( 139 ) 



that the "centres of gravity" of the two dark bands, as well as the 

 bi'ighter space between them, have been alternately shifted to the left 

 and to the right — a phenomenon which needs no further explanation. 



As a matter of course the interposed flame causes the illumination 

 in the plane of the slit S^ to be very irregular, especially with 

 regard to those radiations undergoing anomalous dispersion in the 

 vapour. It is evident that some kinds of rays which are absent 

 from one part of that plane, will be found in excess at another. 

 The distribution of light in this irregular field of radiation might be 

 explored by moving S^, together with the spectroscope, within it. The 

 same object can be obtained with less trouble by means of a thick piece 

 of plate glass, mounted vertically between J3 and S, in such a manner, 

 that it may be moved round a vertical axis. When turning it a little 

 we make the whole radiation- field beyond the plate glass shift parallel 

 to itself, thus causing other parts to cover the slit. This influences 

 the aspect of the dispersion bands very materially. In certain positions 

 apparent emission lines of sodium vapour may happen to be seen, 

 which disappear as soon as the arc-light at S^ is intercepted ^). 



In conclusion we wish to draw attention to a peculiarity we 

 repeatedly observed in the dispersion bands. The dark shading in a 

 dispersion band does not become deeper in proportion as we approach 

 nearer to the central absorption line, but seems to reach its maximum 

 obscurity at certain (though not always equal) distances on both 

 sides of the centre; whilst in the space between, the light appears 

 somewhat intensified just as if a wide absorption band had been 

 partly covered by a narrower emission band, the centre of which 

 is again occupied by the fine absorption line. This phenomenon 

 cannot, however, be attributed to radiation emitted by the absorbing 

 sodium flame; for in our arrangement the intensity of the emission 

 from the flame could bear no comparison with that of the arc for 

 corresponding waves. In order to make sure we tried to photograph 

 the emission spectrum of the flame, exposing the plate during the 

 same length of time and under the same conditions as had been 

 done for obtaining the absorption spectrum; but not a trace of any 

 impression could be detected on the photographic plate. 



The light on both sides of the central line therefore originates 

 in the carbon points and this we explain on the principle of ray- 

 curving. The kinds of rays which are most strongly refracted in the 

 flame may, under certain conditions, be curved twice or even more 



1) These bright lines originate in the same manner as the light of the chromo- 

 sphere. The chromospheric lines are not emission lines, but "bright dispersion bands". 



