on Spectrum Analysis. 99 



I have not employed electrodes fused into the tubes them- 

 selves. As a rule, the tubes have been rendered luminous by 

 means of tinfoil coatings near the ends of the tubes, as 

 employed by Salet, Hasselberg, and others. By the adoption 

 of this method the process of discharge no doubt becomes most 

 nearly like that of Nature, and the temperature is kept as low 

 as possible. The source of electricity was a Holtz machine 

 without Leyden jars, the terminals of which were connected 

 by means of wires with the coating of the tube. The coating 

 thus corresponded to the external coatings of the jars of the 

 machine, the strata of air against the inner wall of the tube 

 serving as the inner coating. 



I have usually employed a spectroscope constructed by 

 Wrede in Stockholm, belonging to the Central Meteorological 

 Institute of Heisingfors, which was used by Prof. Lemstrom* 

 in his observations on the aurora in the years 1871-73. The 

 spectroscope consists of one dispersing-prism and one reflecting- 

 prism. The dispersion is somewhat small, not being sufficient 

 to divide the sodium-] ines. Ten divisions on the head of the 

 micrometer- screw correspond to intervals of 0*000006 and 

 0*0000006 millim. in the extreme red and extreme violet 

 respectively. The scale was constructed as follows : — A thin 

 glass plate (a microscopic cover-glass) was covered on one 

 side with indian-ink and five fine parallel lines ruled through 

 the black coating. The other surface was coated with Bal- 

 main's luminous paint. This plate was then fixed in the eye- 

 piece of the spectroscope, so that the fine lines were parallel to 

 the lines of the spectrum and in focus at the same time and 

 occupied about half of the field of viewj. It was rendered 

 luminous by burning a match before the eyepiece. If a 

 bright line in the spectrum was to be measured, this was done 

 just before the experiment ; but for fainter lines the measure- 

 ment was made after the phosphorescence had become feebler; 

 and with very faint lines it is necessary that the index-lines 

 should only be very slightly luminous, so that they may not 

 overpower the lines of the spectrum. In such cases the 

 adjustment was much easier, because there were five index- 



* This spectroscope is fully described in the Ofoersigt af Finska 

 Vetenskaps-Societetens forhandlingar, xv. pp. 21-23, 1873. Dr. Fucks 

 has described a reflecting direct-vision prism as new in the Zeitschrift fiir 

 Instrumentenkunde, 1881, p. 352; this prism is the most remarkable 

 feature of the Wrede spectroscope, and was described to the Royal 

 Swedish Academy of Sciences so long- ago as 1870. 



t I have described an index of this sort in the Astr. Nachrichten, 

 No. 2430 (1882). I have since learned that Prof. Vogel had employed 

 phosphorescent marks in 1881 for the measurement of spectra of feeble 

 luminosity {Zeitschrift fiir Instrumentenkunde, 1881, p. 20). 



H2 



