250 



NATURE 



[January i6, 1896 



scale similar to those adopted for wind and cloud should have 

 been hitherto used in meteorological reports. Like the above 

 scales, it could only be approximative ; but if observers fixed 

 upon conspicuous objects, such as hills, churches, &c., at known 

 distances for their observations, these ought to be at least as 

 accurate as those for wind and cloud. 

 Nant-y-Glyn, Colwyn Bay. Alfred O. Walker. 



FL UORESCENCE OF SODIUM ANDPOTASSIUM 

 VAPOURS, AND THE IMPORTANCE OF 

 THESE FACTS IN ASTROPHYSICS.^ 



IN the case of unmixed vapours, E. v. Lommel (, Wied. 

 Ann., p. 856, 1883) has proved a fluorescence for 

 iodine vapour, and this has been shown also by us ( Wied. 

 Ann. 56, p. 18, 1895) ^o be true for the vapours of 

 numerous organic substances. In our latest experiments 

 we have studied the behaviour of the vapour of metals 

 in this respect. 



(i) Order of Procedure. — For the experiments the 

 following arrangement was adopted. Rays of sunlight 

 were made to fall on a double convex lens, L, of focal 

 length 5 cm., which brought them to a focus at a point, 

 <r, in the centre of a glass spherical bulb, K, filled with 

 vapour. 



To one side of the bulb was placed a spectroscope 

 with the slit horizontal, the edge of the prism also being 

 horizontal. The spectroscope was fixed by its legs on a 

 board placed vertically, and could be revolved round its 

 axis by means of a movable cone on the inside part of 

 one of the feet. One was able also to adjust the col- 

 limating telescope on the cone of fluorescence proceeding 

 from the inside of the bulb. 



The bulbs, which were made principally of hard glass, 

 were filled with sodium, potassium, and other metals in 

 the following manner. 



In the first place, a small quantity of the metal was 

 placed in the bulb K, the neck then drawn out, and at 

 its extremity connected with a tube of soft glass, leading 

 to a mercury pump. The metal was then placed in 

 position a, and warmed during continuous gentle pump- 

 ing, in order to get rid of all water vapour and CO2 ; 

 each bulb was also several times cleaned out with 

 hydrogen. The heating was then increased, and when 

 a convenient quantity of the metal was distilled the tube 

 was sealed off by means of the blowpipe. In most cases 

 the diameter of the bulb was 5 cm., so that it could be 

 easily enveloped in a flame. We have also employed 

 bulbs made of soft glass and of various sizes, but the 

 optical effects were exactly the same. Difficulties were 



1 Translation of a communication to the " Sitzungsberichten der physikal. 

 med. Societat zu Erlangen." By Eilh Wiedemann .ind G. C. Schmidt. 



NO. 1368, VOL. 53] 



met with when employing the alkali metals ; for, as soon 

 as the vapour began to be formed, the bulbs became 

 browned and blackened, so that new ones had to be 

 substituted. 



(2) Results. —The experiments showed that the fluor- 

 escence of sodium and potassium vapour was bright— 

 the former green, the latter intense red. 



The fluorescence can also be beautifully seen by 

 observing the vapours of these metals in the light of an 

 arc lamp. In cases of the less volatile metals, we have 

 not yet been able with perfect confidence to prove its 

 presence, although the vapour of cadmium undoubtedly 

 at the surface of the melting metal displayed a green 

 fluorescence colour. The failure, however, of these 

 experiments may be explained partly by the poor intensity 

 of the rays of the winter sun. We shall, naturally, repeat 

 them in summer. 



The fluorescence spectrum of sodium vapour was made 

 up as follows : 



675. Boundary of the red. 

 675 — 602 '5. Bright red band. 

 602 •5—540. Dark band. 



540 — 496. Green fluted band, composed of single dark and 

 bright lines. 

 In the blue there is practically nothing visible. 

 At the less refrangible end of the dark band 602 — 540, the 

 yellow sodium line appeared. 



The appearance of the bright line of sodium was not 

 caused by the presence of the flame used for heating, 

 because when this was removed, the line remained still 

 bright ; nor could its presence be explained by the action 

 of chemical processes taking place in the bulb, because 

 the moment the incident Hght was cut off it disappeared. 



The fluorescence spectrum of the vapour of sodium ^ 

 is made up of three parts : (i) the non-fluted band in 

 the red, (2) the fluted band in the green, (3) the bright 

 sodium line in the yellow. 



Whilst in the cases of solids and liquids which exhibit 

 fluorescence the fluorescence spectra consist of broad ill- 

 defined, continuous streaks, we meet here also fluted 

 bands, as shown by other gases under the action of 

 electrical discharges, and single lines. 



The fluorescence spectrum of potassium displays at 

 695 — 615 an intensely red band. 



Adjoining this band the dark interspace is relieved by 

 the somewhat brighter appearance of the green, due, 

 perhaps, to the presence of some sodium vapour. 



The bright lines of potassium could not be proved, but 

 their absence maybe accounted for by the feeble intensity 

 of the incident light. 



The fluorescence spectrum of lithium could not be ob- 

 served, because so soon as the lithium was placed in the bulb 

 and heated, the glass exposed to the light-source became 

 affected. By further heating, the vapour given off" dis- 

 played only the green fluorescence light of sodium. For 

 the same reason experiments to obtain in discharging 

 tubes, the " elekroluminescenz " of hthium were un- 

 successful. 



(3) Validity of Stakes' Law for the Fluorescence Light 

 of Metallic Vapours. — We have also made investigations 

 to find out whether here the law of Stokes — that is, 

 whether the excited rays of light are less refrangible 

 than those exciting — holds good. To this end a spectrum 

 was formed by means of a prism, but only a small strip 

 of this was employed, and led by means of a lens into the 

 bulb filled with vapour. In the case of the vapour of 

 sodium the intense green light radiated was excited, in 



1 A comparison of these fluorescence spectra with that which is obtained 

 by the heating of sodium vapour exhibits certain relations (Evershed, Phil. 

 Mag. (5), 39, p. 460, 1895); the same is the case with both sodium and 

 potassium, if the positions of the fluorescence spectra be compared with the 

 absorption-band-spectra investigated by H. E. Roscoe and A. Schuster 

 (^Proc. Roy. Soc. London, 22, 262, 1894). In both cases the radiation 

 towards the red appears altered. 



