126 
ASTRONOMY: HALE AND BABCOCK 
about one-quarter this scale. The third order, where the linear dis- 
persion is nearly 5 mm. to the angstrom, has been used in some of our 
work. 
2. A strictly differential method of observation, involving the de- 
termination of the width of the same line on contiguous strips of spectra 
photographed in a single exposure with apparatus transmitting light 
polarized in planes at right angles to one another. A long Nicol prism 
mounted over the slit, with a compound half-wave plate above it, made 
of mica strips 2 mm. in width, fully met this requirement. In order 
to eliminate possible absorption effects in individual strips, the half- 
wave plate was made by combining the compound quarter-wave plate 
used in our investigations of the general magnetic field of the smi with 
a long piece of quarter-wave mica. By inverting this between ex- 
posures, a given strip can be made to transmit Kght polarized in either 
plane. 
3. A measuring machine capable of exhibiting the smallest varia- 
tions in the width of the lines on the odd and even strips. A Koch 
registering micro-photometer, recently constructed in our instrument 
shop, served admirably for this purpose. 
A series of photographs of the Ha and 11(3 lines of hydrogen, made 
with the sHt set about 3 mm. within the limb (parallel to a tangent) of 
the large solar image of the 150-foot tower telescope, furnished the required 
observational material. As employed for this work, the Koch machine 
gave photographic curves (reducible to intensity curves) of the Ha and 
H^ lines on a scale fifty times that of the original negatives. Combin- 
ing measures of curves made for several sets of odd and even strips, 
the probable error of the average width of one group of curves is ^2.7 
mm. Thus a difference in mean width of 5 mm., corresponding in the 
second order to 0.034 angstrom, should certainly be discernible by our 
method. Assuming this least appreciable difference in width to be of 
the same order of magnitude as the Stark separation of the compo- 
nents, we may at once determine the maximum electric field present from 
Stark's pubHshed results for Ha. For a field of 28,500 volts per centi- 
meter he obtained a difference in separation of 6.4 angstroms between 
the outer and inner components (polarized parallel and normal to the 
field). As the total separation is directly proportional to the field- 
strength, and as the Ha line shows no appreciable difference in width 
on the odd and even strips, it follows that the intensity of the solar 
electric field at the point of observation cannot exceed 150 volts per 
centimeter. A similar determination for the H^ line in the ttiird order 
