PHYSICS. 387 



Certain of the rare earths, i.e. the oxides of gadohnium, samarium, niobium, 

 praesodymium, and neodymium, also of cerium and erbium, should be added 

 to the list. 



The effect in question has the following characteristics : 



1. It consists of luminescence superimposed upon the ordinary radiation 

 due to temperature. 



2. For each substance there is a definite and rather narrow temperature 

 range within which the effect begins, reaches a maximum, and disappears. 

 In some cases two such definite and well-marked crests have been observed. 



3. At the maximum the brightness of the substance frequently greatly 

 exceeds that of a black body of the same temperature. 



4. The luminescence is highly selective. 



5. The intensity of the effect depends on the mode of heating and partic- 

 ularly on the presence of free oxygen. 



6. It also depends on the previous heat-treatment of the substance. 



7. It is subject to fatigue. 



8. The temperature range is that during which the substance is undergoing 

 some profound physical change or rearrangement and the crest coincides in 

 temperature with some transformation point. 



The Photo-Luminescence of Flames. 



In connection with these studies of the luminescence of incan descent bodies, 

 investigations of the conditions within the flame itself are in progress. 



In the case of the hydrogen flame in air the following quite unexpected 

 effect has been discovered : When such a flame is rendered luminous by the 

 introduction of a salt of lithium, sodium, calcium, barium, or strontium, the 

 intensity of the bright bands of the flame spectrum is modified by exposing the 

 flame to light. 



In the preliminary observations, now in progress, enhancement of these 

 bands (e. g., of the sodium lines, the lithium line at 0.6708//, the red, orange, 

 and green bands in the spectrum of calcium and strontium) has been produced, 

 respectively, by light from the tungsten filament of a nitrogen-filled lamp, 

 from a carbon arc, and from an iron spark. The increase in brightness has 

 in some experiments exceeded 10 per cent of the initial intensity of the band. 



The active rays appear to lie among the shorter wave-lengths, but not in 

 the far ultra-violet. 



Light of the same wave-length as that in a given band, instead of enhancing 

 the band, by resonance, has a quenching effect. Thus sodium light focussed 

 upon a sodium flame tends to inhibit its enhancement by exposure to shorter 

 wave-lengths; light passed through a ruby glass has the same quenching effect 

 on the lithium flame; white light filtered through an ammonio-sulphate solu- 

 tion is more effective as an excitant than the same light, unfiltered, etc. 



The phenomenon is strictly analogous to the well-known quenching effect 

 of infra-red upon the luminescence of certain sulphides, and one is led to ask 

 whether we do not in these experiments have to deal with the luminescence 

 of finely divided solids (possibly particles of Na^O, Li20, CaO, etc., respectively) 

 instead of with the glowing vapors of the elements. 



