EFFECTS OF PRESSURE ON NITRATES. 599 



of expansion may be computed in the regular way. Ill turns out to 

 be more expansible; the average difference over a range of 10000 

 kgm. is 0.00005 cm.^ per gm. per °C. 



Most careful search was made for other modifications to 12500 

 kgm. at 20°, 100°, 150°, and 200°, but no transitions except those 

 found above were detected. 



Potassium Nitrate. — This substance was Kahlbaum's purest, 

 used without further purification. It was fused into an open steel 

 shell, thus effectually driving away the moisture, and then the sides 

 of the shell were pierced with fine holes to ensure equality of pressure 

 in all parts. Pressure was transmitted directly by kerosene. Two 

 samples were used. With the first, 39 gm. in amount, all the high 

 pressure points were determined, 21 in number. The second sample, 

 weighing 32 gm., gave several points at low pressures. 



Considerable trouble was found in getting the Av points, and several 

 repetitions were necessary. The discrepancies were always in the 

 direction of too low values. The reason is probably to be found in a 

 combination of the considerable lag which the transition shows, with 

 the effect of fissures, which formed in great numbers on solidification 

 from the melt. The effect of lag is that it is necessary to push the 

 pressure considerably beyond the transition line to ensure the reaction 

 starting, and the effect of fissures may be to prevent the reaction from 

 running completely throughout the entire mass after it has once 

 started in some parts. To be sure that the transition is complete, 

 the pressure must be pushed far enough beyond the transition line 

 to ensure kernels of the new modification appearing in all the parts. 

 The point at 0°, for example, had to be repeated for this reason. To 

 ensure completion of the reaction, when the point was repeated, 

 pressure was maintained 1000 kgm. beyond the equilibrium line at 

 room temperature for three hours, and only then was the temperature 

 lowered to 0°. The first value of Av, before this procedure was adopted 

 was 25% too low. The sluggishness of the reaction sometimes neces- 

 sitated special procedure. Thus the lower point on the II-III curve 

 was determined by maintaining temperature for some time at 90° 

 to ensure completion of the reaction from II to III, and then lowering 

 temperature at such a pressure as to land in the corner between the 

 II-III and the III-IV lines. The lag varies considerably with the 

 temperature on the several curves. Thus on the upper end of the 

 I I-III curve, III could be considerably subcooled with respect to II, 

 but II could be very little superheated with respect to III. 



The sluggishness of the reaction made it impossible to obtain good 



