POLYMORPHIC TRANSFORMATIONS OF SOLIDS. 59 



the upper cylinder and the connecting tube. If any one should ever 

 have occasion to do work similar to this again, he should use every pre- 

 caution when working with mercury salts. These decompose slightly 

 in contact with the steel, and deposit free mercury, which gradually 

 works its way through the cylinder, even when it is 5 inches in diam- 

 eter, as was that used here. The flaws which resulted from this 

 action of the mercury were not traced to their true cause until two 

 cylinders had been ruptured. The coils with which pressure was 

 measured were also replaced a number of times. These were fre- 

 quently destroyed when the substance under investigation decomposed 

 at the higher temperatures and the decomposition products found 

 their way to all parts of the apparatus. Fortunately a number of well 

 seasoned coils were in hand, so that new ones could be substituted 

 and measurements immediately begun without waiting for seasoning. 

 The new coils were always calibrated in the way already descril)ed 

 by determining the melting point of mercury at 0°. 



The materials were obtained in as pure a state as possible from 

 various chemical houses, and used in most cases without further puri- 

 fication except careful drying. However, those organic substances 

 which had a convenient melting point were further purified by cr^'stal- 

 lization in a thermostat at constant temperature. Slight impurity 

 is of very much less importance in the case of solid transitions than it 

 is for melting, because here the transition coordinates are not affected 

 except in those rare cases where mixed crystals are formed. At least 

 one such case was noticed, however. 



The Effect of Impurities on the Transition Lines. 



In an earlier paper ^ a deduction was given of the temperature 

 depression of a transition point which was applicable at any pressure, 

 since it did not involve the assumption of a vapor phase, as such 

 deductions usually do. To find how a whole transition line is affected 

 by impurity, one has only to apply the formula given there to every 

 point in order to obtain the equation of the modified line in terms of a 

 displacement at every point parallel to the temperature axis. But 

 this process evidently fails if the transition line itself is parallel to the 

 temperature axis. Such transition lines do not occur for melting, 

 but they are comparatively common for solid transitions, and it 

 becomes important to find another expression for the shift of the 

 equilibrium line. 



