1 24 Scientific Proceedings, Royal Dublin Society. 



estimated, using a pocket-lens, care being taken to avoid parallax. The 

 liquid, placed as usual in the inner tube, was slightly supercooled, and then 

 caused to freeze by addition of a minute speck of ice. The highest point 

 reached by the mercury column on rising after supercooling was carefully 

 observed. As the separation of pure solvent in tlie solid state increases the 

 concentration of the remaining liquid, the freezing-point as thus found is 

 theoretically too low, viz. further removed from that of pure water than it 

 ought to be. If the supercooling has not exceeded half a degree, the error 

 will be very slight. Bearing these facts in mind, the routine adopted was to 

 place the tube with the thermometer, after noting the reading, in a beaker 

 of water till the temperature rose to about 2° 0. It was theu free from 

 flakes of ice ; but minute fragments still remained. On replacing the tube in 

 the cooling-space the temperature fell, but remained stationary at the true 

 freezing-point ; for solidification begins immediately the freezing-point is 

 reached, if even a minute particle of the solid phase is present. After a 

 while the mercury column will begin to fall ; this is due to an appreciable 

 alteration in concentration occasioned by the separation of ice. If there was 

 a slight difference between the readings obtained by these two methods, the 

 last was always a smaller depression than the first. In such cases the last 

 determination, and not the mean, was taken, as it is less likely to have any 

 error due to supercooling. But if the supercooling was slight, there was no 

 measurable difference between the results got by the two methods. 



The zero of the thermometer was redetermined from time to time. It 

 was found to fall from 3'64-3"33 in about two years, the change being 

 quicker in summer than in winter. This change in the zero, referred to by 

 most experimenters, is due, I believe, not to changes in the glass, which, are 

 relatively unimportant even with a large, thin bulb, but to the distillation of 

 the mercury from the curved surface of the hanging drop in the top of the 

 reservoir of the Beckmann thermometer, to the flat surface at the lower end. 

 The drop is caused by the expansion of the mercury, on the scale at 0° C, 

 when warmed up to room temperature. Now, although the pressure of 

 mercury vapour at about 15° C. is very slight, yet that of the curved surface 

 always exceeds that of the flat surface. Thus, as both are at the same 

 temperature, slow distillation must go on from the curved to the flat surface. 

 The result is that when cooling to 0° C. causes the drop to withdraw into the 

 narrow capillary stem, it does not stand as high as it did formerly, owing to 

 loss of mercury. 



The drop is more curved in summer, as more mercury leaves the bulb ; 

 hence there is a greater difference in vapour-pressure, and consequently the 

 distillation is more rapid. 



