442 PROCEEDINGS OF THE AMERICAN ACADEMY. 



a rectangular brass box with glass windows. In the experiments on 

 the expansion of ether between ()°C. and 16° C a freezing mixture 

 containing ice, water, and salt, at —2° C, was used instead of ice and 

 water. The liquid in the U-tube was therefore kept near 0° C, while 

 the remainder of the liquid was at a higher temperature. One has 

 here, then, the boundary condition for superheating all the liquid 

 except that contained in the U-shaped part of the capillary tube which 

 dipped into the beaker of ice and water, for it was in contact only with 

 glass or with some of the same sort of liquid at a lower temperature. 



The piezometer used in determining the compressibilities in the 

 superheated state was of the same form as the dilatometer. For 

 the vertical tube E F was substituted a smaller tube of cross section 

 0.0184 (cm.)^. The bulb of the piezometer, which had a capacity of 

 61.11 (cm.)^, was inserted in a cylindrical brass tube, one end of which 

 was closed with a rubber stopper, through the centre of which passed 

 the capillary tube B C. The other end of this brass tube was closed by 

 means of a disk, in the centre of which was a small tube, through which 

 the interior of the brass tube could be made to communicate with the 

 pressure gauge, so that the pressure on the inside and that on the 

 outside of the bulb of the piezometer were the same. The pressure 

 on the inside of the capillary tube was not equal to that on its outside. 

 On account of this inequality of pressure the internal volume of the 

 capillary tube varied somewhat, and the magnitude of the error intro- 

 duced into the data on compressibility from this source had to be 

 considered. By means of the formula given on page 117 of Poynting 

 and Thomson's " Properties of Matter," the change in internal volume 

 for a typical case was calculated, and the error introduced from this 

 source was found not to exceed one fifth of one per cent, and may be 

 neglected. 



Since in the work on expansion that part of the liquid which was 

 in the U-shaped tube remained at 0° C, it introduced no error into 

 the results. In the work on compressibility, however, the observed 

 change of volume, due to a certain change of pressure on the liquid, 

 was the sum of the change of volume of the liquid in the bulb, 

 which was superheated, and the change of volume of the liiiuid in the 

 U-shaped tube, which was not superheated, less, of course, the change 

 in the volume of the glass bulb. Knowing the compressibility and the 

 volume of the liquid in the U-shaped tube, one easily made the nec- 

 essary correction for the error arising from this source, which was 

 about one per cent. There was another correction which had to be 

 applied to the observed change of volume both in compressibility and 

 in expansion. In either of these cases the observed change of volume 



