Spectroscopy in the Vacuum Ultraviolet Region of the Spectrum

J. C. Boyce*)

Contents

1. Introduction. 2. Transparency of gases and requirements for vacuum. 3. Transparency of solid materials. 4. Reflecting power of gratings and mirrors. 5. Photographic technique and other methods of observation. 6. Construction of the spectrograph. 7. Light sources. 8. Wavelength standards. 9. Atomic spectra. 10. Molecular spectra. 11. Spectra of the solid state. 12. Astrophysical and other applications. 13. Literature.

1. Introduction

Spectroscopy in the vacuum ultraviolet, like spectroscopy in the far infrared, differs in its technique from spectroscopy in the more accessible regions of the spectrum. However, the data obtained in the vacuum ultraviolet should be regarded only as a constituent part of the data for all regions of the spectrum, especially since in many cases they play an essential role in establishing the connection between the normal state of an atom and its excited states. In addition, some features of the experimental technique in the vacuum ultraviolet, apart from their general physical interest, may be of interest from the point of view of experiments in neighboring fields of optics, photoelectricity, and photochemistry. Therefore a review of this field should touch upon the physical principles of experimental observations, evaluate the observations already made, and indicate fruitful problems for further investigations.

The study of line spectra provides methods for identifying chemical elements and information about the structure of the atom. For complete observation of the spectrum it is often necessary to include wavelength regions shorter than \(\lambda 1850\)**). In this case, owing to the opacity of oxygen

*) J. C. Boyce, Rev. Modern. Phys., 13, 1, 1941. Translation by N. A. Shishakova.
**) Everywhere in this article wavelengths are denoted in angstroms and are indicated with the addition of the letter \(\lambda\). For nonspecialists in the field of spectroscopy it may be pointed out that the angstrom is defined precisely by the assertion that the red

it is necessary to remove it from the path of the rays by pumping the air out of the optical system. We shall call this region of the spectrum the “vacuum ultraviolet,” and not the “extreme ultraviolet,” since the latter term is less definite. As an introduction to a survey of spectroscopy in this region, it is useful to recall some well-known facts concerning spectroscopy in general.

Measurements of wavelengths in the visible region of the spectrum have been made for almost all elements. Such measurements make it possible to establish, qualitatively, the presence of each of these elements when a definite group of lines belonging to the given element is observed. In the same way, when, for example, in certain astronomical observations or in the case of an ion beam there is a considerable relative velocity between the source of light and the observer, the Doppler displacement of these wavelengths makes it possible to determine this relative velocity. Further empirical studies led to the discovery of various regularities in the distribution of spectral lines. The most important of these is the Ritz combination principle, according to which wave numbers (the number of waves per centimeter) can be arranged as differences between spectroscopic terms. The Bohr–Sommerfeld theory, and later the wave-mechanical theory of the atom, made it possible to identify these spectroscopic terms with the energy states of the atom. For a qualitative or quantitative explanation of the number and positions of these states, the theories mentioned proceeded from atomic models. Thus spectroscopy became a tool for studying the energy states of the atom (and subsequently also of the molecule) and the binding energy of its outer electrons. When the energy states are known in sufficient detail, the spectrum becomes an indicator of the conditions of excitation in the light source, regardless of whether this source is in the atmosphere of a star or in a discharge tube in the laboratory. Finally, when the excitation energies are known in detail, quantitative spectrochemical analysis also has a firm theoretical basis.

The first spectra (designated by the Roman numeral I after the name or symbol of the atom) are the spectra of neutral atoms, while the second spectra (II) are the spectra of singly ionized atoms. Whether a given line spectrum has lines in the vacuum-ultraviolet region or not depends on the ionization potential of the atom (or ion). If the ionization potential of an atom or ion exceeds 6.7 volts ($\lambda 1850$), then at least the higher members of some of its spectral series will lie within the vacuum region. Whether there will be important lines, i.e., lines giving information about terms (co-

The red cadmium line (under certain excitation conditions) has the wavelength $\lambda 6438.4696\,\text{Å}$. For most practical purposes the angstrom may be taken as equal to $10^{-8}\,\text{cm}$. It may also be recalled that the visible region of the spectrum lies approximately between $\lambda 4000$ and $\lambda 7000$.

states of energy), lie in the vacuum region, depends on the arrangement of the states in the atom (or ion). Generally speaking, studies in the vacuum ultraviolet region become important only in those cases when the ionization potential exceeds 10 volts. There are, however, cases in which work with an ionization potential of only 8 volts is also significant. A survey of the accumulated data, with an evaluation of spectra not yet analyzed, indicates that important data lie in the vacuum ultraviolet region for the first spectra of approximately 50% of the elements, for the second spectra of approximately 85% of the elements, and for the higher spectra—of all elements.

The vacuum ultraviolet region extends into the region of soft X-rays and is not distinct from it. Air gradually becomes transparent again for X-rays with wavelengths less than \(\lambda 2\). If radiation arises owing to transitions of one or several electrons from the most weakly bound electron shells present in the emitting atom or ion, then it is convenient to assign it to ultraviolet, or “optical,” radiation. If, however, it arises owing to transitions following the removal of one or several electrons from one of the more tightly bound electron shells in the atom, without removal of weakly bound electrons, then the radiation may be regarded as X-ray. In this sense optical radiations have been investigated down to \(\lambda 12.1\), while soft X-rays have been measured to wavelengths of several hundred angstroms. Beutler succeeded in investigating the most interesting transition between these two, perhaps somewhat arbitrary, classifications of excitation conditions. His work will be considered in the last section.

Obviously, in surveying our present knowledge of line spectra in the vacuum ultraviolet region, many important observations made more than ten years ago will have to be confined to only brief mention. Spectrographs currently available provide considerably greater dispersion, and wavelength standards have thereby been very much improved. Many of the earlier works must now be repeated, especially for those elements that have complex spectra, where inaccurate data can make analysis of the spectrum impossible. It may further be noted that the spectral region first investigated by Schumann (and often called by his name), from \(\lambda 1850\) to \(\lambda 1200\), was studied by later investigators less thoroughly than the region of still shorter wavelengths discovered subsequently.

The historical development of the extension of spectroscopy from the visible region to regions of ever shorter wavelengths will be outlined here only very briefly, since several excellent reviews have already been published on this question. The first studies in the vacuum ultraviolet region were carried out by Schumann.

(1893, 1901). He used a fluorite prism and lenses in a vacuum spectrograph and prepared for himself special photographic plates that now bear his name. He succeeded in reaching wavelengths, as is now known, down to approximately \(\lambda 1200\). He was unable to reach shorter wavelengths because of absorption in the prism and lenses. Owing to the absence of exact data on the dispersion of fluorite, Schumann could not determine the wavelengths of the lines he observed. Lyman (1906) replaced the fluorite prism and lenses by a concave grating. He was able to make measurements of wavelengths in the region discovered by Schumann and to extend it further to \(\lambda 500\). Millikan (1921) introduced a more powerful light source, a hot spark, and thanks to this it proved possible to extend the observations to \(\lambda 200\). Later, Millikan and Bowen carried out extensive investigations of the spectra of isoelectronic series, at a time when these data were needed for the development of the theory of spectra. Compton and Doan (1925) used a grating at grazing incidence for measuring X-ray wavelengths. This method was further developed in the work of Thibaud (1927) with a plane grating and Osgood (1927) with a concave grating for measuring soft X-rays, and in the work of Hoge (1927) for measuring wavelengths in the vacuum ultraviolet region. In this way the gap then existing between the vacuum ultraviolet and X-ray spectral regions was filled. The vacuum spectrograph with grazing incidence received more complete development in the laboratory of Siegbahn in Uppsala, where during the last ten years Edlén carried out extensive investigations, especially of the spectra of multiply ionized atoms. Finally, in any review, however brief, mention must be made of the important astrophysical applications of Bowen’s data on vacuum ultraviolet spectroscopy to the interpretation of the spectra of gaseous nebulae.

2. TRANSPARENCY OF GASES AND VACUUM REQUIREMENTS

To estimate the practical requirements for investigations in the vacuum ultraviolet region, it is useful to summarize the available reliable data on the absorption spectra of the principal constituents of the atmosphere. It is easy to see that oxygen is the most troublesome in this respect. “Forbidden” absorption spectra (i.e. spectra caused by transition processes of very low probability) need not be considered in the case of optical paths measured in meters. For the same reasons, absorption spectra of secondary photochemical products are also usually neglected. However, when both of these factors act through the thickness of the earth’s atmosphere, they establish the short-wavelength limit of the spectrum at \(\lambda 3000\).

Oxygen

The system of \(\mathrm{O}_2\) bands observed by Schumann and Runge was measured by Curry and Herzberg (1934) at high dispersion. It begins near \(\lambda 1950\) and reaches the limit at \(\lambda 1759\), beyond which lies a strong band of continuous absorption. This band system gradually hampers observations below \(\lambda 1950\), with \(\lambda 1850\) usually being taken as the practical limit for observations in air. Owing to the discontinuity of the absorption spectrum in this wavelength region, observation of lines close to \(\lambda 1850\) can in no way serve as a guarantee that other lines, closer to \(\lambda 1950\), cannot be absorbed by some of the discrete bands in the Schumann–Runge system. For convenience, the wavelengths of the band edges in the Schumann–Runge system are given in Appendix A.

Ladenburg, Van Vurgis, and Boyce (1932) and Ladenburg and Van Vurgis (1933) investigated the continuous absorption. The absorption coefficient was determined by them quantitatively as a function of wavelength in the interval from \(\lambda 1670\) to \(\lambda 1330\). The absorption has a maximum at approximately \(\lambda 1450\) and then rapidly decreases with decreasing wavelength. At the wavelength of maximum absorption, the intensity of radiation is reduced in oxygen at normal temperature and pressure by a factor of two over a path length of \(0.0014\ \mathrm{cm}\). This absorption is comparable with absorption in metals and is considerably greater than absorption in the lines of discrete bands. The second band of continuous absorption in oxygen begins approximately at \(\lambda 1100\) and extends at least to \(\lambda 303\), and probably approximately to \(\lambda 160\). Qualitative observations of this second absorption band show that the absorption coefficient here has the same order of magnitude as in the continuous band between \(\lambda 1750\) and \(\lambda 1300\). Ladenburg and Van Vurgis showed that these two absorption bands can explain the form of the dispersion curve of oxygen obtained by Ladenburg and Wolfsohn (1932). The form of the band in the Schumann region is in good agreement with the theoretical calculations of Stückelberg (1932, 1933).

Price and Collins (1935) observed absorption bands in the region between the two continuous absorption bands in oxygen. They had not been able to observe them in earlier investigations, where a line spectrum was used to measure absorption. The bands extend approximately from \(\lambda 1250\) to \(\lambda 650\). The bands between \(\lambda 1250\) and \(\lambda 1000\) are comparatively weak, and they could be observed only when the pressure in the spectrograph was about \(0.1\ \mathrm{mm}\) Hg. The path length of the light in this case was \(1.5\ \mathrm{m}\). The bands below \(\lambda 1000\) were much stronger and appeared when the partial pressure of oxygen, which had been diluted with helium, was \(0.001\ \mathrm{mm}\) Hg.

Nitrogen

Nitrogen proves to be transparent up to $\lambda 1450$. From there begins a series of sharp narrow bands extending to $\lambda 990$. The wavelengths of these bands are given in Appendix A. Beyond $\lambda 990$ continuous absorption occurs.

Carbon dioxide

Leifson’s experiments (1926) established an absorption band beginning at $\lambda 1712$, with very strong absorption in the region below $\lambda 1610$. Rathenau’s investigations (1933) did not succeed in confirming these results, but he found a series of absorption bands beginning at $\lambda 1174$ and passing (near $\lambda 957$) into a band of continuous absorption extending at least to $\lambda 270$. If special precautions are not taken, considerable complications should be expected from absorption by traces of oxygen, which may arise as a result of a photochemical reaction in the path of absorption. Grot (1939) pointed out that the reaction $\mathrm{CO_2} + h\nu \to \mathrm{CO} + \mathrm{O}$ is energetically possible throughout this whole range of wavelengths. Below $\lambda 1640$ this reaction leads to the formation of atomic oxygen in the excited state ${}^1D$. Since Grot (1937) carried out the photochemical decomposition of carbon dioxide with the aid of his xenon lamp ($\lambda 1470 — \lambda 1295$), it is clear that there must be considerable absorption in this region of wavelengths. Of course, further investigations are required in this respect.

Water vapor

Rathenau (1933) found two continuous absorption bands. The upper limit of one of them was at $\lambda 1780$, and of the other at $\lambda 1340$. A series of bands is superposed on each such band. Goldfield (1938) noted diffuse bands extending from $\lambda 1400$ to $\lambda 900$, followed by a continuous band extending toward the shorter wavelengths.

Rare gases

In argon, line absorption begins at $\lambda 1066$, and continuous absorption at approximately $\lambda 800$. In neon, line absorption begins at $\lambda 743$, and continuous absorption follows it, beginning at about $\lambda 575$. The amounts of helium, krypton, and xenon in atmospheric air are insignificant, but they would not be of great importance even if their content were larger. These gases are in essence transparent (apart from a few sharp absorption lines in the two heavy gases) at least to the same degree as nitrogen.

Generally speaking, it is easier to remove air than to get rid of its undesirable constituents. Quantitative data on absorption are available only for oxygen, but there is no reason to think that the other gases give significantly stronger absorption. In order to

absorption in oxygen over an optical path of 4 m (in a typical vacuum spectrograph) did not exceed \(1/2\) at the wavelength of maximum absorption, it is necessary to reduce the partial pressure of oxygen in the apparatus to about one-thousandth of a millimeter of mercury. A convenient pressure of residual gas is a pressure from one ten-thousandth to one hundred-thousandth of a millimeter of mercury. With modern vacuum pumps such a rarefaction is easily attainable, and at it no absorption will be noticeable at any wavelength.

For limited spectral regions it is sometimes convenient to replace air by pure nitrogen. This is done in order to somewhat extend the range of application of quartz prism spectrographs. Nitrogen is taken at a pressure somewhat above atmospheric, so that leakage occurs outward and so that no contamination of any kind arises. Such a method seems very promising for small apparatus, but as their size is increased difficulties arise because of the need to obtain the purest gas in very large quantities. To satisfy the indicated criterion of an oxygen partial pressure of one-thousandth of a millimeter of mercury for a 4-meter optical path, one has to solve a serious problem of gas purification. In this respect helium and hydrogen present other possibilities (hydrogen has absorption bands below \(\lambda 1115\), and a continuous absorption region below \(\lambda 850\)), but the advantage from this is small. Generally speaking, it is much easier to evacuate a spectrograph than to fill it with a transparent gas.

Selwyn (1929) surrounded his light source (an arc or spark) with an atmosphere of nitrogen and admitted radiation into the vacuum spectrograph through a fluorite window. Shenstone (1938) replaced this window by a fluorite lens, so that an image of the source was formed on the slit. He also improved the method of admitting nitrogen, so that a constant gas flow over the surface of the lens and around the arc or spark kept the lens clean. The path of light in nitrogen can be made very short*).

If it is desired to go below \(\lambda 1100\) (the limit of transparency of lithium fluoride), then it is necessary to use some system of differential pumping. One such system was described by K. T. Compton and Boyce (1928). Modern pumping technique should make the system considerably more efficient. Artificial

*) Prof. Shenstone informed the author of the existence of a thin absorption band at wavelengths greater than \(\lambda 1450\) in the spectra of metallic arcs operating in an atmosphere of nitrogen. The structure of the band could easily be observed against the background of emission lines of a metal that are broadened at the expense of autoionization. The nitrogen molecule in its lowest vibrational state is not capable of absorbing in this range of wavelengths, but excitation in the arc may maintain an influx of nitrogen molecules in a state causing the observed absorption. It is still unclear whether such a phenomenon also occurs in a spark in nitrogen.

the pumping system used by Kario and Lochte-Holtgreven (1927), who arranged an argon “window” preventing the diffusion of metallic vapors from the light source to the quartz window. A similar apparatus was used by Beutler (1933), in order not to allow these vapors access to the slit of the vacuum spectrograph.

From the data given above for nitrogen and oxygen it is evident that there exists a region of relative transparency in air between \(\lambda 1300\) and \(\lambda 1100\). This circumstance was first noted by Hopfield (1922) and was subsequently confirmed by Ladenburg, Van Voorhis, and Boyce (1932), and subjected to a more detailed study by Lyman (1935). Lyman found that radiations in the region between \(\lambda 1100\) and \(\lambda 1250\) can penetrate several centimeters at atmospheric pressure. But this transparency is only relative. It is limited, generally speaking, by the tail of the absorption band with a maximum at \(\lambda 1450\), and in some regions it is interrupted by the system of \(N_2\) absorption bands and by the \(O_2\) bands discovered by Price and Collins.

Preston (1940) measured the absorption coefficient of oxygen, nitrogen, dry air, carbon dioxide, and water vapor for one definite wavelength—the hydrogen line at \(\lambda 1215\). The value of the coefficient for nitrogen is very small, and therefore at this wavelength nitrogen may be regarded as a practically transparent substance. The absorption coefficient in oxygen is apparently a function of pressure. This shows that the absorption is caused by pressure broadening of the neighboring \(O_2\) band at \(\lambda 1211\), observed by Price and Collins. Preston finds that the intensity of the hydrogen line \(\lambda 1215\) is reduced by half over a path of \(4.5\ \text{cm}\) in dry, \(CO_2\)-free air at atmospheric pressure. The absorption coefficient of carbon dioxide is considerably larger, and that of water vapor incomparably larger. Schneider (1940) measured the absorption coefficient in dry, \(CO_2\)-free air at 350 points between \(\lambda 1596\) and \(\lambda 382\). The light source was a spark with many lines, and not a continuous spectrum. Obviously, despite the limited resolving power of this method, the continuous absorption by oxygen from \(\lambda 1100\) at least to \(\lambda 300\), mentioned by Ladenburg, Van Voorhis, and Boyce, superposed many absorption bands on the spectrum. Many of these bands were observed by Price and Collins. From the results obtained by Schneider it is clear that even the “window” from \(\lambda 1300\) to \(\lambda 1100\) has absorption bands distributed throughout it.

3. TRANSPARENCY OF SOLID MATERIALS

The number of solid materials sufficiently transparent for the vacuum ultraviolet region is extremely limited. For practical purposes two materials are used: fluorite and lithium fluoride. For a very limited range of wavelengths one may also use cry-

metallic quartz. Different samples of each of these three materials show a wide variety in their spectral-transmission curves. The transparency of quartz was studied by Scheibe (1929) and by Powell (1934). With very thin windows, good samples of crystalline quartz can be used down to \(\lambda 1450\). The considerable thickness required in lenses and prisms greatly limits the use of quartz; thus, even at \(\lambda 1860\), \(1\ \text{cm}\) of quartz absorbs from \(25\%\) to \(30\%\) of the radiation passing through it. Fused quartz is apparently not as transparent as crystalline quartz, but in both cases there are wide variations in the transparency of different samples.

Until the most recent years it was necessary to use fluorite almost exclusively. Its transparency was studied by Schneider (1934) and Powell (1934). Since light is also lost by reflection at each surface of the specimen under investigation, measurements had to be made on many specimens of different thicknesses cut from one and the same piece of fluorite. The true absorption coefficient \(\mu\) is determined from the formula

\[ I = I_0(1-r)^2 e^{-\mu x}, \]

where \(I_0\) is the intensity of the incident radiation, \(I\) is the intensity of the transmitted radiation, \(r\) is the fraction of light lost at the surface, including also reflection losses\(^*\) and all other surface losses of light (at each surface), and \(x\) is the thickness of the specimen. Schneider used photographic photometry, and Powell a photoelectric method. The results they obtained (on the same specimens) proved to be in sufficient agreement. The values of \(\mu\) obtained by Schneider (in \(\text{cm}^{-1}\)) are given as a function of wavelength in Fig. 1. The curve shows that the absorption coefficient increases rapidly at wavelengths shorter than \(\lambda 1250\), and that a fluorite specimen may be considered sufficiently good if it shows appreciable transparency at \(\lambda 1230\). Lyman (1926) succeeded in observing the very intense hydrogen line \(\lambda 1215\) through two thin windows of

Fig. 1. Absorption coefficients in quartz, fluorite, and lithium fluoride as a function of wavelength (Schneider, 1934, 1935, and Powell, 1934).

\(^*\) The reflectivity of fluorite and quartz in the vacuum ultraviolet region was measured by Tousey (1936, 1939, 1940) in studying the optical constants of these substances.

of an exceptionally good specimen of fluorite. In ordinary use the transparency of fluorite seems to be a constant quality, although Palmer (1934) reported that when very strong short-wave ultraviolet radiation emerges from a fluorite window into the air, a surface film forms on the surface of the fluorite. This film strongly absorbs between \(\lambda 1500\) and \(\lambda 1200\). Grot (1936) reports nothing of such a phenomenon in fluorite strongly illuminated by a xenon lamp.

In recent years, thanks to the work of Schneider (1936) and Stockbarger (1936), large crystals of lithium fluoride have become available. With the aid of the technique developed by Bridgman (1925), it proved possible to obtain large homogeneous crystals when the material was subjected to slow and uniform cooling at a definite temperature gradient. The crystals have an interesting optical property, namely that lenses made from them are almost achromatic within the visible part of the spectrum. If extreme precautions are taken in cleaning the material, windows and thin lenses can be obtained that are transparent down to wavelengths somewhat below \(\lambda 1100\). The transparency of lithium fluoride was investigated by Schneider (1936) and Grot (unpublished). Schneider’s values of the absorption coefficient as a function of wavelength are presented in Fig. 1. The absorption coefficient here is everywhere smaller than that of fluorite.

Lyman (1935) developed a very ingenious method for testing the transparency of lithium fluoride without using a vacuum spectrograph. As mentioned in the preceding section, radiations with wavelengths between \(\lambda 1100\) and \(\lambda 1300\) can penetrate several centimeters of air. Such radiations from a spark in air between metallic electrodes (“Entladungsstrahlen” of Wiedemann and Schmidt (1895)) were made to fall on a thermoluminescent sensitive plate. The tested pieces of lithium fluoride were placed on the surface of the plate. After the plate was heated in the dark, immediately after exposure, thermoluminescence appeared in those parts of the plate on which the rays had acted. It also appeared, in a weaker form, on those places of the plate that had been covered with good specimens of lithium fluoride, and did not appear in the places covered with a good specimen of fluorite. Lyman concluded that his thermoluminescent detector was sensitive to radiation in the region from \(\lambda 1100\) to \(\lambda 1250\). Subsequent experiments with a vacuum spectrograph showed that the sensitivity of the thermoelectric detector arises suddenly at approximately \(\lambda 1300\) and extends at least to \(\lambda 900\). Some further details of the thermoluminescence method will be given in the final section.

