1089 
from other substances: this he correctly ascribed to the circumstance 
that air is a mixture, bat the special phenomena which he describes 
were largely due to insufficient mixing. It was therefore necessary 
to repeat the investigation with al] those precautions which in previous 
investigations on mixtures have proved necessary and amongst others 
to try and realise “retrograde” condensation *), which is characteristic 
of mixtures. So far an investigation of that sort had never been 
earried out but at the ordinary and at higher temperatures. 
As expected the investigation proved to be beset with great exper- 
imental difficulties. Generally speaking these difficulties all originate 
in the circumstance that the mixture cannot as a whole be 
cooled down to a low temperature, at least, if the possibility must 
be left open — and this is an essential condition in the experiments 
here contemplated — of changing the volume of the substance 
gradually. A substance which at the very low temperatures could 
play the part which is otherwise fulfilled by mercury, viz. that of 
enclosing a fixed quantity of the substance in a variable volume, is 
unfortunately not known. It is therefore necessary to compress the 
mixture in a small tube which is closed at the bottom and cooled 
to the low temperature, by means of a piezometer, thus using the 
same method as followed with pure substances, so that every time 
a different fraction of the total quantity of substance is present in 
the observation-tube. With pure substances this does not involve any 
fundamental difficulty; by measuring the quantity of gas present in 
the part of the piezometer which is outside the eryostat, the quantity 
in the observation-tube can at each measurement be derived by 
subtraction from the total quantity, even when the substance is partly 
liquefied. With mixtures this is different: in the condensation of a 
mixture new mixtures are formed each time of different composition ; 
in a series of observations in which the mixture is alternately com- 
pressed and expanded the mean composition of the mixture in the 
observation-tube will thereby very soon become different from that 
of the whole and the observations lose all definite meaning as 
referring to mixtures of varying and unknown composition. Taking 
air, the mixture dealt with in our investigation, as an instance, 
when it has been partly liquefied and is now re-evaporated, in the 
beginning the nitrogen will principally boil away and disappear 
from the tube, whereby the mixture, which remains behind, becomes 
richer and richer in oxygen and no longer has the same mean 
composition as air. 
1) J. P. Kuznen. Comm. Leiden 4. 1892, 
