86 
PHYSICS: DUANE AND PATTERSON Proc. N. A. S. 
of using an X-ray tube with a molybdenum target to avoid exactly this 
difficulty. If there had been an appreciable amount of tungsten on the 
anticathode due to sputtering, we would have detected the presence of 
the strong emission lines of the L series. In a more recent paper Coster^ 
states that he has measured the Li and L2 absorption limits of tungsten and 
obtained essentially the same wave-lengths as those given by us. 
Table I contains our new values for the L2 and L5 absorption limits of 
Pt, Au and Bi. These measurements were taken with almost the same 
arrangement of apparatus as before. The X-ray tube was provided with 
an arm which reached nearly to the first slit of the spectrometer, thus re- 
ducing the absorption of X-rays by air. At the end of this arm and also 
on the front of the ionization chamber thin mica windows were employed 
which still further reduced the absorption of X-radiation. Finally the 
sensitivity of our electrometer had been increased many times. We 
were thus supplied with a much more sensitive detector of X-ray spectra 
than in our earlier measurements. This enabled us to use narrower 
slits and to obtain at the same time greater drops in our ionization cur- 
rents for the absorption limits. 
TABLE 1 
Critical Absorption. L Series of X-Rays 
Grating Space for Calcite, 2d = (6.056 ^ 0.004) X 10 Cm. 
CHEMICAL L2 Li 
ELEMENT X X lO* cm. "Aoo ^ X 108 cm. Voo 
Pt .9318 =t= 3 0.9779 .8918 =1=9 1.022 
Au .9008 =t 4 1.0116 .8610=^9 1.058 
Bi .7871 =t: 4 1.1578 .7562 =1= 9 1.205 
In the first column for each limit is given the wave-length in Angstrom 
units; in the second, the wave number divided by the Rydberg constant. 
Our new value for L2 of Au differs from our previous one by 0.17 per 
cent. The error in our previous work was due to the proximity of the 
bromine ionization limit, which has a wave-length = 0.91796 X 10~^cm. 
In our new measurements it was unnecessary to have ethyl bromide in 
the ionization chamber. 
For the L3 limits to Pt and Bi the new values differ from the old ones 
by a considerable amount in excess of our then estimated error. These 
errors were undoubtedly inherent in the faintness of the effect which we 
were endeavoring to measure. 
The other three values are substantially the same as we gave before. 
We do not find the error in L3 for Au suggested by Dauvillier. 
At the time of our previous investigation, the available evidence seemed 
to indicate that the /?5 and 72 emission lines belonged, respectively, to the 
Li and L2 groups — i.e., that /35 was emitted when an electron fell into the 
Li orbit, and 72 when an electron fell into the L2 orbit. As we definitely 
proved that the Li and L2 absorption limits for tungsten were of lower 
