Vol. 6, 1920 
PHYSICS: A. S. KING 
65 
occur with the furnace in vacuo, so that the continuous spectrum cannot 
be regarded as resulting from a high-pressure condition. The possibiUty 
of producing an absorption spectrum accompanied by the suppression 
of many lines which are being emitted strongly at this temperature must 
be considered when dealing with an extended mass of vapor. 
A STUDY OF THE EFFECT OF A MAGNETIC FIELD ON ELEC- 
TRIC FURNACE SPECTRA 
By Arthur S. King 
Mount Wii^son Observatory, Carnegie Institution of Washington 
Communicated by G. E. Hale, December 9, 1919 
Investigations on the splitting of spectrum lines by the magnetic field 
have employed, except in rare instances, the electric spark as the source 
of light. It has proved impracticable to maintain an arc between the 
poles of a magnet, though an apparatus giving a succession of flashes 
has been employed. The flame, because of its weak luminosity, has but 
limited usefulness. The electric furnace, if its size does not prevent its 
use in the magnetic field, will evidently do work not at present taken care 
of. While the furnace does not give the enhanced lines peculiar to the 
spark, the other lines shown by the arc and spark occur for the most part 
in the furnace spectrum. In addition, the furnace is found to show a 
large number of lines which are produced only with great difficulty by either 
the arc or the spark, and for which we have as yet no data as to their mag- 
netic separation. A further advantage over the spark for the lines common 
to both is the extreme sharpness of furnace lines when the apparatus is 
enclosed in a vacuum chamber, a feature which should add materially to 
the definition of the Zeeman components. 
Pending the construction of a more powerful apparatus, a simple tube 
furnace was arranged for use in a magnetic field. A graphite tube, 10 cm. 
long, was placed axially between the poles of a Weiss electro-magnet. 
The jacket enclosing the tube and the contact blocks at the ends were 
water-cooled. A field of 6500 gausses separated the n components of most 
lines sufficiently to permit measurements of fair accuracy and to show the 
characteristic features of the source. The spectra of iron and vanadium 
received the chief attention, and measurements were made for a con- 
siderable number of lines having well-defined components. A comparison 
of lines common to furnace and spark showed no difference either in the 
number and arrangement of components, or in their separation in the two 
sources, the very different excitation in the two cases appearing to have 
no effect on the magnetic characteristics. The furnace and spark can 
thus be used to supplement each other in studying the magnetic behavior 
of all classes of lines. As was expected, separations were obtained for 
