l8 M. Huniboldt on Isothermal Llnes^ and the 
it is liable. In oiir climates, the two extreme terms do not di- 
vide the series of twenty-fours into two equal parts. The maxi- 
mum is an epoch nearly fixed : the rising of the sun retards or 
hastens it three hours. As we ought to take into account the 
duration of the partial temperature, in order to find the quan- 
tity of heat divided between the night and the day, we must 
couple the maocimwn of one day with the minimum of the 
day following, and not be content with taking half the sum of 
all the maxima and minima of a month. In the ordinary me- 
thod, wc determine only the mean temperature of the part of 
the day comprehended between the rising of the sun and two 
o’clock in the afternoon ; and we take it for granted that the 
mean temperature is the same * from two o’clock to sunrise next 
day. This double error, of want of equi-distance and of the 
coupling of observations, does not in general produce errors of 
more than some tenths of degrees, sometimes in excess, and some- 
times in defect, since the warm and cold days are mixed -f*. 
All the calculated results wdll err in defect, if the 365 ordi- 
nates, through which the curve of the year passes, do not 
express an arithmetical progression, and if the partial irre- 
gularities do not sensibly compensate one another. It is only 
on this supposition that w^e can judge by the extreme terms of 
the series, of the sum of the terms, that is, of the partial tempe- 
ratures. It is very obvious, that near the maximum, the in- 
crease ought to be more slow than in other points of the curve, 
and that this increase in the temperature of the air ought to 
depend on the sine of the sun’s altitude, and on the emission of 
the radiant heat of the globe. 
* Example . — At sunrise at 6^, 50°; at 2 o’clock in the afternoon, 62°. 6. At 
sunset, 51°.8 ; at 2^, 66°.2 ; at sunrise, 50°. The true means will be for the first 
24 hours 56°.9, and for the second 59°.0, for- we shall have 
For 8S I (50°.0 -1- 66°.2) X 8 = 450°.4 for 8*^ riL 472°.0 
16' i(51°.8H- 62°.6) X 16 = 915°.2 929°.6 
The method commonly employed gives 4 (50° -f 62°.6) rr 56°.3, and | (66°.2 + 51 “S)* 
=: 58°. 1. The errors being — 0°.6 and -f 0.9, sometimes positive and sometimes 
negative.-^ — H. 
■f The error disappears when days of ec^ual temperature succeed one another. 
It amounts to 1°.8, if the mean temperatures of two successive days differ from 7® 
to 9°, which however very rarely happens.— H. 
