1. The standard deviation of our polynomial fits for the Cu II reference lines in the Pt/Cu lamp was typically 0.0010 �.
2. The standard deviation of our polynomial fits for the Pt lines used as internal standards for measurements in the Pt/Ne lamp was typically 0.0015 �.
3. A comparison of a group of about 100 lines measured by different operators on different plates and taken with different grating rotations in the region 1470 � to 1520 � showed an average deviation of 0.0001 � and a rms difference of 0.0014 �. In general, our separate measurements of the wavelengths of individual lines agreed to about this level of accuracy.
4. A comparison of the wavelengths of 37 lines of Pt II in the region 2247 � to 3700 � that were measured in this work and independently by Engleman [7] shows an average deviation of 0.0003 � and a rms difference of 0.0019 �.
5. For the 508 lines of Pt II whose wavelengths can be calculated from the optimized level values, the rms difference between the calculated and observed wavelengths is about 0.0015 �.
6. A comparison of our measured wavelengths for impurity lines appearing in the Pt/Ne lamp with standard wavelengths for these lines shows an average deviation of 0.0003 � and an rms difference of 0.0015 �.

As mentioned above, the wavelengths of classified lines of Pt II in the atlas which have numbers in the CODE column are those derived from the optimized level values. The uncertainties of these wavelengths are taken to be the square root of the sum of the squares of the uncertainties of the combining levels as given by Reader, Acquista, Sansonetti, and Engleman [7]. They are listed in the far right column under the heading CODE in units of 0.0001 �.

The uncertainties of the photoelectrically measured lines were estimated by comparing the measured wavelengths of Pt II lines observed only in the photoelectric scans with calculated Ritz wavelengths for the same lines. The standard deviation of the differences was about 0.006 � for lines below 2030 � and about 0.015 � for lines at longer wavelengths. Based on these comparisons we estimate the uncertainty to be � 0.01 � for lines below 2030 � and � 0.02 � for lines above 2030 �.

The uncertainties of lines whose wavelengths have been taken from the literature are discussed in some detail in the notes to the atlas. Most of these uncertainties are less than 0.001 � and virtually all are less than 0.002 �.

The cathodes of the lamps used in this work and with GHRS contain isotopes of Pt in their natural abundances. Some lines of Pt I and Pt  II show appreciable isotope and magnetic hyperfine structure (hfs). At the resolution of our spectrograph (and also GHRS) almost all Pt lines appear sharp and symmetric. A few lines show evidence of unresolved structure and appear wide, hazy, or asymmetric on the photographic plates. These lines are noted (W, H, L, or S) adjacent to their intensities in the atlas. Lines showing partially resolved structure are noted in the atlas as being complex (C). A few hyperfine patterns occurred in the photographic data as three fully resolved features and were measured as separate lines.

For GHRS and other instruments with resolving power of 10⁵ or less, the existence of hfs in some lines should present no problem in using the present list of Pt lines for wavelength calibration. To achieve the highest accuracy, lines with notations indicating detectable unresolved structure should not be used. For instruments with resolving limits significantly below 0.02 �, structure may be observed in many additional Pt lines, and our present wavelength list may not be adequate for calibration purposes. Thus, for calibration of spectrographs having much higher resolution, it may be desirable to develop calibration wavelengths based on a lamp whose cathode contains a single even isotope of Pt.