234) . Royal Society t — 



vapour causes its radiation, and therefore its absorption, to assume 

 more and more the character of a continuous spectrum as the 

 thickness is increased. 



It has been shown by Br. IVankland and myself that such a con- 

 dition obtains when the density of a vapour is increased, and my 

 later researches have shown that it is brought about in two ways. 

 G-eneralizing the work I have already done, without intending 

 thereby to imply necessarily that the rule will hold universally, or 

 that it exhausts all the phenomena, it may be stated that metallic 

 elements of low specific gravity approach the continuous spectrum 

 by widening their lines, while metallic elements of high specific gra- 

 vity approach the continuous state by increasing the number of 

 their lines. Hence in the vapours of Na, Ca, Al, and Mg we 

 have a small number of lines which broaden, few short lines being 

 added by increase of density ; in Fe, Co, M, &c. we have many lines 

 which do not so greatly broaden, many short lines being added. 



The observations I made in India during the total solar eclipse of 

 1871 were against the assumption referred to ; and if we are to 

 hold that the lines, both "fundamental " and " short," which we get 

 in a spectrum, are due to atomic impact (defining by the word atom, 

 provisionally, that mass of matter which gives us a line-spectrum), 

 then, as neither the quantity of the impacts nor the quality is 

 necessarily altered by increasing the thickness of the stratum, the 

 assumption seems also devoid of true theoretical foundation. 



One thing is clear, that if the assumed continuous spectrum is 

 ever reached by increased thickness, as by increased density, it must 

 be reached through the " short-line " stage. 



To test this point I have made the following experiments : — 



1. An iron tube about 5 feet long was filled with dry hydro- 

 gen ; pieces of sodium were carefully placed at intervals along 

 the whole length of the tube, except close to the ends. The ends 

 were closed with glass plates. The tube was placed in two gas- 

 furnaces in line and heated. An electric lamp was placed at 

 one end of the tube and a spectroscope at the other. 



When the tube was red-hot and filled with sodium-vapour 

 throughout, as nearly as possible, its whole length, a stream of 

 hydrogen slowly passing through the tube, the line D was seen to 

 be absorbed ; it was no thicker than when seen under similar con- 

 ditions in a test-tube, and far thinner than the line absorbed by 

 sodium-vapour in a test-tube, if the density be only slightly in- 

 creased. 



Only the longest " fundamental" line was absorbed. 



The line was thicker than the D line in the solar spectrum, in which 

 spectrum all the short lines are reversed. 



2. As it was difficult largely to increase either the temperature 

 or the density of the sodium-vapour, I have made another series of 

 experiments with iodine-vapour. 



I have already pointed out the differences indicated by the 

 spectroscope between the quality of the vibrations of the " atom " 



