250 Professor George D. Liveing [Marcli 9, 



dark by comparison with the brightness of the photosphere, are them- 

 selves luminous, even bright, when there is no other still brighter light 

 wherewith to contrast them, so that if there were no other absorbent 

 action the solar spectrum would be continued by the emitted rays of 

 the metallic vapours which produce these lines. Probably then the 

 absorbent is something at a lower temperature, higher in the solar 

 atmosphere. A change of temperature may, and in some cases certainly 

 does imply such a change of state that there may be a corresponding 

 change in the particular vibrations which can be most easily taken up. 



The metals in the liquid and solid states are so very opaque that 

 we should hardly be able to discern their absorption spectra ; never- 

 theless in very thin films they are translucent in different degrees. 

 Gold leaf, as is well-known, transmits a green light, and we have 

 found that a thin film of gold chemically deposited on a plate of quartz 

 is fairly transparent for all the ultra-violet rays, so that its selective 

 absorption is almost wholly of the less refrangible rays. Silver de- 

 posited in a similar way produces a very different effect. It is almost 

 wholly opaque, except for one rather narrow band which begins a little 

 below the solar line P and extends with diminishing transparency to 

 about S. Cornu has before noticed this property of silver, but jdaced 

 the transjDarent band at wave-length 270 instead of 330. Dr. W. A, 

 Miller had observed that the light reflected by gold is equally dis- 

 tributed all through the ultra-violet, but feebler than that reflected 

 by other metals ; while that reflected by silver is characterised by 

 giving a sudden cessation of the photograj)hic image for a certain 

 distance. These characters of the reflected rays he attributed to 

 absorjjtion by the metal. 



"When we examine the absorption produced by the haloid elements, 

 we find that chlorine absorbs a wide band in the ultra-violet with its 

 centre near the solar line P, extending, when the chlorine is in small 

 quantity, from N to T, increasing in width on both sides when the 

 quantity of chlorine is increased, but still leaving the rays above wave- 

 length 2550 unabsorbed. 



Bromine vapour shows an absorption band which begins in the 

 visible spectrum, and extends, when the bromine is in small quantity, 

 up to L, and when the bromine is in greater quantity up to P. From 

 that point, up to about wave-length 2500, the vapour is transparent, 

 but beyond it is again absorbent, the absorption increasing gradually 

 with the refrangibility of the rays. 



Iodine vapour, when thin, is transparent for ultra-violet rays, but 

 produces strong absorption in the violet region. With thicker vapour 

 this absorption extends nearly to H, but the vapour is still transparent 

 for rays more refrangible than H. 



Lecoq de Boisbaudran has observed that in the spectra of similar 

 elements we may trace a shifting of similar lines, or groups of lines, 

 towards the less refrangible side as the atomic weight is increased. 

 Thus the violet pair of lines given by potassium is represented by an 

 indigo pair in the case of rubidium, and a blue pair in the case of 



