1889.J principal Line in the Spectrum of the Aurora. 435 



These values agree closely, and fix within very narrow limits, the 

 position in the spectrum, where we have to seek the chemical origin 

 of the line. 



Gyllenskiold, from his observations of the changes which occur in 

 the spectrum of the Aurora, comes to the conclusion that : " le spectre 

 de 1'Aurore boreale resulte de la superposition de plusieurs spectres 

 differents," and that " la raie principale forme un de ces spectres 

 elementaires ; elle apparait tres souvent seule." A similar view 

 was taken many years ago by Angstrom* and by Yogel.f 



[After consideration, I think that I ought to point out that Mr. 

 Lockyer's recent statement J that : " The characteristic line of the 

 aurora is the remnant of the brightest manganese fluting at 558," is 

 clearly inadmissible, considering the evidence we have of the position 

 of this line. 



In support of this statement Mr. Lockyer says : " Angstrom gave 

 the wave-length of the line as 5567, and since then many observers 

 have given the same wave-length for it, but probably without making 

 independent determinations. Piazzi Smyth, however, gives it as 558, 

 which agrees exactly with the bright edge of the manganese fluting. 

 R. H. Proctor also gives the line as a little less refrangible than 

 Angstrom's determination. He says : ' My own measures give 

 me a wave-length very slightly greater than those of Winlock and 

 Angstrom ' ('Nature,' vol. 3, p. 468)." 



By reference to Gyllenskiold's table it will be seen that the probable 

 errors of the determinations by Piazzi Smyth and Proctor, 5579 + 9'5 

 and 5595 + 25*0 respectively, are too large to entitle these measures 

 to special weight. 



Mr. Lockyer says further : " Gyllenskiold's measures with the 

 Wrede spectroscope also give 5580 as the wave-length of the charac- 

 teristic line. I feel justified, therefore, in disregarding the difference 

 between the wave-length of the edge of the manganese fluting and 

 the generally accepted wave-length of the aurora line." 



Gryllenskiold's single measure of 5580, on which Mr. Lockyer relies, 

 differs widely from the values which Gyllenskiold himself assigns 

 to this line, namely, from observations at Cape Thordsen in 1882, 

 X 5568 + 1'6, and from observations at Upsala in 1884, with the 

 Wrede spectroscope, X 5569 + 6*2. 



Speaking of Krafft's observations, Mr. Lockyer says :|| " The wave- 



* ' Nature,' vol. 10, p. 210. 



t ' Leipzig, Math. Phys. Berichte,' vol. 23, p. 298. 

 t ' Roy. Soc. Proc.,' vol. 45 (1889), p. 234. 



Gyllenskiold's statement of Proctor's value is based on 'Nature,' vol. 3, p. 347 

 and p. 68. 



|| ' Roy. Soc. Proc.,' vol. 45 (1889), p. 241. 



