1025 
windows A with the eylindre B is screwed on, and the glass tube 
() is placed in the vessel 4. After this through the other window 
opening the vessel £ is filled with one of the components, till the 
liquid begins to flow out at S,. The tube C is then closed at S, by 
means of a wooden peg, the second window is screwed on, and at 
last the observation vessel £ is quite filled with liquid through the 
conduit terminating at WW (fig. 5). Then JV is closed. Beforehand 
the tube conduit CC, has been quite filled with mercury, which 
has been poured in at S,, to prevent contact of the observation 
vessel 4 with the oil from the press. The wooden peg is quickly 
removed, and the coupling at S, is effected. The steel tube C’ being 
very narrow, only very little, if any liquid, escapes. Thus the mix- 
ture under examination is quite guarded against the influence of 
contamination, and its concentration changes but exceedingly littie. 
on account of the slight compressibility of the investigated liquids, 
while there can hardly be any question of diffusion through the 
narrow capillary in the course of the observations. Moreover a slight 
change in the concentration could not exert an appreciable influence 
on the results on account of the greatly flattened shape which the 
liquid-liquid plaits always seem to present. 
When the filling is finished, the thermostat is put in its place. 
An intense metal wire incandescent lamp of 300 candles is placed 
behind the window in the back wall. In this way, the mixture, 
particularly the place of the meniscus, can be very clearly observed. 
We have confined ourselves in this investigation to plaitpoint obser- 
vations; as criterion the same phenomenon was taken as was also 
used by one of us (7°) in his observations in Cailletet tubes *): while 
the pressure is kept constant, the temperature is slowly made to 
oscillate round the plaitpoint temperature, the liquid being contin- 
uously stirred. The temperatures are recorded at which the turbi- 
dity resp. the transparency sets in, and the mean of all the obser- 
vations is taken as plaitpoint temperature. Proeeeding in this way 
temperatures of turbidity are always obtained which diverge only 
some hundredths of degrees when the experiment is repeated. lt 
may also be assumed that the temperature of the mixture follows 
that of the thermostat very closely, for also the mean of the tempe- 
ratures of turbidity deviates but a few hundredths of degrees from the 
mean of the temperatures at which transparency sets in in general. 
This circumstance proves at the same time that equilibrium is pro- 
perly secured by the constant stirring of the liquid. This was 
further confirmed when a glass ball was placed in the glass tube 
Y 1) These Proc. XIIL p. 507. 
