Errors in Methods of Measuring Rotary Polarization J 



Such close agreement is of course partly accidental, since con- 

 secutive measurements can not be made by Wiedemann's method 

 closer than o.io°. The magnitude of the error will depend upon 

 what width of space the observer uses. 



As the shift of the space was toward the red, thus giving an 

 apparent positive rotation, it is evident that as the space moves 

 farther into the absorption band of the fuchsin this positive rota- 

 tion would increase until it attained a maximum in the edge of 

 the band. From there on it would gradually decrease and be- 

 come zero at some point in the interior where the absorption is 

 the same for both edges of the space. As the space approaches 

 the other edge of the band from the interior a gradually increas- 

 ing rotation would be observed, attaining a maximvim in the edge 

 of the band, and then decreasing to zero as the space leaves the 

 band. This maximum is, however, in the opposite direction from 

 that previously mentioned. Since the former is a positive, this 

 is a negative rotation ; that is, for solutions it would drop below 

 the rotary dispersion curve of the solvent. 



When the half-shade is used the apparent rotation due to the 

 error increases very perceptibly with, but is not directly propor- 

 tional to, an increase in the strength of the magnetic field, be- 

 cause a doubling of the field gives a corresponding increase in 

 the angle ( ai — a«), all other conditions remaining the same. 

 Jn the black and clear space methods, however, the maximum 

 value of the error varies only very slightly, for small rotations, 

 with the intensity of the field, because the rotation of the sub- 

 stance studied constitutes only a small fraction of the total rota- 

 tion. The thickness of quartz necessary to give the desired space 

 in the spectrum is at least 1.5 mm. This gives an initial rotation 

 of 32.5° at the D line, to which is added the rotation, for example, 

 of I cm. length of water for a field of 8000 lines, making a total 

 34.2°. If the field be doubled the total becomes only 36°, which 

 is sufficient change to affect the maximum error only slightly. 

 In the three methods mentioned the maximum value of the 

 spurious rotation will be nearly directly proportional to the con- 

 centration for dilute solutions, and the sharper the absorbing 

 band the more marked the anomaly. 



261 



