69 
would like to suggest the great desirability of having these 
corrections deduced from a much larger series of observations, 
for although on the whole I find them to agree pretty closely, 
yet at times great differences exist, especially in comparing 
the mean temperature deduced from maximum and mini- 
mum readings with those of a standard thermometer read 
at certain fixed hours. 
The coldest weeks during the 17 years average appear to 
be those ending 7th and 14th January, and the warmest that 
ending July 22nd. 
The week ending December 29th appears to be the one 
in which the greatest variation of mean weekly temperature 
is likely to occur, and the one ending August 19th that in 
which the least variation occurs. 
Taking the mean differences for each month, and we 
obtain the following figures : — 
January 
Mean Differ. 
...... 16-40° 
July 
Mean Differ. 
February ..., 
18-62 
August 
8-20 
March 
13-40 
September... . 
10-26 
April 
12-05 
October 
11-34 
May 
14-37 
November ... . 
..... 13-62 
June ... 
12-42 
December .... 
..... 20-22 
From this table we see that the greatest amount of varia- 
tion occurs in December, and the least amount in August. 
October does not appear to exhibit any abnormal variation, 
although from the amount of barometrical oscillation doing, 
so it might have been expected it would do. 
I should like to see a much longer period observed for 
determining the values of the mean weekly temperature 
here, but hope until such is the case the values I have given 
may be deemed worthy of the confidence so short a period 
can deserve, and I can safely say, that as the great mass of 
the observations were made by myself, every care has been 
taken in order that they might be made correctly* 
