462 
PHYSICS: A. S. KING 
on the two sides of the line. This method was employed by Wood in 
studying the anomalous dispersion of sodium vapor, the prism being 
formed by vapor enclosed in a horizontal steel tube heated from below, 
the upper portion being cooled. 
According to a theory advanced by Julius, when anomalous dispersion 
is active, two vibrations of nearly the same period affect each other in 
such a way as to cause a mutual repulsion between the corresponding 
spectrum lines. It seemed possible to make laboratory tests which 
would set at least an upper limit to the amount of this effect. For this 
purpose, as well as to study the regular phenomena of anomalous dis- 
persion, the writer has recently adapted the tube resistance furnace for 
the production of these effects. A beam of white light from a carbon 
arc outside the furnace chamber was passed through the furnace tube 
containing the metallic vapor to be studied. A strong density gradient 
in the vapor was produced by cutting away the upper portion of the 
tube and passing a water-cooled pipe above it. Examination under 
high dispersion with a plane-grating spectograph gave a means of 
studying the relative anomalous dispersion produced by different spec- 
trum lines, and, by the use of mixed vapors in the furnace tube, to test 
the constancy of wave-lengths of lines when alone and when close to a 
line showing high anomalous dispersion given by another vapor. 
A study of the iron spectrum over a considerable range of wave-length 
gave strong anomalous dispersion effects for those of the stronger lines 
which reverse easily in the arc and other sources. This feature appears 
to be general for the spectra which have been examined. It was found 
further that a certain temperature of the absorbing vapor is required 
to give the most efhcient prism for a particular type of line. Thus the 
calcium line X 4227 and the chromium lines X X 4254, 4275, 4290 can all 
be made to show very strong anomalous dispersion, but the best tem- 
perature for the calcium prism is lower than that for chromium. This 
can be well shown by mixing the two vapors and causing each in turn to 
show anomalous dispersion. When the temperature is adjusted to give 
maximum anomalous dispersion for chromium, the effect usually dis- 
appears for X 4227 of calcium, leaving only a wide absorption line. In 
several experiments, however, a high temperature condition has given a 
curvature of the spectrum adjacent to X 4227 in the opposite direction 
to that prevailing for the chromium lines at the same time, and also the 
reverse of what the calcium line itself shows at a lower temperature. 
This indicates an inversion of the prism of vapor producing X 4227, due 
probably to the cooler vapor above giving the line more strongly than 
the highly heated region below. 