Lithium fluoride has one substantial disadvantage. Under the action of intense radiation, and also under the action of electron…

during bombardment in a gas discharge, discoloration occurs, accompanied by a gradual shift of the absorption edge by several hundred angstroms. This effect was studied by Schneider (1937). If the discoloration is not too strong, it can be eliminated by heating the crystal. Discoloration occurs chiefly on the surface of the crystal, and therefore Stokbarger proposed removing the damaged surface layer by repolishing.

Laird (1920, 1927) found that a celluloid film 300–400 Å thick transmits from 50% to 20% of the radiation from $\lambda 1700$ to $\lambda 900$ and no more than 5% below $\lambda 900$. She also found that silver foil 2000 Å thick transmits strong lines down to $\lambda 900$. O’Brien (1932) investigated absorption in celluloid between $\lambda 1000$ and $\lambda 300$. Films from 60 to 400 Å thick were prepared by dropping a dilute solution of celluloid in amyl acetate onto the surface of water. Small portions of such a film were removed from the surface of the water by means of aluminum frames. The thickness of each film was calculated from the concentration of celluloid in the solution, the weight of the drop that fell onto the water surface, and the surface over which the film spread on the water after evaporation of the amyl acetate. Films of such thickness appear black in reflected light. O’Brien found that the degree of “blackness” is in qualitative agreement with the calculated thickness, and therefore he had to assume that measurement of transparency for visible light can serve as an indicator of film thickness. Comparatively thick films, showing any colors in reflected light, absorb all the energy between $\lambda 1000$ and $\lambda 300$. For celluloid films 100 Å thick, O’Brien compiled a table of transparency extending from 79% at $\lambda 300$ to approximately 30% at $\lambda 900$. At still greater wavelengths (beyond the range investigated by O’Brien) the transparency again increases. The films deteriorate after several days in vacuum and after several weeks in air. Their use as windows is limited by a pressure difference of several millimeters of mercury, but they can be used as substrates for thin condensed films of solid materials in the study of their absorption spectra.

Lyman (1928, p. 66) and Bomke (1937, p. 18) discussed the transparency of a whole series of other materials, but apparently none of them can compare with fluorite and lithium fluoride.

4. REFLECTING POWER OF GRATINGS AND MIRRORS

G. V. Sebain (1939) measured the reflecting power from $\lambda 4000$ to $\lambda 450$ for glass and for condensed films of many metals: aluminum, antimony, beryllium, bismuth, cadmium, chromium, copper, gold, iron, lead, magnesium, manganese, molybdenum,

nickel, palladium, platinum, silver, tellurium, titanium, zinc, and zirconium. These reflectivities (presented graphically as a function of wavelength) do not necessarily correspond to the reflectivities of a clean surface free of all adsorbed gases, but they do correspond to the surface conditions of such mirrors as are usually encountered in an evacuated system. Above \(\lambda 1200\) aluminum is definitely the best reflector, and below \(\lambda 1000\)—platinum. In this respect glass is inferior to platinum throughout the entire range of tests carried out. Fig. 2 shows the results obtained by Sabine for glass, aluminum, and platinum. Sabine’s measurements were made with a light beam incident at an angle of about \(18^\circ\) to the normal.

Fig. 2. Reflection coefficient of aluminum, platinum, and glass as a function of wavelength (G. B. Sabine, 1939).

Glass has an enormous advantage over most metallic surfaces in that it does not tarnish. Its reflectivity may be greatly reduced by adsorbed films, especially oil and grease, but the latter can easily be removed by organic solvents or by peeling off a film of nitrocellulose left after the applied collodion has evaporated. The metallic mirror surface, often used for gratings, should be protected from mercury vapor and from all other vapors containing sulfur (from rubber gaskets and from certain vacuum greases). Under suitable evaporation conditions, thin aluminum and platinum films deposited on glass gratings (and mirrors) can be made sufficiently stable and not subject to tarnishing. But this technique is still so little developed that it is not always possible to achieve constancy of the surface properties of the films.

Gratings ruled directly on aluminum are very useful for the spectral regions from \(\lambda 1850\) to \(\lambda 1000\). Gratings ruled on the surface of a comparatively soft metal, in particular aluminum, have the advantage that they make it possible to modify the shape of the rulings and thus to concentrate the energy in a given order of the spectrum. On glass this is impossible, although Wood succeeded in increasing the intensity of spectra from glass gratings by etching them with hydrofluoric acid after their manufacture. Glass gratings are apparently very useful in preliminary experiments with condensed films, but if it were ...

if it is possible to deposit thin films of platinum on a grating ruled on a softer surface, then at the same time it would be possible to gain the advantage of being able to control the form of the grooves and of obtaining a higher reflectivity for wavelengths shorter than \(\lambda 1100\).

On the basis of deviations from Bragg’s law in the reflection of X-rays from crystals, Stenström (1919) showed that the refractive index of most solids for X-rays is slightly less than unity. A. H. Compton (1923) showed that a beam of X-rays incident almost tangentially on the surface of a polished metal plate undergoes total reflection at that surface. With the greatest success, gratings have been used at grazing incidence both in the region of ordinary X-rays and in the vacuum ultraviolet region with wavelengths shorter than \(\lambda 1000\). When gratings are used at grazing incidence for wavelengths longer than \(\lambda 1000\), the reflectivity increases, but it is still unknown at what upper wavelength limit the phenomenon of total reflection ceases to occur for a given angle of incidence. O’Bryan (1931) pointed out that the quality of a grating is determined by the shape of the grooves and the smoothness of the surface, as well as by the reflectivity, especially when the grating is used nearly at grazing incidence. He concluded that an etched glass grating is better suited for grazing incidence, while a grating on a metal mirror or a lightly ruled glass grating should be preferred at angles close to normal incidence. Other aspects of the choice between normal and grazing incidence for gratings will be considered in the last section.

5. PHOTOGRAPHIC TECHNIQUE AND OTHER METHODS OF OBSERVATION

Even before reaching the vacuum part of the ultraviolet region, photographic difficulties arise. The gelatin of a photographic emulsion absorbs radiation of wavelengths shorter than \(\lambda 2265\). Because of this, the sensitivity and contrast of plates are greatly reduced, since the radiation penetrates only a very thin surface layer of the emulsion. To overcome this difficulty, Schumann first attempted to deposit pure silver bromide directly on a glass plate. Such plates proved sensitive but unstable. It turned out that silver bromide was capable of floating off the plate during fixing. Attempts were then made to deposit a layer of silver bromide on the surface of a gelatin layer. As a result, higher sensitivity and contrast were obtained, as well as greater mechanical stability. After a series of further experiments, Schumann (1901) adopted a procedure that changed very little over the many subsequent years. In principle,

it consists in the fact that a silver-bromide emulsion contains a minimal amount of gelatin. The process of preparing such an emulsion is rather complicated and must be carried out with great care if it is desired not to reduce the sensitivity of the plate. The emulsion applied to glass is very easily wiped off, and therefore the plates must be handled with great caution both before use and afterward, unless, of course, they are subsequently coated with varnish.

There is no need to examine here the technical details of preparing Schumann plates. They were subjected later to critical consideration by Lyman (1928). Gonfield (1922) and then Gonfield and Eppliard (1932) proposed certain modifications of Schumann’s method, consisting chiefly in the fact that the Schumann emulsion was applied to the surface of the gelatin (freed from silver bromide) of an ordinary commercial photographic plate or film. Some investigators still prefer to prepare Schumann plates or films for themselves, but the majority consider it easier to buy them from Hilger in England or from Agfa in Germany.

Duclaux and Jeantet (1921) introduced two innovations. In the first of these, Schumannization of ordinary photographic plates is carried out by treating them with sulfuric acid. In this process most of the gelatin in the surface layer of the emulsion is etched away and the layer is enriched in silver bromide. This method proved very useful, but it did not come into general use, since the resulting emulsion is very fragile. In addition, a whole series of other reagents was tried, including an enzyme for dissolving gelatin, but the best of them still proved to be sulfuric acid.

The second of the methods proposed by Duclaux and Jeantet found wide application. In this method, before exposure a thin layer of fluorescent oil is applied to the surface of an ordinary photographic plate, and is removed before the plate is developed. Oils of the most varied types may be used. If plates are being prepared for work in vacuum, it is best to choose an oil with a comparatively low vapor pressure. For greater effectiveness it is desirable that the wavelength of the fluorescent radiation of the oil correspond to the maximum sensitivity of the photographic plate being used. Duclaux and Jeantet themselves did not test oil-coated plates in the vacuum ultraviolet region. This was first done by Lyman (1922), who reports that he achieved a sensitivity down to \(\lambda 500\).

Photographic plates for observing canal rays must in general have the same properties, since charged particles can penetrate only a short distance into the emulsion of an ordinary photographic plate. By the firm Ilford

In England, special plates were prepared on Aston’s order for isotope research (1931, 1937). Such a plate contains an intermediate amount of gelatin between that in the emulsions of ordinary photographic plates and in Schumann emulsions. By means of a classified manufacturing process, a very high sensitivity of the silver bromide grains is now obtained on the surface of the emulsion. Such plates are sold in three grades—\(Q1\), \(Q2\), and \(Q3\), arranged here in order of increasing sensitivity and grain size.

At present one can order from Eastman spectroscopic plates with special sensitization for the extreme ultraviolet rays. The emulsion is coated at the factory with a fluorescing material, which has to be washed off with ethyl chloride before development. Such a plate is in many respects similar to an oil-coated plate, but is more convenient and more homogeneous. Such sensitization is usually imparted to Eastman type \(O\) spectroscopic plates with various degrees of sensitivity and grain size. It is quite possible to combine this ultraviolet sensitization with other types of Eastman spectroscopic plates, even if it is desired to record on one and the same plate radiation from two widely differing wavelength regions. Such a problem may arise, for example, when one wishes to make direct comparisons between the red lines of cadmium and its lines in the vacuum ultraviolet region.

Comparison of the properties of various plates is not so simple a matter as compiling a list of their sensitivity thresholds at different wavelengths, since for many spectroscopic purposes gradation is more important than the sensitivity threshold. A very interesting study of the properties of various types of plates was carried out by Hunter and Pierce (1938) for a series of wavelengths from \(\lambda 2500\) to \(\lambda 2000\). They found that some plates with a very high sensitivity threshold in this region have an extremely low maximum density and therefore are not suitable for contrast photography. In this spectral range this is especially true for the ordinary type of photographic emulsion. This was to be expected, in view of what has already been said about the opacity of gelatin and the shallow penetration of short-wavelength radiation into ordinary emulsions. Unfortunately, at still shorter wavelengths such a systematic comparison has not yet been carried out. The experiments of the author of this article were limited to Hilger-Schumann plates, Ilford \(Q\) plates, and, more recently, Eastman plates sensitized for ultraviolet rays. The first of these plates have, undoubtedly, the highest sensitivity threshold, but they are the most expensive and the least homogeneous. At present both Ilford and Eastman plates are used. The light sources now employed are not fully suitable for quantitative comparison of such

plates, but the experience available shows that both kinds are satisfactory for general use.

Skinner and Johnston (1937), in investigations of soft X-rays in the region from \(\lambda 300\) to \(\lambda 100\), used Hilger–Schumann plates, Ilford Q plates, and plates with oil films. They found that the Q plates are 100 times more sensitive at \(\lambda 200\) than oil plates in the same region, and are comparable with Schumann plates. Further, they established that, owing to absorption by gelatin, the sensitivity of Q plates at \(\lambda 400\) is only 10 times greater than the sensitivity of plates sensitized with oil.

In the case of short-focus instruments, plates must have high resolving power if it is desired to use the full resolution of the grating. In this property the Schumann plates are the best. Bowen measured on a Schumann plate lines spaced \(0.01\) mm apart. The resolving power of Ilford Q plates is somewhat lower. In the case of oil-coated plates one should expect a considerably lower resolving power. The layer of fluorescent material with which plates sensitized for ultraviolet rays are coated is very thin and therefore cannot greatly reduce the resolving power of the emulsion on which it is deposited. Eastman plates for ultraviolet rays, used by the author up to the present, have emulsions of type O. Apparently, they have sufficient resolution for a two-meter-focus grating with 90,000 lines ruled at 30,000 lines per inch, but they cannot provide the higher resolving power of Schumann plates mentioned above.

Harrison (1925) tested oil-coated plates and films for uniformity of blackening and found that, for photometric purposes, they are superior to Schumann plates. Harrison and Leighton (1930, 1931) found that the characteristics of oil-coated plates are constant over a wide region of the spectrum where the radiation is completely absorbed by the oil, and represent the characteristics of the initial emulsion for the wavelength of fluorescence of the oil. Further, over a broad interval of wavelengths the quantum yield of the fluorescence is constant. From this follows the favorable, almost unique situation according to which the methods of heterochromatic photographic photometry in the vacuum ultraviolet region are considerably simpler than in the more accessible regions of the spectrum.

Photoelectric observation of radiations in the vacuum ultraviolet region of the spectrum is also somewhat simpler than for greater wavelengths, since at such short wavelengths most metals emit electrons, and investigators do not have to confine themselves to alkali metals. With the aid of a suitable choice of metal

one can make the photoelement insensitive to radiation in the visible and near-ultraviolet parts of the spectrum. Platinum is often used for this purpose. If it is thoroughly degassed, its photoelectric threshold is at \(\lambda 1962\). If, however, the platinum has not been degassed, its threshold varies from \(\lambda 2800\) to \(\lambda 3000\). A table of values of the photoelectric threshold for many metals, under various surface conditions, is given in the book by Joos and Doebridge (1932, p. 75). Such photoelectric observation in the vacuum ultraviolet region was used in earlier investigations by Lenard (1902), Franck and Hertz (1914, 1916), and others. In these early investigations on critical potentials, a distinction was not always made between ions accumulating at the anode and photoelectrons emitted by this anode under the action of short-wave radiation. In order to distinguish these two effects, Davisson and Goucher (1917) and others devised combinations of electrodes. Of particular interest in this connection is the experiment of Olmsted and K. T. Compton (1923). These investigators determined the “radiation potentials” of atomic hydrogen, i.e., the energies (in electron-volts) required to excite successive members of the Lyman series of hydrogen. The lines under consideration correspond approximately to \(\lambda 1215\), \(\lambda 1025\), \(\lambda 972\), \(\lambda 949\), etc. Olmsted and Compton detected these radiations by means of electrons emitted by a platinum anode. To be assured of complete dissociation into atomic hydrogen, excitation was carried out in a small furnace at \(2800^\circ\text{K}\).

Photoelectric detection of total radiation not resolved into a spectrum in the vacuum ultraviolet region is a comparatively common procedure, but its combination with a spectrograph has seldom been used. Dojell (1934) used this combination in studying the transparency of fluorite and quartz, and more recently Preston (1940) used it for measurements of the absorption coefficient of certain gases at \(\lambda 1215\).

Pfund (1926) used a thermopile to measure radiation in a study of the reflecting power of materials in the region of these wavelengths, but without spectral resolution.

The photochemical formation of ozone from molecular oxygen under the action of Schumann radiation has been used by many investigators to measure the intensity of radiation. As is known, the quantum yield of this reaction is equal to 2. The radiation enters a vessel containing oxygen in an amount sufficient to ensure complete absorption. The gas flows continuously through the absorption vessel into suitable chemical apparatus for the quantitative determination of ozone. This constitutes the most direct method of determining intensity from the number of quanta per second. We shall return to the question of photochemical reactions in the last section.

It has already been mentioned above that thermoluminescent observation of radiation near \(\lambda 1200\) is possible. Using the process described by Gofman (1897), Lyman (1935) prepared a mixture of calcium sulfate, several percent of manganese sulfate, and a small quantity of water. This mixture was dried and its residue was ground into a powder, which was again mixed with water and in this form applied to copper strips of suitable dimensions. After the water had evaporated, the copper, together with its coating, was heated to red heat for several minutes. The plate thus cooled has the ability to store the energy of short-wave radiations falling upon it and to release it in the form of ordinary light when the plate is heated. As has already been indicated, Lyman found that the long-wave limit of sensitivity of the thermoluminescent plate lies at \(\lambda 1300\). Exposure to short-wave radiation may be carried out in daylight. The luminescence may then be observed in the dark when the copper strip is heated on an electric hot plate heated to \(180^\circ\). After the luminescence has ended, the plate is again ready for use. To ensure continuous recording, a photographic plate may be placed at a distance of one millimeter from the surface of the calcium sulfate; however, it must not touch the sulfate surface.

Lyman’s experiment with a spark and a thermoluminescent plate of the type described here constitutes a very effective lecture demonstration.

6. CONSTRUCTION OF THE SPECTROGRAPH

The most important factors determining the construction of vacuum spectrographs include the requirements of spectral range, dispersion and resolving power, flexibility and convenience in use, as well as the limitations imposed by the prism and grating and by the cost of the apparatus. Three types of instruments are used: the prism spectrograph, the grating spectrograph with normal incidence, and the grating spectrograph likewise using a grating, but with grazing incidence. Typical examples of each of these will be considered in sequence, first as optical instruments, and then additions will be made concerning the corresponding vacuum systems.

The prism instrument is usually used for work in the Schumann region, i.e. up to the transparency limit of fluorite, \(\lambda 1250\), or of lithium fluoride, \(\lambda 1200\)—\(\lambda 1100\), depending on the specimen. A very convenient form was described by Carrio and Schmidt-Ott (1931); it has found wide application. In this instrument a small fluorite prism with an angle of \(60^\circ\) and fluorite lenses 8 mm in diameter with a focal length of 100 mm were used. Such an instrument is quite sufficient for the investigation of simple line spectra.

of spectra and absorption bands in the Schumann region. It does not give sufficient resolution in the study of complex atomic spectra or the rotational structure of bands of diatomic molecules, but for a whole range of spectroscopic observations it is very suitable, including measurements of intensities in simple emission or absorption spectra.

Instruments were also made with prisms of larger dimensions, but they have no advantage over small instruments and, moreover, require a considerably longer time for evacuation. In large prism instruments the freedom from higher orders at short wavelengths does not outweigh the greater wavelength range of the instrument with a grating and its almost constant dispersion.

The theory of the concave grating is well known, and therefore there is no need to repeat it here. It was systematically presented by Kayser (1900), and some of its important features were later considered by Dyke (1933) and MacAdam (1933). In the usual arrangement of the Rowland circle the slit, the grating, and the plate—all lie on one circle, whose diameter is equal to the radius of curvature of the grating. The well-known equation

\[ m\lambda = e(\sin \varphi - \sin \psi) \tag{1} \]

determines the position of the \(m\)-th order of the wavelength \(\lambda\); \(e\) here denotes the distance between successive lines of the grating (for convenience expressed in angstroms), \(\varphi\) is the angle of incidence and \(\psi\) the angle of diffraction. (In this form of the equation, \(\varphi\) and \(\psi\) are taken as positive on opposite sides of the normal.) If \(s\) denotes the distance from the direct image of the slit \((\varphi=\psi)\), measured along the arc of the Rowland circle to some point in the spectrum, then the linear dispersion \(ds/d\lambda\) at this point is given by the equation

\[ ds/d\lambda = \rho/e \cos \psi, \tag{2} \]

where \(\rho\) is the radius of curvature of the grating. For a grating with 30,000 lines per inch and with a radius of curvature of \(2\ m\), both these equations have the numerical values:

\[ m\lambda = 8466(\sin \varphi - \sin \psi) \tag{1a} \]

\[ ds/d\lambda = 0.2362/\cos \psi\;(\text{in mm}/\text{\AA}). \tag{2a} \]

It should be noted that the dispersion is relatively constant near the normal to the grating, but increases rapidly at large values of \(\psi\). Geometrically, the dispersion at any point of the circle is inversely proportional to the length of the chord drawn from this point to the center of the grating. Of course, it does not depend on \(\varphi\), and therefore the value of \(\varphi\) may be chosen so as to bring a given wavelength into a position giving the desired dispersion. Beutler (1940) developed a convenient method

to find the value of \(\varphi\) corresponding to any given spectral interval.

A typical arrangement of the vacuum spectrograph of K. T. Compton and Boyce (1934), with a grating and normal incidence of the beam, is shown in Fig. 3. A glass grating with 30,000 lines per inch has a radius of 2 m. The dispersion (0.2362 mm/Å at the normal) and the resolving power are apparently sufficient for complex line spectra and for the study of the rotational structure of diatomic molecules. The instrument can be evacuated by means of a vacuum system (a Distillation Products pump with its own distillation, having a speed of 200 l/sec, operating from a Cenco Hypervac fore-vacuum pump) in the course of one hour. For spectrographs of considerably larger size, pumps are avail—

Fig. 3. Spectrograph with normal incidence \((\varphi = 13.7^\circ)\) (K. T. Compton and J. K. Boyce, 1934).

able with a greater pumping speed. Existing spectrographs more than 3 m long usually have to sacrifice wavelength range in order to gain high dispersion. But large spectrographs may have considerable advantages in the accuracy of determination of wavelength standards and in the study of rotational structure in the spectra of polyatomic molecules (Duncan and Harrison, 1936).

In the apparatus shown in Fig. 3, one photographic plate 24 inches long covers the region from the direct image of the slit \((\varphi = \psi)\) to \(\lambda 2500\). With this instrument one can observe spectra down to wavelengths somewhat below \(\lambda 300\). It should be noted that the angle of incidence here was chosen equal to \(13.7^\circ\), the basis for which was the wavelength region for which operation was planned when the spectrograph was designed in 1929. Subsequent investigations by Dyke (1933) showed that an angle of incidence of \(13^\circ\) gives the greatest astigmatism. It is very convenient in adjusting the spectrograph for obtaining a direct image of the slit on the plate. However, it would be useful, using the design of Simeon (1923),

add a second slit at a somewhat larger angle of incidence, \(22.2^\circ\). Then, with one and the same position of the plate, it would be possible to obtain also a second interval from \(\lambda 1200\) to \(\lambda 3700\). This would be advantageous for photographing the Schumann region simultaneously in the first and second orders in this new series, and also for passing from one spectral series to another without changing the optical arrangement. For gratings with a radius of curvature of \(1\ \text{m}\), several vacuum spectrographs with normal incidence were constructed. In this case, however, difficulties arise because of the need to have glass plates corresponding to a focal curvature of \(50\ \text{cm}\). With still smaller instruments it is necessary to use films. Other difficulties are connected with the necessity of ruling gratings on a surface with a very small radius of curvature.

The vacuum spectrograph with grazing incidence has been widely used in many laboratories, chiefly in Uppsala. Of the various constructions, only the oldest and the newest may be described here. Erikson and Edlén (1930) described a spectrograph constructed by Siegbahn. The apparatus has an angle of incidence of \(80^\circ\) on a grating of radius of curvature \(1\ \text{m}\), containing 571 lines per millimeter; it is suitable for wavelengths down to \(\lambda 75\).

Fig. 4. Spectrograph with grazing incidence \((\varphi = 80^\circ)\) (diagram).

