( 942 ) 
for temperatures between its boiling point and its critical point, so 
that measurements for the most important portion of the diameter 
viz. that part lying between the boiling point and the critical point, 
are not yet possible, and only that part of the diameter lying below 
the boiling point, which we may term the produced diameter, is 
available for measurement. 
A final reason for the use of oxygen is that it can be easily 
prepared in a perfectly pure condition. This was done by heating 
potassium permanganate contained in tubes forming part of an 
apparatus made completely from glass. The oxygen liquefied in 
another part of the apparatus that was immersed in liquid air, and 
was then distilled from this apparatus into a cylindrical reservoir 
(see § 4) that was in its turn cooled in liquid air. The quantity of 
liquid condensed in this reservoir was so regulated that when the 
ordinary temperature was reached again, the safe pressure allowed 
by the construction of this copper cylinder was not exceeded. 
§ 2. First Method. Densimeter. We deduced the constants for the 
diameter from measurements of the densities of the liquid and of 
the vapour at a series of different temperatures. At every density 
determination for both liquid and vapour the phase whose density 
was being determined was kept in equilibrium with a small quantity 
of the coexisting phase. 
Fig. 1 Pl. IL represents an apparatus with which the diameter 
can be directly determined. Two equal reservoirs A and 5 are 
connected by means of a graduated capillary c; the upper reservoir 
ends in a narrow capillary coupled to a tap f. The internal volume 
of the apparatus as far as one of the divisions d is twice the volume of the 
lower reservoir as far as one of the divisions on the capillary c. 
The dilatometer is now filled so that at the lowest temperature 
employed, the Jiquid meniscus stands at the central mark: the tem- 
perature of the apparatus is then raised and each time this is done 
so much of the vapour is allowed to escape through the tap f that 
the liquid meniscus always touches the central line. The quantity 
of oxygen contained in the apparatus at the lowest temperature and 
the quantities which are successively allowed to escape on proceding 
to other temperatures are measured, and corrections are applied for 
the narrow capillary e, for the difference between the nominally 
equal volumes V4 and Vg of the two portions A and B and also 
for the difference between the liquid level and the central mark, 
and then the data for the diameter are at once deduced from the 
equation VA Olig + Vn Ovap = 
Va (lig + Gray), Where V4 is known, 
