1893 - 94 .] Prof. TdAt on Compression of Ordinary Liqiiids. 287 
of the region in which the constants of Van der Waals’ equation 
are non-real ; though they are, as a rule, nearer to the upper than 
to the lower limit. 
But it is well to inquire what values A assumes at the limits of 
this region, when it has just become real. A rough calculation 
shows that when^/g = 2'23 we have A= - 18*lg (a tension)', and 
for = 2'82 , A = 20^ . Outside these limits A has of course two 
values. 
It thus appears that Van der Waals’ equation becomes altogether 
meaningless except for liquids in which the compressibility alters 
very much with increase of pressure : — i.e. for substances which 
have just assumed the liquid form under considerable pressure. 
For, of course, under the lower limit we are dealing with substances 
naturally in a state of tension. As I said in my previous paper, 
this state of things is due mainly to the factor with which A 
(if taken as corresponding to my II) is affected. There is little 
doubt that the II term in my formula does increase as the volume 
is diminished, but much more slowly than in the inverse ratio of 
the square of the volume. 
(Added, 6/6/94.) It may be interesting to look at the above 
result from a different point of view, so as to find why it is im- 
possible to reconcile the general equation of Yan der Waals with 
the experiments of Amagat. 
For this purpose let us take jS as independent variable, and 
(using the same data as before) find the value of pjq. Eliminating 
BT and A, we obtain the equation 
from which, at once, 
^ - o-)(a - - e‘)(a - - P ) 
In the further discussion of this equation we may neglect the 
last term (which is usually very much smaller than the preceding 
term, and becomes infinite for the same values of ^). Its only 
