( 578 ) 
adjustable diaphragm, which allows only a beam of a cross-section 
of about 0.2 em?. to enter the teiescope lens. 
When next the arc light is allowed to cross the flame, the spec- 
trum of fig. 3 appears with such intensity, that the bright sodium 
lines are undistinguishable in the centre of the dark bands. In the 
upper and lower parts of the field of vision however, they can yet 
be seen as continuations of the four bright arrows of light which 
are, as it were, flashed forth from the horizontal spectrum into the dark. 
By repeatedly intercepting and re-admitting the light of the main 
source, I have actually convinced myself that the intense arrow-light, 
with the dispersion used, gradually passes into the faint light of the 
emission-lines, both with respect to intensity and place in the spec- 
trum. A flat ROWLAND grating with 47000 lines was used in the 
spectroscope; one spectrum of the first order being extremely brilliant. 
The crosswires of a micrometer eye-piece (65 divisons of which cor- 
responding to the distance of the D-lines in the first diffraction spec- 
trum) were repeatedly adjusted as close as possible to the extreme 
part that was yet distinctly visible of such an arrow, the so- 
dium lines of the flame being all but invisible. I next removed the 
diaphragm near the flame, intercepted the main light, so that the 
sodium lines now became clearly visible, and took a number of the 
readings of the emission line. The mean readings of two series of 
observations did not mutually differ by one division; the arrow, 
therefore, approached the D-line to within 0.01 uu. 
From the data furnished by BreeQuereL (C. R. 128. p. 146) it 
can be inferred, that the distance between the D-lines and the most 
deflected parts of the arrows upon which, in his experiments, the 
cross-wires could still be adjusted, was on an average greater than 
Ol wee. 
I am not quite sure how this difference in the results must be 
accounted for; perhaps BecQuEREL’s flame contained more sodium 
than mine; anyhow so much sodium is not wanted to produce strong 
anomalous dispersion. 
The following experiment convinced me how narrow was in reality 
the absorption-region of each of the sodium-hnes. An additional lens 
of 20 em focal distance was placed between the telescope lens and 
the vertical slit, in such a manner that on this shit was thrown 
the image of the prismatic part of the sodium flame, and not that 
of the horizontal slit, as before. In this image, therefore, all rays 
that had passed the flame and had been refracted in different direc- 
tions, must be found re-united. The absorption-lines were now actually 
ery narrow, the emission-lines in some places all but covering them. 
