( 545 ) 



academical thesis, which will soon be published, further details 



will be given. 



For the investigation of the alkalies, the metallic salts were volatilised 

 in the arc in the ordinary way. The very complicated band-spectrum, 

 emitted by the arc when no metallic vapour is present, extends far 

 into the ultra-red. But this interferes in no way with the investi- 

 gation of the metals, since it is entirely superseded when the arc con- 

 tains a sufficient quantity of metal. On the other hand the continuous 

 spectrum, emitted by the incandescent particles in the arc, makes it 

 somewhat difficult to observe some feebler lines; besides, the radiation 

 of carbonic acid, the product of combustion of the carbons, (with a 

 maximum near 4,44 (i) persists with almost unchanged intensity. 



The image of the arc is projected by a concave mirror on the 

 slit of a reflecting-spectroineter ; the rays are analysed by a rock- 

 salt prism and part of the so formed spectrum falls on a linear 

 thermopile. This thermopile, like that of Rubens, is built up of iron 

 and constantan ; all the dimensions were chosen smaller than in the 

 original pattern and a great sensitiveness was obtained. As well the 

 emitting slit as the thermopile are mounted in fixed positions; in 

 order to throw on this latter different parts of the spectrum in 

 succession, the prism can be rotated through small angles. A 

 Wadwortii combination of prism and plane mirror maintains minimum- 

 deviation during rotation. 



In chosing and designing the instruments, the desirability ^was kept 

 in mind of replacing the very tiring reading of the galvanometer 

 and the simultaneous noting of the corresponding position of the 

 prism, by an automatical recording-device. I had in mind the 

 splendid arrangement by which Langley has for years recorded the 

 intensity-curve of the ultra-red solar spectrum on a photographic 

 plate. That this method has not been followed for recording heat- 

 spectra instead of the time-absorbing "visual observations, must be 

 ascribed in the first place to a very complicated mechanism being 

 required for obtaining complete correspondence between the linear 

 displacement of the photographic plate and the rotation of' the spectro- 

 meter, and secondly to the difficulty of keeping the surrounding 

 temperature perfectly equal during the observations. 



With very simple means I devised a method of recording, which 

 avoids these two difficulties, while yet it warrants a sure "corres- 

 pondence", and yields accurate results also when changes in the 

 surrounding temperature cannot be prevented. For this purpose the 

 continuous recording has been replaced by the marking of a series 



37 



Proceedings Boyal Acad. Amsterdam. Vol. IX. 



