254 
ME. J. NORMAN LOCKYER ON SPECTRUM-ANALYSIS 
sodium lines and exactly occupying the centre of each. Next, this thin black line was ob- 
served to thicken at the top, where the spectrum of the lower denser vapours was observed, 
and to advance downwards along the D line, until arriving at the bottom they both became 
black throughout ; and if now the heat was still applied, thus increasing the density of 
the various layers of the sodium vapour, the lines began to broaden until, in spite of 
considerable dispersion, the two lines blended into one. The source of heat being now 
removed, the same changes occurred in inverse order ; the broad band split into two 
lines, gradually the black thread alone was left, and finally that vanished, and the two 
bright lines were restored. 
(2) This experiment was then varied in the following way. Some pieces of metallic 
sodium were introduced into a test-tube, and a long glass tube conveying coal-gas 
passed to the bottom, an exit for the gas being also provided at the top. The sodium was 
now heated and the flow of coal-gas stopped. In a short time the reversal of the D lines 
was complete. The gas was now admitted, and a small quantity only had passed when 
the black lines were reduced to threads. 
In my former communications to the Royal Society I have pointed out the extreme 
importance of these facts in connexion with solar and stellar physics. In observing the 
sun by the new method, we get various Fraunhofer lines thickened in the spots and 
thinned in the chromosphere and prominences ; and in these latter, in some instances, 
notably in the case of F, we find the lines gradually widening as they approach the limb 
of the sun. 
While this may be remarked as a solar demonstration of the correctness of the 
conclusion at which Dr. Frankland and myself had arrived, it is to be noted that bright 
line prominences may occasionally be seen on the sun’s disk over or near spots in the 
spectrum of which the same lines are thick, while this phenomenon could not exist if 
the thickening of the lines were due to temperature alone. 
Method employed *. 
The method of observing spectra to which I have already referred, and which has 
been adopted in the work of which I now propose to give an account, consists in 
throwing an image of the spark on the slit of a spectroscope in the laboratory experi- 
ments in exactly the same manner in which I proposed, in 1866, that an image of the 
sun should be thrown on the slit in order to spectroscopically examine minute portions 
of the sun and his surrounding atmosphere. 
It is obvious that in this method the image of the slit will be associated in the 
spectroscope with an image of a section of the spark, and that if from any cause there be 
various shells of vapour surrounding each pole, which shells give different spectra, then 
these spectra will be sorted out so that their variations may be traced from pole to poie. 
* This method was first exhibited at a lecture at the Royal Institution, April 2nd, 1870. The same method 
has more recently been employed with great success by M. Salet in a research on the spectra of the metalloids. 
