290 New Lines in the Spark- Spectrum of Aluminium. 



coil and a small leyden jar only two of them are visible, and 

 they appear as short lines ; but when using a large coil giving 

 a 4 to 10-inch spark and a large jar these lines appear long 

 and bright, and are easily seen. One essential point is that 

 the electrodes should be thick and rounded off at the ends. 



The wave-lengths of the two brighter lines were measured 

 in two ways, viz. by a concave Rowland grating (25,000 lines 

 per inch) and by the ordinary prism-spectroscope. With the 

 grating the two lines appeared rather weak, and were difficult 

 to measure even in the first-order spectrum. The reference- 

 lines I used were those of potassium, of wave-lengths 6911*2 

 and 6938*8 (Kaiser and Runge), which are obtained in the 

 arc. The relative distances were measured by means of an 

 eyepiece-micrometer, which was fitted with a pointer. Eight 

 sets of measurements were made, and the wave-lengths cal- 

 culated by simple proportion. The values obtained were : — 



7042-3, I 



7056-7. J 

 With a prism of ordinary flint glass having an angle of 60° 

 the aluminium-lines appeared very bright and sharp. For 

 reference I used lithium 6708-1 (Rowland), and the two 

 potassium-lines 7665*6 and 7699'3 (Ederand Valenta). Some 

 of the iron-lines would probably have been more convenient 

 for the purpose, but there was no arc-light available at the 

 time. Greater dispersion than that obtained by one prism 

 could not be employed very well on account of the great 

 distance between the Li and K lines, and therefore the values 

 for the wave-lengths cannot be considered very accurate, but 

 they are near enough for identification. Ten sets of micro- 

 meter-readings of the apparent distances were taken on three 

 different days, which agree fairly well with each other. From 

 the mean values of these readings the wave-lengths were 

 determined by the method of inverse squares and were found 

 to be as follows : — 



7058-8 I w ^ reference ^ Li 6708*1 and K 7665*6. 

 7058-3 1 w ^ n inference to Li 6708-1 and K 7699*3. 



The value for the first — the brighter one — agrees very well 

 with that obtained by the grating. 



For the next three lines a dense prism of Jena glass was 

 employed, and in place of the potassium-lines the newly 

 determined aluminium-lines were used for reference together 

 with Li 6708*1. The greater dispersion — which is more 

 convenient in observing close double lines— made this neces- 



