Showing their identity luith tvio lines in the Solar Spectrum. 3 



Plate III. — A second plate with the same two spectra, a band of the solar 

 spectrum running through the middle of the spectrum, and a 

 second one touching the edges of the lines. 



Plate IV. — (1) An arc spectrum of pure iron, the metal being prepared from 

 potassium ferrocyanide, by fusion with potassium carbonate. 



(2) The same. 



(3) The same, with a large proportion of the residue obtained on 



ignition of gallium ferrocyanide. 



(4) Similar to (3), but with a smaller proportion of gallium. 



(5) Solar spectrum photographed on the succeeding day, the sun 



at the time being too low to show possible coincidences. 

 In (3) and (4), the gallium lines are beautifully reversed ; but in (4), the lines 

 are broad and the reversals much less marked. The reversed lines of gallium are 

 clearly seen to correspond with reversals in the solar spectrum ; but the reversals 

 may probably be those of iron lines very closely adjacent to those of gallium. 



Plate V. — (1) Arc spectrum of pure iron from ferrocyanide, witli the addition 

 of a gallium compound, on the middle portion only. The 

 solar spectrum is taken with the middle portion cut out. 

 (2) Arc spectrum of a small quantity of a gallium compound and a 

 small quantity of the iron also, with the solar spectrum 

 as in (1). 



In the oxyhydrogen flame, arc and spark spectra of substances both poor and 

 rich in gallium, the line 4172 is always stronger than 4033. 



By measurements of the iron lines and the gallium lines in arc spectra of 

 materials containing different proportions of the two metals, the wave-lengths of 

 the two gallium lines were determined by interpolation from the iron lines. The 

 wave-lengths of the latter used were those determined by Rowland in the solar 

 spectrum. By this method, the wave-lengths of the reversed gallium lines are 

 found to be 4172-214 and 4033-125. These numbers are higher than those 

 obtained by Lecocq de Boisbaudran in the spark, and higher also than our 

 measurements of the lines in the oxyhydrogen spectra photographed with very 

 small dispersion, namely, 4171-6 and 4032-7; but they have been verified to the 

 second decimal place by different measurements. 



In Rowland's Table of Solar Spectrum Wave-lengths, published in the" Astro- 

 Physical Journal," vol. i., pp. 139 and 225, there are two lines corresponding to 

 these; but, to judge of the probability of these lines belonging to the element 

 gallium, it is necessary to consider their relative intensities. Rowland measures 



