The Anovialovs Dispersion of Sodium Vapour, 165 



length of which could be changed by turning the prism, which was 

 furnished with a mirror in the manner described by Wadsworth. 



On heating the tube the black vapour was seen to spread out and 

 fill the prismatic clear space between the iron tubes, while the lateral 

 deviation of the image of the slit, as observed in the telescope, indicated 

 that, to a certain extent at least, the desired prismatic form had been 

 secured. In the extreme red the deviation was very slight, but as the 

 spectrum was advanced across the slit by slowly turning the prism the 

 image in the telescope moved off to one side, the deviation in this 

 direction reaching its maximum value just before the wave-length 

 reached that of the sodium lines. At this point the image jumped 

 abruptly to the other side just as we should expect it to do on crossing 

 the D lines in the spectrum, and from now on the image slowly crawled 

 back to its undeviated position. The focal length of the telescope 

 was 460 mm., and the maximum deviation of the rays adjacent to 

 the D lines on the red side, as measured by an eye-piece filar micro- 

 meter, was but 1 mm., while the deviation in the opposite direction of 

 the rays on the other side of the D lines was 1-2 mm. The angle of 

 the prism was 130, from which data we get the following values for 

 the refractive index of the vapour for these two wave-lengths (relative 

 to hydrogen) /x = 1-0005 and /* = 0-9994. 



Similar results were obtained with the device shown in fig. 6, when 

 two elliptical pieces of perforated sheet iron were used for moulding 

 the vapour. The images formed in this case were blurred by 

 diffraction. 



FIG. 6. 



I place very little value however on these figures, for I am of the 

 opinion that the effective angle of the sodium prism is much less than 

 the angle between the ends of the tubes, it seeming probable that the 

 vapour bulges out into the tubes, especially near the centre. That 

 this is to a certain extent the case is indicated by the fact that the 

 image of the slit is not very sharp, though this may well be caused by 

 the varying density of the vapour. I have not yet despaired of getting 

 a prism bounded by plates of glass, about which there can be raised 

 no question, though the problem is a difficult one, and observations 

 will have to be made with great rapidity. 



The deviations obtained by this method are very much less than 

 those obtained with the dispersion tubes, indicating either that the 

 equivalent angle is very large in the latter case (it may be nearly 180 

 for all we know) or that the angle of the prism formed by the iron 

 tubes was less than the estimated value. 



VOL. LXIX. N 



