ENGN ас

100 054 no.32

Suppl.

Library of Congress Catalog Card Number: 61-64759

The calibration of the intensity scale in the NBS Tables of Spectral-Line Intensities by Meggers, Corliss, and Scribner (1961) was made by comparison with the radiation from a standard tungsten ribbon filament lamp in the region between 3300 and 9000 Å. The intensity of the lamp at 5500 Å is 40 times its intensity at 3300 Å and 300 times its intensity at 2800 A. This fact introduced a progressively increasing error from scattered light of the intense visible radiation as the work proceeded toward short wavelengths, below 3300 Å.

Since the ribbon filament lamp was too faint in the region below 3300 Å to serve as a standard, we took recourse to a hydrogen arc lamp. Output from this lamp was compared by R. Stair in the Radiometry Section of the Bureau with a standard tungsten-in-quartz lamp and a standard mercury arc in the region from 2500 to 3800 Å; this provided an independent overlapping calibration which carried us down to 2500 Å.

At that time there was no practicable method available for making an intensity calibration at wavelengths short of 2500 Å. As the authors then stated, "Lacking any reliable energy calibration for shorter waves, the intensity estimates from 2500 to 2000 Å were necessarily adjusted by judicious extrapolation, guided by the declining densities of background in the spectrograms, caused by the increasing absorption in the apparatus and in the air at shorter wavelengths."

Soon after the appearance of the Intensity Tables, Penkin and Slavenas [1963] published absolute oscillator strengths for lines of tin and lead that extend down to 2170 Å. They made their measurements with the hook method, which does not require an intensity calibration of any sort. This being the case, there should be no wavelengthdependent error in their results. Now, Corliss and Bozman [1962] had already derived absolute oscillator strengths for these spectra from the relative intensities of Meggers, Corliss and Scribner. A plot of the ratio of the oscillator strengths of Corliss and Bozman to those of Penkin and Slavenas is shown in figure 1. The figure shows that the ratio is not wavelength dependent between 2900 and 2450 Å but that Corliss and Bozman's scale declines by a factor of 30 between 2450 and 2150 Å. Since oscillator strengths are proportional to intensities, this plot can be used to calibrate the intensity

scale at the short wavelength end of the Intensity Tables.

There are additional data available below 2500 Å that support the calibration derived from Penkin and Slavenas' work on Sn I and Pb I. For example, Penkin and Shabonova [1963a] have measured by the hook method oscillator strengths for Al I, Ga I and In I and the same authors [1963b] for Tl 1, which give results more or less parallel to those for Sn I and Pb I, but the data do not extend much below 2300 Å. Gruzdev [1962b] has calculated relative oscillator strengths for the transition 3d7(4F)4s - 3d7 (4F)4p in Co II using intermediate coupling. A plot of the ratio of Corliss and Bozman's values to those of Gruzdev is given in figure 2, which exhibits a wavelength dependence in substantial agreement with that of figure 1. There seems little possibility of a wavelength dependent error in theoretically determined oscillator strengths. Gruzdev [1962a] has also calculated oscillator strengths for Fe II which are in agreement with values measured in Fe II by Morosova and Startsev [1965]. Both of these sets of values yield curves in qualitative agreement with the calibration based on SnI and Pb I, although the slope is somewhat steeper. Similar curves are also obtained using the oscillator strengths measured for Ni II by Bell, Paquette, and Wiese [1966], although we have only four lines of Ni II in our data.