1027 
on the supposition, however, that the glass windows are again 
horizontal not vertical, as in fig. 2, but in the same position as it 
is represented on a small scale in fig. 3). So A is again the glass 
window; the other parts too, for so far as they are visible, are 
denoted by the same letters. It is clear from the figure that besides 
the main conduit, in which the observation tube O lies, and the 
branch conduit, which connects this main conduit with DC, there 
is another branch conduit in the observation vessel /, at right angles 
to the two first mentioned. This branch conduit is used for the 
filling; it is then closed by a tigat stopper W, which is again 
provided with a cone, and which is pressed against the vessel 7 
with steel-to-steel closure by means of an oval flange G,:laid over 
this cone, with the nuts and bolts belonging to it. This closure is further 
made clear by fig. 6, where the flange G, is 
represented seen from above, the line ZZ’ corre- 
sponding with ZZ of fig. 2. Fig. 6 also shows 
the general form of the flange plates G. Fig. 5 
finally shows the octagon L,...L, of the 
observation. vessel Z, which serves to fix the whole piece sufficiently 
firmly, when the steel cylindres B are serewed into it, which 
cylindres themselves have of course also a hexagon. The closure of 
B on L takes of course again place steel to steel by means of the 
raised hardened rim of 4, which fig. 2 clearly shows in section. 
We will avail ourselves of this opportunity to express our indeb- 
tedness to the instrumentmaker of the Van per Waars-fund, Mr. U. 
H. SrurveNBERG, for his intelligent assistance in the construction of 
this apparatus, and particularly in the grinding of the windows. 
$ 3. Description of the observations. 
When the apparatus described in the preceding paragraphs, is 
used, care should be taken in the first place that the composition 
of the examined mixture does not change, and that no impurities 
can appear, which might have a preponderating influence on the 
course of the phenomena. This result may be attained by enclosing 
the mixtures to be examined in a glass tube, closed at the two ends 
and provided on the side by a capillary (fig. 2, QO) as long as 
possible, which causes the pressure on the liquid inside to be the 
same as that on the surrounding liquid. This tube has beforehand 
been filled with a mixture of the required concentration, that of the 
critical endpoint. 
Then the steel tube C' is disconnected at $,, being connected with 
the observation vessel 4 at the same time. Now one of the two 
