( 480 ) 
the meniscus with increase of temperature was seen last and reap- 
peared when the temperature was lowered; hence very nearly the 
critical temperature. It will be seen that most of the curves of 
equal density, when the temperature is raised, leave the point where 
the meniscus was seen last, bend away from that point more and more 
rapidly, turn round at about the highest temperature observed and 
return to the same point, which only few, however, reach when the 
temperature is fallen to the eritical temperature. 
From this last circumstance we conclude that the course of 
the curves of equal density is not only governed by the variation 
of the temperature but also by diffusion. Both through increase of tem- 
perature and also through diffusion, the distribution of the substance 
becomes more regular, and hence the curves of equal density ascend 
and would finally project beyond the drawing, if not the decrease 
of temperature in the second part of the experiment caused the 
withdrawing curves, partly at least, to return. But the very fact that 
the curves of equal density in the second part le higher than those 
at equal temperatures in the first, is a proof that the progressing 
diffusion opposes the influence of the temperature; the following 
numbers may show which is about the course of the greatest difference 
of density in the tube throughout the series of experiment : 
t — 282° 283° 284° 285° 286° 288° 286° 284° 283° 282° 
A —= 50 DA AQ) ie 30 he 25 mie 20 He 15 Os HED 20 SA 2D 30 bi 
It will be seen that the difference in density first decreases, then 
inereases, but the values at equal temperatures are lower in the 
second part than in the first and the deviation increases; from this 
appears the influence of diffusion. 
The value of 4 is not even smallest at the highest temperatures ; 
the smallest value is not reached until the temperature is falling, in 
harmony with which is the fact that the bulbs 0.555 and 0.578 have 
reached their highest position not at 281°.1 C. but at 286°.0 C., 
hence during the period of decreasing temperature. This proves that, 
at least at the beginning of the decrease of temperature, the diffusion 
has a preponderating influence. That the heaviest bulbs did not show 
this peculiarity must probably be ascribed to the circumstance that 
in the lower part of the tube, where the substance is much denser, 
the diffusion takes place much more slowly; in those lower curves of 
density, however, we can clearly distinguish a point of inflection, 
which also, though less striking, points at the progressing diffusion. 
That these circumstances can actually be explained by the diffusion 
of impurities I have tried to demonstrate by calculating and by 
