244 PROCEEDINGS OP THE AMERICAN ACADEMY. 



spectrum. Even with its slits very narrow, the wave-length range 

 of the emitted light was wide enough to cover several of the absorp- 

 tion lines of the vapor. The earlier work had shown that the light of 

 the cadmium spark was capable of exciting fluorescence, and ex- 

 periments were accordingly started with metallic arcs. Just at this 

 time came the very opportune invention of the fused quartz metallic 

 arc lamps by Stark, working in the Heraeus laboratory at Hanau. 

 Two of these lamps were immediately ordered, one filled with cad- 

 mium, the other with zinc. Their form is shown on Plate 3, Figure 1. 

 The lamp is kept in communication with a mercury pump during its 

 operation and stands in a disk of water. The cadmium lamp worked 

 well on a circuit of 110 volts direct, but the zinc lamp gave better 

 results on the 220. They are started by a small induction coil, one 

 terminal of which is connected to the negative pole of the lamp, 

 the other twisted around the quartz U tube. A blast lamp is directed 

 against the tube until the portion above the metal electrodes is 

 red hot, the coil is then started, and the arc usually forms at once. 

 As exposures of eight or ten hours were often necessary, and the lamps 

 have a trick of going out every half- hour or so, an automatic starter 

 was devised, which turned on the coil the moment the lamp went out. 

 As soon as the arc struck again, the coil was stopped. This arrange- 

 ment is figured on Plate 3, Figure 2, and consisted of a small electro- 

 magnet in circuit with the lamp, which pulled a steel spring away 

 from a brass screw as long as the lamp burnt. The spring and screw 

 were inserted in the primary circuit of the coil. 



The cadmium lamp burns with a greenish-blue light of dazzling bril- 

 liancy, the zinc lamp with a curious purple light, which causes all the 

 woodwork in the room to appear blood-red, while most other objects 

 appear bluish white or purple. Both lamps excite a brilliant fluores- 

 cence of the sodium vapor when their images are thrown upon the oval 

 aperture of the retort. In this case the fluorescence is excited by 

 several different radiations. Various devices were used for picking out 

 one line at a time. The cadmium radiations which are capable of 

 exciting fluorescence have wave-lengths 5086, 4800, and 4678. Color 

 screens and the Fuess monochromatic illuminator, as well as the thin 

 crystals of chlorate of potash (described in the Phil. Mag. for June), 

 were tried ; also a block of quartz, cut perpendicular to the axis, placed 

 between two nicols. The arrangement which gave the best results, and 

 appeared to be accompanied with the least loss of light, is the one 

 figured on Plate 3, Figure 2. One vertical tube of the lamp is used as 

 the source, the light from which, after collimation, passes through a 

 large bisulphide of carbon prism, and is focussed upon the retort by an 



