640 REGNAULT’S HYGROMETRICAL RESEARCHES. 
4. The ether employed is never anhydrous zther. The ordi- 
nary zther of commerce contains as much as one-tenth of its 
weight of water. This water is carried in great part by the va- 
pour of zther into a space very close to that in which the deposit 
of dew is determined. This circumstance again tends therefore 
to change the hygrometric state. 
5. If the temperature is high and the air very dry, it is im- 
possible to produce the deposit of dew upon the bulb A, even 
by pouring large quantities of zther upon the bulb B; so that 
in this case the instrument ceases to act. It is evident, more- 
over, that the inconveniences which I have noticed in 3 and 4, are 
considerable in proportion to the quantity of zther evaporated. 
A great number of modifications of Daniell’s apparatus have 
been suggested ; several experimenters have proposed to observe 
the deposit of dew upon the bulb of the thermometer itself. 
They have bent the stem of the thermometer, as shown in fig. 7, 
and adjusted exactly upon the upper part of the bulb a metallic 
cup, into which the ether intended to produce the cooling is 
poured ; the deposit of the dew is observed upon the naked part 
of the bulb. It is evident that this arrangement has no advan- 
tage over that of Daniell; the indication of the thermometer 
corresponds to the mean temperature of the different points of 
the mercury of the reservoir, and not to that of the portion of 
its envelope on which the deposit of dew is observed, and every 
one will see that a very marked difference may exist between 
these two temperatures, especially during the descending and 
always very rapid course of the thermometer. 
The same objections apply to the constructions proposed by 
M. Pouillet under the names of Capsule hygrometer and Virole 
hygrometer*, and to the metallic hygrometer of M. Savaryf. 
All these instruments are very inconvenient in one respect, which 
it is absolutely necessary to avoid. The surface upon which the 
deposit of the dew is observed is very near, often even the 
middle of the space in which the vapour of zther is developed, 
destined to produce the cooling. In all these instruments the 
object has been to increase the rapidity of the descending course 
of the thermometer, whereas it is evidently the contrary problem 
which must be proposed; this course must be rendered very 
slow, in order to be sure that only a very small difference exists 
* Elémens de Physique, 4th edition, vol. ii. p. 635. 
+ Annales de Chimie et de Physique, 3rd Series, vol. ii. p. 581. 
