Regular Reflexion of Light by an Absorbing Gas. 341 



that an iron line one Angstrom unit on the short wave- 

 length side of the 2536 mercury line, was much more power- 

 fully reflected than a pair of iron lines on the long wave- 

 length side situated at 0*1 and 0'4 A.U. from the mercury 

 line. No explanation of this was given in the paper, but 

 the suggestion was made later in Wood's ' Physical Optics ' 

 (second edition), page 432, that it undoubtedly resulted from 

 the sudden change in the refractive index of the vapour in 

 the vicinity of the absorption-line. 



"The 2536 line shows powerful selective dispersion and the 

 refractive index, in its immediate vicinity on the short wave- 

 length side, is much below unity, probably as low as 0*5, 

 or even much less close to the line. In the case of light 

 going from a rare to a dense medium, a high value of the- 

 refractive index for the latter is accompanied by strong 

 reflexion. When, however, the ray goes from dense to rare 

 (quartz-mercury vapour) as in the present case, a low value 

 of the index for the latter is accompanied by strong 

 reflexion. 



" On the long wave-length side, for a region very close to 

 the line, the index of the dense mercury vapour may rise to a, 

 value as high as that of quartz, in which case there will be 

 no reflexion at all." 



It would appear then that, if we could employ light of two 

 frequencies, one slightly higher and the other slightly lower 

 than the frequency of the 2536 line, the former would be- 

 powerfully reflected and the latter not at all. This condition 

 was realized by employing as our source of light a quartz- 

 mercury arc operated at a potential just sufficient to distinctly 

 double the 2536 line by self-reversal. On Plate VI. fig. 7 

 we have four views of the 2536 line taken with a small Fuess 

 quartz spectrograph, very accurately focussed. This line has 

 a faint companion on the short wave-length side, indicated 

 by an arrow on the photographs. If the light is first passed 

 through mercury vapour in a heated quartz tube, the main 

 line is weakened or removed by absorption, and the faint 

 companion remains, as shown by fig. 7 (a), in which the 

 upper and lower figures represent the line without and with 

 mercury absorption. Fig. 7 (b) shows the appearance of the 

 line when the quartz arc, designed to operate at a potential 

 drop of 170 volts, is run at 30 volts, while (c) and (d) show it 

 reversed at 60 and 80 volts. 



We made our experiment as follows: — The light of the 

 lamp running at 80 volts was reflected from the inner surface 

 of the prismatic plate of the quartz bulb into the quartz 

 spectrograph, the slit of which was reduced to a length of 



