422 



Capt. H. P. Dawson. 



[liar. 13, 



of fact, one part of magnesium was detected in 10,000,000,000 parts of 

 water, the lines seen under these circumstances being those with 

 wave-lengths 2801 "6 and 2794'1. The spectrum reaction of arsenic 

 is the weakest, those of antimony and tellurium are also weak, 

 while that of bismuth is not strong. In fact it is noticeable that the 

 more strongly basic elements are those with the most persistent 

 lines. 



Evidence is afforded in the case of the aluminium spectrum that it 

 is not invariably the longest or strongest line which is the most 

 persistent. The line with wave-length 3584*4 is both longer and 

 stronger than a pair of lines adjacent thereto with wave-lengths 

 3612*4 and 3601*1, but whereas the first is not seen in solutions 

 containing 0*1 per cent, of aluminium, the pair are still visible in 

 solutions containing 0*01 per cent. Under certain conditions this 

 single line appears the longest in the whole spectrum, whereas other- 

 wise, and under most circumstances, the lines with wave-lengths 

 3960-9 and 3943*4 are longest. 



As a rule, even the longest lines are shortened by great dilution of 

 the solutions, but there is a pair of lines in the spectrum of copper with 

 wave-lengths 3273'2 and 3246*9 which become greatly attenuated, yet 

 nevertheless remain long lines till they finally disappear. 



It is shown by one or two examples how the tables of spectra and 

 accompanying maps may be employed in rendering quantitative 

 results. The special applications of this method it is proposed to 

 describe in a further communication. 



II. " On the Mean Diurnal Variation of Magnetic Declination, 

 from Hourly Observations at Fort Rae." Communicated 

 by Professor G. G. Stokes, Sec. R.S. By Captain H. P. 

 Dawson, E.A. Received February 28, 1884. 



The accompanying diagrams represent the mean diurnal variation 

 of declination for each month from September, 1882, to August, 1883, 

 at the Circumpolar station at Fort Rae, British North America. 



These curves are plotted from the hourly means for each month, 

 and include the effect of disturbances which have greatly influenced 

 tli em during the winter months, especially in November. 



One effect of an unusual amount of disturbance is an increase 

 of the mean declination ; the easterly disturbance, or that tending to 

 increase the declination, being always in excess. 