Recently two almost identical instruments were constructed with gratings of 5-meter radius of curvature. One of them was used for the investigation of soft X-rays and was described by Siegbahn and Magnusson (1935). The other was used by Edlén (1936 and following) in his investigations of the spectra of very highly ionized atoms. The grating has 37,440 lines, 576 lines per millimeter. The angle of incidence is \(\varphi = 86^\circ\). The long-wavelength limit of the plate falls at \(\lambda 200\), and the lower limit at \(\lambda 40\). In another instrument, described by Tyren (1936), the angle of incidence was \(89^\circ\). At this extreme angle Tyren (1938) observed the Fe XVII line at \(\lambda 12.1\). The largest of all the grazing-incidence instruments was constructed by Kruger (1933); in this instrument a grating with a radius of curvature of 21 feet was used.

The gratings used in Uppsala have other periods and radii of curvature than the grating in the normal-incidence instrument shown in Fig. 3. For convenience of comparison, Fig. 4 is made to the same scale; it illustrates the use of a grating of 2-meter—

of curvature, with 30,000 lines per inch at an angle of incidence \(\varphi = 80^\circ\). Let us also give a numerical comparison. In an instrument with (nearly) normal incidence (Fig. 3, \(\varphi = 13.7^\circ\)) the distance from the direct image to the position of the first order at \(\lambda 500\) is 119 mm, and the dispersion at \(\lambda 500\) is \(0.240\) mm/Å. In the case of an instrument with grazing incidence (Fig. 4, \(\varphi = 80^\circ\)) the same spectral range would cover 426 mm on the plate, and the dispersion at \(\lambda 500\) would be \(0.625\) mm/Å. Since at grazing incidence the spectra of high orders are strong, which can greatly complicate observations at large wavelengths, in practice the angle of incidence has to be made only large enough to obtain total reflection for the shortest wavelengths that it is desired to obtain. For each individual grating the angle may also vary. Edlén (1934) gives the case of one metallic reflecting grating which could be used at normal incidence only down to \(\lambda 320\). But at an angle \(\varphi = 60^\circ\) it could be used down to \(\lambda 160\), at \(80^\circ\) down to \(\lambda 75\), and at \(84.6^\circ\) down to \(\lambda 53\).

The investigations of Mack, Sten, and Edlén (1932), Bowen (1933), and also Anderson and Mack (1934) showed that the resolving power of a concave grating increases with increasing angle of incidence. For nearly grazing incidence there exists an optimum grating width for maximum resolving power and intensity. It depends on the angle of incidence, the radius of curvature, the position of the grating surface, the wavelength, and the order of the spectrum. A convenient graph for finding this optimum width was given by Mack and Sten (1933), but Bowen’s calculations showed that when the plate is slightly displaced from the Rowland circle, a somewhat wider grating may be used. The most complete and rigorous treatment of this question was given by Anderson and Mack. The aberrations of a concave grating, characteristic of large angles of incidence, hinder the use under these conditions of the outer orders of the spectrum between the grating and the direct image. When a large grating used at grazing incidence is stopped down in order to satisfy this requirement, a double benefit is obtained. Not only is the resolution improved, but the type of ghosts that arise at grazing incidence is also eliminated. Sometimes it is necessary to stop down the ruled surface of the grating somewhat from above and below as well as at the sides, since these ghosts are apparently produced by the outer edges of the ruled surface. Methods for adjusting instruments with grazing incidence were described in detail by Ratenau and Pericampo (1935).

Spectrographs with both normal and grazing incidence have their own exclusive advantages. In the Schumann region the nearly linear dispersion and the saving of light caused by the smaller astigmatism are decidedly favorable for instruments with normal incidence. Below \(\lambda 600\) the considerably increased efficiency

the grating at normal incidence definitely favors this type of instrument, and below \(\lambda 300\) it alone will operate. In the region from \(\lambda 1200\) to \(\lambda 600\) both instruments are usable. In this region, if a precise determination of wavelength standards is required, the instrument with normal incidence has the advantage that the light falls on the photographic plate almost normally and the wavelength scale is very little distorted by small irregularities in the curvature of the plate. If observation of weak lines is required, then in this spectral region a considerably greater intensity is obtained in the case of the instrument with normal incidence. It also appears that the instrument with normal incidence is more satisfactory for a range of wavelengths in which molecular spectra of rotational structure are studied. It should be noted that, within the range of wavelengths calculated for instruments with normal incidence, there are still considerable variations in the angle of incidence. Because of this, two instruments or two sets of parts for one instrument may be required. Edlén (1934) described an instrument with an interchangeable slit mounting, calculated for angles of incidence of \(70^\circ\) and \(80^\circ\).

For housing the optical system in a vacuum spectrograph, the most varied types of housings are used. It is not necessary here to go into a description of the details, and one may confine oneself to only a few general remarks. The housing should be as simple as possible, and all seams or joints should be as accessible as possible, so that they can be tested separately in the event of a “leak.”

A comparison of Figs. 3 and 4 shows that at large angles of incidence it is necessary to evacuate a volume with a smaller fraction of the area of the Rowland circle. However, although the volume to be evacuated in the case of an instrument with grazing incidence is smaller, this does not mean that the instrument takes up less space in the laboratory. In an instrument of this type, it is apparently more convenient to mount the grating and the holder for the plate on a crossbar in such a way that the larger part of the housing can roll away from the crossbar. For this a rail is required that is twice as long as the optical part of the instrument.

In an instrument of any type, the housing must have wide internal clearances so that, if desired, adjustment or an unforeseen replacement of some optical parts by others with somewhat different dimensions or focal properties can be carried out. The optical devices must have good indicating mechanisms so that the setting can be reproduced with sufficient accuracy. This is important because, in the course of focusing, it is often necessary to evacuate the instrument for each separate test. It is necessary to ensure the possibility of carrying out any adjustment and setting inside the spectrograph without any inconvenience. It is also necessary to ensure the possibility of removing the grating for cleaning and of replacing it with a minimum number of applications to the former position.

The slit must be readily accessible for cleaning and adjustment, if possible without loss of vacuum in the spectrograph. Adjustment of the slit width and its rotation in order to ensure its parallelism to the ruling of the grating are of extremely great importance for attaining maximum resolving power. These two adjustments must be independent of one another. The slit adjustment must be such that its width can be reduced to several thousandths of a millimeter. All these requirements relating to the slit should be satisfied while at the same time economizing space, so that the tube serving as the light source can be fixed in the immediate vicinity of the slit.

With many light sources, and especially in the case of vacuum operation, the slit becomes coated with a deposit of electrode material. In the case of a vacuum spark the slit requires frequent and thorough cleaning. Bowen finds it very advantageous to make the edge of the slit of stellite (an alloy of tungsten with nickel and cobalt), since it can be regularly and reliably cleaned with any of the usual concentrated acids. Moreover, it is very hard and therefore can also withstand vigorous mechanical cleaning well. Slit edges made of this material were used by Bowen for years and remained intact, whereas slits made of other materials require frequent replacement.

To isolate the slit and the source from the rest of the spectrograph it is desirable to use some sort of valve, but there is no necessity to make such a device so that the plate can be removed from the apparatus without admitting air into it, at least in the case of a 2-meter apparatus. It is incomparably simpler to use a plate wide enough that two exposures can be obtained on it. External adjustment of a mechanism designed to move the plate between exposures was described by Bomke (1937, p. 37). The use of any vacuum gauges—ionization, thermal, or McLeod gauges—is fully offset by the saving of the time that otherwise would have to be spent searching for possible leaks. To monitor the operation of the diffusion pump it is quite sufficient to use a gauge in the fore-vacuum section. In vacuum systems for cyclotrons a thermoelectric gauge is used for this purpose.

7. LIGHT SOURCES

Light sources for studying emission spectra in the vacuum ultraviolet region do not differ very greatly from the light sources used when working with longer wavelengths.

The only necessary additional condition is merely to ensure that the radiation can pass through the region of space immediately adjacent to the light source without excessive absorption of the rays by the gases and vapors present there. In the case of an arc or

of sparks ordinarily operating in air at atmospheric pressure, it is necessary to replace the air by some gas transparent at short wavelengths. As already mentioned, this technique was developed by Selwyn (1929) and subsequently improved by Shenstone (1938). Depending on the range of wavelengths to be investigated, nitrogen, argon, or helium are suitable for this purpose; but if monatomic gases are used, the characteristics of the discharge may change appreciably. Data on the transparency of these gases have already been given in Section II. Another problem has to be faced in investigating the spectra of ionized atoms in the region of the absorption wavelengths of the corresponding neutral atom. This difficulty is partly removed by reducing the density of the absorbing atoms, either by diluting them with a gas or vapor transparent in this region, or by manipulation at low total pressure.

If it were possible to indicate, by means of one or several references, where information on spectroscopic light sources in general might be found, and then to rework this information with allowance for its application in the vacuum-ultraviolet region, this section would be the shortest. However, in accordance with established spectroscopic tradition, such information is transmitted from one investigator to another for the most part orally; and if it were set out in an article, it would more likely be empirical in character. At the risk of imprecision in defining what, in the present case, should be understood as general knowledge, one may nevertheless attempt to give a general idea of the mechanism of excitation of spectra and, on this basis, to consider the light sources required for the excitation of various spectra.

Atoms and molecules can emit light whenever they possess excess energy. The probability of emission of this radiation has widely varying values depending on the state to which the atom (or molecule) has been excited. In some states this probability is so high that radiation almost certainly occurs in the short time during which the atom or molecule remains undisturbed between collisions. In other (metastable) states this probability of emission is very small, and therefore the occurrence of radiation is extremely unlikely, unless one counts such conditions when some collision frees the atom or molecule from the excess energy. If we disregard nuclear processes, in which gamma rays or X-rays are emitted, an atom or molecule can receive energy in the following four ways: (1) in collision with an electron of sufficient kinetic energy (collisions of the first kind); (2) by absorption of radiation of suitable frequency; (3) in collisions with other atoms or molecules that have sufficient quantities of energy (collisions of the second kind); (4) by thermal excitation, in which part of the intrա

energy of the system is used to excite some of the atoms or molecules composing it. It should be noted that, in thermal excitation, all these processes occur. Excitation may be regarded as thermal when all the processes are in sufficient equilibrium to justify a thermal rather than an individual interpretation. In the spectra of flames and furnaces the means of excitation is (4). In fluorescence spectra the chief role is played by (2). But for the majority of sources, where the energy is supplied electrically, the primary mechanism is (1), even if something is changed as a result of other processes.

The general properties of electric discharges in gases were considered by K. T. Compton and Langmuir (1930, 1931), and more recently by Druyvesteyn and Penning (1940), and therefore there is no need to discuss them here again. It will suffice only to say that the amount of excess energy that can be stored in an atom by excitation or ionization, or by both these means, ultimately depends on how often atoms can receive impacts from electrons. How strongly an electron can strike an atom will depend on the amount of kinetic energy acquired by the electron during its accelerated motion in the electric field and lost upon slowing down as a result of collisions with atoms. As a result of this contest, the average electron attains a certain final velocity, which depends on the intensity of the electric field, on the gas density, and on the cross section of the atom for collisions with electrons of this velocity. Subsequent streams of electrons striking an atom before it has sufficient time to discharge its excess energy by radiation or by collisions lead, cumulatively, to appreciably higher energies than in simple collisions at the very same final velocity.

Another part of the competition between the processes by which an atom gains energy and the processes by which energy is lost in the form of radiation or in collisions determines the statistical distribution of atoms between those in the normal state and those that have various states of excess energy. This distribution, together with the probabilities of atomic transitions associated with each process of radiation from each excited state, determines the spectrum of the emitted radiation. In mixtures of gases or vapors, where the pressure is sufficient for collisions between molecules to occur more often than impacts of molecules against the walls of the vessel, the spectrum will consist chiefly of the spectrum of the element with low states of excitation. This occurs for two reasons: impacts (of the first kind) between these atoms limit the final velocity of the electrons, and impacts (of the second kind) rapidly transfer energy when it is acquired by an atom with higher states of excitation. Metastable states represent

are the most important stages in the processes of cumulative excitation and are the most effective for storing energy during its transfer in collisions of the second kind. Metastable states occur in many atoms (and molecules) and are especially important in the inert gases, mercury, and also in molecular and atomic nitrogen.

Before an electric discharge begins to pass through a gas, the electric field is determined by geometrical factors and by the applied potential difference. Once the discharge has begun, however, this field distribution changes considerably as a result of the concentration of positive ions. (Even at high frequencies of an oscillatory discharge, the concentration of ions plays a role in creating the field distribution.) The layer of ions formed near the cathode constitutes the boundary of the so-called cathode fall. If the discharge takes place in an arc, the potential difference within the cathode fall is of the order of ten volts. If the discharge is a glow discharge, the cathode fall is measured in hundreds of volts. However, whether we are dealing with an arc or with a glow discharge, the character of the discharge is determined by a combination of factors, among which the leading role belongs to the constants of the external circuit. In the cathode fall of an arc, the electrons acquire final velocities of the order of the ionization or excitation potential of the gas; in the region of the cathode fall in a glow discharge these velocities are considerably higher. The remaining free space between the electrodes (which is supplementary to that occupied by the cathode fall) is occupied by the negative glow, the Faraday dark space, and the positive column, which appears when sufficient space is available. The positive column is a region of considerably smaller electric field than the cathode fall; at a given pressure and a given current it is the same for an arc and for a glow discharge. The final velocities of the electrons here are lower than in the region of the cathode fall and in the other part of the negative glow, but they have a Maxwellian distribution about the mean value, so that some of them have sufficient energy to cause ionization. The electron temperature characteristic of this Maxwellian distribution of electron velocities determines the distribution of excited states produced by simple collisions of electrons with atoms (or molecules) of the gas. For gas pressures of the order of millimeters of mercury, there is no equilibrium between the high temperature of the electrons and the much lower temperature of the gas, but for pressures of the order of one atmosphere this equilibrium is established, and then, according to process (4), excitation can be calculated on a thermodynamic basis.

T. Compton and Boyce (1928) investigated the spectra obtained in simple collisions with electrons of known energy in gases at low pressures. As was to be expected, spectral lines appeared only in the case when the electrons had sufficient

the exact energy for exciting a neutral atom from its ground state to the corresponding excited state of the neutral atom, or directly to a definite excited state of the singly ionized atom. For the gases studied—neon, argon, and nitrogen—in not a single case did the spectrum of a doubly ionized atom appear, even when the energy was sufficient for their excitation. It follows from these investigations that, if excitation were due only to simple collisions, one could expect to obtain the spectra of neutral and singly ionized atoms in the negative glow of a low-pressure glow discharge, the spectra of neutral atoms only in the negative glow of a low-pressure arc, and in the positive column of both types of discharge. In fact, however, spectra are observed for higher states of ionization than those indicated here, both in the negative glow and, less intensely, in the positive column. The obvious conclusion, that excitation is cumulative, is confirmed by further evidence that excitation is stronger for those atoms in which conveniently situated metastable states provide intermediate “resting places” in a two-stage process. At high pressures, where thermal equilibrium is established between the electrons and the gas molecules, the conditions are very favorable for cumulative processes.

Everything just set forth applies to discharges with alternating current, provided only that the frequency is not too high, and also to discharges with direct current. In the case of alternating current, the negative glow and (if there is space) the positive column are established every half-period in the proper positions. The light from any place in the tube, whether observed by the eye or by means of a spectrograph, is a mixture of what is obtained from the half-periods of each polarity. When the frequency is so high that the path of an electron during a half-period becomes smaller than the distance between the electrodes, an individual electron can oscillate back and forth in the tube for many periods. Such a high-frequency discharge will occur even when the electrodes are outside the vessel containing the gas, provided, of course, that the oscillating electric field includes the region occupied by the gas. If the oscillating electromagnetic field is produced by a coil located outside the tube, then long curved paths become possible for the electrons, and an electrodeless “ring discharge” can be maintained even at comparatively low frequencies.

For the excitation of solid materials the vacuum furnace of King (1908, 1922) is used. It was widely used by King for the selective excitation of groups of emission lines from successively increasing states of excitation in the neutral atom, which data proved of inestimable help in the analysis of complex spectra.

Because the second ionization potential of the rare earths is only slightly higher than the first ionization potential, both the first and second spectra of these elements appear in an arc source. The vacuum furnace gives only the first spectrum, on the basis of which other lines in the arc can be assigned to the second spectrum. King’s furnace was not used for emission spectra in the vacuum ultraviolet region, but was used by Paul (1937) in obtaining vapor for absorption spectra in this region.

An arc between solid electrodes, operating at atmospheric pressure, usually gives the spectrum of the neutral atom in the positive column of the arc. Here the excitation is thermal.

For exact measurements of wavelengths, light emerging from points near the electrodes must be excluded from the spectrograph, since strong electric fields near the electrodes cause a “polar shift” in wavelength. This is the Stark effect (Adam, 1932). The strong field near the cathode usually excites (in the negative glow) the spectrum of the singly ionized atom in addition to the spectrum of the neutral atom. The excitation near the electrodes is not thermal, and because of the small magnitude of the cathode fall, the spectrum of the ionized atom is evidently excited by a cumulative process. The line spectrum of certain elements is complicated by bands of oxides and nitrides. They can be suppressed if the arc operates in another atmosphere, for example in nitrogen (for oxide bands), hydrogen, argon, and helium. The use of these gases can increase excitation beyond the excitation in an arc in air. Lines excited in an arc at atmospheric pressure have a slightly asymmetric broadening due to molecular collisions. This shift under the action of pressure can be eliminated if the arc operates at reduced pressures. Such a vacuum arc was investigated as a source of standard wavelengths, but did not become widespread (International Astronomical Union, 1932). If the arc can be maintained at a sufficiently low pressure, when the vapor temperature does not reach equilibrium with the electron temperature, then the reduced Doppler broadening will increase the sharpness of the spectral lines.

In a spark between solid electrodes we have a nonstationary discharge, which then reaches equilibrium in the form of an arc or a glow discharge. At atmospheric pressure this equilibrium is reached in thousandths of a second, and at lower pressures in proportionally shorter intervals of time. In the initial phase of the spark the potential distribution is determined by geometrical factors, but then it gradually changes under the influence of changes in ion concentrations, until a distribution characteristic of one or another state of equilibrium is reached. The spectrograph gives the integral light from all phases. The first phase favors the appearance of “hard” electrons throughout the entire space

between the electrodes, and then higher excitation takes place. To increase the intensity of this useful phase of the discharge, a capacitor is connected in parallel with the spark gap. This capacitor accumulates an electric charge before the initial discharge and acts as a reservoir for it during the oscillations of the discharge. If the other variables remain the same, an increase in capacitance increases the number of “hard” electrons and favors high excitation in cumulative processes. An increase in self-inductance, on the other hand, lengthens the period of the individual oscillations. Owing to this, the current density is reduced, and at the same time so is the time during which the initial field distribution predominates. For these two reasons the excitation decreases. Fowler (1925) considered the question of using variable self-inductance to explain spark lines by different stages of ionization. Gibbs, Fivég, and Gartlein (1929) described its use in connection with the vacuum spectrograph. The discharge voltage of the spark is another variable that can be used to regulate excitation by changing the pressure of the gas in which the spark discharge takes place. The “softest” spark occurs at a minimum discharge gas pressure (for the most part, of the order of tenths of a millimeter of mercury), but the increase in discharge potential obtained in passing to atmospheric pressure is not so great as when the pressure is reduced to about one ten-thousandth of a millimeter of mercury. The “vacuum spark,” introduced by Millikan and his collaborators and subsequently used by Edlén (1936), gives record excitation of eighteen-times-ionized copper (Cu XIX). Numerical values may be cited for two limiting cases of excitation by a spark. In a spark between iron electrodes in nitrogen at atmospheric pressure, Fe II and Fe III are readily excited. In this case a capacitor of 0.1 microfarad is used, charged to approximately 5,000 volts. For Cu XIX Edlén used copper electrodes in a vacuum spark with a 0.4-microfarad capacitor, charged to approximately 50,000 volts. If the vapor pressure of the electrodes (or the pressure of the permanent gas) is too high to maintain the desired initial voltage, it can be maintained artificially by connecting a “hard” spark gap, or even some kind of switch, in series with the existing spark gap.

Not all elements are suitable or available for use as solid electrodes. Solid substances may be placed in the hollow core of an electrode made of a more suitable material, or may be added in solution to a carbon electrode, or alloyed with other substances, or pressed with silver powder or powders of other metals. Electrodes of the last type may also be prepared without sintering. Care should be taken to avoid compounds

or mixtures that may explode under pressure*). Evaporation always occurs at the cathode, and sometimes just as well at the anode. In a carbon arc it is stronger at the anode.

The hollow cathode of the Schuler tube is especially well suited for the complete development of spectra of singly ionized atoms (Sawyer and Paschen, 1927). It is used for investigations in the vacuum ultraviolet region by many researchers, chiefly by Shenstone and co-workers (Shenstone, 1936, 1938; Green, 1939, and others). In such a discharge the negative glow is concentrated inside the hollow cathode, and the positive column is absent. The cathode material (or material placed inside it) evaporates under bombardment by positive ions. The excitation can be regulated by changing the gas pressure, by changes in the external circuit, and by the choice of gas (helium, neon, or argon). Collisions of the second kind play a significant role in the excitation process. Green (1939) used a special external direct-current circuit for the Fe II spectrum, as Shenstone (1938) had done for Pt II. Subsequently this circuit was described by Green and Cooper (1940).

In the spectra of the Schuler tube, lines of the lowest excitation of the third spectrum (doubly ionized atoms) often appear. The excitation of this third spectrum was increased by Gartlein and Gibbs (1931) by introducing a spark gap, initially closed, in series with the Schuler tube. When, during prolonged operation of the Schuler tube, heating of the cathode occurred, the gap was left only slightly open.

Materials with a considerable vapor pressure (for example, mercury, or even chlorides of some metals) can be used in an electrodeless discharge if it is initiated in an inert gas. When the tube is heated, the salt evaporates in approximately the same way as sodium evaporates during the initial discharge in neon in an ordinary commercial sodium lamp.

If the element under investigation can be introduced in the form of a gas or vapor—provided, of course, that it does not act on the walls of the tube, the windows, or the electrodes—then excitation becomes comparatively easy. In this case electrodes of large area are often used in order to obtain strong currents, and the spectrum of the positive column is observed in a narrowed part of the tube to increase the current density. For the vacuum ultraviolet region this part is usually viewed along the tube. The excitation changes with changes in the external circuit and with changes in gas pressure. The various possibilities may be illustrated by two numerical examples. In commercial neon tubes the neon pressure is about one centimeter of mercury. For small tubes, several feet long, up to—

* An introduction to the possibilities of powder metallurgy may be found in Proc. of the Conference on Powder Metallurgy, Massachusetts Institute of Technology, 1940.

it is sufficient to use a transformer of 5–10 thousand volts without employing a capacitor. In this case a spectrum is obtained almost exclusively from the neutral neon atom (Ne I). For the spectrum of quadruply ionized neon (Ne V), Paul (1939) used a mechanical switch to discharge a capacitor of one microfarad through a tube containing neon at approximately one hundredth of a millimeter of mercury. A serious limitation in the excitation of spectra of highly ionized gases is contamination, chiefly by oxygen and silicon, arising from bombardment of the walls of the discharge tube.

