290 Dr. II. Herwig a Investigations on the Conformity 



is shown by a mutual comparison of the maximum tensions at dif- 

 ferent temperatures. To the sum of the errors in tension is still 

 to be added the influence of the second of the above-named cir- 

 cumstances, viz. the variation in the position of the cathetometer 

 when placed successively on the six quicksilver-levels, which can- 

 not have been of precisely similar form in all respects. But 

 in general we found under the maximum tension a deviation of 

 only 0*5 of a niillim. from the mean ; the greatest deviation that 

 occurred is *G of a niillim. in the case of alcohol at G2° - 9, where 

 the mean of eight measurements of maximum tension amounted 

 to 396 - 83 millims., while the measurement in which the aberra- 

 tion was greatest was 397*43. 



From a variation in adjusting the cathetometer on the mercury- 

 level in the tube containing the vapour, and from placing the 

 tube in a position not exactly vertical, a further error in taking 

 the volume might be committed, to the amount, perhaps, of 0*3 

 of a cubic centimetre. In order to check the errors arising from 

 this source, we had to see how much one of the products p v, 

 which for any one temperature already obeyed Mariotte's law and 

 were constant, deviated from the mean of all these p v's, and, 

 moreover, how widely this mean deviated from the mean values 

 holding for other temperatures, differently from what is required 

 by Gay-Lussac's law. We had also to take the mean of the 

 vapour-densities for the different temperatures which are derived 

 from the constant^?; of each temperature, and calculate accord- 

 ingly the true mean values of the constants p v for each tempe- 

 rature, and then seek for the greatest deviation therefrom. Besides 

 these errors in volume, the errors in tension already spoken of 

 would also naturally come into consideration. But we invariably 

 found much smaller deviations than the extreme deviation, which 

 arises in the case of alcohol at 69°*9, where, with a volume of 93 

 cubic centims. and a tension of 127*54 millims., the product 

 11861 was calculated instead of the true mean value 11797. If 

 we here assume an error of *3 cubic centim. in volume, the ad- 

 ditional error in tension will only amount to 0*3 millim., which 

 is far within the specified limits. Upon the whole it follows, 

 then, that the greatest errors in tension are to be taken at most 

 at 0*G millim., and of volume at "3 cubic centim., and that these 

 limits were reached in very exceptional cases only. 



§ 4. Examination of the Vapour of Alcohol. 

 The first numbers found, according to the method sketched out, 

 were those given in the following Table for alcohol. They con- 

 tain the values of the volume v (in cubic centims.) and of the 

 tension p (in millimetres of mercury) corresponding to the eight 

 temperatures examined. There are also given the products p v. 

 The cessation of saturation, as well as the occurrence of Ma- 

 law, is indicated on each occasion by the horizontal lines. 



