160 
ON THE ATMOSPHERIC LINES OE THE SOLAR SPECTRUM. 
the latter sank behind the former ; so that in the first autumn of my observations the 
appearance of the haze obliged me to close work for the season. Subsequently it 
occurred to me that the gradual growth in height of this haze gave me exactly the very 
test I required, viz. sunset varying one day with another from a depression of 1^° to an 
altitude of some 3^°. Accordingly I watched the corresponding effect on the air-lines, 
and found beyond all question that as the bank rose and the corresponding sunset 
occurred higher, the variable and air-lines all disappeared, each in its turn. This test 
is of course most effectually applied to lines which require the lowest of sunsets to be 
developed, and the behaviour of all lines is by no means the same. For instance, 813 is 
almost as good as a clock to me, commencing to change so early as 2 or 3 p.m. ; whereas 
712 (which is, in fact, the more prominent line eventually , and is, I believe, noticed here 
for the first time) hardly presents the smallest change until the sun is under 1° of 
altitude or thereabout. Similarly, my air-wall (above noticed) requires a low sunset, 
but not so low as 712. The test just explained gave a visible connexion between the 
atmospheric lines and the terrestrial atmosphere ; i. e. the higher the sunset, the more 
the air-lines were absent. I state the fact thus briefly, notwithstanding that I tested 
it day after day, and that I possess abundance of notes on the subject; these notes, 
however, are in the main repetitions, which have no interest once the fact has been 
announced, and I therefore refrain from transcribing them. 
7. It will be seen that in the portion extreme red to D there are some slight discre- 
pancies between my map of 1870 and my present map; these are solely due to the far 
greater powers of the new spectroscope, and not to any want of care in preparing the 
earlier map : the additions are chiefly due to the same cause, and to some extent are the 
results of greater skill, which experience may have brought me. Amongst the new 
lines or bands are group 315-352, the additions about A, 460, 730, 950, and else- 
where, not forgetting 712 (which, I repeat, is a very prominent air-line, but only so at a 
very low sunset). For further discussion of the map now submitted to the Royal 
Society, and for comparison with other maps which have preceded it, I must await a 
more favourable opportunity — merely remarking at present, that while looking for air- 
lines I happened to detect a few other lines which do not vary, but which are not given 
in Kirchhoff’s map; of this class are 1006, the pair 1310, and some others. I may 
repeat my conviction, already stated in my paper dated 25th April, 1870, that besides 
other changes in the light, as the sun approaches the horizon, there is this peculiarity, 
that rays of less refrangibility become visible, so that the spectrum appears to be extended 
towards the red end. My search, as will be seen from the map, has not as yet extended 
rigorously beyond F ; indeed examination of the spectrum beyond this line is hardly 
practicable for the detection of air-lines without some additional provision for collecting 
light, which, however, I think I could contrive ; with my present means, but little light 
reaches beyond F when the earth’s atmosphere intervenes to a depth which may be 
expected to produce an effect, the brightest part of the spectrum being the portion that 
is last visible at sunset. 