Other gaseous radiation sources are also useful. For the first and second spectra of gases one may use a Schuler tube, provided that the current is kept sufficiently low that evaporation of the cathode material does not occur. For excitation of the second, third, and fourth spectra of gases, an electrodeless discharge is useful. Some selection of ionization stages can be obtained by varying the pressure from the optimum value (perhaps to one tenth of a millimeter of mercury) for maximum excitation. In connection with gaseous sources, one should not forget the general remarks made in the first paragraph of this section concerning transparency in the spectral regions under investigation.

Investigations of absorption spectra in the vacuum ultraviolet region require a source of a continuous spectrum in this region. A discharge in hydrogen at high current gives a continuous spectrum extending from the visible region to \(\lambda 1600\). Goldfield (1930) found a similar continuous part of the spectrum in helium in the region from \(\lambda 900\) to \(\lambda 600\). Beutler (1933) used this continuous part for his absorption spectra. Lyman (1924, 1928, p. 49) used the pulsed discharge of a capacitor through a capillary tube with an external gap connected in series in order to obtain a high discharge voltage. This yields a continuous spectrum from the visible region to \(\lambda 900\). Rathenau (1933) increased the electrical power and extended Lyman’s continuous part of the spectrum to \(\lambda 270\). He came to the conclusion that the excitation is of two kinds. At long wavelengths black-body radiation arises from particles flying out from the walls of the tube, whereas at shorter wavelengths there is continuous soft X-ray radiation produced by electron impacts during the discharge. The nature of the gas in the tube plays no essential role, since a large part of the continuous spectrum apparently arises from the decomposition products of the tube walls. Naturally, the capillary tube has a short lifetime. The usefulness of this source was considerably increased when Collins and Price (1934) modified the technique by introducing replaceable quartz capillary inserts into a permanent tube.

Finkelnburg (1933) considered the question of the mechanism by which a gas can give a continuous spectrum.

8. Wavelength Standards

The wavelengths of the spectra of neutral atomic hydrogen and singly ionized helium can be calculated with very great accuracy. Even when wavelengths are calculated on the basis of the simple Bohr model with circular orbits, the results obtained agree very closely with the results of more exact calculations. The calculated wavelengths are very widely used as standards in the vacuum region. The best values were given by Paschen (1929) and Penney (1930). For convenience, Paschen’s data are reproduced in Table 3 of Appendix B.

Although these calculated wavelengths of hydrogen-like atoms (Li III, Be IV, etc., may be used at still shorter wavelengths) for the most part satisfy the requirements of preliminary investigations, they fluctuate too widely for many purposes. The hydrogen lines have the property of being somewhat broadened, which interferes with exact measurements. The lines of ionized helium are difficult to excite, and, moreover, helium of considerable purity must be used. Bowen and Ingram (1926) proposed as standards certain lines of C, N, O, and Al between \(\lambda 1990\) and \(\lambda 599\), and determined their wavelengths by comparing their higher orders with the first order of iron lines in the visible and ordinary ultraviolet regions of the spectrum. Edlén (1933, 1934) used the same method to obtain standard lines of C, O, and Al in the region from \(\lambda 1371\) to \(\lambda 160\). Boyce and Rieke (1935), and subsequently Beber and Watson (1936), used the same method, but at higher dispersion, for a new determination of certain standards (C, N, O, Ar) in the region from \(\lambda 1930\) to \(\lambda 580\).

All methods involving comparison of spectra of different orders introduce a certain error. This error arises from the line profile in different orders. Although this effect may be absent, or be very small, in the case of an ideal grating, it nevertheless exists in practice, as the following consideration shows, and one must try to eliminate it. When a certain spectral line, formed by a concave grating, is examined with one or another side of the grating masked off, it is usually found that the line profile changes slightly. This may be due to a gradual change in the shape of the groove, such as is produced during the ruling of the grating, or even to a slight change in the angle of incidence. The angle of incidence and the angle of diffraction change in going from one side of a concave grating to the other, and the change in the influence of each grating element with this angle must be a function of the wavelength. Therefore, for example, the first order of a certain wavelength and the second order of half that wavelength, falling at one and the same place on the photographic plate, should have slightly different profiles as a result of the different “weight” of the elementary action of many grating elements on the different orders. Under favorable conditions, in the case of some-

...of such gratings, this error is negligible; but since a similar error may also arise from imperfections of focusing, in those cases where the method of overlapping orders is used in measurements it is necessary to know whether such an error is present. Moreover, this check must be made every time the grating setting is changed. This type of error has long been known to spectroscopists, but has sometimes not been taken into account*).

A possible error in comparing overlapping orders can be avoided by means of a method based on the combination principle and first used by A. Fowler. The method may be illustrated schematically as follows. Let \(A\), \(B^0\), \(C\), and \(D^0\) be spectral terms arranged in such a way that all the lines \(A — B^0\), \(B^0 — C\), and \(C — D^0\) can be measured (in one and the same order) with a certain definite accuracy relative to secondary wavelength standards. Then the transition \(A — D^0\) will occur at a considerably shorter wavelength, and the corresponding wavelength can be calculated with great accuracy. For example, if these three lines lie near \(\lambda 5000\) \((20\,000\ \mathrm{cm}^{-1})\) and can be measured with an accuracy of \(0.05\) Å \((0.2\ \mathrm{cm}^{-1})\), then the fourth line will lie near \(\lambda 1666\) \((60\,000\ \mathrm{cm}^{-1})\) and can be measured with an accuracy of \(0.016\) Å \((0.6\ \mathrm{cm}^{-1})\). Shenstone (1936) made extensive use of this method in his investigation of the second spectrum of copper. He listed about 95 lines between \(\lambda 1663\) and \(\lambda 685\), which he succeeded in observing in the spectrum of a hollow cathode in helium and for which he calculated wavelengths by means of the combination principle. The approximate error for these 95 lines ranged from \(0.001\) Å to \(0.004\) Å. Shenstone’s standard copper lines are listed in Table 1 of Appendix B. Only a few of these lines appear in the copper spark in nitrogen. Green (1939) investigated the vacuum ultraviolet part of the second spectrum of iron, excited in an iron spark in nitrogen. He found 110 lines (see Table 2 of Appendix B) between \(\lambda 1960\) and \(\lambda 1550\), whose wavelengths could be calculated by the same method and with the same accuracy. It is easier to work with the spark in nitrogen than with the hollow cathode. It would be very useful to find similar standards, in order to fill at least part of the copper series, among the lines of some other element that could be excited in a spark in nitrogen.

Mour and Rieke (1936) compared the standards of Boyce and Rieke, as well as of Webber and Watson, with Cu II standards, using lines of one and the same spectral order. In this case no ob—

*) A second, related type of error is better known. It may arise when the grating is illuminated by the source of the standard spectrum not in exactly the same way as by the source of the spectrum under investigation. This error is apparently also caused by differences in the line profile. It can be avoided by simultaneous excitation of the lines of both spectra.

no systematic error in the coincidences was found. Therefore Mour and Rieke recommended using certain mean values from their own results and from the results of two earlier investigations. They are given in Table 4 of Appendix B. Such mean values have probable errors from 0.002 to 0.005 Å and, generally speaking, should be regarded as less accurate than the values for the copper and iron standards of Shenstone and Green, although for many purposes they are quite sufficient.

MacAdam (1936) attempted to use a reflecting echelon for the interferometric determination of standards in the vacuum ultraviolet region. He succeeded in obtaining interference patterns for some of the Cu II lines studied by Shenstone and confirmed by the calculated wavelengths. The continuation of MacAdam’s work is included in the program of further work in the author’s laboratory, but it must be admitted that the work will be very difficult, since one will have to deal with high orders of interference. Apparently no other type of interferometer is suitable for this region of wavelengths.

It is very probable that in the near future a large part of the wavelength standards for the vacuum ultraviolet region will be determined by means of the combination principle or, as in the case of the investigations now being carried out by Edlén, by successive applications of the combination principle in order to reach still shorter waves. Edlén is measuring the spectra of a whole series of light elements in the interval covered by Green’s iron standards. Therefore, the application of the combination principle within the spectra of these light elements will make it possible to determine new standards at considerably shorter wavelengths.

A method based on the combination principle can yield the greatest advantages if the visible and near-ultraviolet lines are measured interferometrically. Fortunately, this is being done for many copper and iron lines. It is also very important that the spectrum under study be as free as possible from hyperfine structure.

9. ATOMIC SPECTRA

Most of the experimental data already obtained concerning the vacuum ultraviolet region relate to atomic spectra. A critical bibliography of these data was compiled by the author with the valuable assistance of Dr. Lora Mish. An earlier bibliography of atomic spectra in all wavelength regions was published by Gibbs (1932). Only a few of the papers listed by Gibbs contain data on wavelengths shorter than \(\lambda 2000\). Our bibliography refers only to wavelengths below this, perhaps arbitrary, boundary, but even in it those of the listed—*

Boice articles, which contain material that later became known with greater accuracy.

The articles are listed here by elements, and, where possible, also by degrees of ionization of the given element. In addition to the usual references to authors and to the places of publication of the articles, some lines and series of wavelengths are also given, and, moreover, generally speaking, it is indicated whether the lines have been classified in the term system or not. Where possible, the latest values of the ionization potential are also included. Articles on emission spectra are marked with the symbols A, B, C, denoting the evaluation that the author gives them on the basis of the number and accuracy of the wavelengths contained in the article.

The symbol A denotes an investigation that may be regarded as decisive in the sense that it contains, with sufficient accuracy, such data on the vacuum ultraviolet region as are necessary for a complete analysis of the given spectrum. In some cases, for example in the case of the first spectrum of the inert gases, only a few lines are required in order to satisfy this criterion, since all energy states of the atom, with the exception of the ground state, are well determined from data obtained from the more accessible parts of the spectrum. At the same time account is taken of the circumstance that difficulties with excitation limit the comparative development of spectra of successively higher degrees of ionization. Articles marked with the symbol B constitute only a partly completed description of the spectrum in the vacuum ultraviolet region. In such articles the wavelengths cited must be considered sufficiently accurate, but the number of lines insufficient for carrying out a complete analysis of the spectrum. In some cases the lines are not even assigned states of ionization. Articles marked with the symbol C are regarded only as serving to make the survey complete and as insufficient for application to analysis, either because of insufficient dispersion or because of inadequate wavelength standards. Articles dealing with the peculiarities of individual lines are listed here without a designation.

All these evaluations apply only to observations in the vacuum ultraviolet region and have no direct relation to the degree of completeness of the analysis of the spectrum. This question lies outside the scope of the present article and has already been considered by Russell (1935) and, after him, by Shenstone (1939).

For the first five spectra (I, II, III, IV, V and for unidentified lines u) of each element these symbols are collected in Table 2. The table is divided into two parts in order to express the contrast between the comparatively complete knowledge of the spectra of atoms with atomic numbers up to 20 (calcium) and the comparatively insufficient knowledge of the spectra of atoms with higher atomic numbers. It is quite natural that the simple spectra were investigated first and more thoroughly. It should be remembered that in the case of the element with atomic number 21 (scandium) the d-electron introduces great complexity into atomic

spectra. Many places in Table 2 have been left blank, since in this spectrum no observations at all have been made for wavelengths shorter than \(\lambda 2000\). One of three other symbols (L, X, O) has been introduced in order to represent the probable state of such spectra. L denotes spectra for which it is known, or considered probable, that no ordinary emission lines lie at wavelengths shorter than \(\lambda 2000\). X denotes spectra for which it is known, or considered probable, that no important lines lie in this spectral region, i.e. there are no lines that would give information about terms. It should be noted that such spectra may contain lines with wavelengths shorter than \(\lambda 2000\), which may prove useful as wavelength standards, for such wavelengths can be calculated by means of the combination principle, as was explained above. O denotes spectra for which there are no observations, but which contain presumed important lines of wavelengths less than \(\lambda 2000\). Since A, L, and X represent spectra least in need of further observations of emission in the vacuum ultraviolet region, they are indicated in Table 1 by bold letters. The symbols B, C, and O, set in the table in ordinary type and occurring in very large numbers, show that much still remains to be done. Statistical data extracted from Table 2 might be of interest, and therefore they are presented separately in Table 3.

Consideration of the diagonal lines (from upper left to lower right) of Table 2 shows the isoelectronic sequences of the first five spectra (the neutral atom and the first four stages of ionization) where sufficient data are available. The isoelectronic sequences are shown more clearly in Table 1, where missing members of a sequence are enclosed in parentheses. The available information on each member of such long sequences may be found in the literature listed for the corresponding spectrum in Appendix C.

Up to this point we have considered only emission spectra. As for absorption spectra, they have been investigated for only 16 elements: Mg, Si, Ar, K, Mn, Zu, Kr, Rb, Ag, Cd, J, Xe, Cs, Hg, Te, and Pb; the articles relating to each of them are listed in Appendix C. Beutler’s work (1933), which deserves special consideration, has already been mentioned in discussing the question of spectra that constitute a transition between optical and X-ray spectra. Perhaps the best way to approach a discussion of this work is to make use of Beutler’s first paper and to consider the essential difference between the process of absorption in the optical spectrum and in the X-ray absorption spectrum. In the first case the most weakly bound electron is displaced to a free “orbit” (subject to the usual restrictions determined by changes in angular momentum and spin) or is entirely removed from the atom (still subject to the same restrictions).

These processes lead to the appearance of series of lines and, in going over to shorter wavelengths, bands of continuous absorption. In the case of X-ray absorption, the energy required to remove one of the inner electrons into one of the vacant “orbits” is almost the same as that required for the complete removal of the electron from the atom. For example, in the \(K\)-absorption of argon this difference, as found by Koster and van der Tuuk (1936), is 1.7 electron-volts for a total change in energy of approximately 10,000 electron-volts. The fine structure, not fully resolved, at the long-wave edge of the \(K\)-absorption band of X-rays by argon corresponds to an optical absorption line broadened by about 200 Å.

If, by means of optical absorption spectra, it were possible to obtain complete information about the vacant orbits in a neutral atom, then these other processes might be of only secondary interest. But, owing to differences in screening and because both the energy and sometimes the term structure of the atom are different when the normally vacant “orbit” is temporarily filled in these two ways, establishing the structure in the absorption band in the region of ordinary X-rays makes it possible to estimate the resolving power; in the case of the vacuum ultraviolet region, however, comparative data must be available concerning absorption spectra associated with excitation in intermediate electron shells. Then one might expect energy differences, compared with the total energy, sufficient for a clear separation of line absorption from the continuous absorption accompanying it. In a number of works (listed under the elements in Appendix C), Beutler and his collaborators observed precisely such absorption spectra for the following seven elements: K, Zn, Rb, Cd, Cs, Hg, and Tl. For the most part they lie in the region \(\lambda\) 1200—\(\lambda\) 600, and all arise from excitation of an electron from the least firmly bound filled electron shell, for example from the \(4d\) electron in Cd and the \(5p\) electron in Cs. While, technically, by their excitation, these spectra are X-ray spectra, they clearly display all the features of multiplets and series of a certain line spectrum. For this reason it may be said that such spectra constitute a transition from optical spectra to X-ray spectra.

These new absorption spectra reveal all the multiplicities and positions of terms that are to be expected on the basis of Hund’s theory. The term values bear a definite relation to these values in the arc spectrum of the following element with the next higher atomic number, but with different multiplicities. Examples may be found for \((Z, S)\), \((j, j)\), and intermediate types of coupling. Since these spectra are, evidently, spectra of the neutral atom, Beutler proposes to designate them as \(I^b\) spectra, leaving the symbol \(I^a\) to designate the simpler spectrum of the neutral

of the atom. Similar superscripts may also be added to subsequent Roman numerals, if it is desirable to designate as separate spectra certain known examples of excitation of inner electrons in spark spectra. In Tl there was observed a spectrum \(I^b\), caused by excitation of the \(6s\) electron, in the region \(\lambda 2500\)—\(\lambda 1400\), and also a spectrum \(I^b\), excited by the \(5d\) electron, in the region \(\lambda 900\)—\(\lambda 600\). The term system \(I^a\) converges to the ground state of spectrum II of the same atom, whereas (in the alkali metals) the term system \(I^b\) converges to the excited state \(p^5s\) in spectrum II. If it were desired to extend these designations, the \(K\)-absorption of X-rays in argon should be denoted by \(I^b\).

Most of the terms in the spectrum \(I^b\) lie above the ionization limit in the corresponding spectrum \(I^a\). Such terms (and the corresponding absorption lines) are considerably broadened owing to autoionization, when, in accordance with the predictions of Shenstone (1931), beyond the limit of \(I^a\) there exists a continuous region having the same values \(L, S, j\). The broadening is apparently proportional to the intensity of the continuum arising at the same wavelength; it, apparently, also prevents observation of the emission spectra \(I^b\). Shenstone’s rules for autoionization are clearly justified here, if one does not count a few doubtful cases with intermediate coupling. The same interaction between the wave functions of the \(I^b\) terms and the associated \(I^a\) continuum, which causes this broadening, greatly increases the intensity of the \(I^b\) absorption lines. It seems probable that these intense absorption processes determine the form of the dispersion curve of such elements. Indeed, Beutler (1935) succeeded in showing this for the rare gases—argon, krypton, and xenon. Here the anomalously strong absorption lines (also broadened) are not lines of the \(I^b\) spectrum. The ordinary (\(I^a\)) spectrum in these rare gases consists of separate groups of lines converging to each of two limits—states \(2P^0_{1/2}\) and \(2P^0_{3/2}\) of the ion. The lines converging to the limit \(2P^0_{1/2}\), which lies above \(2P^0_{3/2}\), are broadened. They are much stronger in absorption lines than either the unbroadened lines or the group below the limit \({}^0P^0_{1/2}\), or the continuum behind each limit. The dispersion curves for these rare gases, as it turns out, predict the principal absorption precisely at the wavelengths of the strengthened and broadened lines.

10. MOLECULAR SPECTRA

Observations in the vacuum ultraviolet region of both emission spectra and absorption spectra have made it possible to obtain important data on molecular spectra. However, no attempt had yet been made to give a review of the work in this field; Stoner (1935) tabulated the data then available on molecular spect-

relatively few works on the spectra of diatomic molecules have appeared. Many more investigations have been carried out on the spectra of polyatomic molecules. The article by Sponer and Teller contains a review of the available data on the spectra of polyatomic molecules in all wavelength regions.

Sponer’s tables (1935) show that by that time the spectra of the following diatomic molecules or molecular ions had been investigated in the vacuum ultraviolet region: \(H_2\), \(N_2^+\), \(N_2\), \(O_2\), \(S_2\), \(J_2\), BrCl, JCl, JBr, CO, NO. Most of these molecules have comparatively high heats of dissociation. Whether or not the spectrum of a molecule (or molecular ion) extends into the vacuum ultraviolet region depends chiefly on the energy distribution of the possible electronic states in the molecule. Where the electronic structure of the molecule in its unexcited state “forms a closed shell,” the excitation energy of its first excited state will be comparatively high, and the transition from it to the unexcited state will give bands in the vacuum ultraviolet region. This case is analogous to the situation with the atomic spectra of the inert gases. The gases \(H_2\), \(N_2\), and CO have such “closed shells” of electrons.

It is very regrettable that the increased interest in the rotational structure in molecular spectra, to the detriment of the vibrational structure, has been an obstacle to repeated measurements, with modern accuracy, of some previously studied vibrational systems. Even the bands of \(O_2\) and \(H_2\), listed in Appendix A, could successfully be measured again.

Observations of molecular spectra require comparatively little resolving power if they are confined to vibrational systems in emission or absorption, or to spectra of continuous absorption. Investigations of the rotational structure of bands of diatomic molecules require a resolving power of the order of 30,000. With the practically achieved resolving power of 300,000, Liberman (1940) was able to resolve the rotational structure of the \(CS_2\) bands at \(\lambda 3200\). It is very difficult to obtain such resolving power at shorter wavelengths. For resolving the rotational structure in the bands of still more complex polyatomic molecules there is perhaps only very little hope.

Bands in the Schumann region of \(N_2\) and CO (the spectrum of the latter is more easily excited in an electric discharge in \(CO_2\)) are convenient light sources for the final optical adjustment of a spectrograph, since they give groups of very closely spaced lines of comparatively uniform intensity.

A few remarks should be made about bands that sometimes appear in other investigations. The bands of \(N_2\), \(O_2\), and NO can appear in the spectra of electric discharges and in other gases,

if there is a very slight access of air into the system. If the access is large, then the spectrum disappears over a considerable part of the vacuum ultraviolet region. CO bands may appear in gas-discharge spectra if the electrodes are not sufficiently degassed. This is especially pronounced in the case of electrodes made of iron or nickel.

Very great difficulties are caused by band spectra which have the appearance of line spectra. Owing to the very small moment of inertia of the hydrogen molecule, there are very wide intervals between the lines of the rotational structure of its spectrum. The characteristic band structure is not obvious, and the spectrum is called the “many-line” spectrum of hydrogen (see Richardson, 1934). There is a single band in helium near \(\lambda 600\), first identified by Sommer (1927) and subsequently studied by Nickerson (1935). (Some features of its origin have so far remained unexplained.) Under certain excitation conditions, close doublets appear in nitrogen near \(\lambda 1000\). They were identified by Birge and Hopfield (1928) as a partial development of the ordinary band spectrum of nitrogen in this region.

For completeness it should be mentioned that in atomic spectra there are often lines which look like unresolved bands. These are lines arising from terms above the limit of ordinary ionization in the atom and broadened by autoionization.

11. SOLID-STATE SPECTRA

Skinner (1939, 1940) considered soft X-ray spectroscopy of the solid state in some detail. This question is very similar in its technique to our general question and, moreover, has a sufficiently related theory; it is therefore appropriate to make a brief mention of it here. According to modern views, valence electrons in a solid move in the crystal lattice with energies distributed in a discrete band. No more than two electrons, corresponding to the two possible spin orientations, can have the same energies. All energy levels within this band are filled at absolute zero, but in a conducting body at temperatures above absolute zero some of the electrons may have higher energy in allowed, but unfilled, energy bands lying immediately above the filled band.

This diffusion of electrons into the unfilled band on the energy diagram is, of course, a function of temperature. However, in a nonconducting solid there is a forbidden band of energy values between the filled and unfilled bands. An electron can reach the unfilled band only by absorbing energy sufficient to carry it across the forbidden band. A crystal

is transparent for a range of wavelengths corresponding to quantum energies insufficient for jumping across the forbidden band, but is opaque for shorter wavelengths.

Figs. 5 and 6 present (schematically) an illustration of this essential difference between conductors and nonconductors, and also show the spectroscopic possibilities for obtaining information about these energy bands, both filled and unfilled. The transition from filled bands to unfilled ones causes the opacity of conductors and absorption below the characteristic wavelength limit in nonconductors. The visible emission spectra of solids, arising from the return of electrons from the unfilled band to temporarily vacant places in the filled band,

Fig. 5. Energy bands of electrons in conductors (the normally filled energy band is indicated by hatching, and the normally empty energy band by dots).

