7 
J ieratures—tlie two results should be alike, or nearly so ; and when 
tbj was not the case, to note all instances of discordance. 
len to examine a second time all Greenwich records on those days, 
to! .’ace the source of discordance; the errors thus found were mostly 
b< aging to the morning or noon observations. 
be last step was to combine the results found by the two methods 
tc ther, for the determination of the most probable mean temperature 
oii rery day, as found from all the observations taken that day, and in 
t] way Tables I. to XII. were formed. 
he numbers in the first column are the days of the month, those in 
t] following forty-four are the mean temperatures of the same day of 
t] month in the successive years ; the forty-sixth column contains the 
n n temperature of every day, as deduced from the forty-four years 
o ovations, and each value, therefore, is based upon about 220 obser- 
v ons, spread equally over the period. The remaining columns contain 
t highest and lowest mean daily temperatures within the periods, 
a the last column the difference between them. 
, he numbers in the bottom line are the means of all the numbers in 
tj columns above them, and therefore are the mean temperatures of 
e i month; each result is based upon 150 observations nearly; the 
v )le number of observations treated of in this paper exceed 80,000. 
I 
i 
