The mass-absorption coefficients of a number of elements from C to U for homogeneous rays from $\lambda .709$ to 1.933A and for filtered general rays from 1.93 to 4.0A have been measured. A Shearer gas tube, continuously pumped and arranged with a controlled side leak was used. The data from these observations, together with those from the two previous papers by the author at wave-lengths from $\lambda .08$ to.709A taken with the same apparatus and on the same samples throughout, form a set of continuous results from.08 to 4A, from which, after a careful analysis, the following conclusions can be stated. Over a wide range of wave-lengths, in which the mass-scattering coefficients $\frac{\sigma }{\rho }$ can be neglected, the fluorescent absorptin coefficients $\frac{\tau }{\rho }$ cannot always be represented by the generally accepted formula, $\frac{o\tau }{\rho }=C{\lambda }^n$ with $n$ and $C$ constant, but each element must be considered by itself, with values of $n$ and $C$ changing rather abruptly in certain regions of wave-length. In general the values of $n$ which best fit the results are a high value 2.92 and a low value 2.6. Nowhere does ${\lambda }^3$ over a long range accurately represent the data. Over the complete range of wave-length, the absorption of Al appears to be quite accurately represented by the formula, $\frac{\mu }{\rho }=(13.9\mathrm{} or 14.0){\lambda }^{2.92}+(.\mathrm{}14\mathrm{} \mathrm{to} .18)$ and is therefore well suited as a standard substance for the determination of effective wave-lengths from absorption coefficients. The $\frac{N^4}{A}$ law, although holding fairly well in those cases where $n$ is the same for all the elements involved, can at best be only approximate, and it therefore does not seem possible to express the absorption of x-rays by a universal formula of the form $\frac{\tau }{\rho }=\frac{C{\lambda }^3{N}^4}{A}$, in which $C$ is a constant. The discontinuity magnitudes, $J_K=\frac{C_K}{C_L}$, extended to Fe, show steadily increasing values with decrease of atomic number $N$, reaching a maximum value for Fe of about 10. For the L-M series, $J_L(=\frac{C_L}{C_M})$ shows values following the same law as $J_K$ does, with probably somewhat higher values for the same elements than $J_K$. For the M-N series, $J_M(=\frac{C_M}{C_N}=10\mathrm{} \mathrm{to} 12 \mathrm{in}\mathrm{the}\mathrm{case}\mathrm{of}\mathrm{U})$ apparently indicates that $J_M$ with five discontinuities has much higher values than $J_L$ or $J_K$ for the same elements. The effect of impurities in the cases of Al, Cu, Sn and Pb was ascertained, and only the presence of Fe in the Al could make an error as great as 1 percent.

• Received 1 June 1926

DOI:https://doi.org/10.1103/PhysRev.28.907