Fig. 6. Energy bands of electrons in nonconductors (the normally filled energy band is indicated by inclined hatching, the forbidden band by cross-hatching, and the normally empty band by dots).

were observed by Möller and Becker (1931). But transitions between two broad energy bands give broad continuous spectra that are difficult to interpret. More useful are emission spectra corresponding to transitions from filled energy bands to temporary vacancies in the outer electron shell, associated with individual atoms of the lattice, and absorption spectra corresponding to the ejection of electrons from these atomic shells into unfilled energy bands. From spectra associated with transitions between these bands and the inner shells of atomic electrons, one can obtain some data. But, as in the case of comparing Beutler spectra of atomic absorption with X-ray absorption spectra of gases, the energy interval in the bands, in comparison...

tion with the total energy of hard X-rays determines the resolving power. The difficulties with resolving power are connected both with instrumental limitations and with the broadening of lines as a result of radiation damping. Starting from the hydrogen-like atom, Skinner indicates that this radiative broadening amounts to \(4 \times 10^{-4}\) volt at \(\lambda 100\), \(4 \times 10^{-2}\) volt at \(\lambda 10\), and 4 volts at \(\lambda 1\).

Skinner gives an extensive bibliography of the observed emission and absorption spectra of solids in the soft X-ray region and, on the basis of data borrowed from many observers, gives experimental curves showing the density of energy levels in the filled and unfilled zones of lithium, magnesium, and aluminum. In all cases of metals for which the necessary data are available, the short-wavelength edge of the emission band coincides with the long-wavelength edge of the absorption bands. This was to be expected if the filled and unfilled zones are adjacent to one another.

In his observations with an anticathode at \(110^\circ\text{K}\), \(300^\circ\text{K}\), and \(680^\circ\text{K}\), Skinner showed a broadening of the emission bands in aluminum, caused by the thermal change in the energy distribution of its electrons.

Observations of the spectra of soft X-rays from solid substances are still for the most part limited to elements with small atomic number, but further investigations should fill this gap. The results obtained with several alloys and several compounds are also promising for future work.

It is now quite clear why earlier investigations of soft X-rays from metals, based on the method of critical potentials (Kurt, 1921; Richardson and Chalklin, 1926; K. T. Compton and Thomas, 1926, etc.), yielded results that are difficult to interpret. In addition to the experimental difficulties with surface contamination, the scatter of the energy values of the valence electron was unknown. When the emission spectrum of soft X-rays was observed for the first time, the line width was unexpected (Tibo, 1927; Zederman, 1929).

If the aforementioned experimental difficulties were eliminated, measurements of absorption in this region would be greatly facilitated. The most interesting region, \(\lambda 300 — \lambda 100\), is too short-wavelength for its investigation with a gas discharge as a source of a continuous spectrum, and too long-wavelength for a continuous X-ray spectrum of sufficient intensity to be obtained. It was therefore necessary to use many lines of the spectrum of a hot spark as a source for absorption measurements. Hence complications arose with photographic photometry, as well as limitations of resolving power.

12. ASTROPHYSICAL AND OTHER APPLICATIONS

The data obtained in laboratory investigations of atomic spectra in the vacuum ultraviolet region proved to be extremely important in interpreting the spectra of galactic (gaseous) nebulae. Some of these objects are detected by the light they scatter from neighboring stars, while others have their own emission spectrum. These latter nebulae are always associated with very hot stars of an earlier spectral type than B I (Hubble, 1922). Menzel (1926) and Zanstra (1927) independently suggested that absorption of very short-wavelength ultraviolet radiation coming from very hot stars leads to excitation of an emission spectrum. From the intensity of the Balmer lines of hydrogen in this emission spectrum it proved possible to estimate the total intensity of the star’s ultraviolet continuous spectrum beyond \(\lambda 912\) of the Lyman series of hydrogen. Comparison of it with the intensity of the continuous spectrum in the observed regions gives the surface “black” temperature of the exciting star. In this way values of about \(100\,000^\circ\mathrm{K}\) were obtained. Nebulae give a considerable density of hydrogen resonance radiation (\(\lambda 1215\)), for which they prove to be strongly absorbing. For longer wavelengths nebulae are very transparent.

Some lines in the spectra of nebulae, including the strongest lines, as was shown by Bowen (1928), arise as the result of “forbidden” transitions in N II, O II, and O III. These transitions originate from metastable states in these ions, to which they had been excited by collisions with photoelectrons produced in the nebulae by the frequent primary radiation of the star with wavelengths shorter than \(\lambda 912\). Negligible numbers of these photoelectrons have sufficient energy to excite the spectra of these elements. Recombination of ions and electrons is least frequent. Transitions from metastable states have very small probabilities, of the order of \(1\ \mathrm{sec}^{-1}\) and even less (Condon, 1934). Under laboratory conditions it is impossible to obtain time intervals between collisions greater than one thousandth of a second, but in nebulae these time intervals range from hours to weeks. Subsequent investigations by Bowen and others led to the identification of a considerable number of forbidden lines in nebular spectra, including, besides the previously discovered N I and O I lines, lines of fluorine, neon, magnesium, sulfur, chlorine, argon, calcium, and iron, and possibly also silicon and potassium. The best data were summarized by Bowen (1935), Boyce (1936), and also by Bowen and Wyse (1939); the physical aspects of forbidden lines were discussed by Bowen (1936). Apparently, about two-thirds of all the lines listed by Bowen and Wyse may be regarded as definitely identified, but the remaining lines are doubt-

...because they were very weak. The forbidden Fe II lines were discovered by Merrill (1928) in the spectrum of $\eta$ Carinae and were subsequently also found in the spectra of some new stars. The identification of the spectra of new stars is made difficult by the excessive breadth of the lines, but sometimes in these spectra significant successive changes occur as the new stars change.

Knowledge of the spectra in the vacuum ultraviolet region not only leads to an understanding of the mechanism of excitation in nebulae, but it also provides data for calculating the positions of forbidden lines. Fig. 7, which is a portion of the Grotrian diagram for O I, may explain this. The forbidden lines of the light elements may be divided into two main types: nebular and auroral (for neutral oxygen both types of these transitions occur both in nebulae and in the aurora borealis). The type of transition is named according to the place where the given oxygen line is the stronger. The wave number of the auroral O I line is equal to the difference between the wave numbers of two vacuum ultraviolet lines at $\lambda 1217$ and $\lambda 999$. The difference of the wave numbers of two nebular O I lines is equal to the difference between the wave numbers of two vacuum ultraviolet lines at $\lambda 1304$ and $\lambda 1302$. The absolute wave number of a line of the nebular type can be calculated only if intercombination lines can be found somewhere in the spectrum that would make it possible to find the connection between the triplet and singlet systems of terms. In some cases these intercombination lines have confirmed the preliminary trial identification of nebular lines, but they have not yet been discovered in O I. In the absence of intercombination lines, calculations of the absolute wave numbers of lines of the nebular type may be carried out in various ways: by interpolation or extrapolation along an isoelectronic sequence or from well-determined series in both systems converging to the same limit, or (in the case of ions) from well-determined series in the next lower stage of ioniza-

Fig. 7. Partial Grotrian diagram of the neutral oxygen atom, showing the connection of forbidden lines with lines in the vacuum ultraviolet region.

…lines converging both to the normal and to the metastable state of the ion. This latter approximation is, of course, impossible in the case of a neutral atom. The auroral line in neutral oxygen was obtained in the laboratory by McLennan and Shrum (1925) and was subsequently studied by McLennan and McQuarrie (1927). Nebular lines in neutral oxygen were obtained in the laboratory by Hopfield (1931). In these experiments, in order to compensate for the low probability of emission of the lines, the simple influence of the number of metastable atoms at their high concentration was used. It would be incomparably more difficult to apply this method to ionized atoms.

The data from the vacuum ultraviolet region have also been used for interpreting the spectra of nebulae in another way. Besides several “allowed” lines present in the spectra of nebulae and not belonging to hydrogen and helium, there are there some partial multiplets of O III and N III. In the case of the oxygen lines, they all arise from a single high odd level or from odd levels reached by radiative transitions from this high odd level. Other (laboratory) members of the multiplet are absent. Bowen (1935) showed that this high odd level is selectively excited by absorption of the resonance radiation He II at \(\lambda 303\). Further, as a result of another coincidence of wavelengths, the O III line at \(\lambda 374\), which follows the emission of certain pairs of lines in the partial multiplets, itself leads to the selective excitation of one level in N III, with subsequent emission of partial multiplets of this spectrum.

These various mechanisms of excitation in the ultraviolet region can evidently play a significant role in the spectra of novae, Wolf–Rayet stars, and, in general, stars with bright line spectra. Even the Sun apparently emits a much larger amount of ultraviolet radiation than could be emitted by a black body at its effective surface temperature. It is still unclear what part of the excess ultraviolet rays comes rather from “hot spots” on the solar surface than from the entire surface as a whole. Evidence for this excess ultraviolet radiation is provided by a whole series of phenomena, which may be briefly mentioned here. The formation of ozone in the atmosphere depends mainly on absorption of radiation in the Schumann region. At least a substantial part of the ionization in the atmosphere is produced by solar ultraviolet light. Certain types of rapid ionospheric disturbances are apparently caused by additional flashes of light on active areas of the Sun’s surface. Regions of ionization at various heights in the atmosphere are probably caused by various processes, as a result of which the constituent parts of the atmosphere become ionized. The absorption coefficient for each of these processes determines the depth…

under the upper part of the atmosphere, where it is most effective. It is clear that all or almost all of these processes are caused by radiations in the vacuum ultraviolet region of the spectrum, but it has not yet been possible to determine fully the exact relation between the individual processes and the regions of the ionosphere. This question has been considered by Chapman and Price (1937), Wulf and Deming (1938), Preston (1940), and others.

The evidence connected with aurorae is still more remarkable, as was pointed out by Shakh (1937).

Observations by Slipher (1933) showed that the “negative” bands of nitrogen (caused by \(N_2^+\)), which are very weakly represented in the night sky, flare up strongly when the rays of the returning Sun begin to touch the upper layers of the atmosphere. The bands arise from the so-called excited state \(A\) of the ionized nitrogen molecule. Hopfield (1930) was unable to find any traces of continuous absorption corresponding to the formation of the ionized nitrogen molecule in its normal state, but a continuum beginning at \(\lambda 660\) was detected, leading to the formation of \(N_2^+\) in its excited \((A)\) state. It might be objected that the action of sunlight consists simply in the excitation of \(N_2^+\) ions already present in the upper atmosphere. But then it would be difficult to understand how molecular ions could avoid recombination during the night (flashes of the negative nitrogen bands, comparable in intensity, cease immediately after the last rays of the setting Sun leave the top of the atmosphere). Since it seems impossible to find any other ionizing agent besides sunlight to maintain the existence of molecular nitrogen ions, and since, at the time when these ions are formed upon absorption of light, they are already ready to emit the negative bands, Shakh’s arguments appear very convincing.

Further evidence for an excess of solar ultraviolet rays was adduced by Shakh in connection with the bands of \(CO^+\) and \(N_2^+\) in comet spectra. However, because the low density of comets is unfavorable for recombination, this argument is apparently not so convincing. Excitation in the solar chromosphere also strongly indicates an excess of ultraviolet radiation (Sillé and Menzel, 1935). Hemmendinger (1939) pointed to selective effects in the chromosphere which are most readily explained by the presence of strong radiation in the Lyman series of hydrogen (\(\lambda 1215\), \(\lambda 1205\), \(\lambda 972\), etc.). The strict parallel between the intensities of the lines of the solar corona (in various positions above the limb of the eclipsed Sun) and the intensities of helium lines in the lower parts of the chromosphere indicates a common source of excitation energy.

The considerable energies characteristic of radiation quanta in the vacuum ultraviolet region make possible photochemical processes for which radiation of longer wavelengths is powerless. Many investigators have used radiations in the Schumann region. When monochromators are used, the intensities prove to be too low.*) The xenon lamp of Harteck and Oppenheimer (1932), further developed by Groth (1936, 1937), made it possible to extend somewhat the possibilities of research. The lamp gives strong radiation in two resonance lines of xenon, \(\lambda 1470\) and \(\lambda 1295\), each of which passes through fluorite windows. A convenient filter cell, which can contain oxygen or can be evacuated, makes it possible to carry out observations either with only the line \(\lambda 1295\), or with both lines. In a series of papers by Harteck, Groth, and others from the Hamburg laboratory, there are descriptions of photochemical investigations carried out with this lamp. A review of these results was published by Groth (1939). As a quantitative method for measuring intensities, the photochemical production of ozone was used; this, as is known from the work of Kistiakowsky and Smith (1935), gives a photochemical yield of 2 molecules of ozone per absorbed quantum. The ozone obtained was determined chemically, with the aid of iodine liberated from a potassium iodide solution, or physically, by means of its absorption (in a long tube with quartz windows) of mercury radiation at \(\lambda 2537\). Among the considerable number of reactions studied is the production of formaldehyde from a mixture of carbon monoxide and hydrogen (Falthings, Harteck, and Groth, 1938).

Other comparatively monochromatic light sources also prove applicable, despite the fact that they readily cause discoloration of lithium fluoride (Schneider, 1937). Groth (1939) reports unpublished experiments by Groth and Harteck with a krypton lamp giving \(\lambda 1236\) and \(\lambda 1165\). A hydrogen lamp for \(\lambda 1215\) can best operate with a mixture of hydrogen with helium and neon.

As already mentioned, photoelectric cells were used for detecting vacuum ultraviolet radiations. It is quite possible that investigation of the photoelectric effect in this spectral region would be of interest for the study of metals having a high work function (provided the surface is clean). P. E. Sabine (1917) tested Einstein’s photoelectric equation in the Schumann region. Vacuum ultraviolet radiations from electric discharges in gases (Thomson, 1925) probably play a role in the photoelectric liberation of electrons from the cathode in a cold-cathode discharge.

*) In measurements of the transparency of crystals, a fluorite double monochromator described by Gilsh and Pohl (1930) was used.

Lyman (1928, pp. 123, 124) considered the question of the bactericidal and abiotic action of radiation in the Schumann region. Blank and Arnold (1935) described experiments with bacillus subtilis which showed the advantage of the “window” in air between \(\lambda 1300\) and \(\lambda 1100\). In the presence of living matter, any radiation in this region must, of course, pass through the absorption band of water vapor (Patenay, 1933). Recent measurements by Preston (1940) show that \(\lambda 1215\) is strongly absorbed by water vapor. The penetration of such radiations into living matter is less deep than that of ordinary ultraviolet rays or X-rays. From the known points of view this would also be advantageous, since it would be known exactly where the energy is absorbed.

Table 1. Long isoelectronic series

Number of electrons	Series	Missing members
1	H I—C VI
2	He I—C V
3	Li I—Na IX	(Ne VIII)
4	Be I—Al X	(Ne VII)
5	B I—Al IX	(Ne VI)
6	C I—Al VIII
7	N I—Cl XI	(P IX, S X)
8	O I—Cl X	(P VIII,* S IX)
9	F I—Ca XII	(P VII, A X)
10	Ne I—Co XVIII	(P VI, S VII, A IX)
11	Na I—Cu XIX
12	Mg I—Co XVI
13	Al I—Co XV
14	Si I—Sc VIII
15	P I—V IX
16	S I—Fe XI*
17	Cl I—Co XI
18	A I—Fe IX
19	K I—Fe VIII
20	Ca I—Ni IX
21	Sc I—Fe VI
22	Ti I—Ni VII
23	V I—Cu VII
28	Ni I—Br VIII
29	Cu I—Br VII
30	Zn I—Br VI
35	Br I—Zn VI
37	Rb I—Mo VI
46	Pd I—J VIII
47	Ag I—Te VI
78	Pt I—Bi VI

Table 2. Experimental data on atomic emission spectra in the wavelength region shorter than \(\lambda 2000\)

No.	Element	u	I	II	III	IV	V	No.	Element	u	I	II	III	IV	V
1	H		A					47	Ag	B	A	C	C	O	O
2	He		A	A				48	Cd	B	O	C	B	B	O
3	Li		L	B	B			49	In		L	A	C	C	O
4	Be		A	A	B	B		50	Sn		A	A	C	C	C
5	B		C	A	A	B	B	51	Sb		C	B	B	B	B
6	C		A	A	A	A	B	52	Te	B	C	O	B	B	B
7	N		A	A	A	A	B	53	J	B	C	B	B	O	O
8	O		B	A	A	A	A	54	Xe		A	B	A	C	O
9	F		B	B	A	A	A	55	Cs		L	C	C	O	O
10	Ne		A	A	A	C	C	56	Ba		L	A	O	C	O
11	Na		L	B	B	B	B	57	La		L	X	B	O	O
12	Mg		C	C	A	A	B	58	Ce		L	X	B	B	O
13	Al		C	B	B	A	B	59	Pr		L	X	O	O	O
14	Si		A	B	A	A	B	60	Nd		L	X	O	O	O
15	P		B	A	A	A	A	61	—		—	—	—	—	—
16	S		A	B	B	B	B	62	Sa		L	X	O	O	O
17	Cl		C	C	A	B	B	63	Eu		L	X	O	O	O
18	A		A	A	A	B	B	64	Gd		L	X	O	O	O
19	K		L	B	B	B	B	65	Tb		L	X	O	O	O
20	Ca		L	C	B	B	B	66	Dy		L	X	O	O	O
21	Sc	B	X	O	B	B	B	67	Ho		L	X	O	O	O
22	Ti		X	C	C	C	B	68	Er		L	X	O	O	O
23	V		C	A	C	B	B	69	Tu		L	X	O	O	O
24	Cr		X	B	A	C	C	70	Yb		L	X	O	O	O
25	Mn		X	A	B	A	B	71	Lu		L	X	O	O	O
26	Fe		B	B	A	B	A	72	Hf		O	B	O	O	O
27	Co	C	C	B	O	O	B	73	Ta		O	O	O	O	O
28	Ni	C	C	C	O	O	O	74	W	C	O	B	O	O	O
29	Cu	B	C	A	C	O	O	75	Re		O	O	O	O	O
30	Zn	B	C	C	B	B	O	76	Os		O	O	O	O	O
31	Ga	C	L	C	C	C	O	77	Ir		O	O	O	O	O
32	Ge		C	C	C	C	B	78	Pt		C	A	O	O	O
33	As		C	C	B	B	C	79	Au	B	C	C	O	O	O
34	Se		A	A	B	B	B	80	Hg	B	B	B	B	B	O
35	Br		C	C	B	B	B	81	Tl	B	L	C	C	C	O
36	Kr		A	A	A	B	O	82	Pb		C	B	B	B	B
37	Rb		L	C	B	O	O	83	Bi		C	B	B	B	B
38	Sr		L	A	O	B	O	84	Po		—	—	—	—	—
39	Y		X	X	C	O	B	85	—		—	—	—	—	—
40	Zr		X	A	A	B	O	86	Rn		O	O	O	O	O
41	Cb	C	O	O	B	A	B	87	—		—	—	—	—	—
42	Mo		X	O	O	B	A	88	Ra		L	C	O	O	O
43	—		—	—	—	—	—	89	Ac		—	—	—	—	—
44	Ru		X	O	O	O	O	90	Th		O	O	O	B	O
45	Rh		X	O	O	O	O	91	Pa		—	—	—	—	—
46	Pd		C	C	O	O	O	92	U		O	O	O	O	O

Table 3. Statistics of data for wavelengths shorter than \(\lambda 2000\)

	I	II	III	IV	V
A	14	20	16	10	5
B	4	17	24	27	25
C	20	18	11	10	5
L	26	0	0	0	0
X	10	15	0	0	0
O	11	14	32	35	46
A + X + L	50	35	16	10	5
B + C + O	35	49	67	72	76
% A + X + L	59	42	19	12	6

REFERENCES TO THE TEXT

Basic

1. S. Bowen, J. Opt. Soc. Am. and Rev. Sci. Inst., 13, 89 (1926).
   T. Lyman, Spectroscopy of the Extreme Ultraviolet, second edition (Longmans, Green and Co., New York, 1928).
   H. Bonke, Vakuumspektroskopie (J. A. Barth, Leipzig 1937); Physik. Zeits. 37, 222—260 (1936).
   R. W. B. Pearse, Reports on Progress in Physics (London) 4, 311 (1938).

1893
V. Schumann, Akad. Wiss. Wien, 102, 2A, 625. (A complete list of Schumann’s works is given by Lyman: Spectroscopy of the Extreme Ultraviolet.)

1895
E. Wiedemann and G. C. Schmidt, Wied. Ann. 56, 237.

1897
M. W. Hoffmann, Wied. Ann. 60, 269.

1900
H. Kayser, Handbuch der Spektroskopie (Herzel, Leipzig), vol. I, p. 450.

1901
V. Schumann, Ann. d. Physik 5, 349.

1902
P. Lenard, Ann. d. Physik 8, 149.

1906
T. Lyman, Astrophys. J. 5, 349.

1908
A. S. King, Astrophys. J. 28, 335.

1914
J. Franck and G. Hertz, Verh. d. D. Phys. Ges. 16, 10.

1916
J. Franck and G. Hertz, Physik. Zeits. 17, 409.

1917
B. Davis and F. S. Goucher, Phys. Rev. 10, 101.
P. E. Sabine, Phys. Rev. 9, 210.

1919
W. Stenström, Dissertation, Lund.

1920
E. R. Laird, Phys. Rev. 15, 543.

1921
J. Duclaux and P. Jeantet, J. de phys. et rad. 2, 154.
E. H. Kurth, Phys. Rev. 18, 461.

R. A. Millikan, Proc. Nat. Acad. Sci. 7, 289.

1922
J. J. Hopfield, Phys. Rev. 20, 573.
E. Hubble, Astrophys. J. 56, 162, 400.
A. S. King, Astrophys. J. 55, 103.
T. Lyman, Nature 110, 278.

1923
A. H. Compton, Phil. Mag. 45, 1121.
P. S. Olmstead and K. T. Compton, Phys. Rev. 22, 559.
F. Simeon, Proc. Roy. Soc. 102, 484.

1924
S. Lyman, Astrophys. J. 60, 1.

1925
P. W. Bridgman, Proc. Am. Acad. 60, 305.
A. H. Compton and R. L. Doan, Proc. Nat. Acad. Sci. 11, 598.
A. Fowler, Trans. Roy. Soc. Phil. A225, 1.
G. R. Harrison, J. Opt. Soc. Am. and Rev. Sci. Inst. 11, 113, 341.
J. C. McLennan and G. M. Shrum, Proc. Roy. Soc. A108, 501.
J. J. Thomson, Phil. Mag. 49, 761.

