150 
ON THE PERIODIC AND NON-PERlODlC VARIATIONS 
observations at 9“ or 10'’— 10" instead of from hourly observations, are shown on 
the average of six years to be as follows the sign + denoting that the mean daily 
temperature in the month is given too high, and - that it is given too low, by the 
mean of the two compared with the mean of the twenty-four observations. 
^ ^ ^ ^ 
January 
O 
— 0-4 
July —0*4 
January O’O 
July 
—0-2 
February 
— 0-8 
August --0‘3 
February —O'l 
August 
“1” O' 1 
March 
-0*5 
September — 0'2 
March +0‘1 
September 
+0-3 
April 
May 
June 
-0-4 
October —01 
April —O'l 
October 
-f-0'5 
— O'l 
November — 0’2 
May +0-3 
November 
+ 0-4 
-0-3 
December — 0’6 
June — 0'2 
December 
O'O 
Mean of the year 
_ 0 o. 3 g Mean of the year +0 ’09 
Difference of the hottest and coldest months : 
Too great by Q-'d Too great by O"- 1 5 
For the purpose of combining with an approximate mean temperature of the day, 
an approximation also to the hottest and coldest hours of the day, and to the hours 
of maximum and minimum of other meteorological elements, three equidistant obser- 
vations are frequently adopted in preference to a binary system, and the hours of 
6 A.M., 2 p.M. and 10 P.M. appear to be usually preferred. These hours are still within 
the command of a single observer, though we often find substituted for them the 
non-equidistant hours of 7 a.m., 2 p.m. and 9 p.m., doubtless because they suit better 
the convenience of observers. On comparing the mean temperatures in the different 
months derived from 6", 2", 10", or 7 ", 2", 9", with the full complement of twenty- 
four hours, we find that the approximation to the mean temperature obtained by 
7" 91. 9" is not quite so good as by 6", 2", 10" ; and that either of the triplets give a 
less correct mean temperature than 10"-10" : 6i, 2, 9i would appear a more suitable 
combination as far as regards approximation to the mean temperature. 
6 A.M., 2 P.M., 10 P.M. 
7 A.M., 2 P.M., 9 P.M. 
r 
January 
+6-3 
^ 
July — 0'6 
( 
January 
4-0*4 July 
O 
+ 1*0 
February 
— O'l 
August —0*9 
February 
0*0 August 
+0*5 
March 
-0-4 
September -0*7 
March 
4-0*1 September 
+ 0*5 
April 
May 
June 
-0-6 
October — 0*2 
April 
4-0*4 October 
+ 0*3 
— 0-5 
November +0'2 
May 
+0*8 November 
+ 0*3 
-0-7 
December +0*1 
June 
+ 0*7 December 
+ 0*2 
Mean of the year 
— 0°'34 
Difference of the hottest and 
Mean of the year +0*43 
the coldest months : 
Too small by 0»-4 Too great by r-0 
Three equidistant observations in the twenty-four hours are the utmost that can 
