reported to have 

 frozen in a thermometer as near as Ste. Rose, (20 miles), while during 

 the winter just passed no such condîtion existed although the tempera- 

 ture at the observatory as shown by the monthly averages was about the 

 same for the tiwo winters. 



Below we give the monthly average differences for February, March 

 and April and for comparison we reproduce those for the corresponding 

 months of the year previous : 



Relative Temperature Differences. 



In general we find that the position of the trace, whether rising or 

 falling, is of the greatest importance in determining the character of the 

 temperature conditions at the lower station for the following 10 or -12 

 hours. In nearly all cases of a sudden change of temperature we find 

 that the high level instrument is aifected several hours ahead. This is 

 shown by the slow travelling of the pen across the sheet. A continued 

 trace of positive differences (higher at the upper station) means weather 

 conditions growing steadily milder at the low level. The following 

 conditions have been observed to hold fairly accurately. 



1. Steady positive difference shows weather conditions growing 

 milder. 



2. Increasing positive ditt'erence, or decreasing negative, indicates 

 a change to milder weather. 



3. Decreasing positive difference means steady temperature con- 

 ditions possibly growing colder. 



4. Increasing negative difference indicates temperature conditions 

 growing colder at the lower level. 



5. Steady negative difference indicates steady and continued tem- 

 perature conditions at the low level. 



All these represent changes at the low level following from 5 to 34 

 hours after the difference is observed to change. A few examples taken 

 from our last year's records at random will serve as an illustration : On 

 February 9th, a warm wave was indicated 9 hours before it reached the ob- 



