1902.] on the Constitution of Crystals. 143 



of all the results derived from the work. We will illustrate it by the 

 case of caesium magnesium selenate. The curves which you now see 

 on the screen are those for the magnesium triplet of double selenates. 

 On arrival at the caesium salt the convergence has actually brought 

 the curve for the minimum refraction into contact with that for the 

 intermediate refraction, so that we have here a crystal whose total 

 double refraction is exceptionally small, and in which the velocities 

 along two of the three ellipsoidal axes are approximately equal, which 

 is characteristic of tetragonal and hexagonal crystals, but a perfect 

 anomaly in a monoclinic crystal. Let us see, however, if the iden- 

 tity is a fact for all wave-lengths, as would be the case in a truly 

 uniaxial crystal of tetragonal or hexagonal symmetry. We will re- 

 produce for you the appearance in the spectrometer afforded by a 

 prism so cut as to give us the two spectra corresponding to these 

 two intersecting curves, that is, whose vibration directions are those 

 of the two apparently equal axes of the velocity ellipsoid. 



The images in three colours which you see are those corresponding 

 to the wave-lengths of red lithium light, green thallium light and 

 violet hydrogen light. You observe that they completely overlap, 

 and if we only used the same magnification as we used for the other 

 images we saw just now, you would say that they are absolutely 

 identical images. We are attempting to faithfully reproduce the 

 whole phenomena by the use of separate slides, shown by two equal 

 lanterns, and when one of them is shut off, the effect is exactly as 

 when a Nicol is introduced at 0° in the spectrometer, which extin- 

 guishes one of the spectra. When the other is shut off instead, what 

 you observe is the same as if we rotated the Nicol to 90°, which would 

 extinguish the second spectrum. Commencing with both lanterns 

 on, as when no Nicol is being used, or if it is, it is arranged at 45°, 

 and examining the images closely, we notice that the red and violet 

 images are distinctly double, while the central green image is a truly 

 single one. Now we cut off one lantern, and you please imagine that 

 we are introducing a Nicol at 0° ; the effect has been that the outer- 

 most image in the case of both red and violet has disappeared. Now 

 shutting off the other lantern instead, and you please imagine we are 

 rotating the Nicol to 90°, the innermost images disappear. All this 

 time the green image remains fixed and single. This evidently 

 means that for the middle part of the spectrum there is absolute 

 identity of refraction and therefore of velocity, while for the red end 

 there is a minute difference of refraction in one sense, and for the 

 violet end a similar small difference in the opposite sense. The fact 

 is, the two images, corresponding to the two ellipsoidal axes, for red 

 are slightly separated ; they approach and coalesce for green ; then 

 they pass each other and re-separate on the opposite side of each 

 other as violet is approached. 



Thus our crystal is only truly uniaxial for one wave-length of 

 light, and this apparent anomalous refraction is merely the effect of 

 the operation of the rule of progression. 



