CLARK : MOLECULAR WEIGHT DETERMINATIONS. 251 



In both tables, two sets of values for the molecular weight 

 are given : one obtained from data in which the boiling-point 

 elevations are corrected for change of barometer, after the 

 barometer readings have been corrected for change of room 

 temperature ; and another in which these two corrections are 

 made, and also another one, depending upon the variation of 

 volume of the liquid, as will be explained farther on. This 

 second set of values is the one referred to in each case in the 

 preceding parts of this paper. None of these corrections are 

 excessive. The first mentioned are so small as to be negligible 

 in the molecular-weight determinations, and the second are not 

 large. According to Landolt and Boernstein, a change of 1 mm. 

 in the height of the barometer causes an appreciable change in 

 .the boiling-point of pure water (0.037°) , and a change of 1° in 

 the temperature of the barometer produces a change of 0.12 

 mm. in its readings, between the limits of temperature and 

 pressure observed in this work. The maximum variation of 

 barometer and of room temperature during each set of readings 

 is given at the head of the corresponding table. The molecular- 

 weight determinations were made without using the platinum 

 cylinder. 



I have confirmed the observations of Franklin and Kraus, 

 that the boiling-point of the liquid in the tube increased with 

 the amount of liquid present. To ascertain the volume correc- 

 tion, three determinations of the variation of the boiling-point 

 of pure water with the amount of water (see tables C, D, and 

 E ) were taken without the platinum cylinder, and two with it 

 (see tables F and G), besides several preliminary determina- 

 tions. The separate determinations for the same condition are 

 consistent with one another, as will be seen by comparing tables 

 C, D and E with one another, and F with G. 



This rise in the boiling-point of a solvent with increasing 

 depth is obviously the effect of the weight of the upper layers 

 of liquid in increasing the temperature of equilibrium between 

 the liquid and vapor phases in the liquid surrounding the ther- 

 mometer bulb. The rise in pressure due to an increase in the 

 height of water in the tube of one centimeter would raise the 

 boiling-point of the liquid surrounding the thermometer bulb 

 by 0.027°, on the assumption that no mixing of the lower, hotter 

 layers with the upper, cooler ones takes place. In practice, of 



