4. Description of the Atlas
The wavelengths and intensities of Rowland ghosts (spurious lines caused by imperfections in the ruling of the grating) were predicted from the known properties of the gratings. Ghost lines are marked on the plots with a carat instead of a tic mark to distinguish them from true spectral lines. They are not listed in the table.
Wavelengths of lines measured on our photographic plates, taken from the literature, or calculated from optimized Pt II energy levels are given to four decimal places. Lines measured in the photoelectric data only are given to two decimal places. Wavelengths below 2000 Å are given in vacuum; wavelengths above 2000 Å are given in air. For lines originally observed in vacuum, conversion of the wavelengths from vacuum to standard air was carried out by using the three-term formula of Peck and Reeder  for the index of refraction of air.
Also listed in the table under the column heading CODE are the sources for wavelengths of various lines emitted by the Pt/Ne lamp that we have taken from the literature, mainly Pt I, Ne I, and Ne II. Most of these lines were used as wavelength standards. Literature values were also substituted for lines of impurity species such as H I, C I, O I, Si I, Al I, and Al II. The presence of additional impurity lines of Mg I, Mg II, Fe I, Cr I, Pd I, Rh I, Au I, Ag I, Ni I, Ca I and Ca II were subsequently pointed out by J. Blaise. These lines are identified in the table. Literature values for their wavelengths have been substituted only for Ca II and Fe I.
The intensity of impurity lines varies greatly from lamp to lamp. For example, we did not observe the intense Al I lines at 3944 Å and 3961 Å on our photographic plates. However, in a lower wavelength exposure using a different lamp the normally less intense lines at 3082 Å and 3092 Å did appear. For this reason we have given no intensities for the impurity lines.
The energy level designations for classified lines of Pt I and Pt II correspond to the integer parts of the level energies and are given with the even parity level first. Classifications and wavelengths for Pt I lines with CODES D and E were taken from Engleman . Pt I lines with CODE N and Pt II lines with CODE K are newly classified by Blaise and Wyart ; the wavelengths are from the present work. Classifications for other Pt II lines were taken from Shenstone , with level values given by Reader, Acquista, Sansonetti, and Engleman ; the number given in the CODE column is the wavelength uncertainty of the Ritz wavelength in units of 0.0001 Å (see next section).
The intensities in the atlas are a uniform set of relative values covering the entire region of observation. For lines that were blended on the photoelectric scans but resolved or nearly resolved on the photographic exposures, the intensities were estimated visually from the photographic plates by comparison with nearby well-resolved lines. In a few places a real spectral line is blended with a grating ghost. This is noted with an M in the CODE column in the table. The intensities measured for such lines are probably affected by the presence of the ghost. As mentioned, the spectral sensitivity of the spectrometer and detector combination was taken into account by using the accurate radiance values of Klose  for about 80 of the lines to normalize the observations. From the reproducibility of our measurements and comparisons with the data of Klose we estimate the relative intensities for a given species (element and stage of ionization) to be accurate to about 20%. A prime factor in possible variation of the relative intensities is the length of time that a particular lamp has been used. Over many hours of use the intensities of the Ne lines are observed to change relative to the Pt lines. However, for a given atom and ionization stage the relative intensities should be reliable within our estimated uncertainty. For most lines the present intensities are identical to those given by Reader, Acquista, Sansonetti, and Sansonetti . The intensities of a few lines have been slightly revised in the present work.
Our relative intensities for lines emitted by the Pt/Ne lamp are potentially useful for calibration of the spectral response of spectrographic systems in other laboratories. In general, the values are sufficiently reliable to provide a good semi-quantitative calibration. Of course the accuracy that can be obtained is limited by the degree to which other Pt/Ne lamps might vary from those we used. We found only small variations in the relative intensities of lines in our lamps, all of which were purchased separately over a five year period. Nevertheless, it is not certain that other lamps would exhibit identical properties. In particular, comparison of lines in the 1130 Å to 1300 Å region with lines in higher wavelength regions could be affected by variation in the low wavelength transmission of the magnesium fluoride windows of different lamps. Since we used only a small number of lamps and did not scan each lamp over the entire spectral region, we can make no definitive statement regarding lamp to lamp variation. Further investigation would be needed to evaluate the importance of such systematic variations.