46 Rubens and Trowbridge — Dispersion, etc. 



already shows a visible absorption at 12/jl for this thickness of 

 layer, and that beyond 20/^ the absorption is almost complete. 

 Sylvine (curve b) begins first to absorb to a marked degree 

 beyond 13/x and still shows even at X = 23"7/x the same trans- 

 mission that rock salt shows at X = 18'5/t. The curve of 

 transmission for fluorite falls sharply and touches the axis of 

 abscissas close beyond X = 11.//,. Paschen* has also examined, 

 in the communication cited above, the absorption of a fluorite 

 plate to the wave length X = 9*43/t. If the transmission of a 

 layer l cm thick is calculated from the data obtained by him, 

 and the points are plotted in fig. 5, they show a very good 

 agreement with the points observed by us. In order to dis- 

 tinguish between the points calculated from Mr. Paschen's 

 values and those obtained from our results, the former have 

 been plotted as crosses (+) and the latter as circles (o). 



This agreement furnishes at the same time a proof of the 

 pureness of our spectrum and of the correctness of our 

 measurements on the dispersion of the rock salt and sylvine 

 prisms.f 



The dotted line (a) in fig. 5 gives the transmission of the 

 chloride of silver plate 3 mm thick as a function of the wave 

 length. A correction for the loss of energy due to reflection 

 has not been applied to it. 



In reviewing the results communicated in this paper, the 

 conclusion is reached that it is possible with double dispersion, 

 using two acute-angled prisms made of rock salt, to obtain 

 sufficient energy in the infra-red spectrum out as far as X =18//-, 

 and that using sylvine prisms, one may attain to still greater 

 wave lengths (about X = 23/t). To be sure, the amount of 

 energy obtained after this method is much less than that 

 obtained by repeated reflection on quartz or fluorite surfaces. 



Physical Institute of the Polytechnic, 

 Charlottenburg, Berlin. 



* F. Paschen, Wied. Ann., li, p. 333. 



f As a further test of the correctness of our measurements on dispersion, we 

 took, in the above mentioned manner, measurements on the dispersion in an acute 

 angled fluorite prism in the field between /u = 9'Ojn and A = 10*0// which were in 

 good agreement with Mr. Paschen's results. The greatest departure from his 

 values was 0'4 per cent in the wave length. 



