THE ULTRA-VIOLET SPECTRA OF THE ELEMENTS. 209 
Indeed, the measurement of plates taken successively without movement of any part 
of the apparatus, except the photographic slide, showed that there was such a dis¬ 
placement of the axis, and that it might make an error of fit 10" in the measurement 
of the angle, or dr'13 on an average in the value of the wave-length. 
Another source of error is the want of sharp definition of some of the lines. Some 
of the lines are really diffuse, and in every case, as already observed, if the image is 
correbtly focussed on the plate when the telescope is on one side of the collimator, it is 
always a little out of focus when the telescope is moved round to the other side. Hence 
there may easdy be an error in measuring the distance between the lines which may 
easily amount to dr8” of angular measure, or dz’l in the value of the wave-length. 
Still we do not think the probable error exceeds dz'25 of a tenth-metre. 
Determination of the intermediate lines. 
The wave-lengths of a sufficient number of lines of reference having been measured 
by the grating, the intermediate lines have been mapped by means of prisms. For this 
part of the work a calcite prism of 30°, cut so that one face is perpendicular to the 
axis of the crystal, was fixed to the end of the collimator, and a similar prism to the end 
of the telescope, while between them another calcite prism of 60°, cut so that the faces 
are equally inclined to the axis of the crystal, was maintained by a simple system of 
linkage at the position of minimum deviation, which is also that of single refraction. 
Photographs were taken at short intervals all down the scale of the spectrum of the 
arc and spark simultaneously. The image of the arc was focussed on the slit by the 
quartz lens already mentioned, and thrown just under the centre. At the same time 
the spark was made to pass horizontally close in front of the slit, without the inter¬ 
position of a lens, but just above the centre. In this way two images were impressed 
on the plate overlapping one another in the middle. The distances between the lines 
was afterwards measured under the microscope, and the inverse squares of the wave¬ 
lengths of the intermediate lines deduced by the graphic method of interpolation 
between those of the lines of reference. A table of inverse squares was used for the 
reduction to simple wave-lengths. 
The following table gives the results. In many cases there is much difference in the 
relative intensities of the same line in the arc and spark, and in some cases lines are 
visible in the photograph of the arc which are not in that of the spark, and vice versd . 
Beyond the wave-length 2327 no spark lines seem to have made any impression on 
the plates, but the arc lines continue with, however, a sensible falling off in intensity 
up to the end of the region observed. For this region the copper lines of reference 
were used, as already explained, and for the highest part of it, above wave-length 2230, 
quartz prisms were substituted for those of calcite with advantage as regards the 
amount of light transmitted, but with some loss of dispersion and more of definition. 
In the following table the second column gives the wave-length, and the lines 
MDCCCLXXXIIL 2 E 
