THE A15S0UITION AND EMISSION Ol' AIR. 11 



this diminution of tliickness is due not only tlieir sensitiveness for the most 

 i'efi;uigil)le ultra-violet rays, but also their power of promoting the optical efficiency 

 of the photographic api)aratus. What adds to tlie importance of this fact is that 

 if the spectrum is to be sharply defined over the longest possil)le poi'tion of it, tlien 

 the rays will not traverse the sensitive layer i)erpendicularly, but will, on tlie 

 contrary, have an angle of incidence of 60° to 69°. The short focal distance (the 

 focus at 589 /'/' is 12 cm.) and weak dispersion (with a fluorite prism of 60°) of 

 ray spectrograp)h make the last named property of the ultra-violet plates incompara- 

 bly more important in the pivsent case than with instruments of stronger disper- 

 sion. And it is not the least of the advantages 1 owe to that property that 

 the very small negatives of the part of the spectrum in question, in consequence 

 of their uncommon capacity for enlargement, nevertheless answer the demands of 

 exact spectroscopy. A numerical example will best show what my spectrograph 

 was in this way able to do. In a jiart of the hydrogen spectrum about 160 /'/' it 

 resolved a length of 13 millimeters into more than 300 clearly marked lines. I 

 need not say that very exact edges and very narrow opening of the collimatoi'-slit 

 wei'e needed for this. My negatives measure 37 by 12.7 mm. The image of the 

 spectrum appears upon them upon a ground as clear as glass. It bears a far greater 

 enlargement than is possible with ordinary gelatine jtlates. 



Filling tlie absorption and emission tubes. — The absor[)tion-tubes were filled at 

 the atmospheric pressure, at fiist merely by drawing the gas through them for 

 a long time before the final filling. But it was found that the air was not in this 

 way as completely expelled from the tubes as my experiments required, since even 

 a small residue notal)ly altered the transmissivity of the gas concerned. Conse- 

 quently in the later work, the tubes were pumped out before the gas was allowed 

 to enter, and were then repeatedly washed out with the gas. The connections of 

 the apparatus for the evolution of the different gases with the absorption-tube 

 would have been much easier manipulated if india-rubber could have been used for 

 them. But my experiments showed that such tubing, because of its exhalations, if 

 for no other reason, may introduce such impurities into the tube as to render it not 

 fit for use, giving the spectrum a totally different appearance. I also ascertained 

 that leakage of air was to be feared even with the thickest india-rubber tubes. 

 For example, through an exhausted Para-rubber tube of 40 centimeters length, 1.15 

 cm. external' diameter, and 0.35 cm. internal diameter, 2.4 (cm.)^ of air passed 

 in fifteen hours. But I attribute only a subordinate prejudicial effect to the air 

 that was carried in through this tube along with the gas, because the washing out 

 and filling of the absorption-tube was at that time done at the ordinary pressure, 

 so that but a very small amount of air would enter, and because there was nothing 

 in the appearance of the spectrum as altered by the influence of the rubber to 

 suo-o-est the influence of air. This opinion was confirmed by the circumstance 

 that° when this tube of Para-rubber was replaced by a tube of black rubber, 

 such as is often used to make connections of chemical apparatus, the distribution 

 of enero-y in the spectrum underwent a new change : namely, while with the former 

 tube th^'e spectrum was weakened generally, the use of the black tube completely 