1926
I. S. Bowen and S. B. Ingram, Phys. Rev. 28, 444.
K. T. Compton and C. H. Thomas, Phys. Rev. 28, 601.
D. Coster and J. H. van der Tuuk, Zeits. f. Physik 37, 367.
S. W. Leifson, Astrophys. J. 63, 73.
D. H. Menzel, Pub. Astron. Soc. Pacific 38, 295.
A. H. Pfund, J. Opt. Soc. Am. and Rev. Sci. Inst. 12, 467.
O. W. Richardson and F. C. Chalklin, Proc. Roy. Soc. A110, 247.

1927
G. Cario and W. Lochte—Holtgreven, Zeits. f. Physik 42, 22.
J. B. Hoag, Astrophys. J. 66, 225.
E. R. Laird, Phys. Rev. 29, 41.
J. C. McLennan, J. H. McLeod and W. C. McQuarrie, Proc. Roy. Soc. A114, 1.
T. H. Osgood, Phys. Rev. 30, 567.
R. A. Sawyer and F. Paschen, Ann. d. Physik 84, 1.
L. A. Sommer, Proc. Nat. Acad. Sci. 13, 213.
J. Thibaud, J. de phys. et rad. 8, 13.
H. Zanstra, Astrophys. J. 65, 50.

1928
R. T. Birge and J. J. Hopfield, Astrophys. J. 68, 257.
I. S. Bowen, Astrophys. J. 67, 1.
K. T. Compton and J. C. Boyce, J. Frank. Inst. 205, 497.
T. Lyman, Spectroscopy of the Extreme Ultraviolet, second edition (Longmans, Green and Co., New York).
P. W. Merrill, Astrophys. J. 67, 391.

1929
R. C. Gibbs, A. M. Vieweg and C. W. Gartlein, Phys. Rev. 34, 406.
F. Paschen, Preuss. Akad. Wiss., Berlin Ber., p. 662.
G. Scheibe, Zeits. f. physik. Chemie B5, 355.
E. W. H. Selwyn, Proc. Phys. Soc. (London) 41, 392.
M. Söderman, Zeits. f. Physik 52, 795.

1930
K. T. Compton and I. Langmuir, Rev. Mod. Phys. 2, 123.
A. Ericson and B. Edlén, Zeits. f. Physik 59, 656.
G. R. Harrison and P. A. Leighton, J. Opt. Soc. Am. 20, 313.
R. Hilsch and R. W. Pohl, Zeits. f. Physik 59, 812.

J. J. Hopfield, Astrophys. J. 72, 139 (continuous spectrum).
J. J. Hopfield, Phys. Rev. 36, 784A (nitrogen spectrum).
W. G. Penney, Phil. Mag. 9, 661.

1931
F. W. Aston, Proc. Roy. Soc. A132, 487.
G. Cario and H. D. Schmidt-Ott, Zeits. f. Physik 69, 719.
K. T. Compton and I. Langmuir, Rev. Mod. Phys. 3, 191.
C. W. Cartlein and R. C. Gibbs, Phys. Rev. 38, 1907.
G. R. Harrison and P. A. Leighton, Phys. Rev. 38, 899.
J. J. Hopfield, Phys. Rev. 37, 160.
F. L. Mohler and C. Boeckner, Bur. Stand. J. Research 6, 673; 7, 751.
H. M. O’Bryan, Phys. Rev. 38, 32.
A. G. Shenstone, Phys. Rev. 38, 873.

1932
M. Adam, Ann. d. Physik 15, 568.
R. C. Gibbs, Rev. Mod. Phys. 4, 278.
P. Harteck and F. Oppenheimer, Zeits. f. physik. Chemie, B16, 77.
J. J. Hopfield and Appleyard, J. Opt. Soc. Am. 22, 488.
A. L. Hughes and L. A. DuBridge, Photoelectric Phenomena (McGraw-Hill, New York).
R. Ladenburg, C. C. Van Voorhis and J. C. Boyce, Phys. Rev. 40, 1018.
R. Ladenburg and G. Wolfsohn, Zeits. f. Physik 79, 42.
J. E. Mack, J. R. Stehn, and B. Edlén, J. Opt. Soc. Am. 22, 245.
H. M. O’Bryan, J. Opt. Soc. Am. 22, 739.
E. C. G. Stueckelberg, Phys. Rev. 42, 518.
Trans. Int. Astron. Union 4, 62.

1933
H. Beutler, Zeits. f. Physik 86, 495 and subsequent works indicated by element in Appendix C.
I. S. Bowen, J. Opt. Soc. Am. 23, 313.
G. H. Dieke, J. Opt. Soc. Am. 23, 274.
B. Edlén, Zeits. f. Physik 85, 85.
W. Finkelnburg, Physik. Zeits. 34, 529.
P. G. Kruger, Rev. Sci. Inst. 4, 128.
R. Ladenburg and C. C. Van Voorhis, Phys. Rev. 43, 315.
D. L. MacAdam, J. Opt. Soc. Am. 23, 178.
J. E. Mack and J. R. Stehn, J. Opt. Soc. Am. 23, 184.
W. A. Noyes, Jr., Rev. Mod. Phys. 5, 280.
G. Rathenau, Zeits. f. Physik 87, 32.
V. M. Slipher, Roy. Astron. Soc., M. N. 93, 657.
E. C. G. Stueckelberg, Phys. Rev. 44, 234.

1934
R. O. Anderson and J. E. Mack; J. Opt. Soc. Am. 24, 292.
G. Collins and W. C. Price, Rev. Sci. Inst. 5, 423.
K. T. Compton and J. C. Boyce, Rev. Sci. Inst. 5, 218.
E. U. Condon, Astrophys. J. 79, 217.
J. Curry and G. Herzberg, Ann. d. Physik 19, 800.
B. Edlén, Nova Acta Reg. Soc. Sci. Ups. Series IV, 9, No. 6.
F. Palmer, Jr., J. Chem. Phys. 2, 296.
W. M. Powell, Jr., Phys. Rev. 45, 154 (fluorite); 46, 43 (quartz).
O. W. Richardson, Molecular Hydrogen and its Spectrum. (Yale University Press, New Haven.)
F. G. Schneider, Phys. Rev. 45, 152.

1935
H. Beutler, Zeits. f. Physik 93, 177.
I. H. Blank and W. Arnold, J. Bacteriol. 30, 503.

I. S. Bowen, Astrophys. J. 81, 1.
J. C. Boyce and C. A. Rieke, Phys. Rev. 47, 653.
G. G. Cillié and D. H. Menzel, Harvard Circular, 410.
G. B. Kistiakowsky and H. A. Smith, J. Am. Chem. Soc. 57, 835.
T. Lyman, Phys. Rev. 48, 149.
J. L. Nickerson, Phys. Rev. 47, 707.
W. C. Price and G. Collins, Phys. Rev. 48, 714.
G. Rathenau and P. K. Peerlkamp, Physica 2, 125.
H. N. Russel, Roy. Astron. Soc., M. N. 95, 610.
M. Siegbahn and T. Magnusson, Zeits. f. Physik 95, 133.
H. Sponer, Molekülspektren und ihre Anwendung auf Chemische Probleme (Julius Springer, Berlin).

1936

I. S. Bowen, Rev. Mod. Phys. 8, 55.
J. C. Boyce, Roy. Astron. Soc., M. N. 96, 690.
A. B. F. Duncan and G. R. Harrison, Phys. Rev. 49, 211.
B. Edlén, Zeits. f. Physik 100, 621.
W. Groth, Zeits. f. Electrochemie 42, 533.
D. L. MacAdam, Phys. Rev. 50, 185.
K. R. More and C. A. Rieke, Phys. Rev. 50, 1054.
E. G. Schneider, Phys. Rev. 49, 341.
A. G. Shenstone, Phil. Trans. Roy. Soc. A235, 145.
D. C. Stockbarger, Phys. Rev. 49, 200A.
R. Tousey, Phys. Rev. 50, 1057.
F. Tyrén, Zeits. f. Physik 98, 768.
R. L. Webber and W. W. Watson, J. Opt. Soc. Am. 26, 307.

1937

F. W. Aston, Proc. Roy. Soc. A163, 391.
H. Bomke, Vakuumspektroskopie (J. A. Barth, Leipzig).
S. Chapman and W. C. Price, Reports on Progress in Physics (London) 3, 42.
W. Groth, Zeits. f. physik. Chemie, B37, 307 and 315.
F. W. Paul, Phys. Rev. 52, 923.
M. N. Saha, Proc. Roy. Soc. A160, 155.
E. G. Schneider, J. Opt. Soc. Am. 27, 72.
H. W. B. Skinner and J. E. Johnston, Proc. Roy. Soc. A161, 420.

1938

K. Faltings, W. Groth and P. Harteck, Zeits. f. physik. Chemie, B41, 15.
R. W. Pohl, Physik. Zeits. 39, 36.
J. J. Hopfield, Phys. Rev. 53, 931.
A. Hunter and R. W. B. Pearse, Proc. Phys. Soc. (London) 50, 256.
A. G. Shenstone, Phil. Trans. Roy. Soc. A237, 453.
F. Tyrén, Zeits. f. Physik 111, 314.
O. R. Wulf and L. S. Deming, Terr. Mag. 43, 283.

1939

I. S. Bowen and A. B. Wyse, Lick Obs. Bull. 495.
L. C. Green, Phys. Rev. 55, 1209.
W. Groth, Zeits. f. Elektrochemie 45, 268.
H. Hemmendinger, Thesis, Princeton, Phys. Rev. 55, 1119A.
F. W. Paul, Phys. Rev. 56, 1067.
G. B. Sabine, Phys. Rev. 55, 1064.
A. G. Shenstone, Reports on Progress in Physics (London) 5, 210.
H. W. B. Skinner, Reports on Progress in Physics (London) 5, 257.
R. Tousey, J. Opt. Soc. Am. 29, 235.

1940

H. Beutler, Phys. Rev. 57, 1073A. (Detailed work is being prepared for J. Opt. Soc. Am.).
M. J. Druyvesteyn and F. M. Penning, Rev. Mod. Phys. 12, 87.
L. C. Green and J. B. H. Kuper, Rev. Sci. Inst. 11, 250.
L. N. Liebermann, Phys. Rev. 58, 183.
M. F. Manning and M. E. Bell, Rev. Mod. Phys. 12, 215.
W. M. Preston, Phys. Rev. 57, 887.
E. G. Schneider, J. Opt. Soc. Am. 30, 128.
H. W. B. Skinner, Phil. Trans. Roy. Soc. A239, 95; see also
H. M. O’Bryan and H. W. B. Skinner, Proc. Roy. Soc. A176, 229.
R. Tousey, Phys. Rev. 57, 1060.

APPENDIX A.

Principal absorption bands of oxygen and nitrogen.

Oxygen.—Schumann–Runge system.
Bands from the lower vibrational state, S. W. Leifson, Astrophys. J., 63, 73 (1926).

λ	λ	λ	λ
(1971.4)	1883.0	1816.8	1775.9
1946.8	1864.2	1804.3	1769.2
1924.8	1846.9	1793.4	1763.8
1903.1	1831.1	1783.9	1759.6
			1757.7

Additional bands from the next lower vibrational state were found by Dieke and Jantz (C. R. 173, 581 (1921)) in the somewhat longer-wavelength region.

λ	1960.1	1938.6	1918.9	1900.7

The rotational structure of some of these bands was studied by a number of investigators, including Osebrüggen [Zeits. f. Phys., 49, 167 (1928)] and Curry and Herzberg (Ann. d. Phys., 19, 800 (1934)]. The bands of this system become weaker toward longer wavelengths.

Nitrogen.

Birge and Hopfield, Astrophys. J. 68, 257 (1928).

λ	λ	λ
1450.09	1353.61	1249.25
1415.88	1325.16	1226.63
1333.76	1298.37	1205.27
	1273.13

Watson and Koontz [Phys. Rev. 46, 32 (1934)] measured the rotational structure of some of the bands of this system. The bands of this system become weaker toward longer wavelengths.

J. K. BOICE

APPENDIX B.

Standard wavelengths in the vacuum ultraviolet region

Table 1. Standard wavelengths for copper
(Shenstone, 1936).

$\lambda$ (calculated)	$I$	Possible error in 0.001 Å	$\lambda$ (calculated)	$I$	Possible error in 0.001 Å
2000.339	60	4	1351.837	25	2
1989.849	30	4	1350.592	15	2
1979.947	50	4	1326.394	10	2
1970.489	15	4	1325.511	3	2
1944.586	25	4	1314.335	30	2
1663.003	30	3	1314.147	15	2
1660.005	20	3	1309.463	15	2
1656.326	20	3	1308.296	30	2
1649.457	25	3	1299.267	10	2
1621.426	60	3	1298.394	15	2
1617.914	20	3	1297.549	2	2
1610.298	15	3	1281.458	8	4
1608.638	25	3	1275.570	30	2
1606.834	40	3	1274.463	3	2
1604.848	20	3	1274.069	3	2
1602.387	40	3	1266.308	10	1
1598.402	40	3	1265.504	15	1
1593.557	60	3	1250.045	10	2
1590.164	40	3	1248.790	5	2
1569.216	10	2	1241.961	2	1
1566.411	40	2	1219.332	1	2
1565.925	40	2	1214.553	1	2
1558.344	30	2	1185.899	2	2
1541.701	75	2	1109.742	1	2
1540.391	30	2	1106.446	3	2
1535.004	25	2	1088.393	20	2
1519.491	50	2	1069.193	50	2
1517.630	20	2	1066.133	20	2
1496.686	35	4	1065.781	20	1
1473.976	25	2	1059.094	60	1
1444.131	2	2	1058.796	40	2
1442.136	15	2	1055.795	40	2
1402.776	15	2	1054.690	60	1
1399.355	3	2	1049.754	50	1
1393.126	10	2	1049.363	20	2
1371.840	20	2	1044.742	80	1
1370.558	2	2	1039.345	60	2
1363.501	5	2	1036.468	60	1
1362.598	20	2	1035.160	8	1
1359.935	5	2	1031.764	8	1
1359.010	20	2	1028.326	25	1
1355.304	15	2	1027.830	50	1

Table I (continued)

$\lambda$ (computed)	$I$	Possible error in 0.001 Å	$\lambda$ (computed)	$I$	Possible error in 0.001 Å
1022.100	5	1	883.837	5	2
1019.652	15	1	876.719	20	2
1018.705	50	1	866.440	5	2
1012.595	25	1	826.995	30	2
1011.433	2	1	813.882	20	2
1008.568	30	1	810.997	15	2
1004.053	30	1	736.031	25	2
1001.010	8	1	735.519	20	2
992.951	25	1	724.487	15	2
912.022	0	2	685.396	2	2
911.654	1	2	685.139	8	2
884.824	5	2

Table II. Standard wavelengths for iron
(Green, 1939)

$\lambda$ (computed)	Intensity, Schuler helium lamp	Intensity, spark	$\lambda$ (computed)	Intensity, Schuler helium lamp	Intensity, spark
2001.025	30	30	1670.990	1	—
1904.785	15	5	1663.220	15	2
1903.384	1	—	1659.470	20	10
1898.535	10	2	1658.771	15	2
1859.744	15	10	1654.476	5	1
1851.526	1	—	1652.482	0	—
1848.771	12	2	1643.576	15	2
1842.238	0	—	1640.150	12	2
1833.073	0	—	1637.398	15	2
1826.994	1	1	1633.906	15	2
1818.516	2	1	1632.665	1	—
1815.411	0	1	1625.520	20	8
1726.391	12	8	1623.090	8	1
1724.962	8	1	1612.805	20	8
1720.611	20	20	1584.949	15	1
1718.100	2	—	1577.167	1	—
1712.998	20	25	1574.921	20	1
1709.551	0	—	1574.769	0	—
1702.044	25	25	1573.826	5	—
1699.195	2	—	1572.754	1	—
1696.794	8	—	1570.244	20	1
1693.935	0	—	1569.674	12	—
1693.475	0	—	1568.017	8	—
1691.272	8	1	1566.821	20	1
1686.454	8	1	1563.788	25	2
1685.952	5	1	1559.084	20	2
1676.854	1	—	1550.273	1	—
1674.254	2	1

Table III. Standard wavelengths for hydrogen and helium
(Paschen, 1929)

Hydrogen (calculated)	Helium (He II) (calculated)	Helium (He II) (calculated)
1215.664	1640.409	303.7788
1025.717	1215.129	256.3145
972.532	1084.940	243.0244
949.739	1025.270	237.3297
937.799	992.361	234.3452
930.745	972.109
926.222	958.696
923.148	949.326

Table IV. Standard wavelengths for carbon, oxygen, and nitrogen
(More and Rieke, 1936).

Spectrum	\(\lambda\)	\(I\)	Spectrum	\(\lambda\)	\(I\)
CI	1658.135	8d	NI	1200.707	5
CI	1657.908	8d	NI	1200.218	6
CI	1657.388	5d	NI	1199.550	7
CI	1656.994	15d	NI	1176.502	1
CI	1656.271	8d	NI	1134.979	4
CI	1560.702	15d	NI	1134.417	3
OI	1560.313	8d	NI	1134.168	3
NI	1494.670	4	NII	1084.582	3
CII	1335.703	18d	NII	1083.996	2
CII	1334.534	15d	OI	999.493	2
CI	1329.099	5	OI	990.794	3
CI	1328.820	3	OI	990.205	4d
OI	1306.023	6	OII	834.467	2
OI	1304.858	8	OII	833.332	1
OI	1302.174	8	OII	832.764	0
OI	1217.643	2

APPENDIX C.

Literature on atomic spectra in the vacuum ultraviolet

(Compiled with the participation of Dr. Lora Mish).

The two works cited most often appeared in a Swedish journal under detailed Latin titles. Since they are large articles, they are cited simply as Edlén, Monograph and Söderquist, Monograph. The full reference is given below. Both works are written in German.

Edlén, Reg. Soc. Sci. Upsaliensis Nova Acta, Series IV, vol. 9, No. 6 (1934).

Söderquist, Reg. Soc. Sci. Upsaliensis Nova Acta, Series IV, vol. 9, No. 7 (1934).

The observed wavelengths are evaluated as follows:

A—The accuracy is sufficient, and the number of lines is sufficient for analysis of the spectrum.
B—The accuracy is sufficient, but the number of lines is insufficient for analysis of the spectrum.
C—The accuracy is insufficient.

The symbols A, B, C apply only to measurements, not to analyses. If no observations were made for the given spectrum below \(\lambda 2000\), one of the following symbols is used:

L—the series limit is above \(\lambda 2000\). The absence of ordinary lines below \(\lambda 2000\) is expected.
X—the series limit is below \(\lambda 2000\), but all terms can be fixed by means of lines above \(\lambda 2000\). Observations below \(\lambda 2000\) have not yet been made.
O—the series limit is below \(\lambda 2000\). Measurements below \(\lambda 2000\) have not yet been made.

The values of the ionization potential in electron-volts (hereafter denoted everywhere by I. P.) are taken from the first of the works named, unless another source is indicated.

The following letters indicate frequently cited sources:

B—Bomke, Vakuumspektroskopie (Leipzig, 1937).
BG—Bacher and Goudsmit, Atomic Energy States (New York, 1932).
E—Edlén, Monograph.
E1—Edlén, Zeits. f. Physik 104, 188 (1937).
E2—Edlén, Zeits. f. Physik 104, 407 (1937).
KS—Kruger and Shoupp, Phys. Rev. 46, 124 (1934).
KW1—Kruger and Weissberg, Phys. Rev. 48, 659 (1935).
KW2—Kruger and Weissberg, Phys. Rev. 52, 314 (1937).
M—Moore, Term Designations for Excitation Potentials (Princeton, 1934).
R—Robinson, Phys. Rev. 51, 14 (1937).
RM—Russell and Meggers, Bur. Stand. J. Res. 9, 625 (1932).
S—Söderquist, Monograph.

Hydrogen 1.

H I
I.P. 13.530 (E)

Paschen, Preuss. Akad. Wiss. Berlin Ber. 30, 662 (1929). Calculation of wavelengths, series \(\lambda 1215\). See Appendix B, Table III.

Suga, Sci. Pap. Inst. Phys. Chem. Research Tokyo 34, 7 (1937). \(\lambda 914\)—\(\lambda 938\).
Class. 18 lines. (A)

K. R. Rao and Badami, Proc. Roy. Soc. A138, 540 (1932). \(\lambda 915\)—\(\lambda 950\).
Class. 12 lines. Excitation conditions.

Frerichs, Ann. d. Physik (10) 19, 1 (1933). Series \(\lambda 1215\). Stark effect.

Frerichs and Bomke, Physik. Zeits. 35, 349 (1934), 35, 549 (1934).
Series \(\lambda 1215\). Stark effect.

The molecular spectrum of hydrogen is more “many-lined” than banded and can easily be confused with a line spectrum. See Section X.

Helium 2.

He I
I.P. 24.465 (E), 24.463 (R)

Paschen, Preuss. Akad. Wiss. Berlin Ber., p. 662 (1929), \(\lambda 506\)—\(\lambda 591\).
Class. 12 lines. (A)

Kruger, Phys. Rev. 36, 855 (1930). \(\lambda 320\)—\(\lambda 601\). Class. 3 lines. Excitation conditions. (B)

Suga, Sci. Pap. Inst. Phys. and Chem. Research Tokyo 34, 7 (1937). λ506—λ584. Class 14 lines; λ510—λ601, 2 inner-system combinations and 7 forbidden lines. Details of line contours, reversibility, and bands are established, but the wavelengths are probably erroneous. (C)

Bomke, Physik. Zeits. 36, 158 (1935). Observation of the Stark effect. There is no splitting in the λ584 series; \(S—S\) and \(S—D\) lines appear. Helium has several bands in the far ultraviolet spectrum which, at small dispersion, may appear as lines. See Section X.

He II
I.P. 54.144 (E)

Paschen, Preuss. Akad. Wiss. Berlin, Ber., p. 662 (1929). Wavelengths of the λ1640 and λ303 series calculated. See Appendix B, Table III.

Suga, Sci. Pap. Inst. Phys. and Chem. Research Tokyo 34, 7 (1937). λ228—λ303, Class 13 lines. (A)

Lyman, Astrophys. J. 60, 1 (1924). 3 members of the λ1640 series and 2—of the λ303 series. (C)

Kruger, Phys. Rev. 36, 855 (1930). λ230—λ303, Class 9 lines. A discrepancy has been established between the calculated and observed wavelengths. (C)

Lithium 3.

Li I (L)
I.P. 5.364 (E)

Li II
I.P. 75.256, 75.259*

Edlén, Monograph (1934), p. 31. λ172—λ199, Class 3 lines. (B)

*Robinson, Phys. Rev. 51, 14 (1937). λ172—λ199, Class 3 lines. (B)

Werner, Nature 116, 574 (1925); 118, 154 (1926). λ1198—λ1754, Class 5 lines (C).

Ericson and Edlén, Zeits. f. Physik 59, 656 (1930). λ172—λ199, Class 3 lines. The wavelengths are included in the monograph.

