BAND OF AMMONIA, AND ITS OCCURRENCE IN THE SOLAR SPECTRUM. 363 
The coincidences in the case of the central maximum as a whole are scarcely less 
striking. The majority of the solar lines in this region in fact appear to be due to 
ammonia, as will be seen from the table, where the solar lines from 3363'4 to 3358'0 
are tabulated in full in order to bring out this feature. In this small region there 
are 47 solar lines, at an average distance apart of 0'117A, and 37 ammonia band lines, 
at an average distance of O'149A. Thirty-four of the ammonia lines are either 
represented directly by reasonably appropriate lines in Rowland’s table, or fall upon 
solar lines of recognised or probable metallic origins. The absence of one of the 
remaining three lines is satisfactorily accounted for by its low intensity. In the case 
of the outstanding lines 3361'61 and 3359'89, the solar line corresponding to the first 
appears to be masked by dark ground extending between the adjacent solar lines, 
while the second is probably included in the nebulous line 3359'936. The coincidences 
are clearly too numerous and too systematic to be considered accidental. # Confirmative 
evidence of their reality is afforded by the identity of narrow bright interspaces in the 
two spectra, especially those at 3361'0, 3360'9, 3360'5, 3360'0, 3359'7, and 3359'5, 
which will be clearly seen on reference to the photographs, these appear to be produced 
by patches of continuous background which are common to the two spectra. The 
discussion of the central maximum may thus be considered to establish the presence 
of ammonia in the sun beyond all doubt. 
Analysis of the secondary maximum, occupying the region 3371 to 3378, leads to a 
similar conclusion, and there can be no doubt that the majority of the coincidences of 
lines of ammonia with faint solar lines in this part of the spectrum also have a real 
significance. 
As regards the groups of three, forming the series a to £, there is also a close general 
agreement, and the few irregularities may well be caused by imperfect estimates of 
the wave-lengths and intensities in the two spectra, or by the approximate super¬ 
position of lines of other substances. If it were not for interference by metallic 
lines in the case of the sun, a good test would be provided by the series investigations. 
Supposing the coincidences to be genuine, the wave-lengths of the solar lines should 
show the series relations with the same order of accuracy as those of the ammonia lines 
themselves, and the intensities of the lines should be consistent with the series 
relationship. This test cannot be completely applied to any one of the six series, on 
account of near coincidences with metallic lines, but it may be worth while to take 
the chief lines of the $ series as an illustration. The facts with regard to these are 
collected in Table IV. 
It will be observed that while three of the metallic lines (3341'967, 3320783, 
* In order to get a rough idea of the proportion of coincidences which might be merely accidental, the 
37 ammonia wave-lengths in the region 3363 • 5 to 3358 ■ 0 were compared with the 47 solar lines from 
3563 • 5 to 3558 - 0. The total number of approximate coincidences was 14, as compared with 34 in the 
true region, and of these, 5 were with lines ■ assigned to metals. There was no systematic agreement 
of intensities in this case. 
