564 
MR. J. NORMAN LOCKYER ON SPECTRUM-ANALYSIS 
Two curves (A and B) were employed in the construction of this section. The 
first curve (A) was obtained from an enlargement on glass, and was available from 
about W.L. 386'0 to 399 - 98 millionths millim. The distance between K and H in 
this photograph is 3'5 centims., and it was employed in preference to Rutherfurd’s 
print, because being in much better focus it was more easy, in consequence, to compare 
it with Cornu’s map. 
The second curve (B) was constructed from Rutherfurd’s print, commencing at 
the point where the photograph referred to in the last paragraph fails ; this curve 
extends from W.L. 397'045 to W.L. 400'925 millionths millim. 
The wave-lengths are expressed in x th metres, and are given in two places of 
decimals. This has been found necessary in order to distinguish between lines very 
close together. The greatest error in the method of determining wave-lengths by 
graphical interpolation does not much exceed 2 millims. of the present scale millim. 
Cornu’s scale = _ . n millim. in the actual wave-length. 
20 , 000,00 0 O 
I have not thought it necessary to reproduce the curves, but at the end of the 
paper I give two tables which will show the accuracy which it is possible to secure by 
the method above described. The experience gained in constructing these curves 
(I had already completed the region W.L. 390-440 millionths millim. before I received 
the grating, of which mention has already been made, from Mr. Rutherfurd) leads 
to the following conclusions :—- 
I. The photograph of the solar spectrum employed should be on as large a scale 
as possible. 
II. The wave-length scale giving the abscissae should never be smaller than that 
used here. 
III. The photograph should be tested for distortion after enlargement. 
IV. The fundamental lines should be copied from the glass and not from an 
enlargement on paper. 
ft. Construction of the Map. 
The photograph used for the construction of the original refraction map is the 
original negative varnished and examined by a lens, or placed under a simple micro¬ 
scope of low magnifying power. Considerable loss of detail and sharpness, leading 
to the obliteration of faint lines and the fusion together of close groups, has been found 
to be an invariable effect of photographic enlargements. Thus in section W.L. 390-400 
millionths millim., there were mapped 333 lines in the refraction map as completed, 
whereas in the map of the same section exhibited to the Society in November, 1875', 
which was drawn from an enlargement, there were but 221 lines. I may add that 
the same holds good with metallic spectra. Thus the spectrum of cerium as first 
mapped from enlargements contained several broad and nebulous bands which the 
negatives now in use resolve into groups of distinct but closely-packed lines. 