Werner, Studies over Spektroskopiske Lyskilder, Copenhagen, 1927. The work was not available. Cited according to Ericson and Edlén as a list of two additional lines λ1166 and λ1132.

Li III
I.P. 121.840

Edlén, Monograph (1934), p. 28. λ108—λ135, Class 3 lines. (B)

Ericson and Edlén, Nature 125, 233 (1930). λ114—λ135, Class 2 lines (C).

Gale and Hoag, Phys. Rev. 37, 1703 (1931). λ104—λ135, 5 lines of the Lyman series. First line of the Balmer series λ729 (C).

Beryllium 4.

Be I
I.P. 9.276 (E), 9.281

Paschen and Kruger, Ann. d. Physik (5) 8, 1005 (1931). λ1487—λ1998, Class 11 lines. (A)

Whitelaw and Mack, Phys. Rev. 47, 677 (1935); terms.

Be II
I.P. 18.119 (E), 18.13*

Paschen and Kruger, Ann. d. Physik (5) 8, 1005 (1931). λ726—λ1776, Class 18 lines. (A)

*Edlén and Ericson, Zeits. f. Physik 59, 656 (1930). λ1036—λ1143, Class 2 lines. (B)

Bowen and Millikan, Phys. Rev. 28, 256 (1926). λ842—λ1776, Class 8 lines. (C)

Be III
I.P. 153.118*; 153.012**; 153.108***

Edlén, Nature 127, 405 (1931). λ82—λ100, Class 5 lines. (B)

*Robinson, Phys. Rev. 51, 14 (1937). λ82—λ100, Class 6 lines. (B)

**Kruger and Cooper, Phys. Rev. 44, 418 (1933). λ85—λ100, Class 3 lines. (B)

Edlén, Phys. Rev. *44, 778 (1933), Monograph (1934), p. 31. λ85—λ100, class. 3 lines. (B)

Be IV  I. P. 216.628
Robinson, Phys. Rev. 50, 99 (1936). λ58—λ76, 6 lines. Computed wavelengths are given. (B)
Edlén, Nature 127, 405 (1931). λ64, λ76—2 lines. Computed wavelengths are given, included in the monograph. (B)
Edlén, Monograph (1934), p. 28. λ64, λ76—2 lines observed. Wavelengths computed for 3 lines. λ61—λ76. (B)

Boron 5.

B Fundamental.
Selwyn, Proc. Phys. Soc. 41, 392 (1929). λ1558—λ1843, 18 lines. Class. 14, B I, II. (C)

B I  I.P. 8.245 (E)
Selwyn, Proc. Phys. Soc. 41, 392 (1929). λ1567—λ1827, class. 10 lines. (C)

B II  I. P. 24.998
Edlén, Monograph (1934), p. 52. λ694—λ1843, class. 18 lines. (A)
Edlén, Zeits. f. Physik 73, 476 (1931). λ694—λ1843, class. 17 lines. Apparently the wavelengths are included in the monograph.

B III  I. P. 37.740
Edlén, Monograph (1934), p. 37. λ412—λ759, class. 8 lines. (A)
Ericson and Edlén, Zeits. f. Physik 59, 656 (1930). λ376—λ759, class. 11 lines. (C)
Edlén, Zeits. f. Physik 72, 763 (1931). λ518—λ759, class. 5 lines. Apparently the wavelengths are included in the monograph.

B IV  I.P. 258.028, 258.064
Edlén, Nature 127, 405 (1931). Monograph (1934), p. 31. λ53, λ60. Class. 2 lines. (B)
Robinson, Phys. Rev. 51, 14 (1937). λ49—λ60, class. 4 lines. (B)

B V  I. P. 338.525
Edlén, Nature 127, 405 (1931). λ49, 1 line. A computed wavelength is given. Included in the monograph. (B)
Edlén, Monograph (1934), p. 28. λ49, 1 line observed. Wavelengths computed for 3 lines—λ39—λ49. (B)

Carbon 6.

C. I  I. P. 11.212, 11.203
Paschen and Kruger, Ann. d. Physik (5) 7, 1 (1930). λ1112—λ1994, class. 113 lines. (A)
Edlén, Monograph (1934), p. 104. λ945—λ946, class. 3 lines. (A) λ1112—λ1602, 67 lines, λ Paschen and Kruger, classification for several lines changed.
Edlén, Zeits. f. Physik 85, 85 (1933). λ945—λ1659, 15 lines. (B)
Boyce and Rieke, Phys. Rev. 47, 653 (1935). λ945—λ1931, 26 lines. (B)
Webber and Watson, J. Opt. Soc. Am. 26, 307 (1936). λ1140—λ1931, 10 lines. (B)
More and Rieke, Phys. Rev. 50, 1054 (1936). λ1329—λ1658, 14 lines. (B)
Fowler and Selwyn, Proc. Roy. Soc. A118, 34 (1928). λ1260—λ1932, class. 22 lines. (C)
Edlén, Zeits. f. Physik 84, 746 (1933); terms.
Ufford, Phys. Rev. 53, 568 (1938); terms.

C II
I. P. 24.260

Edlén, Monograph (1934), p. 74. λ 425—λ 1761, class. 96 lines. (A) λ 1988, 1 line, λ Fowler and Selwyn (Proc. Roy. Soc. A120, 312 (1928).
Edlén, Zeits. f. Physik 85, 85 (1933). λ 425—λ 1761, 96 lines, included in the monograph. (B)
Boyce and Rieke, Phys. Rev. 47, 653 (1935). λ 858—λ 1761, 21 lines. (B)
Webber and Watson, J. Opt. Soc. Am. 26, 307 (1936). λ 1335, λ 1336, 2 lines. (B)
More and Rieke, Phys. Rev. 50, 1054 (1936). λ 1335, λ 1336—2 lines. (B)
Edlén, Zeits. f. Physik 84, 746 (1933); terms.

C III
I. P. 47.637

Edlén, Zeits. f. Physik 85, 85 (1933), Monograph (1934), p. 51. λ 265—λ 1923, 151 lines. Classification monograph. (A)
Kruger and Shoupp, Phys. Rev. 44, 105 (1933). λ 459—λ 535, class. 6 lines. (B)
Boyce and Rieke, Phys. Rev. 47, 653 (1935). λ 977—λ 1247, 8 lines. (B)
Fowler and Selwyn, Proc. Roy. Soc. A120, 312 (1928). λ 1894—λ 1980, 3 lines. (C)
Bowen, Phys. Rev. 38, 128 (1931). λ 319—λ 1531, class. 49 lines. (C)
Edlén, Zeits. f. Physik 84, 746 (1933); terms.
Whitelaw and Mack, Phys. Rev. 47, 677 (1935); terms.

C IV
I.P. 64.169

Edlén, Monograph (1934), p. 40. λ 198—λ 1551, class. 43 lines. (A)
Edlén, Zeits. f. Physik 85, 85 (1933). λ 200—λ 1551. 41 lines, included in the monograph. (B)
Boyce and Rieke, Phys. Rev. 47, 653 (1935). λ 1548—λ 1551, 2 lines. (B)
Edlén, Zeits. f. Physik 84, 746 (1933); terms.

C V
I. P. 390.018*; 390.120**

Edlén, Nature 127, 405 (1931). λ 40, class. 1 line, included in the monograph. (B)
Siegbahn and Söderman, Nature 129, 21 (1932). λ 34—λ 40, class. 3 lines. (B)
Edlén, Zeits. f. Physik 85, 85 (1933). λ 249, class. 2 lines, included in the monograph. (B)
*Edlén, Monograph (1934), p. 31. λ 40—λ 249, class. 3 lines. (B)
*Robinson, Phys. Rev. 51, 14 (1937). λ 33—λ 40, class. 4 lines. (B)

C VI
I. P. 487.550*

Siegbahn and Söderman, Nature 129, 21 (1932). λ 34, 1 line. The calculated wavelength is given. (B)
*Edlén, Monograph (1934), p. 28. 3 calculated lines λ 30—λ 34.

Nitrogen 7.

N I
I. P. 14.460 (E)

Ekefors, Zeits. f. Physik 63, 437 (1930). λ 1009—λ 1889, 130 lines, 51 class. (A)
Edlén, Zeits. f. Physik 85, 85 (1933). λ 1134—λ 1201, 6 lines. (B)
Boyce and Rieke, Phys. Rev. 47, 653 (1935). λ 1134—λ 1745, 12 lines. (B)
Webber and Watson, J. Opt. Soc. Am. 26, 307 (1936), λ 1134—λ 1745, 21 lines. (B)
More and Rieke, Phys. Rev. 50, 1054 (1936). λ 1134—λ 1495, 17 lines. (B).
Kamiyama, Sci. Pap. Inst. Chem. and Phys. Research Tokyo, No. 933, p. 375. λ 850—λ 1850, 120 new lines, partly classified.

N II

I. P. 29.443

Edlén, Monograph (1934), p. 109. λ 453—λ 1086, class. 64 lines, included in Zeits. f. Physik, 85, 85 (1933). (A) λ 1275—λ 1277, 3 lines λ and class. Bowen, Physik. Rev. 29, 231 (1927). λ 1343—λ 1887, 14 lines λ and class. Fowler and Freeman (see below).

Edlén, Zeits. f. Physik 85, 85 (1933). λ 453—λ 1086, 66 lines. (B)
Boyce and Rieke, Phys. Rev. 47, 653 (1935). λ 776—λ 1086, 9 lines. (B)
Weber and Watson, J. Opt. Soc. Am. 26, 307 (1936). λ 1084—λ 1743, 13 lines. (B)
More and Rieke, Phys. Rev. 50, 1054 (1936). λ 1084—λ 1085, 2 lines. (B)
Fowler and Freeman, Proc. Roy. Soc. A114, 652 (1927). λ 533—λ 1887, 39 lines, 19 class. (C)
Freeman, Proc. Roy. Soc. A124, 654 (1929). λ 1841—λ 1887, class. 10 lines. (C)
Edlén, Zeits. f. Physik 84, 746 (1933); terms.

N III

I. P. 47.201

Edlén, Monograph (1934), p. 78. λ 265—λ 1185, class. 98 lines. (A) λ 1387—λ 1921, class. 15 lines. Freeman (see below).
Edlén, Zeits. f. Physik 85, 85 (1933). λ 265—λ 1185, 98 lines, included in the monograph. (B)
Cady, Phys. Rev. 44, 821 (1933). λ 305—λ 434, class. 3 lines. (B)
Boyce and Rieke, Phys. Rev. 47, 653 (1935). λ 990—λ 992, 2 lines. (B)
Freeman, Proc. Roy. Soc. A121, 318 (1928). λ 1324—λ 1954, 48 lines, 32 class. (C)
Edlén, Zeits. f. Physik 84, 746 (1933); terms.
Edlén, Zeits. f. Physik 98, 561 (1936). Classification, corrections to the monograph.

N IV

I. P. 77.038

Edlén, Monograph (1934), p. 62. λ 182—λ 1719, class. 65 lines. (A)
Edlén, Zeits. f. Physik 85, 85 (1933). λ 182—λ 955, 64 lines, included in the monograph. (B)
Cady, Phys. Rev. 44, 821 (1933). λ 283—λ 323, class. 5 lines. (B)
Edlén, Zeits. f. Physik 73, 476 (1931). λ 1131—λ 1168, class. 3 lines. (C)
Edlén, Zeits. f. Physik 84, 746 (1933); terms.

N V

I. P. 97.397*

Cady, Phys. Rev. 44, 821 (1933). λ 134—λ 266, class. 24 lines. (B)
*Edlén, Zeits. f. Physik 85, 85 (1933); Monograph (1934), p. 41. λ 140—λ 1243, 18 lines, classification monograph. (B)
Edlén, Zeits. f. Physik 84, 746 (1933); terms.

N VI

I. P. 549.081 (E), 549.41 (R)

N VII

I. P. 663.728 (E)

Oxygen 8.

O. Basic.
Edlén, Zeits. f. Physik 85, 85 (1933). λ 188—λ 663, 7 lines. (B)

O I

I. P. 13.549 (E)

Hopfield, Phys. Rev. 37, 160 (1931). λ 1302—λ 1303, class. 3 lines. (B)
Edlén, Zeits. f. Physik 85, 85 (1933). λ 878—λ 1152, 6 lines. (B)
Boyce and Rieke, Phys. Rev. 47, 653 (1935). λ 935—λ 1306, 11 lines. (B)
Webber and Watson, J. Opt. Soc. Am. 26, 307 (1936). λ 1305, 1 line. (B)
More and Rieke, Phys. Rev. 50, 1054 (1936). λ 972—λ 1306, 12 lines. (B)
Frerichs, Phys. Rev. 36, 398 (1930). λ 748—λ 1218, 34 lines, 21 class. (C)

O II

I. P. 34.941

Edlén, Monograph (1934), p. 136. λ 377—λ 834, class. 84 lines, included in Zeits. f. Phys. 85, 85 (1933). (A)

Edlén, Zeits. f. Physik 85, 85 (1933). λ 377—λ 834, 86 lines. (B)
Edlén, Zeits. f. Physik 93, 726 (1935). λ 740—λ 741, class. 3 lines. (B)
Boyce and Rieke, Phys. Rev. 47, 653 (1935). λ 534—λ 834, 8 lines. (B)
More and Rieke, Phys. Rev. 50, 1054 (1936). λ 833—λ 834, 3 lines. (B)
Fowler, Proc. Roy. Soc. A110, 476 (1926). λ 1956—λ 1964, 5 lines, classified.
Bowen in: Phys. Rev. 29, 231 (1927). (C)
Edlén, Zeits. f. Physik 84, 746 (1933); terms.

O III  I. P. 54.625
Edlén, Monograph (1934), p. 115. λ 226—λ 1154, class. 161 lines, included in: Zeits. f. Phys. 85, 85 (1933). (A)
Edlén, Zeits. f. Phys. 85, 85 (1933). λ 226—λ 1154, 168 lines. (B)
Kruger and Shoupp, Phys. Rev. 44, 105 (1933). λ 303—λ 374, class. 17 lines. (B)
Cady, Phys. Rev. 44, 821 (1933). λ 359, class. 3 lines. (B)
Edlén, Zeits. f. Physik 93, 726 (1935). λ 554—λ 659, class. 4 lines. (B)
Boyce and Rieke, Phys. Rev. 47, 653 (1935). λ 600—λ 899, 6 lines. (B)
Fowler, Proc. Roy. Soc. A117, 317 (1928). λ 1903—λ 1917, class. 3 lines. (C)
Edlén, Zeits. f. Physik 84, 746 (1933); terms.

O IV  I. P. 77.026
Edlén, Monograph (1934), p. 87. λ 152—λ 1344, class. 201 lines, included in: Zeits. f. Physik 85, 85 (1933). (A)
Edlén, Zeits. f. Physik, 85, 85 (1933). λ 152—λ 1344, 205 lines. (B)
Kruger and Shoupp, Phys. Rev. 44, 105 (1933). λ 231—λ 280, class. 23 lines. (B)
Cady, Phys. Rev. 44, 821 (1933). λ 174—λ 196, class. 6 lines. (B)
Edlén, Zeits. f. Physik 93, 726 (1935). λ 153—λ 279, class. 9 lines. (B)
Edlén, Zeits. f. Physik 84, 746 (1933); terms.
Whitelaw and Mack, Phys. Rev. 47, 677 (1935); terms.

O V  I. P. 113.298
Edlén, Zeits. f. Physik 85, 85 (1933), Monograph (1934), p. 62, λ 122—λ 1371, 108 lines, classification monograph. (A)
Kruger and Shoupp, Phys. Rev. 44, 105 (1933). λ 151—λ 215, class. 9 lines. (B)
Cady, Phys. Rev. 44, 821 (1933). λ 134—λ 774, class. 26 lines. (B)
Edlén, Zeits. f. Physik 73, 476 (1931). λ 630—λ 966, class. 6 lines. (C)
Edlén, Zeits. f. Physik 84, 746 (1933); terms.

O VI  I. P. 137.424
Edlén, Zeits. f. Physik 85, 85 (1933), Monograph (1934), p. 44. λ 104—λ 1038, 24 lines, classification monograph. (A)
Edlén, Zeits. f. Physik 84, 746 (1933); terms.

O VII  I. P. 735.218 (E), 735.74 (R)
O VIII  I. P. 867.087 (E)

Fluorine 9.

F Basic.
Edlén, Zeits. f. Physik 94, 47 (1935). λ 87—λ 1140, 525 lines, 516 indicated as weak. (B)

F I  I. P. 17.34
Edlén, Zeits. f. Physik 93, 433 (1935). λ 952—λ 959, class. 4 lines. (B)
Edlén, Zeits. f. Physik 94, 47 (1935). λ 807—λ 959, 6 lines. (C + B)
Bowen, Phys. Rev. 29, 231 (1927). λ 807—λ 958, class. 6 lines. (C)

F II                I. P. 34.81*

Bowen, Phys. Rev. 45, 82 (1934). λ353—λ608, class. 41 lines. (B)
* Edlén, Zeits. f. Physik 94, 47 (1935). λ349—λ608, 35 lines. (B)
Dingle, Proc. Roy. Soc. A128, 600 (1930). λ1702—λ1747, class. 5 lines. (C)
Edlén, Zeits. f. Physik 93, 433 (1935); terms.

F III                I. P. 62.35

Bowen, Phys. Rev. 45, 82 (1934). λ215—λ1104, class. 59 lines. (B)
Edlén, Zeits. f. Physik 93, 433 (1935); 94, 47 (1935). λ215—λ743, 96 lines, classification in Zeits. f. Physik 93. (A)

F IV                I. P. 87.34, 86.72*

Bowen, Phys. Rev. 45, 82 (1934). λ200—λ679, class. 29 lines. (B)
* Edlén, Zeits. f. Physik 92, 19 (1934); 94, 47 (1935). λ141—λ679, 169 lines, classification in Zeits. f. Physik 92. (A)

F V                I. P. 113.670

Edlén, Zeits. f. Physik 89, 597 (1934). λ120—λ1088, class. 123 lines. (A)
Edlén, Zeits. f. Physik 94, 47 (1935). λ120—λ1088, 123 lines, the same as in Zeits. f. Physik 89, λ122—λ197, class. 16 lines. (B)

F VI                I. P. 156.369

Edlén, Zeits. f. Physik 89, 179 (1934); 94, 47 (1935). λ99—λ1140, 57 lines, classification in Zeits. f. Physik 89. (A)

F VII                I. P. 184.261

Edlén, Zeits. f. Physik 89, 179 (1934); 94, 47 (1935). λ87—λ135, 9 lines, classification in Zeits. f. Physik 89. (A)

F VIII

Flemberg, Zeits. f. Physik 111, 747 (1939). λ16, class. 2 lines (wavelengths in X-units). (B)

Neon 10.

Ne I                I. P. 21.47

Boyce, Phys. Rev. 46, 378 (1934). λ743—λ587, class. 16 lines. (A)
Suga, Sci. Pap. Inst. Phys. and Chem. Research Tokyo 34, 7 (1937). λ743—λ576, class. 32 lines. Excitation study. (C)

Ne II                I. P. 40.91

Boyce, Phys. Rev. 46, 378 (1934). λ324—λ1938, class. 46 lines. (A)

Ne III                I. P. 63.3

Boyce, Phys. Rev. 46, 378 (1934). λ282—λ1257, class. 17 lines. (A)
von Keussler, Zeits. f. Physik 85, 1 (1933). λ251—λ313, class. 13 lines. (B)

Ne IV                I. P. (97)

Boyce, Phys. Rev. 46, 378 (1934). λ358—λ543, class. 11 lines. The doublets in this spectrum are very close, and a considerably greater resolving power is necessary. (C)

Ne V

Paul, Phys. Rev. 56, 1067 (1939). λ480—λ572, class. 7 lines. (C)

Sodium 11.

Na Ground.

Söderquist, Monograph (1934), p. 13. λ105—λ516, 19 lines. (B)

Na I                (L) I. P. 5.113 (S)

Na II                I. P. 47.0.65
Söderquist, Monograph (1934), p. 26. \(\lambda 281—\lambda 376\), class. 6 lines. (B)
Vance, Phys. Rev. 41, 430 (1932). \(\lambda 270—\lambda 302\), class. 9 lines. (C)

Na III                I. P. 71.307
Söderquist, Zeits. f. Physik 76, 316 (1932). \(\lambda 188—\lambda 381\), class. 26 lines, included in the monograph. (B)
* Söderquist, Monograph (1934), p. 43. \(\lambda 183—\lambda 1986\), class. 41 lines. (B)
Tomboulian, Phys. Rev. 54, 347 (1938). \(\lambda 1100—\lambda 1996\), class. 106 lines. (B)
Vance, Phys. Rev. 41*, 480 (1932). \(\lambda 230—\lambda 273\), class. 9 lines. (C)

Na IV                I. P. 98.409
Söderquist, Zeits. f. Physik 79, 634 (1932). \(\lambda 150—\lambda 413\), class. 43 lines, included in the monograph. (B)
Söderquist, Monograph (1934), p. 54. \(\lambda 129—\lambda 413\), class. 72 lines. (B)
Vance, Phys. Rev. 41*, 480 (1932). \(\lambda 320—\lambda 412\), class. 7 lines. (C)

Na V                I. P. 137.638
Söderquist, Monograph (1934), p. 68, \(\lambda 106—\lambda 515\), class. 72 lines. (B)

Na VI                I. P. 171.374
Söderquist, Monograph (1934), p. 79. \(\lambda 88—\lambda 639\), class. 62 lines (B)

Na VII               I. P. 208.235
Söderquist, Monograph (1934), p. 89. \(\lambda 86—\lambda 492\), class. 33 lines. (B)

Na VIII              I. P. 262.969
Söderquist, Monograph (1934), p. 96, \(\lambda 77—\lambda 497\), class. 15 lines. (B)

Na IX               I. P. 298.388
Söderquist, Monograph (1934), p. 100. \(\lambda 77—\lambda 682\), class. 3 lines. (B)

Magnesium 12.

Mg. Principal
Söderquist, Monograph (1934), pp. 16, 101. \(\lambda 68—\lambda 491\), 10 lines. (B)

Mg.                I. P. 7.608 (S)
Seiwyn, Proc. Phys. Soc. 41, 392 (1929). \(\lambda 1668—\lambda 1828\), class. 5 lines. (C)

Mg. II               I. P. 14.959 (S)
Lyman, Science 60, 388 (1924). \(\lambda 947—\lambda 1241\), class. 4 lines. (C)

Mg. III              I. P. 79.736
Söderquist, Monograph (1934), p. 27. \(\lambda 182—\lambda 1979\), class. 58 lines. (A)
Söderquist, Zeits. f. Physik 79, 634 (1932). \(\lambda 164—\lambda 234\), class. 12 lines, included in the monograph. (B)

Mg. IV              I. P. 108.774
Söderquist, Monograph (1934), p. 44. \(\lambda 123—\lambda 1957\), class. 73 lines. (A)
Söderquist, Zeits. f. Physik 76, 756 (1932). \(\lambda 137—\lambda 324\), class. 38 lines. (B)

Mg V               I. P. 140.466
Söderquist, Zeits. f. Physik 79, 634 (1932). \(\lambda 110—\lambda 356\), class. 39 lines, included in the monograph. (B)

Söderquist, Monograph (1934), p. 55. λ95—λ356, Class. 53 lines. (B)

Mg VI                I. P. 185.566
Söderquist, Monograph (1934), p. 69. λ79—λ404, Class. 46 lines. (B)

Mg VII               I. P. 224.309
Söderquist, Monograph (1934), p. 80. λ79—λ435, Class. 29 lines. (B)

Mg VIII              I. P. 265.580
Söderquist, Monograph (1934), p. 89. λ74—λ437, Class. 18 lines. (B)

Mg IX               I. P. 326.507
Söderquist, Monograph (1934), p. 96. λ67—λ444, Class. 9 lines. (B)

Absorption.

