and Ahsorption K:<pectni of Sodium Wipour. 365 



The marked resemblance which the fluted .spectrum bears 

 to the absorption at once suggests that it may be due to the 

 tact that the fluorescent Hght has to pass through a certain 

 amount of vapour before reaching the spectroscope, in other 

 words that it does not belong to the fluorescent spectrum at 

 all, but is the result of absorption. To determine whether or 

 not this was the case an absorption comparison spectrum was 

 formed by throwing some of the light which had passed 

 through the tube into the instrument by means of a pair 

 of mirrors and a small right-angle prism. It was at once 

 apparent that the bright lines and bands of the fluorescent 

 spectrum were exactly out-of-step with those of the absorp- 

 tion spectrum. To secure a fixed record of this fact a camera 

 was attached to the spectroscope and the two spectra photo- 

 graphed. The spectrogram confirmed the visual observations 

 in every respect, but the dispersion was too small to make 

 the pictures very satisfactory. 



A Rowland concave grating with 15,000 lines to the inch 

 of one metre radius was then used in place of the spectro- 

 scope, and some excellent photographs obtained with an 

 exposure of less than an hour. The fluorescent spectrum 

 was found to extend from wave-length 5340 to wave-length 

 -1-600 in the green and blue region. All of the photographs 

 show in addition a faint solar spectrum extending from the 

 end of the fluorescent spectrum down to the H and K lines. 

 This is due to a small amount of white light which is scattered 

 by occasional specks of oxide, or perhaps reflected from the 

 wall of the tube. So far as we have been able to determine 

 the fluorescent spectrum is not contaminated with solar lines 

 since it is located in a less actinic region, and the scattered 

 light is not of sufficient intensity to leave any appreciable 

 record in this region. 



These photographs brought out the remarkable fact that 

 the fluorescent spectrum is the exact complement of the 

 absorption spectrum. The two spectra were photographed 

 in contact on the same film, and either one might have been 

 a contact print taken from the other. In the upper spectrum, 

 for example, there were t'.vo broad light bands with a fine 

 dark line down the centre, while in the lower spectrum 

 occurred two broad dark bands with bright lines down the 

 centre. The same thing was true for all of the irregularities 

 of shading in the very complicated fluted spectrum. A 

 number of these ])hotograptis are reproduced in Plate XIV. 

 figs. 1 & 2. As most .of the fine detail will doubtless be lost 

 in the process of reproduction a very careful drawing of the 

 two spectra has been prepared from the original negative, 

 which is reproduced with the direct records (Plate XIV. fig. 3, 



