THE AURORA BOREALIS. 



743 



with this arrangement the position of the faint aurora-lines could 

 be exactly determined. 



For the wave-lengths corresponding to the Unes in the spec- 

 trum, Angstr5m's values were made use of, and the wave-lengths 

 have been deduced from the readings of the micrometer screw by the 

 following formula, which is applicable to the rays from D to H : — - 



X = • 00046489 + • 000009540(a — 1 0) + • 0000004564(a - 1 0)2 

 + 0-00000002184(a-.10)3, 



where a is the reading of the spectroscope, and x the length of 

 the wave. 



Observations on the solar spectrum were made during the 

 autumn of 1872 and the spring of 1873, and they have shown 

 that during this period no change of the zero took place. 



During the winter the instrument was tested by means of the 

 faint part of the flame of a candle. In a few experiments there 

 was an accidental disjilacement, probably because a small quantity 

 of dust or ice had penetrated into the apparatus, and in these cases 

 the values determined have been enclosed in brackets, and entirely 

 omitted from the final result. A variation of one-tenth of a revo- 

 lution of the screw corresponds to from 0- 0000019 to 0-0000007 

 of a wave-length in the part of the spectrum where the Aurora 

 lines occur. 



The first table contains the numbers found by M. Wijkander 

 for the different series of lines grouped according to the different 

 days. The measures of the most brilliant ray of the Aurora are 

 not included in the table, since it is bright enough to be easily 

 determined by instruments of greater dispersive power. The 

 time when the Aurora was most brilliant was spent in observing' 

 the other rays, and the brightest lines have been measured in the 

 intervals. The mean value found for the wave-length of the 

 brightest ray was 0-0005572. 



The following are the lengths of wave, with their means and 

 probable errors, taking the millionth of a millimetre as the unit : — 



The following Table contains the measurements made by 

 M. Wijkander. 



1872. 



1. 



2. 3. 



4. 



5. 



6. 



7. 



8. 



October 24 - 



I SO " '' 

 November 3 



December 5 



1873. 



Ja^uary 18 - 



»» » " 



536-46 

 534-80 



535*46 



535-46 

 537-46 



526*89 



527-99 

 528-63 

 530-23 



531-85 

 527-68 



523-76 

 524-07 

 526-28 

 524-85 

 521-93 

 522-69 



498-85 

 503-20 

 496-29 



497-31 



502-26 



487-13 



470-77 

 468-78 

 470-77 

 468-88 

 469-08 

 469-87 



467-80 

 468-29 



436*65 



428-54 



427*86 

 426*60 



429-02 



Mean 



Probable Error - 



535-9 

 + 0-31 



528-9 

 + 0-51 



523-9 

 + 0-43 



499-6 

 + 0-91 



487-1 



469-2 

 + 0-25 



436-6 



428-0 

 + 0-35 



