726 Prof. R. W. Wood on Anomalous Dispersion of 



done by Bates is well shown by the tables of readings which 

 accompany the paper, and there seems to be no means of 

 avoiding the conclusion that these substances are without 

 rotatory power. No reply to these criticisms has appeared 

 to my knowledge, the two pieces of work standing without 

 comment, each on its own merits. Having already come to 

 conclusions similar to those of Bates, I was the more ready 

 to accept his experimental refutation of any anomalous 

 dispersion effects in the magnetic rotation of these substances. 

 As the method which he employed was much more sensitive 

 than the one which I had made use of, publication of my 

 own results was of course superfluous. 



I have recently made a series of experiments with a 

 saturated solution of praseodymium chloride, and have found 

 what appears to me to be indisputable evidence of the effect. 

 Curves were obtained for the substance by two different 

 methods, and both showed anomalous effects at the edges of 

 the absorption-band. The solution was contained in a cell 

 1 cm. in thickness, furnished with windows of the thinnest 

 cover- glass. The light from the monochromatic illuminator 

 (spectroscope with slit in place of eyepiece) was sent through 

 the axis of the magnet and the rotations measured by setting 

 the analysing nicol for extinction. As the monochromatic 

 light furnished by this apparatus is extremely brilliant, the 

 setting could be made to within a tenth of a degree without 

 difficulty. A Savart plate was also employed, which increased 

 the accuracy of the observations, and the curves accompanying 

 this paper were plotted from readings obtained with this 

 form of polarimeter. 



Strong elliptical polarization was observed when working 

 in the vicinity of the absorption-band, due to the unequal 

 absorbing power of the magnetized medium for the two 

 circular components. This made accurate readings difficult, 

 for it was impossible to completely extinguish the light when 

 employing merely the two nicols, or to cause the complete 

 disappearance of the Savart fringes. It was found that the 

 readings were prejudiced by a knowledge of what was 

 expected. The first curves were made by progressively 

 changing the wave-length of the light emitted by the mono- 

 chromatic illuminator. The rotation, which amounted to 

 about three degrees for the extreme red, increased nearly a 

 whole degree as the absorption-band was approached, dropping 

 back half a degree or so on crossing the band, and then rising 

 again to a value of 4*5 in the blue-oreen region. A curve 

 obtained in this manner is shown in fig. 1 (Curve 1). On 

 taking a number of readings at random close to the edges of 



