IN THE PRESENCE OP HUST-FREE AIR AND OTHER GASES. 
267 
jjroving beyond doubt that condensation may be made to take place by increasing 
the supersaturation alone, as so many of the conditions besides the degree of 
supersaturation must vary as the initial pressure of the steam, and consequently the 
velocity with whicli it escapes from the nozzle are increased. 
Conditions to he satisfied hy the Expansion Apparatus. 
To obtain unequivocal proof of the production of condensation in moist air, free 
from all extraneous nuclei, it is necessary that we should not be dependent upon any 
process of filtering, for it might always be objected that the filtering apparatus only 
removed those particles which exceeded a certain size. 
If, however, we expand repeatedly the same sample of moist air, while protecting 
it from all chance of contamination, we are able to test whether all nuclei of a 
permanent kind have been removed. For by making an expansion rather greater 
than is sufficient to cause condensation, and allowing the drops formed to settle, we 
remove in this way a certain proportion at least, and if the drops be few and large, 
almost the whole of the nuclei which are able to cause condensation witli this degree 
of supersaturation. 
If this process can be repeated indefinitely without any diminution in the number 
of drops formed, we are justified in concluding that the nuclei are being replaced by 
others as fast as they are removed, and are thus an essential part of the structure of 
the moist gas. 
It is desirable also that the expansion should admit of accurate measurement and 
be exceedingly rapid, so that the lowest temperature and maximum supersaturation 
reached may be calculated with as small an error as possible d\ie to the influx of heat 
during the expansion. 
In a note"^ read before the Cambridge Philosophical Society, I gave an account of 
some preliminary results obtained with a form of apparatus which I believed to 
satisfy these conditions. It was there stated that condensation results from the 
sudden expansion of saturated dust-free air when exceeds a value not differing 
much from 1‘258, where v-^, are the volumes of the air before and after expansion. 
No description of the apparatus was published, as it was then in quite a rudi¬ 
mentary condition. 
The first series of experiments to be described here was carried out with an 
improved apparatus of the same type. 
Apparatus used in the first series of Experiments. 
This is represented in vertical section in fig. 1. 
The air to be expanded is contained in the inverted cylindrical glass vessel A, 
* ‘ Proo. Canili. Phil. Soc.,’ vol. 8, p. 306, 1895. 
2 H 2 
