I think this edge-like centre has actually a line coincident with the line I refer to; but if so, its intensity little exceeds that of the band itself. _Spark over Water._ [Sidenote: Spark over water: spectrum described.] To complete the set of air-experiments, I examined the same spark taken from the surface of a small meniscus of water, placed in a glass cup upon the lower platinum wire. In this case the air-spectrum was plainly, but not brightly, seen at the violet end of the spectrum—the red, yellow, green, and blue being filled with a continuous spectrum, through which some of the air-lines faintly showed (see Plate XV. spectrum 7). _Phosphoretted-Hydrogen Flame._ [Sidenote: Phosphoretted-hydrogen flame.] This was obtained from a hydrogen-bottle fitted with glass tubing, two or three minute pieces of phosphorus being placed with the zinc. The flame was of a bright yellow colour, with a cone of vivid green light in its centre. [Sidenote: Spectrum described.] The spectrum was found to consist mainly of three bright bands in the yellow, green, and green-blue respectively (see Plate XVI. spectrum 3). [Sidenote: Mons. Lecoq de Boisbaudran’s remarks on the spectrum increasing in brilliancy when the flame is cooled.] The central band was very striking in its emerald-green colour, while all the bands were remarkable as being very broad in proportion to the slit (which, however, was not fine). The yellow band had a rich glow of colour. My spectrum was mapped out at ordinary temperature, and I found the bands sufficiently bright; but Mons. Lecoq de Boisbaudran, in his ‘Spectres Lumineux’ (texte, p. 188), has described how the brilliancy of these bands is increased when the flame is artificially cooled (_refroidie_). The idea of cooling the flame was due to M. Salet, who effected it either by a jet of water or by an air-blast. The less refrangible bands seem the most susceptible to increase of brilliancy. Mons. Boisbaudran also makes the important remark that the relative intensities of the bands are in such case altered, adding:—“La plus importante de ces modifications consiste en un renforcement très-considérable de la bande rouge δ 97·03 (W.L. 5994) qui devient vive de presque invisible qu’elle était en l’absence du refroidissement artificiel de la flamme.” Full details of the changes are given by M. de Boisbaudran. The bearing of these observations as connected with the variable character of the red line in the Aurora-spectrum seems to me in the highest degree noteworthy. _Iron-Spectrum._ [Sidenote: Iron-spectrum.] A comparison of this spectrum suggested itself, partly from the suspected relations between the Aurora and solar corona, and partly from a consideration of the views expressed by M. Gronemann and others in favour of the Aurora having its origin in the fall of an incandescent meteoric powder. [Sidenote: How obtained. Spectrum described. Mons. Lecoq de Boisbaudran’s spectra also given.]