1032 M. L. Dunoyer and Prof. Wood on Photometric 



silvering one surface of a double convex lens. Under these 

 conditions the magnesia triangle, and the resonance radiation 

 (which may be termed secondary) which surrounded it had 

 practically the same intensity. In fact it was only with 

 difficulty that the outline of the triangle could be seen, the 

 black dot being surrounded by a uniform glow of li^ht of 

 oval outline (fig. 3, b) . 



Photographs taken of the phenomena are reproduced on 

 PI. XVII. figs. 8 and 9, the latter showing the disappearance 

 of the magnesia triangle, when the area is illuminated by 

 resonance light reflected from the mirror. The brilliantly 

 illuminated area to the left is the primary resonance excited 

 by the rays from the flame. A narrow dark line partially 

 outlines the triangle ; this is due to the shadow thrown upon 

 the resonating vapour by the edges of the layer of magnesia. 



The complete disappearance of the triangle was observed 

 only when the flame for exciting the primary resonance was 

 very poor in sodium. We thus have a ratio equal to unity 

 when the exciting rays are sufficiently homogeneous, and 

 can safely say that no true absorption exists in the case of 

 sodium vapour at very low density and in a high vacuum, 

 though the spectroscope would of course show an absorption 

 line. All of the energy abstracted from the primary beam 

 is re-emitted by the molecules, precisely as was found for 

 mercury vapour. 



Probable width of the resonance lines. 



The experiments which we have just described show that 

 the resonance radiation of sodium is excited by the narrow 

 central regions of the D lines. 



Let ABC of fig. 4 represent the intensity curve of one of 

 the exciting lines, and the dotted curve DBE the region 

 effective in exciting resonance. The intensity curve of the 

 emitted resonance radiation will be of similar dimensions 

 and may be represented by F. The ratio of the area of the 

 curve ABC to the area of DBE is obviously the ratio found 

 by the photometric measurements, and if we know the form 

 of the curves, and the actual dimensions of ABC (?'. e. the 

 width of the line in the flame spectrum), we can, from our 

 experimentally found ratio of 4 : 1, determine the width of 

 DBE, the line of the resonance radiation. 



The interferential measurements of Fabry and Buisson 

 have shown that the widths of the D lines emitted by a flame 

 poor in sodium are 0*08 A. 



