8 THK AHSOKn'IoX AND KMISSION OF AIK. 



The cxhauslioii of the sitectrograph was effected Ijy a Geissler mfivurial air- 

 piiiiip. The soiiice of light wjis usually a Geissler tube of the " end-on" form, the 

 lays entering and leaving the tube through plates of (quartz or fluorspar. Occa- 

 sionally, electi-ic sparks from metals were employed. The discharge was furnished 

 by a Kuhnikorff coil giving si)arks of 20 cm., the primary current being derived 

 from a numljer of Grove cells. The electro-motive force and the intensity of the 

 piimaiy current were accurately measured. To furnish layers of the different 

 gases foi- observations of absorption, air-tight brass tubes were interposed length- 

 wise, coa.vially with the collimator, between the source of light and the spectro- 

 graph. They could be e.xhausted and filled with the gas whose absorption was to 

 be studied. The lengths of the tubes therefore correspond to the thickness of the 

 stiata of gas which measured 65, 30, 25, 20, 15, 10, 5, 2^ cm. Most gases appear, 

 even in very thin strata, strongly to absorb the rays of the most refrangible parts 



Fig. io. Absorption Ai-i-aratcs. Half Natlral Size. 



of the ultraviolet specti-um. Some gases exercise a general absorption, while otheis 

 give characteristic lisiug and falling selective absorption-curves. Since these curves 

 interest physicists foi' various reasons, I have constructed a special apparatus for 

 the exhibition of them in the photographic way. This is shown in Fig. 10 in half 

 its natural size. The principle of it is as follows : 



A Geisslei' tuVje filled with hydrogen, or other gas, sends its rays to the prism 

 through a chamber of variable de[)th provided with two opposite and i)arallel walls 

 of fluorspar, and filled with the gas to be investigated at atmospheric pressure. 

 One of the windows remains fixed, while the other can be so shifted in the direc- 

 tion of the rays that the distance aj)art of the two can be continuously varied from 

 (t to 15 mm. This distance is evidently the thickness of the [)lane-parallel gas- 

 stratum to l^e investigated. This whole apparatus is put in the place of the cover 

 of the slit-carriage marked e', in the foregoing descriptioji of the vacuura-spectro- 

 graph, so that the capillary is in line with the produced geometiical axis of the colli- 

 mator. The apparatus is constructed in detail as follows : The cover, e (Fig. 10), 



