190 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1918. 



higher pressures there were, quite frequently, discrepancies which 

 could not be explained by errors in the apparatus. The tempera- 

 ture of these measurements was that of the room, about 20° C. 



Apparently the only possible explanation of the irregularities 

 shown was that the water had been frozen by the high pressure, so 

 that measurements of the volume at high pressures were sometimes 

 being made on the liquid and sometimes on the solid. This explana- 

 tion, if it were the true one, indicated a very remarkable state of 

 affairs, as the application of ordinary pressures to ice causes it to 

 melt. One would expect to be able to melt ice by high pressure, 

 therefore, and not to freeze water. 



However, the discoveries of Tammann materially assist in provid- 

 ing an explanation of the irregularities found in the compressibility 

 measurements referred to above. 



Tammann had discovered that at high pressures there are two modi- 

 fications of ice, each of which is denser than water. It would be 

 expected that the freezing point of the modified form would be 

 raised by the application of pressure, so that possibly the irregulari- 

 ties could be explained by the freezing of water to this new form of 

 ice at 20° C. under the very high pressures reached in this work, 

 which were about five times those reached by Tammann. But the 

 fault in this explanation is that Tammann had predicted from meas- 

 urements on this new kind of ice that no pressure, however great, 

 could possibly raise the freezing point of water higher than about 

 — -17°, and a temperature of 4-20°C. was being employed. Care- 

 ful investigation of the whole matter was therefore called for, and 

 special apparatus had to be designed to attack the new problem. 



To state that it is possible in the experiments to ascertain whether 

 the water has frozen to ice or not may appear strange, when it is 

 considered that the ice is inclosed in a cylinder and can never be 

 seen, because as soon as the pressure is removed and the cylinder 

 opened the ice immediately liquefies. As a matter of fact, this can 

 not be ascertained, except indirectly. When the water freezes to 

 ice, there is a decrease in volume, and this is shown by a drop in 

 pressure. Conversely, too, when ice melts to water the volume in- 

 creases, which is indicated by an increase of pressure. 



In the actual measurements the temperature of the water was kept 

 constant. In order to increase the pressure, the piston was pushed 

 into the cylinder, the distance being measured, and the displacement 

 of the piston plotted against the increase of pressure produced. The 

 pressure at first increased regularly as the displacement, but when 

 the pressure reached a value high enough to freeze the water at the 

 particular temperature of the experiment the volume suddenly de- 

 creased without the pressure rising at all. Then, after freezing was 

 completed, so that there was only solid ice in the apparatus, the pres- 



