LOW TEMPERATURES AND REFRIGERATION—MARCHIS. PAI lab 
It can be readily seen, therefore, that if the two phases, liquid 
and gaseous, formed by the progressive liquefaction of air are main- 
tained in contact, it is impossible to prepare gaseous nitrogen of a 
sufficient degree of purity. 
It is a different matter, however, when, under constant pressure 
conditions, the liquid phase is eliminated as fast as it is produced. 
We find ourselves here in the presence of a phenomenon inverse to 
that observed when liquid air is distilled under constant pressure. 
In such a case the contents of gaseous and liquid phases increase 
continuously. The phases both tend toward a composition of pure 
oxygen. At the same time the temperature of boiling rises from 
a value in the neighborhood of that of pure nitrogen to the boiling 
point of pure oxygen. Inversely, if we progressively condense air 
under constant pressure, eliminating the liquid phase as fast as it 
is formed, there are obtained gaseous residues less and less rich in 
oxygen; at the same time the temperature of condensation becomes 
lower and tends toward the boiling point of pure nitrogen at the 
pressure employed. We obtain, therefore, much more rapidly than in 
the process considered above a gaseous mixture richer in nitrogen. 
To obtain a gaseous residue practically free from oxygen, however, 
it is necessary, if this method is used, to almost completely liquefy 
the air. 
A much better result is obtained by the use of a device designed 
by M. Claude which he calls a retour en arriére (reflux apparatus). 
Let us imagine that the liquid after separating from the gas en- 
counters a gaseous mass richer in oxygen. The liquid is colder on 
account of the large proportion of nitrogen it contains. <A part of 
the more condensable oxygen of the mixture will therefore be lique- 
fied and take the place of nitrogen which will vaporize. Thus by 
circulating in opposite directions a liquid and a gas having different 
contents, there is obtained on one hand a liquid very rich in oxygen 
and on the other practically pure gaseous nitrogen. This process 
utilizes, moreover, a large amount of air which never needs to be 
liquefied. 
M. Claude has utilized this principle of reflux in the following 
way: A sort of small tubular boiler is arranged so that its axis is 
vertical. The tubes are surrounded on the outside with lquid air 
and into the lower ends of these tubes is led a current of cold com- 
pressed air. This liquefies progressively, giving liquids poorer and 
poorer in oxygen; these liquids, in falling into a receptacle below, 
encounter gases rich in oxygen and produce the gradual dilution, 
the principle of which we have described. There finally separates 
out at the top of the group of tubes practically pure nitrogen, while 
liquid with a high percentage of oxygen is continually drawn out 
of the lower part. 
