192 



ANNUAL REPORT SMITHSONIAN INSTITUTION, 1918. 



perimenters. The only important difference is that the new packing 

 which I devised makes it possible to obtain a piston which has ab- 

 solutely no leak, even at the highest pressures, and so renders possible 

 accurate measurements of the change of volume. This has. I believe, 

 not been possible before. In all previous experiments there has been 

 some leakage around the piston, which made it impossible to obtain 

 accurate measurements of the change of volume. 



To return now to the compressibility measurements and the dis- 

 crepancies found at high pressure, the application of the present 

 method of experiment to the study of water showed that there did 

 exist a new variety of ice at the high pressures, as had been sus- 

 pected. It was found that the new variety of ice was not one of those 



80 



60 



-40 



-60 



■so 



10 



14 



16 



20 



,-3 



Pressure, kgm. per sq. cm. x 10" 

 The equilibrium diagram between the liquid and the five solid modifications of water. 



Fig. 2. 



two kinds previously discovered by Tammann, but was, instead, con- 

 siderably denser than either of the varieties found b} 7 him. In ad- 

 dition to this new kind, which is stable at high temperatures and pres- 

 sures, I discovered still another kind, not previously known, inter- 

 mediate between the new high-pressure ice and the two varieties 

 found by Tammann, making four varieties of ice denser than water. 

 There are, therefore, in all at least five different kinds of ice, only one 

 of which we are ordinarily familiar with. 



Figure 2 shows more clearly the relation between these different 

 kinds of ice. It will be noted that in this figure there are five 

 regions, numbered according to the kind of ice to be found within 

 the region. Thus, for example, if in an experiment the pressure be 

 raised to 10,000 or 10 4 kgm. and the temperature maintained at 0°, 



