of Metallic Spectra in the Ultra-Violet. 347 



measured lines of certain metals. When one metal ceases to 

 give spectral lines another must be selected. To test the 

 relative accuracy of what we have termed the old method of 

 measurement with that of the new, we have measured the 

 lines of electrolytic copper, and have compared our results 

 with those of previous observers in regard to the distribution 

 of errors. Besides the comparison of accuracy, we have ex- 

 amined the limit of the spectrum of copper in the ultra-violet, 

 in order to see if that given by previous observers could be 

 extended. 



Apparatus. 



The apparatus consisted of a concave grating of 21 ft. 6 in. 

 radius, mounted in the manner described by Professor Row- 

 land. The camera was provided with a shutter, which enabled 

 us to expose different portions of the sensitive plate at 

 pleasure. An alternating dynamo machine was employed, 

 together with a Ruhmkorf coil. The alternating machine gave 

 from eight to ten thousand reversals per second. With a 

 battery of from six to ten two-quart Leyden jars, a powerful 

 spark was obtained between the metallic terminals which we 

 employed. The spark was produced close to the slit of the 

 apparatus, and the time of exposure varied from one to two 

 hours. At various times endeavours were made to substitute 

 the more powerful carbon electric light for the electric 

 spark, in the hope of shortening the time of exposure ; but 

 these efforts were not successful. If they had been, we 

 should have been obliged to struggle with the question of 

 impurities in the carbons. An exposure of fifteen minutes 

 to the ultra-violet spectra of metals burned in the electric 

 light produced no image below wave-length 3000. A quartz 

 condensing lens was employed with the arc light, and there- 

 fore no light was lost by selective absorption. With the 

 spark no lens was necessary. 



By curving the photographic plate all parts of it remain 

 in focus, and distances on the plate are closely proportioned 

 to wave-lengths. If Y= wave-length, we have Y = C + «a', 

 where C and a are constants and x is the distance along the 

 plate. 



The determination of the wave-lengths of lines extending 

 over a range of three hundred tenth metres involved the 

 taking of three negatives. The sensitive dry plate (2 x 10 

 inches) was pressed by springs against the " forms " of the 

 plate-holder into an arc of a circle. Having placed the 

 plate-holder on the camera box, the girder bearing the camera 

 and grating was moved along its tracks until the position of 



