( 298 ) 



ginning i), and a mixture with 26.0 percent, on which only a few 

 observations have been made not published hitherto. For these 

 mixtures form (II) required three terms, as was the case with 

 oxygen. 



For judging of the accordance of the formulae we have always 

 calculated the probable error of an observation of the weight one, 

 according (o the rules of the method of least squares. In the 

 two first columns of the following table these probable errors are 

 to be found. They are computed by taking three terms in the form 

 II for oxygen and mixtures containing oxygen as said above, and two 

 terms for the other gases, and belong to the rotatory constants 

 expressed in minutes, multiplied by lO*^ ami for the pressure and 

 the tempeiatiire mentioned in the observations. 



P 11 O B A B L E ERRORS. 



An investigation of the errors, which are to be anticipated in the 

 different adjustments and readings, always yielded for the e"ror to 

 be expected in the rotatory constant numbers larger than those of 

 the table given above, e.g. for oxygen 11 in the violet and 6 in 

 the red. With a view to this fact the accordance for all formulae 

 mentioned may be considered satisfactory. Yet in the gases without 

 oxygen the errors of the form (II) wiih two terms are always 

 smaller than those of (I), so that if in choosing between the 

 two forms we had to consider this only, we should prefer for these 

 gases the form (II). But it is not excluded that other forms of for- 

 mulae for interpolation should yield a still smaller error. This 

 appeared to be the case with the form 



>) Versl. Kou. .\kad. 1894/95, p. 236. Comm. Pliys. Lab. nt Leiden X". 15, p. 26. 

 2) Observations with gas from two cylinders of unequal purity. 



