THE AliSOKin'ION AND KMI.SSIOX OK AlK. 19 



of this gas at different times. I accordingly abstain from making further statements 

 upon this subject. 



The spectrum of liydrogvn ends, according to my latest results, f;ir beyond the 

 limit which I obtained in 1893 with the apparatus I tlien used, and which by a 

 merely estimated wave-length I placed at 100 /'/'. This estimate, as will be later 

 shown, was by no means correct. The number of lines in the i)hotograph, which, 

 magnified 20 times, has a length of 1.4 meters, and of which a portion is shown in 

 Plates I-III, I estimate as at least 1500. Hydrogen develo^js its highest photo- 

 graphic efficiency at 162 /(/'. At the supposed wave-length of 100 nn its effect sud- 

 denly diminishes, and this is still more marked if in the path of the light from the 

 tube to the photogia[)hic plate, one or more fluor-si)ar plates are interposed, a thing 

 whicli the arrangements of the apj)aratus not infre(;[uently call for. A single such 

 plate 1 mm. thick may weaken the photogi'aphic effect at 100 /</i by one half. The 

 thickness of the plate is a secondary matter, for the loss of light depends chiefly 

 upon the number of reflecting surfaces. We may conjecture that the cause of the 

 enfeeblemeut lies in metallic reflection, possibly aided by absorption of light exei-- 

 cised by gases which these surfaces condense. The weakening is by no means 

 limited to the region mentioned, but with longer waves it is greatly diminished. 

 It may be conjectured that, on the other hand, it increases with shorter waves. 



I proceed to give photographic rei)roductions twenty times enlarged from my 

 photographed spectra. They are the only photographs of the hydrogen specti-um 

 which have been obtained up to this time in vacuo or in an atmosphere of hydro- 

 gen. In consideration of the not inconsiderable difficulties which had to be over- 

 come in their production, and which still oppose the I'epetition of the experiments, 

 it may not be deemed superfluous to enter into the details of the taking of this 

 photograph further than, for ordinary photographs of spectra, is either necessaiy 

 ..r nsu^d. Vacuun.-spectrography can be accomplished even now only under con- 

 ditions of great difficulty. , ■ i 



The spectrum that is mapped is that of the light of an end-on (Teissier tube 

 with a narrow capillary, such as is shown in Spectrum 5, Plate II. This tube 

 was connected with the collimator of the spectrograph. The four originals of the 

 first eight enlarged photographs of this spectral series were obtained with a fluor- 

 spar window interposed between the tube and the collimator. H^^ other our 

 were made without the window, since the spectra that they show would be undu y 

 weakened by the window, and the time of exposure thus rendered extraordinan y 

 lon^ In the former case, with the window in, the tube and spectrograph could 

 be filled and exhausted independently of one another. But in the lattei; case, 

 where tube and spectrograph were in communication through the open hole in the 

 slit-cover, both contained hydrogen at such a pressure as the electrical discharge 

 demanded. Since this pressure is only that of some '^^^'f TZJ^.l 

 hi<.hest, and since hydrogen is very transparent, since the depth of h>d_,ogen 

 tlrou.l which the mys had to travel was only 30 cm., since the deviation 

 o t h: rays by the hydrogen must have been extremely small owing to is smal 

 f acti i'y, and the taking up of any impurities by the hydrogen in the interior of 



