(577 ) 
exhibited the anomalous dispersion. On either side of the two dark 
sodium lines the originally horizontal spectrum-band was boldly 
curved, so that for rays with wave-lengths, slightly larger than 
Ap, or Ap, the sodium vapour appeared to possess an index of 
refraction rapidly increasing in the neighbourhood of an absorption- 
line; whereas for rays of wave-lengths, slightly smaller than Ap, or 
Ap, the index of refraction rapidiy decreased when approaching the 
absorption-lines. The amount of the anomalous dispersion near Ds 
exceeded that near Dj. 
In repeating this experiment I obtained materially the same results. 
Moreover, I noticed a peculiarity in the phenomenon, not mentioned 
by BECQURREL, and not exhibited in the diagrams accompanying 
his paper. BrCQUEREL states, that when he introduced a flame, 
rich in sodium, the lines D, and D, appeared 
te J ( as broad, dark bands, and that on either 
de DD side of both bands the spectrum was curved. 
According to his diagrams these displace- 
ments only refer to light, outside the bands; 
the rays inside this region, in the more 
immediate neighbourhood of the D-lines, are 
NN totally wanting. Fig. 1 refers to a pris- 
ee matic part of the flame, edge upwards; 
fig. 2, to a prismatic part, edge downwards. 
Both cases represent the image as seen in a telescope, so, reversed. 
Fig. 1 and 2. 
I myself, however, have observed the phenomenon in the form of 
i fig. 3. The dotted lines 
indicate the places of D, 
and D,. When the electric 
light is intercepted by 
means of a screen intro- 
duced between the flame 
and the horizontal slit, 
the D-lines appear in those 
places as two faintly lumi- 
nous, sharply defined slit- 
images. The light is faint 
because the flame is placed 
at a distance of more than 
70 cms. from the vertical 
slit, and its radiation is 
all but intercepted by the 
42% 
ee wee ee ee 
