of Reducing Observations of Temperature. 29 
fan a. é. 
Jan, 29 Feb. 33 March 29 gives 46°91 10°87 83° 37’ 
Meee 26. 4. 34. 2 4688-10-81 © 83° 19" 
Meee 31 | 80+, 4690 1076 83° 33" 
Calendar months give a . 46:90 10:78 83° 27’ 
Hence, I think we may conclude that the probable error pro- 
duced by using the calendar months as twelfth parts of a year 
is about a tenth of a day in phase, and about a hundredth of a 
degree in amplitude. These differences are rather greater than 
those which were detailed in last paragraph ; but they are not 
peculiar to the present method, and will generally tend to de- 
stroy one another in making comparisons. 
Apart from this small source of error, the conclusions de- 
duced from monthly means are as accurate as those from daily 
means. Practically they will be found more accurate, because 
the comparative steadiness of monthly means renders their 
treatment more easy and certain. 
Supposing the operations to be correctly performed, the value 
of e will be the same from daily means as from monthly, and 
that of @ will be greater in the constant ratio of 1-:0115 to 1, 
as I have ascertained by a mathematical investigation.* 
The labour of computation involved in the present method 
is so small, that when the monthly means have been written 
down, the values of a and e can be found in 10 minutes. 
It is not necessary that 1 should show in detail the advan- 
tages which meteorology may be expected to derive from the 
extensive application of the method of reduction here pro- 
posed. The superiority of definite measures to mere general 
estimates is universally recognised by those who have to deal 
with statistics ; and yet no such measure has been usually, if 
at all, applied to the important element “date of phase;” and 
the measures which are usually applied to determine range 
are subject to an error which affects different places very un- 
equally. Some such method as the present is therefore de- 
manded by the requirements of science. t 
%* [Professor Everett seems not to have seen Principal Forbes’ paper in which 
a similar correction is made for the annual range.—EDITOR Edin. New Phil. 
Journal.]} 
{ [The “ Date of Phase” was calculated by the same method by Kamst and 
Herschel, and was lately calculated by Principal Forbes for each of the forty 
