264 PROCEEDINGS OF THE AMERICAN ACADEIVIY. 



millimeters, is presumably the most reliable. The result of this run 

 is therefore 21.5 millimeters at 1074° C. 



The results of the above experiments are collected in the table 

 below, and are plotted in Figure 5. The ordinates are pressures in 

 centimeters of mercury and the abscissae are the temperatures in 

 centigrade degrees. The pressures are not reduced to zero degrees 

 as this correction is much less than the experimental error. 



It is remarkable that the relation between the temperature and pres^ 

 sure is linear through such a great range in temperature. It would 

 seem that at lower temperatures the line representing this relation must 

 become concave upward, and it is hoped that in the near future it will 

 be possible to carry out some experiments on this equilibrium below 

 1000 degrees, to test this point. 



In order to compute the heat Q, evolved by the reaction CaCo + 

 N2 :^ CaCN2 + C, taken from left to right, it is not allowable to use 

 the integrated van't Hofif equation 



4.57log|j = «(i--J.) 



because this equation is derived on the assumption that Q is constant, 

 which is not the case in this reaction. If Q is computed from the 

 differential eijuation 



