172 
of the scale that was to be read, was placed behind an opening in 
the isolation material. Besides this thermometer opening only two 
small apertures were made in the isolation layer, through which 
the two small panes remained visible. 
4. Results. The results obtained by the methods described in the 
preceding paragraph, have been collected in Table I. They have 
been made with five apparatus; the determinations in the neighbour- 
hood of the critical temperature have been carried out with an 
apparatus which was about half filled with liquid at the ordinary 
temperature,. and the volume of which was therefore somewhat 
smaller than the critical volume. From our determinations at the 
temperatures in the immediate neighbourhood of 7; we have deter- 
mined the critical pressure graphically. The extrapolation that is 
required for it, can certainly be executed within the error of one 
atmosphere. Yet we think that we must consider the critical pressure 
accurate up to two atmospheres. We have namely no perfect certainty 
that the observations at the highest temperatures refer to the hetero- 
geneous equilibrium. The possibility cannot be entirely excluded that 
these observations represent a line in the homogeneous liquid region, 
though these determinations yield a practically continuously progressing 
curve with those at lower temperatures; if this should be the case, 
the deviation from the real vapour-tension curve is so slight, that 
the accurate value of the critial pressure could only be found by 
means of an extrapolation formula, drawn up from observations at 
lower temperature. In this, however, we also meet with difficulties, 
as then the extrapolation would have to take place over a greater 
range of temperature; we return to this extrapolation in a following 
paragraph. So we find 100 atmospheres for the critical pressure, in 
which we must consider a maximum deviation of two atmospheres 
possible. It will, moreover, be difficult to reach a greater accuracy, 
‘as it will not be possible to observe the critical pressure at the 
same time with the measurement of the pressure without complicating 
the arrangement considerably. Besides this would give rise to new 
experimental difficulties, because the critical phenomenon in itself 
is sO very difficult to observe. A manometer which was filled for 
about two thirds with liquid, presented a sudden deviation from 
the vapour tension line at about 140°; the pressure rose abnormally 
rapidly (about 6 atm. per degree) with slight rise of temperature, 
much more rapidly than the vapour tension line, even in the neigh- 
bourhood of the critical circumstances. So this apparatus is quite 
filled with liquid at 140°, and the abnormal rise of pressure was 
