296 
PHYSICS: ST. JOHN AND BABCOCK 
the light should not be taken from a point less than 2 mm. from the 
pole of the arc. The astigmatism of the concave grating in the usual 
mounting introduces more or less pole effect and to that degree vitiates 
results involving the lines under consideration. The practice of re- 
versing the current in the arc in order to overcome the tendency to 
produce wedge-shaped lines, when the slit and the axis of the arc are 
parallel, obscures but does not eHminate the pole effect. Since the 
redeterminations aim at a precision of 0.002 to 0.003 angstrom, it is 
important to take the pole effect into consideration. 
4. Lines of the type considered are not Hmited to iron but occur 
in the spectra of other elements, the detailed investigation of which is 
necessary before safe deductions can be made from their use in astro- 
physical investigations. 
Bearing upon the explanation of the pole effect, our recent investiga- 
tions yield the following results. The absence of a general increase 
in pressure at the negative pole, evidenced by the unchanged wave- 
length of whole classes of lines known to be affected by pressure, led 
us to suggest an increased density of the radiating vapor as a possible 
cause, but our measurements of furnace spectra taken with a 10 fold 
increase in the quantity of iron vaporized show no change of wave- 
length for these Hnes. Furthermore, the lines of manganese appearing 
as an impurity in the iron poles show displacements of the same order 
as neighboring iron Hnes; it is difficult to imagine a high density, even 
at the pole, for manganese vapor appearing only as a trace. 
The effect of temperature was examined by comparing the wave- 
lengths of these lines at temperatures as widely different as is practi- 
cable in the furnace (2100°-2600°C.). No evidence was found of any 
dependence of wave-length upon temperature. 
As the result of a study of the behavior of these lines in an arc in 
vacuo, it was found that the pole effect in general does not occur at 
pressures below 10 cm. of mercury. The arc which we employ in vacuo 
is of the same type and length and carries the same current as that 
used at normal pressure, but in appearance it is strikingly different. 
Under normal pressure the vapor mainly concerned in the production 
of these lines issues from a point source on the electrode, is highly 
luminous and is confined to a comparatively limited volume; in vacuo 
the luminosity covers the entire surface of the pole more uniformly 
and is much less intense, while definite structure is for the most part 
lacking. The disappearance of the pole effect indicates that the elec- 
trical conditions play a subordinate role, if any, in producing the dis- 
placements, but a more definitive investigation is about to be undertaken 
