94 Scientific Proceedings, Royal Dublin Society. 
maximum vapour-pressure; and, as it is quite impossible to form 
an estimate, it is safest to assume that no mercury-vapour diffuses 
through the liquid into the space above. 
Tt must be admitted, however, that a small error is thus 
introduced, because, after the determinations of vapour-pressure 
at the highest temperatures, when the condensed liquid was 
carefully examined with a lens, minute globules of mercury 
could be seen adhering to the walls of the experimental tube. 
It is evident, then, that, during the determination of vapour- 
pressure, minute quantities of mercury are constantly diffusing 
through the liquid into the vapour above. When the readings 
are taken during evaporation, both the total amount of mercury- 
vapour and the volume occupied by it will be increasing, and the 
partial pressure will remain small and fairly constant. But if, 
after the expansion, we again take readings during condensation, 
additional mercury-vapour will continue to diffuse through the 
liquid, while that in the space above will become more and more 
compressed. The partial pressure will, therefore, increase from both 
causes, and the error will be much greater at the smallest volume 
than at the largest. 
We have thus an additional reason for preferring readings 
taken during evaporation to those taken during condensation, 
and also for expecting larger errors at high temperatures than 
at low. 
3. Insufficient time for equilibrium before taking readings.— 
When the volume is increased, evaporation takes place and 
heat is absorbed. The liquid in the experimental tube is thus 
cooled, and time must be allowed for heat to be received from 
the jacketing vapour. But when the experimental tube is cooler 
than the saturated vapour outside it, condensation of this vapour 
takes place and the heat thus evolved is rapidly communicated to 
the tube, and through it to the liquid inside. Experience has 
shown that a very few minutes suffice to bring about equilibrium 
even when the experimental tube is at first many degrees cooler 
than the vapour. On the other hand, when readings are taken 
during condensation, heat is evolved and the temperature rises 
slightly. But the fall in temperature of a superheated tube 
surrounded by the vapour of a boiling liquid is a very slow 
