372 - . L, DeForest on Correcting an error 
ference of seven days in the length of February causes only a 
difference of 03 in the mean annual temperature A,, and 18 or 
ral times greater than this, as will appear hereafter. 
The length of the year being taken at 3654 days, one twelfth 
part of it will be 30,7, days, which for convenience’s sake will 
call a mean month. I propose to show how to find the mean 
temperatures of the mean months, and to derive from them the 
equation of the curve, which will thus be corrected for the ine- 
quality of the months; then by comparing such equations with 
uncorrected ones found by using calendar months only, the real 
amount of the errors in question will become known. 
Reckoning from the beginning of January, and allowing 284 
days to the calendar month of February, we shall find that 
The 2nd mean month begins +8, days before February begins. 
“ 3rd “ “ “ begins. 
] 1g days after March 
“ 4th iT “ 66 1 as +“ 7 April 
“ 5th “ “ “ 1 4 6 “« eM ay “ 
* 6th “6 rT: “ 4 eS “ &“ June bd 
“ "th ‘6 “ “ 1 $ “ “ Ju sd 
“ 8th 3 3 ‘“ ES “ ‘“ un t “ 
“ gt u“ “ “ Yi “i ‘“ September & 
“ 10th “ “ “ 4 4 ss “ October $6 
“ 1th “ “ “ 4 ‘“ “ Noy ember 6“ 
ts 
Tf we have the daily mean temperatures at a given place i 
ulated for every day in the year, as in the table for Greenwi¢ 
43°6 which is the daily mean for April Ist, and 2°76 which 1s 
one-sixteenth of the mean for April 2nd; then there m 
added 50:0 the mean for May Ist, and 25-25 which is one-half the 
mean for May 2nd. This gives 1415-99 as the sum of all the 
daily means for the fourth mean month, and dividing by 30% 
we get 46°52 for its mean temperature, he 
The mean temperatures for the calendar months at Greenwich 
are, from 43 years’ observations, ‘ 
