BWS,
Ce thn, } Spectrum of Carbon. 155
’ der into fused potash, and subsequent treatment with aqua
T H E C H E M I C A L N E Ww S. regia, by prolonged ignition in a current of chlorine, and
: by treatment with hydrofluoric acid. The well-washed
powder was afterwards compressed into blocks by hydraulic
pressure between platinum plates, and from these blocks
the electrodes employed were cut. Notwithstanding the
purification the photographs of the spark between these
zle@irodes still showed very distinctly lines of magnesium
and iren. This fact shows the extreme difficulty of getting
rid of all impurity, and the caution which is requisite in
any reasoning depending on the assumption of chemical
purity in the materials employed. It is very possible that
:he magnesium and even the iron in this case may have
seen due to oxides of those metals in the floating dust of
‘he laboratory, which we know always contains sodium
compounds, and which at Cambridge—where the water,
soil, and bricks contain sensible quantities of lithium—
almost always show traces of that element.
The wave-lengths of the strongest carbon lines were
letermined by means of a Rutherford diffraction grating
naving 17,296 lines to the inch. The measures were made
-n the following way :—The collimator and telescope of
the goniometer were first centred by the instrument
naker’s marks. The telescope was then more carefully
adjusted for centre by directing it on to a distant mark,
taking the reading of the circle, turning the arm carrying
the telescope through 180° and reversing the telescope,
whereby the mark was again brought into the field of
view, and adjustments were then made until the mark had
the same position on the cross-wires in both positions of
the telescope. The grating was next placed in position,
and, after adjustment to the vertical plane, was brought
very nearly at right angles to the axis of the collimator by
turning it until the sodium D lines in the spe@ra of the
second order were observed to fall at equal distances on
sither side of the collimator. The small photographic
slide, containing the sensitive plate, fitted the telescope
in place of the eye-piece, and so could easily be turned
about an axis coincident, or nearly so, with the optic axis
of the telescope. In taking a measurement of the poeition
 of a line the approximate wave-length was first found
by interpolating between the nearest cadmium or other
lines of known wave-length in photographs taken with
calcite prisms. The telescope was then set to the angle
corresponding to this approximate wave-length for the
spectrum of the fourth order. The lower half of the slit
was closed by a shutter, and, the photographic slide having
been adjusted for level, the plate was exposed to the light
which came through the upper half of thg slit, and gave
an image of the lines in the lower half of the field. When
this exposure was completed the photographic slide was
turned round through 180° about the axis of the telescope,
zo as to bring to the top that part of the sensitive plate
which -had been before lowest. It was then exposed a
second time, and thus two images of the same line were
:mpressed on the plate, which were necessarily at equal
distances on either side of the point where the axis of the
telescope met the plate. By a subsequent measurement
with a micrometer under a microscope of the distance between
 the two images, and the conversion of this distance
into angular measure, a connection was found, which was
added to or subtracted from the reading of the circle to
get the exact deviation of the ray producing the line under
observation, Another photograph of the same line was
aext taken in the same way as before, except that the
telescope was placed at the corresponding angle on the
sther side of the collimator. From the two angles thus
found the wave-length of the line was calculated. The
process was repeated three or four times for each line, and
the mean wave-length thus found for carbon lines were
2297'4, 2478°4, 25087, 2511'6, 2836°3, and 2837°3. The
numbers deduced from the different photographs of the
same line differed from one another in the last figure only,
so that we are justified in assuming the first four figures
to be accurate in each case. The wave-lengths of the
remaining lines were obtained by interpolation from mea-Tur

 spectroscopic investigations we have communicated
to the Society * On the Reversals of the Lines of Metallic
Vapours ”” have shown the importance of a thorough and
accurate knowledge of the ultra-violet spe@ra of the ele.
ments, for it is in the lines of short wave-length as a rule
ihat the greatest emissive power is manifested, and they
are therefore most readily reversed. Thus we have suc.
ceeded in reversing upwards of a hundred lines in the
altra-violet spectrum of iron (Proc. Roy. Soc., vol, xxxii.,
p. 404). The necessity for accurate data in regard to this
region of the spectrum led us to make a long study of the
spectrum of magnesium, and the results of this investigation
 appeared in the volume of the Proc. Roy. Soc. just
cited. Having had occasion to examine the origin of the
different fluted spectra of carbon, it became apparent that
a complete knowledge of the relations of these spectra to
the simple spectrum of the element could only be reached
by the help of a complete record of the line spectrum.
Angstrdm and Thalén, in their memoir ¢* On the Spedra
of the Metalloids ” (Nova Acta Reg. Soc. Upsal., Ser. 3,
vol. ix.}, give a map and table of wave-lengths of the
lines due to carbon in the visible part of the spectrum, as
distinguished from the fluted spectra given by compounds
of carbon,—namely, carbonic oxide, cyanogen, and acetylene.
 These. lines, they state, always appeared when
very powerful indu&ion sparks were passed through the
vapour of any compound of carbon, or between carbon
ele@trodes. This line spectrum is remarkable for simpli.
tity, consisting of eleven lines, of which the single line in
the yellow, followed by a triple group in the green, and a
very strong line'in the blue, recall vividly the spe@rum of
magnesium ; and as we know two modifications of the
spectrum of magnesium which seem to be due respectively
to the oxide and a hydride, the parallel between the behaviour
 of the two elements is the more striking. The
plates of the ultra-violet spedra of the metals by the late
Prof. W, A. Miller (Phil. Trans., 1864) include plates of
the spark taken between metallic ele@trodes in different
compounds of carbon, which show with sufficient clear.
ness that there are some five groups of lines in the ultraviolet
 spedrum of this element. In the observations here
described we have preferred taking intense indu&ion
sparks between pure graphite poles in different gases.
A figure accompanying the paper represents the ultra.
violet spectrum of carbon to a scale of wave-lengths within
the range of the rays transmitted through calcite. The
lines figured have been observed in photographs of the
spark of a large induction coil, having a large Leyden jar
In connection with the secondary coil, between poles of
purified graphite in air, carbonic acid gas, hydrogen, and
coal-gas. The same lines have been observed. in photographs
 of the spark between iron, and between aluminium
poles in carbonic acid gas. By comparing the photo-8raphs
 taken under these different circumstances, we have,
we believe, eliminated the air lines, which are numerous
and strong in the region between H and T, and will form
the subject of a future communication, and also the me:
tallic lines which graphite, purified with the utmost care,
still exhibited,
The graphite was purified by being stirred in fine pow:
By rr AW muy SURAT TR po
gg, Drag of a Paper read before the Royal Society, March gth