POLYMORPHIC TRANSFORMATIONS OF SOLIDS. 109 



temperature had been chosen a Kttle higher than 4600, so as to exactly 

 equal the maximum, no trace of a second transition would have been 

 found. 



The existence of the maximum involves properties that at first 

 sight seem so unnatural, that perhaps this elaborate procedure to 

 establish it has been worth while. A maximum would seem unlikely 

 because at this point the change of volume is zero; at lower pressure 

 the high temperature form with the larger volume is the more com- 

 pressible, which is of course natural, but at pressures higher than the 

 maximum, the high temperature form has the smaller volume and a 

 higher compressibility than the low temperature phase. Now this 

 peculiar relation of the compressibilities was shown by experiment 

 to actually exist. On the falling branch of the curve, the compressi- 

 bility of the phase with the smaller volume was the higher. This 

 could be determined from the difference of the slope of the volume 

 isothermals above and lielow the transition. Some attempt will be 

 made later to find the magnitude of this difl'erence; it cannot be 

 determined accurately, but the fact is absolutely beyond question. 

 This curious relation of the compressibilities is not an absolutely new 

 thing; it is shown by ice and water, but there the explanation is proba- 

 bly to be found in the greater rigidity of the crystalline framework as 

 contrasted with the structureless liquid. 



Regarding the accuracy of the results, the values given here are 

 probably more irregular than those for any other substance which does 

 not decompose. The explanation, of course, is the great width of the 

 band of indifference, and the sluggishness of the reaction. The irregu- 

 larities in the AV values are considerably greater than in the values 

 for pressure and temperature. The values of AV probably have their 

 greatest error on the rising branch of the curve. The AV curve as 

 shown was so drawn as to nearly pass through the directly determined 

 value at low pressure, and also to pass through zero at the pressure 

 of the maximum temperature. 



In spite of the abnormally wide band of indifference, the pressure 

 approached a stationary value after the reaction had once started as 

 rapidly for this as for many other substances; the usual time for 

 reaching a stationary pressure being only about 15 minutes. This 

 suggests most strongly that there is no direct connection between the 

 width of the band of indifference and the velocity of reaction; there 

 seems every reason to think that even in infinite time the apparent 

 width of the band of indift'erence would not become materially less. 



There are no other values at high pressures for comparison. The 



