36 PROCEEDINGS OF THE AMERICAN ACADEMY. 



graphically from the curves of dilatation against temperature at con- 

 stant pressure, and were integrated mechanically. The results so 

 found agree fairly well with the values found by the other more accu- 

 rate method. The magnitude of the discrepancies might be as much 

 as 10%, but all the essential characteristics of the curve as gi\'en by 

 one method were reproduced by the other also, such as the maxima 

 and the minima, and the points of inflection. Of course the pressure 

 of maximum or minimum was sometimes displaced, as was to be 

 expected. 



Specific Heat at Constant Volume. — From the specific heat at 

 constant pressure it is now possible to compute the specific heat at 

 constant volume by the well known formula for the difference of the 



two specific heats, namely Cp — C, = — ... . ^ . This formula 



\dp)r 



involves only quantities which have already been determined, so that 

 Cj may be found immediately. The values of [-rj and ( v~) 



used in this computation were taken from the tables, not from the 

 diagrams. Just as for the specific heat at constant pressure, the values 

 found in this way are the differences between C\ at atmospheric 

 pressure and the pressure in question. The differences are such that 

 a positive value means that the specific heat is greater at atmospheric 

 pressure than at the pressure in question. A decreasing curve indi- 

 cates, therefore, that the specific heat is increasing with increasing 

 pressure. 



In the paper on water other quantities of thermodynamic interest 

 were plotted. These are the temperature effect of compression and 

 the adiabatic compressibility. They may be easily calculated from 

 the data given in this paper. While they are of interest in themselves, 

 they do not seem to be of such fundamental importance as the quan- 

 tities already listed in suggesting the possible internal mechanism 

 of the liquid. It was felt, therefore, that to give them would unduly 

 increase the volume of this paper, and they have accordingly not 

 been computed. 



A word seems called for as to the general character of the curves. 

 In many cases there are slight irregularities which may very well not 

 correspond to the actual facts, the irregularities being beyond the 

 limit of experimental accuracy of the work. It is true that if each 

 of these ciuantities were being given for itself alone, without con- 



