944 REPORT—1885. 
number unit. Its procession remains in sensibly exact agreement with this rule 
to the seventeenth and eighteenth lines, still visible, and measured in the map. 
But it grows quickly fainter from the sixth line onwards, and presents signs of 
division and duplicity in its lines from about the ninth line onwards. 
The other rank of lines (which completes the whole group) is identical in its 
intervals with the first, but its lines are not double, and, fading more slowly in bright- 
ness than those of the other rank, they form in the open part of the band its strongest 
array or main-line troop. Almost exact coincidence occurs of its tenth line 
with the eieventh line of the other rank, and the line intervals being there pretty 
wide, an appearance of crossing of the two series and of condensed brightness in the 
nearly doubled line forms the easily recognised and marked feature of the band’ 
already mentioned, not very far from its bright edge. Nearer to the edge than this 
the lines of the two ranks, nearly equal to each other in brightness, are unaccordant 
and intermixed in place with little apparent signs of orderly relation. But another 
crossing-point yet happens here (an arithmetical consequence of the former one),. 
where the two ranks have a spectral line in common with each other. This is the 
leading line of the second or single-line rank, which coincides and is, as far as the 
observations can determine, identical with the fifth line of the projecting band-edge- 
or split-line rank. 
‘The new common line of the two progressions is not brighter by superposition 
than adjoining lines, preceding and following it, of the two ranks in which it falls; 
but the second rank of lines takes its origin in the fifth line from the band-edge or 
front line (inclusive) of the other rank, and being thus four intervals or ten unit- 
spaces (1+2+3+4) distantfrom that edge, all its lines fall thus far in advance of 
those corresponding to them in the other leading or projecting file. It thus happens 
that its tenth line, instead of coinciding with the other file’s tenth line, is advanced 
ten unit spaces of the progression from it, and falls into agreement, therefore, with 
the eleventh line of that rank, which is also ten unit spaces distant from the last 
preceding or tenth line of its serial progression, as the two lines are found to do 
quite accurately at the crossing-place of the band. For in the faithfully depicted’ 
map, the single or main line at this point exactly hides or takes the place of one of | 
the two half-lines, which together should form a close pair of the split-line spectrum 
at that place. Both the great accuracy of the observations and the perfect correct-- 
ness of the presumed law of line-interval succession seem to be alike proved, and 
established satisfactorily by this unintentional and yet most remarkably exact 
agreement. 
The two partial linelet spectra of this band, as far as they were measured, agree 
with each other when superposed rather more perfectly than with the theoretical 
progressive points computed for them on the micrometric scale of the map; but 
this is to be expected, as the scale of prismatic dispersion, or of micrometric parts, is 
not identical with that of wave-numbers, in relation to which alone the rule of 
succession of the lines in intervals forming an arithmetical progression is in all 
probability exactly true. But the real identity of the formula, whatever it may be, 
for the two partial spectra is shown quite free from any such deceptions by con- 
fronting a copy of the lines of one with the lines of the other spectrum on the 
map, and observing the almost perfect superpositions which this yields, in each 
and every line. It seems allowable to speculate whether the occasional choice of 
valency which chemists find it necessary to ascribe to carbon, between bivalency 
and tetravalency in its varying power of forming saturated compounds, may not 
be connected with this twofold repetition, which here certainly exists, of one 
and the same rank of spectral lines in a single spectral band, thus shown by one of 
that element’s abnormally saturated compounds under very vigorously electrified. 
conditions, 
