BRIDGMAN. — THERMODYNAMIC PROPERTIES OF LIQUIDS. 37 



nection with other quantities, it would not have been justifiable to 

 retain all the irregularities which some of the curves show. The 

 reason for retaining the irregularities is that the attempt has been 

 made to present a set of data which should be thermodynamically 

 consistent. Let us suppose, for example, that the compressibility 

 and the dilatation were both determined from the original tables of 

 volumes and that they have been plotted against pressure. Both 

 of these curves show irregularities which may be smoothed off by 

 drawing smoother curves through the points, thus giving values of the 

 compressibility and the dilatation which doubtless in themselves 

 represent with greater probability the actual compressibility and 

 dilatation. But each of these modified values for the compressibility 

 and the dilatation will have a reflex effect on the table of volumes, 

 which has now become inconsistent with the better values of the 

 compressibility and the dilatation, and must therefore be altered 

 slightly so as to be in accord with the new values. The alteration 

 in the table necessary to accomplish this may be produced by changes 

 less than the possible experimental error. But the point is this. 

 Either the revised value of the compressibility or of the dilatation is 

 sufficient of itself to completely revise the table of volume. If we 

 are to adjust the compressibility or the dilatation we must do it so 

 that both have the same reflex action on the table. Furthermore, 

 all seven thermodynamic quantities must be adjusted in the same 

 way. It is evident that this is a task of no small difficulty. To 

 perform it, the only method seems to be a tedious one of trial and 

 error. The labor of such an adjustment would be far beyond the 

 labor of making the measurements with greater accuracy, and the 

 labor had much better be so used in performing new experiments. 

 It must furthermore be remembered that the values in the tables 

 have been smoothed once with respect to both temperature and 

 pressure. Any further changes would amount simply to slight 

 changes in this smoothing; changes which were not justifiable by an 

 examination of the data themselves but are rendered probable only 

 by an examination of certain derived quantities. The choice has 

 been made in this paper, therefore, to present results which may be 

 slightly in error when taken by themselves, but which are neverthe- 

 less consistent thermodynamically. 



IV. Numerical Details of Experiment and Computation. 



In the detailed discussion and presentation of the results for the 

 twelve liquids which is to follow, there will be given the experimental 



