_ 
END-ON VIEWS OF GAS SPECTRA UNDER HIGH DISPERSION. 155 
direct comparison of the new measures in the latter kind of wave-numbers (entered in the 
fourth and translated into metrical wave-lengths in the first column) with the chief linelet’s 
wave-positions actually recorded in the Swedish observers’ Memoir. The metrical wave- 
lengths of AncsTr6m’s Table have also been translated accurately into British wave-numbers ; 
but they are only preserved in their original form (for the purpose of admitting a direct com- 
parison) in the case of the leading-lines a of the triplet groups. Among the other triplet- 
lines, comparisons of metrical wave-lengths are for brevity omitted, and only the equivalent 
values are retained in wave-numbers to an inch. Throughout all the comparisons presented in 
the Table, ANcsTROm’s and THALEN’s measures in their metrical or British values occupy the 
upper line in the space accorded to the position-cbservations of the several doubly-measured 
groups. The next, or middle line in duplicated groups, gives the measurement with the aurora- 
spectroscope; and under both are written average values of the two independently observed 
positions. Finally, the interval of the average value of each subordinate line’s wave-number 
from that of its leading triplet-line, is inserted in the next following column contiguous to it, 
so that the regularity or degree of variation of these intervals in the successive groups and 
sections of the spectrum can be apprehended at a glance. The last of the intervals, on the 
right hand side of the Table, denotes the increase of wave-number from the first line of one 
linear group to the first line of that following it (repeated in position for this purpose in 
the next preceding column), or devotes the distance in wave-numbers between the first lines of 
successive groups. 
A mere inspection of the Table shows that while all the other intervals which it presents 
are sensibly invariable, or practically constant, the latter interval between the leading lines of 
successive groups is a constantly decreasing one. There are large and marked steps of this 
decrease at the 4th and 16th groups of the Table, followed by a rapid fall for one or two groups, 
and then a nearly constant interval for a long period of six or ten groups afterwards. In the 
remaining half of the flutings mapped by AncsTrOm and THALEN, but not now remeasured as 
they merge into the carbon-citron baud, the same phenomenon is presented. At one or two 
groups after the carbon intrusion, the group interval falls abruptly from about 360 to 300, and 
remains at the average value of 295 for the remaining fifteen or sixteen groups to the green end 
of the series. Throughout the whole range of about forty groups mapped by Anestr6m, the 
interval ac, nevertheless preserves a constant average value of 163 English wave-numbers 
between the two peaks of the double-notched serration. As far as ANastROm’s and THALEN’s 
groups are remeasured in the present Table, the common averages of the subordinate intervals 
of them are placed at the foot of their columns; and it appears that the line-intervals ab, ab’ on 
the first slope have an arithmetical progression from the edge of the slope, or a common interval 
from line to line of about 52:3. The interval ac is perhaps not conformable to the progression, 
as although the tendency in measuring a bright edge is to place the pointer rather far upon it, 
and the excessive value of the interval ac=163°5, shown in the average of the Table, might in 
this way possibly be accounted for, yet a different account of the discordance may perhaps be 
given from the following consideration: while in the remaining eight or ten complete ranks of 
Anesrrow’s list (following those here remeasured) the interval ac has a pretty constant value, 
as before, of about 162, the interval ad, though very uniform, has an average value of only 42°5, 
and this is not more conformable to the space of 162 or 163, than the former common difference 
of 52°3; while it differs entirely from that regular interval in the earlier portion of the 
spectrum. 
It appears probable, therefore from this review, that the two notch-edges of the nitrogen 
