192 PROCEEDINGS OF THE AMERICAN ACADEMY. 



vary from about 1.8 ccm. to 1 ccm. If, now, a considerable amount of 

 the water were present iu the vapor space, the specific- volume values 

 obtained would, of course, be larger the smaller that space. As a matter 

 of fact, however, the values obtained with the 1 ccm. vapor space were 

 as often smaller as they were larger than those obtained with the 1.8 ccm. 

 vapor space. In other words, no difference greater than the variable 

 experimental error was observed. The error due to this source is there- 

 fore probably less than 0.1 per cent. 



7. The temperature measurements may be regarded as accurate to 

 within 0.2° ; and this of itself introduces an uncertainty of only 0.07 per 

 cent in the worst case, that of the 306° values. That the bomb and its 

 contents actually attained the temperature of the surrounding vapor is 

 shown by the fiict that the extremely sensitive reading of the upper elec- 

 trode remained constant indefinitely after it had once become so ; and by 

 the fact that there could not be a continuous loss of heat of appreciable 

 magnitude from the bomb to the surroundings, since upon the sides the 

 bomb was protected against radiation and cold convection-currents by 

 the iron shield with the vapor outside, and since above there was always 

 a layer of vapor 10 cm. in height, and since the dropping back of con- 

 densed liquid on to the bomb was prevented by the mica shield ; more- 

 over, the copper lead-wires were only 1.2 mm. in diameter and passed 

 through the upper layer of vapor before emerging. 



8. Another possible source of error might be the gathering of vapor 

 bubbles on the under surface of the cover, whereby the apparent volume 

 of the liquid would be increased. That this did not occur was shown by 

 removing the bomb from the heater, shaking vigorously, immediately re- 

 placing it, and taking conductivity readings as soon as the temperature 

 had again become constant, whereby the same readings were obtained as 

 before the shaking. 



9. The air was not entirely removed from the bomb at the start, and, 

 as the solution expands and the temperature rises, the air pressure in- 

 creases. Assuming that the preliminary boiling had removed all of the 

 air from the solution in the beginning, and that there is no solubility of 

 the air in the liquid at the high temperatures, its pressure can be calcu- 

 lated by the gas laws. At the temperatures of 218, 281, and 306°, it 

 would thus amount to about 1, 2, and 2.5 atmospheres respectively. 

 The effect of these air pressures on the specific-volume values cannot be 

 calculated, since the compressibility at these temperatures is not known ; 

 but for these small pressures it is undoubtedly less than the errors of 

 observation. 



