30 
PHYSICS: D. L, WEBSTER 
Proc. N. a. S. 
tralize the slight remaining fluctuation, as at 2 meters distance, with Li 
and L2 parallel, M = 100 henries. ' In this case, however, no attempt was 
made to adjust M accurately, but L2 was tipped to a nearly vertical posi- 
tion, 2 meters from Li, which was horizontal, and Li was set perpendicular 
to the 500 ^ transformer, and also to the 60 ^ one, which were both 
horizontal and far apart. 
An important point is the condensers across the secondary of the trans- 
former. To eliminate unsteadiness due to fluctuations on the 115-volt 
D. C. line, that ran the motor-generator, the generator field was excited 
by a storage battery and the variable condenser was then adjusted for 
resonance. Under these conditions, the potential on Ci being a maximum 
with respect to speed changes, a small change in the speed of the motor 
could affect it only by a second order amount, always negative, therefore, 
not such as to bring out spectrum lines falsely. This arrangement proved 
invaluable in making long exposures. The only fluctuations that ever 
made trouble were small ones due to current changes produced by the 
traces of gas that cannot be removed from the tube without melting the 
platinum target to do so. 
I wish here to express my thanks to Mr. R. M. Frye for valuable assis- 
tance in the construction of this apparatus. 
The X-ray tube was the same one used in the previous work, but, through 
the kindness of Dr. Coolidge, it had been provided with a thin window of 
a type used by Blake and Duane,^ made by blowing a thin, spherical shell, 
convex inward. This greatly reduced the exposures required. 
Results. — Considering each line separately, the results are as follows: 
ai. This line was definitely shown by the ionization method to be- 
long to Li, and fixed the critical potential of Li most accurately, the average 
of the best measiurements being 11.47 =^ 0.05 kv. Since conditions were 
adapted for intensity measurements but no attempt was made to measure 
wave-lengths, the value of h found there will not be used, but rather the 
value recommended by Birge,^ which is 6.554 X 10 erg sec, with Milli- 
kan's e. This gives X^^ = 1.076 =^ 0.005 A. The films show ai very 
strong when X^in = 1-01 A, and not at all at 1.08. 
a2- In the mica films this line is resolved from ai only in the 5th order, 
but as this order is strong, it shows clearly when ^^in is around 0.90 A, 
and definitely, though faintly, at 1.01. Since the critical wave-length 
of L2 is 0.935 A (see this places 0:2 in Li. The ratio of its intensity to 
that of ai was tested at Harvard by the ionization method, with results 
shown in figure 3. Since the ratio is the same at 40.0 kv. as at 17.85, 
the law of constant intensity ratios for lines of the same series holds in 
this case. 
Ir ai. This line, due to an impurity of iridium in the target, shows 
faintly but definitely when \min = 1-01 A, but not at 1.08. 
/. This line is too faint to show near its critical potential in any order 
