362 PROF. A. FOWLER AND MR. C. C. L. GREGORY ON THE ULTRA-VIOLET 
full. A direct comparison of the two spectra in the neighbourhood of the central 
maximum is given in fig. 5, Plate 1, where the ammonia spectrum is reproduced as a 
negative in order to represent its appearance in absorption. 
The comparison is complicated by the approximate coincidences of many of the 
ammonia lines with lines of metallic origin in the solar spectrum, or with unidentified 
lines which have intensities too great to allow of their being assigned to ammonia alone. 
As Rowland’s limit of resolution appears to be about 0'04A, ammonia lines may 
evidently be masked in this way by solar lines showing considerable differences in 
wave-length. Direct evidence of the presence of all the ammonia lines in the sun, 
such as would be afforded by identity of wave-lengths throughout, is therefore not to 
be expected. 
There is also some uncertainty as to the completeness of Rowland’s list of wave¬ 
lengths, and as to the uniformity of his estimates of intensity. In this connection, a 
photograph of the solar spectrum in the Higgs’ collection of the Royal Astronomical 
Society has been to some extent utilised as a general check on Rowland in the region 
of the ammonia band. While the tabulated intensities on the whole were closely 
confirmed, there are several lines for which Rowland’s estimates appear to need 
revision. Attention is drawn to some of these in the column of remarks in Table V. 
As regards the wave-lengths, the relation between the international scale and the 
scale of Rowland is by no means simple. Comparison of the solar lines with the 
positions on the international scale given for the iron and nickel lines by Burns 
shows that whilst the average difference in the region of the ammonia band is about 
0‘14A, there is no consistent agreement among the different lines. Deviations from the 
mean in a selected region frequently amount to nearly O'OlA, and there are occasional 
variations of 0'02A and upwards. These irregularities are not necessarily due wholly 
to errors of measurement, but may also be caused by differences in the effective level 
at which the various lines are produced in the solar atmosphere. It would seem, 
however, that if due regard be paid to intensities, coincidences within 0'02A may not 
be without significance. 
Notwithstanding the difficulties affecting the comparison, there is abundant 
evidence that the ammonia band is present in the solar spectrum. The most 
convincing proof is perhaps afforded by the strongest part of the central maximum, 
extending from 3360'45 to 3360'08 (Rowland’s scale). As will be seen from fig. V., 
Plate 2, and from Table V., there is a complete correspondence of the solar and 
laboratory spectra as regards this group, except that the line 3360'45 may be slightly 
reinforced in the sun by a line of nickel. # This agreement is emphasised by the 
presence of a background of dark continuous spectrum which is sharply bounded in 
each of the two spectra by the outer two of the five lines involved. There is a similar 
dark ground covering the adjacent group 3360'82 to 3360'63, which is also clearly 
common to the two spectra. 
* A nickel line in this position is given by Rowland, but not by Exner and Haschek. 
