210 
SECOND REPORT- 1832. 
able to show that no sensible change has taken place upon the 
climate of Florence between that period and the present, which 
had been suspected. 
The metallic thermometers of M. Breguet have not been so 
much used as might have been expected ; they are however 
rather adapted to delicate experiments on heat than to Meteo¬ 
rology in general. The same remark applies to the beautiful 
thermo-multiplier recently invented by Signor Nobili*, and 
which he has applied to the investigation of some of the most 
delicate phenomena of radiant heat f. 
In the use of the thermometer, the sources of error from 
terrestrial and solar radiation have not been sufficiently attended 
to. Some hints on the subject may be found in the Article 
Thermometer in the Edinburgh Encyclopaedia. The mode 
of exposure of the thermometer of the Observatory at Paris 
is described by Pouillet in his Elemens de Physique . 
These various sources of error go far to diminish our confi¬ 
dence in the nice accuracy of thermometric results, where they 
have not been the subject of particular attention. It is surpri¬ 
sing at what a distance a sensible portion of heat is conveyed 
from soil and walls, or even from grass, illuminated by the sun; 
the maxima of temperature are thus generally too great, and 
from the near contact in which thermometers are generally 
placed with large difficultly conducting masses, such as walls, 
the temperature during the night is kept up, and the minima 
are thus also too high. 
This however has been by no means the greatest difficulty 
in determining the mean temperature of a given spot. Since it 
is difficult to have the thermometer observed oftener than twice 
or thrice in a day, it becomes an object of great importance to 
determine those hours the mean temperatures of which will give 
that of the whole twenty-four. This however involves an ac¬ 
curate determination of the curve of diurnal temperature ; and 
as this varies with the seasons, its connexion with the curve 
of annual mean temperature must also be assigned. It is ob¬ 
vious that for these objects a very extended scale of observations 
is requisite, but that when once attained, the results will be sub¬ 
ject to the same general law throughout a considerable extent 
of country. The mean of the maximum and minimum tempe¬ 
rature measured by a register thermometer is one of the best 
approximations ; it is however by no means absolutely accurate. 
The multiplication of hourly observations has only in one or 
two instances been resorted to for filling up this blank ; but it 
* Bibliotheque Universelle, N. S. ii. 225. 
f Annales de Chimie, 1831. 