Kremenewsky, Physik. Zeits. d. Sowjetunion 2, 491 (1931). λ1629—λ1826, Class. 15 lines.

Aluminum 13.

Al Principal.
Söderquist, Monograph (1934), pp. 17 and 101. λ58—λ229, 21 lines. λ1150—λ1464, 13 lines. (B)

Al I                I. P. 5.956. (S)
Selwyn, Proc. Phys. Soc. 41, 392 (1929). λ1763—λ1769, 3 lines. Class. \(pp'\) groups. (C)

Al II               I. P. 18.733 (S)
Zumstein, Phys. Rev. 38, 2214 (1931). λ1670—λ1863, Class. 12 lines. (B)
Sawyer and Paschen, Ann. d. Physik 84, 1 (1927). λ933—λ1991, 99 lines, class. 92. (C)
Paschen, Ann. d. Physik (5) 12, 509 (1932). λ1858—λ1889, Class. 4 lines. Classification and corrections to Ann. d. Physik 84. (C)
Pincnerle, Accad. dei Lincei 18, 35 (1932). Perturbed lines in Al II. The journal was not available.

Al III              I. P. 28.306 (S)
Zumstein, Phys. Rev. 38, 2214 (1931). λ1854—λ1863, Class. 2 lines. (B)
Ekefors, Zeits. f. Physik 51, 471 (1928). λ487—λ1863, Class. 21 lines. (B)

Al IV              I. P. 119.386
Söderquist, Monograph (1934), p. 29. λ85—λ1881, Class. 66 lines. (A)
Söderquist, Zeits. f. Physik 79, 634 (1932). λ111—λ162, Class. 12 lines. (B)

Al V               I. P. 153.109
Söderquist, Zeits. f. Physik 76*, 756 (1932). λ95—λ281, Class. 38 lines, included in the monograph. (B)
* Söderquist, Monograph (1934), p. 45. λ86—λ282, Class. 56 lines. (B)

Al VI              I. P. 189.347
Söderquist, Zeits. f. Physik 79*, 634 (1932). λ87—λ313, Class. 30 lines, included in the monograph. (B)
Söderquist, Monograph (1934), p. 56. λ70—λ313, 44 lines, class. 42. (B)

Al VII            I. P. 240.450
Söderquist, Monograph (1934), p. 70. λ62—λ357, Class. 34 lines. (B)

Al VIII           I. P. 283.747
Söderquist, Monograph (1934), p. 80. λ64—λ387, Class. 19 lines. (B)

Al IX            I. P. 330.465
Söderquist, Monograph (1934), p. 90. λ60—λ393, Class. 6 lines. (B)

Al. X            I. P. 397.177
Söderquist, Monograph (1934), p. 96. λ55—λ333, Class. 5 lines. (B)

Silicon 14.

Si Principal.
Söderquist, Monograph (1934), p. 18. λ62—λ79, 3 lines. (B)

Si I            I. P. 8.077 (S), 8.11
Kiess, J. Research Nat. Bur. Stand. 21, 185 (1938). λ1565—λ1991, Class.
  148 lines. (A)
Fowler, Proc. Roy. Soc. A123, 422 (1929). λ1590—λ1992, 133 lines,
  Class. 70. (A)

Si II            I. P. 16.261 (S)
Zumstein, Phys. Rev. 38, 2214 (1931). λ1527—λ1817, Class. 4 lines. (B)
Bowen, Phys. Rev. 39, 8 (1932). λ1246—λ1354, Class. 9 lines. (B)
Kiess, J. Research Nat. Bur. Stand. 21, 185 (1938). λ1527—λ1817, Class.
  5 lines. (B)
Bowen and Millikan, Phys. Rev. 26, 150 (1925). λ1190—λ1197, Class.
  4 lines. (C)
Fowler, Phil. Trans. Roy. Soc. A225, 1 (1925). λ990—λ1818, Class. 17
  lines. (C)
Bowen, Phys. Rev. 31, 34 (1928). λ1246—1309, Class. 6 lines. (C)

Si III           I. P. 33.329 (S)
Bowen, Phys. Rev. 39, 8 (1932). λ566—λ1896, Class. 59 lines. (A)

Si IV           I. P. 44.915
Edlén and Söderquist, Zeits. f. Physik 87, 217 (1933). λ815—λ1798,
  Class. 11 lines. (A)

Si V            I. P. 165.660
Söderquist, Monograph (1934), p. 30. λ85—λ119, Class. 9 lines. (B)

Si VI           I. P. 203.835
Söderquist, Monograph (1934), p. 46. λ77—λ250, 25 lines, Class. 22. (B)

Si VII           I. P. 244.635
Söderquist, Monograph (1934), p. 56. λ68—λ279, Class. 21 lines. (B)

Si VIII          I. P. 302.720
Söderquist, Monograph (1934), p. 70. λ61—λ320, Class. 16 lines. (B)

Absorption.

Takamine, Suga and Kamiyama, Sci. Pap. Inst. Phys. and Chem.
  Research Tokyo 33, 247 (1937). λ1630—λ2000.

Phosphorus 15.

P Principal.
Quepey, J. de phys. et rad. (6) 10, 299 (1929). λ 1189—λ 1987, 192 lines. (C)

P I                                             I. P. 10.9
Robinson, Phys. Rev. 49, 297 (1936). λ 1324—λ 1908, Class. 41 lines. (B)

P II                                            I. P. 19.59
Robinson, Phys. Rev. 49, 297 (1936). λ 783—λ 1880, Class. 102 lines. (A)
Robinson, Phys. Rev. 51, 726 (1937). λ 927—λ 930, Class. 3 lines. Corrections to Phys. Rev. 49 (B)

P III                                           I. P. 30.012
Robinson, Phys. Rev. 51, 726 (1937). λ 497—λ 1758, 88, Class. 88 lines. (A)
Bowen, Phys. Rev. 39, 8 (1932). λ 781—λ 978, Class. 24 lines. (B)

P IV                                            I. P. 51.106
Robinson, Phys. Rev. 51, 726 (1937). λ 283—λ 1910, Class. 88 lines. (A)
Bowen, Phys. Rev. 39, 8 (1932). λ 628—λ 1910, Class. 46 lines. (B)

P V                                             I. P. 69.698
Robinson, Phys. Rev. 51, 726 (1937). λ 210—λ 1611, Class. 31 lines. (A)

Sulfur 16.

S Principal.
Bloch and Bloch, J. de phys. et rad. (7) 6, 441 (1935). λ 328—λ 1259, 404 lines, 147 indicated as weak. (B)
Lacroute, J. de phys. et rad. (6) 9, 180 (1928). λ 1251—λ 1999, 50 lines. (C)

S I                                             I. P. 10.31 (M)
Ruedy, Phys. Rev. 44, 757 (1933). λ 1000—λ 1915, 83 lines, class. 53. (A)
Bloch and Bloch, J. de phys. et rad. (7) 6, 441 (1935). λ 1204, 1 line. (B)

S II                                            I. P. 23.3
Bloch and Bloch, J. de phys. et rad. (7) 6, 441 (1935). λ 640—λ 1260, 38 lines. (B)
* Ingram, Phys. Rev. 32, 172 (1928). λ 640—λ 1260, Class. 43 lines. (C)
Gilles*, Thèse Paris (1930) Sér. A No. 1285, 2154; Classification.

S III                                           I. P. 34.9¹
Bloch and Bloch, J. de phys. et rad. (7) 6, 441 (1935). λ 475—λ 1202, 55 lines, class. 15. (B)
Robinson, Phys. Rev. 52, 724 (1937). λ 500—λ 1077, class. 17 lines. (B)

Ingram, Phys. Rev. 33, 907 (1929). λ 484—λ 1391, Class. 41 lines. (C)

S IV                                            I. P. 47.08 (BG)
Bowen, Phys. Rev. 39, 8 (1932). λ 519—λ 804, Class. 21 lines. (B)
Bloch and Bloch, J. de phys. et rad. (7) 6, 441 (1935). λ 519—λ 1074, 42 lines, class. 6. (B)
Millikan and Bowen, Phys. Rev. 25, 600 (1925). λ 551—λ 1745, Class. 16 lines. (C)
Bowen, Phys. Rev. 31, 34 (1928). λ 836—λ 1297, Class. 10 lines. (B)

S V                                             I. P. 63 (BG)
Bowen, Phys. Rev. 39, 8 (1932). λ 437—λ 906, Class. 30 lines. (B)
Bloch and Bloch, J. de phys. et rad. (7) 6, 441 (1935). λ 458—λ 861, 11 lines. (B)

Bowen and Millikan, Phys. Rev. 25, 591 (1925). λ873—λ876, class. 2 lines. (C)

S VI　　　　　　　　　　　　　　　　I. P. 87.610
Robinson, Phys. Rev. 52, 724 (1937). λ171—λ945, class. 26 lines. (A)
Bowen and Millikan, Phys. Rev. 25, 295 (1925). λ289—λ1118, class. 9 lines. (C)

S VIII　　　　　　　　　　　　　　　I. P.
Robinson, Phys. Rev. 52, 724 (1937). λ199—λ203, class. 2 lines. (B)

Chlorine 17.

Cl Basic.
Edlén, Zeits. f. Physik 100, 726 (1936). λ40—λ48, 4 lines. (B).
Vaudet, Comptes Rendus 185, 1270 (1927). λ1298—λ1999, 59 lines. (C)

Cl I　　　　　　　　　　　　　　　　I. P. 12.96 (BG)
Bowen, Phys. Rev. 31, 497 (1928). λ1336—λ1364, class. 4 lines. (C)

Cl II　　　　　　　　　　　　　　　　I. P. 23.70
Kiess and de Bruin, J. Research Nat. Bur. Stand. 23, 443 (1939). λ558—λ1923, class. 119 lines. (C)
Murakawa, Zeits. f. Physik 109, 162 (1938). Classification.
Bowen, Phys. Rev. 31, 34 (1928). λ634—λ1080, class. 26 lines. (C)

Cl III　　　　　　　　　　　　　　　I. P. 39.7 (BG)
Bowen, Phys. Rev. 45, 401 (1934). λ406—λ1984, class. 90 lines. (A)
Bowen, Phys. Rev. 31, 34 (1928). λ573—λ1015, class. 31 lines. (C)

Cl IV　　　　　　　　　　　　　　　I. P. 53.16 (B)
Bowen, Phys. Rev. 45, 401 (1934). λ318—λ1652, class. 43 lines. (B)
Bowen, Phys. Rev. 46, 377 (1934). λ332—λ757, class. 10 lines. (B)
Bowen, Phys. Rev. 31, 34 (1928). λ463—λ986, class. 26 lines. (C)
Deb, Proc. Acad. Sci. U. P. India 2, 43 (1932). Terms. The journal was not available.

Cl V　　　　　　　　　　　　　　　　I. P. 67.4 (B)
Bowen, Phys. Rev. 45, 401 (1934). λ286—λ556, class. 22 lines. (B)
Millikan and Bowen, Phys. Rev. 25, 600 (1925). λ538—λ639, class. 10 lines. (C)
Bowen, Phys. Rev. 31, 34 (1928). λ676—λ895, class. 17 lines. (C)
Deb, Proc. Acad. Sci. U. P. India 2, 43 (1932). Terms. The journal was not available.

Cl VI　　　　　　　　　　　　　　　I. P. 88.6 (B)
Bowen and Millikan, Phys. Rev. 25, 591 (1925). λ671—λ737, class. 6 lines. (C)
Parker and Phillips, Phys. Rev. 57, 140 (1940). λ194—λ737, class. 23 lines. (B)

Cl VII　　　　　　　　　　　　　　　I. P. 113.73
Phillips, Phys. Rev. 53, 248 (1938). λ174—λ813, class. 22 lines. (A)

Cl VIII　　　　　　　　　　　　　　I. P. 346.6
Edlén, Zeits. f. Physik 100, 726 (1936). λ39—λ60, class. 13 lines. (B)

Cl IX　　　　　　　　　　　　　　　I. P. 398.8
Edlén, Zeits. f. Physik 100, 726 (1936). λ42—λ54, class. 35 lines. (B)

Cl X
I. P. 453.1
Edién, Zeits. f. Physik 100, 726 (1936). \(\lambda 39\)—\(\lambda 48\), classified. 15 lines. (B)

Cl XI
Edién, Zeits. f. Physik 100, 726 (1936). \(\lambda 40\)—\(\lambda 41\), classified. 2 lines. (B)

Argon 18.

A I
I. P. 15.69
Boyce, Phys. Rev. 48, 396 (1935). \(\lambda 866\)—\(\lambda 1066\), classified. 6 lines. (A)
Dorgelo and Abbink, Zeits. f. Physik 41, 753 (1927). \(\lambda 797\)—\(\lambda 1066\), 22 lines. (C)

A II
I. P. 27.49
Boyce, Phys. Rev. 48, 396 (1935). \(\lambda 487\)—\(\lambda 1978\), classified. 80 lines. (A)
Edién, Zeits. f. Physik 104*, 407 (1937). Corrections to Boyce, concerning the analysis, but not the wavelengths.

A III
I. P. 40.48
von Keussler, Zeits. f. Physik 84, 42 (1933). \(\lambda 508\)—\(\lambda 1836\), 20 lines, classified. 17. (A)
Boyce, Phys. Rev. 48, 396 (1935), 49*, 351 (1936). \(\lambda 395\)—\(\lambda 1973\), classified. 87 lines. (A)

A IV
I. P. (61)
Boyce, Phys. Rev. 48, 396 (1935). \(\lambda 396\)—\(\lambda 1197\), classified. 26 lines. (B)
de Bruin, Physica 3, 809 (1936). Extension of the analysis.

A V
I. P. (78)
Boyce, Phys. Rev. 48, 396 (1935). \(\lambda 705\)—\(\lambda 836\), classified. 10 lines, two multiplets. (B)
Parker and Phillips, Phys. Rev. 58, 93 (1940). \(\lambda 522\)—\(\lambda 527\), classified. 3 lines. (B)

A VI
Parker and Phillips, Phys. Rev. 58, 93 (1940). \(\lambda 544\)—\(\lambda 596\), classified. 6 lines. (B)

A VII
Parker and Phillips, Phys. Rev. 58, 93 (1940). \(\lambda 191\)—\(\lambda 479\), classified. 9 lines. (B)

A VIII
Parker and Phillips, Phys. Rev. 58, 93 (1940). \(\lambda 229\)—\(\lambda 526\), classified. 7 lines. (B)

Absorption.

Beutler, Zeits. f. Physik 93, 177 (1935). \(\lambda 781\)—\(\lambda 1067\), classified. 20 lines.

Potassium 19.

K Principal
Ekefors, Zeits. f. Physik 71, 53 (1931). \(\lambda 156\)—\(\lambda 1034\), 659 lines, 483 indicated as weak. (B)

K I (L)
I. P. 4.32 (BG)

K II
I. P. 31.67 (KW), 31.7
Ekefors, Zeits. f. Physik 71, 53 (1931). \(\lambda 261\)—\(\lambda 613\), 26 lines. (B)
Bowen Phys. Rev. 31, 497 (1928). \(\lambda 601\)—\(\lambda 615\), classified. 4 lines. (C)

K III              I. P. 45.5*

Ekefors, Zeits. f. Physik 71, 53 (1931). λ204—λ874, classified 77 lines. (B)
Bowen, Phys. Rev. 31, 497 (1928). λ467—λ779, classified 6 lines. (C)
Ram, Ind. J. Phys. 8, 151 (1933). Classification.
Kruger and Phillips, Phys. Rev. 51, 1087 (1937). Classification.
Edlén, Zeits. f. Physik 104, 407 (1937). Classification.

K IV              I. P. 62.5 (B)

Ekefors, Zeits. f. Physik 71, 53 (1931). λ166—λ1026, 62 lines. (B)
Bowen, Phys. Rev. 46, 791 (1934). λ272—λ417, classified 19 lines. (B)
Bowen, Phys. Rev. 31, 497 (1928). λ737—λ755, classified 6 lines. (C)
Ram, Ind. J. Phys. 8, 151 (1933). Classification.
Whitford, Phys. Rev. 46, 793 (1934). Classification.
Robinson, Phys. Rev. 52, 724 (1937). Classification.

K V

Ekefors, Zeits. f. Physik 71, 53 (1931). λ213—λ1027, 131 lines. (B)
Bowen, Phys. Rev. 46, 791 (1934). λ295—λ645, classified 26 lines. (B)

K VI

Ekefors, Zeits. f. Physik 71, 53 (1931). λ156—λ1000, 187 lines. (B)
Ram, Ind. J. Phys. 8, 151 (1933). Classification.
Robinson, Phys. Rev. 52, 724 (1937). Classification.

•

K VII

Whitford, Phys. Rev. 46, 793 (1934). Classification.

K VIII

Whitford, Phys. Rev. 46, 793 (1934). Classification.
Parker and Phillips, Phys. Rev. 57, 140 (1940). λ156—λ573, classified 23 lines. (B)

K IX              I. P. 174.90

Edlén, Zeits. f. Physik 100, 621 (1936). λ185—λ636, classified 6 lines. (A)
Whitford, Phys. Rev. 46, 793 (1934). Classification.

K X              I. P. 501.4

Edlén and Tyrén, Zeits. f. Physik 101, 206 (1935). λ30—λ42, classified 11 lines. (B)

K XI

Edlén and Tyrén, Zeits. f. Physik 101, 206 (1936). λ32—λ38, classified 8 lines. (B)

•

Absorption.

Beutler and Guggenheimer, Zeits. f. Physik 87, 188 (1933). λ653, λ662, classified 2 lines.

Calcium 20.

Ca Basic.
Ekefors, Zeits. f. Physik 71, 53 (1931). λ136—λ1035, 728 lines, 476 indicated as weak. (B)

Ca I (L)            I. P. 6.09 (BG)
Ca II              I. P. 11.82 (BG)

Saunders and Russell, Astrophys. J. 62, 51 (1925). λ1369—λ1851, classified 20 lines. (C)
Selwyn, Proc. Phys. Soc. 41, 392 (1929). λ1649—λ1841, classified 20 lines. (C)

Ca III I. P. 51.0
Ekefors, Zeits. f. Physik 71, 53 (1931). λ 242—λ 1035, 94 lines. (B)
Bowen, Phys. Rev. 31, 497 (1928). λ 404—λ 2000, class. 64 lines. (C)

Ca IV I. P. 67 (E 1)
Ekefors, Zeits. f. Physik 71, 53 (1931). λ 249—λ 1031, class. 52 lines. (B)
Kruger and Phillips, Phys. Rev. 51, 1087 (1937). λ 297—λ 670, class. 16 lines. (B)
Ram, Ind. J. Phys. 8, 163 (1933). Classification.

Ca V
Ekefors, Zeits. f. Physik 71, 53 (1931). λ 231—λ 1022, 83 lines. (B)
Bowen, Phys. Rev. 46, 791 (1934). λ 184—λ 387, class. 36 lines. (B)
Bowen, Phys. Rev. 31, 497 (1928). λ 638—λ 657, class. 6 lines. (C)
Ram, Ind. J. Phys. 8, 163 (1933). Classification.

Ca VI
Ekefors, Zeits. f. Physik 71, 53 (1931). λ 228—λ 1033, 102 lines. (B)
Bowen, Phys. Rev. 46, 791 (1934). λ 228—λ 642, class. 31 lines. (B)

Ca VII
Ekefors, Zeits. f. Physik 71, 53 (1931). λ 250—λ 1034, 136 lines. (B)
Whitford, Phys. Rev. 46, 793 (1934). Classification.
Robinson, Phys. Rev. 52, 724 (1937). Classification.

Ca VIII
Whitford, Phys. Rev. 46, 793 (1934). Classification.

Ca IX
Edlén, Zeits. f. Physik 103, 535 (1936). Classification.
Parker and Phillips, Phys. Rev. 57, 140 (1940). λ 101—λ 516, class. 25 lines. (B)

Ca X I. P. 120.21
Edlén, Zeits. f. Physik 100, 621 (1936). λ 152—λ 574, class. 6 lines. (A)

Ca XI I. P. 589.0
Edlén and Tyrén, Zeits. f. Physik 101, 206 (1936). λ 25—λ 36, class. 11 lines. (B)

Ca XII
Edlén and Tyrén, Zeits. f. Physik 101, 206 (1936). λ 27—λ 33, class. 9 lines. (B)

Scandium 21.

Sc Neutral
Beckman, Thesis, Upsala (1937). λ 62—λ 1613, 770 lines, 189 indicated as weak. (B)
McLennan and Liggett, Trans. Roy. Soc. Canada 20, 377 (1926). λ 1598—λ 1994, 8 lines. (C)

Sc I (X) I. P. 6.7 (BG)
Sc II (O) I. P. 12.8 (BG)

Sc III I. P. 24.635 (KW 2)
Beckman, Thesis, Upsala (1937). λ 627—λ 1610, 17 lines. (B)
Smith, Proc. Nat. Acad. Sci. 13, 65 (1927). λ 730—λ 1994, class. 6 lines. (C)

Sc IV I. P. 73.6*
Beckman, Thesis, Upsala (1937). λ 216—λ 376, 22 lines, class. 4. (B)

Kruger, Weissberg and Phillips, Phys. Rev. 51, 1090 (1937). λ215—λ298, class. 4 lines. (B)

Sc V            I. P. 91 (E 2)
Beckman, Thesis, Upsala (1937). λ229—λ588, 21 lines, class. 13. (B)
Kruger and Phillips, Phys. Rev. 51, 1087 (1937). λ229—λ588, class. 16 lines. (B)

Sc VI           I. P. 110.5 (E 1)
Beckman, Thesis, Upsala (1937). λ203—λ581, 31 lines, class. 23. (B)
Kruger and Pattin, Phys. Rev. 52, 621 (1937). λ201—λ581, class. 29 lines. (B)

Sc VII
Beckman, Thesis, Upsala (1937). λ183—λ568. 25 lines, class. 18. (B)
Kruger and Pattin, Phys. Rev. 52, 621 (1937). λ186—λ571, class. 16 lines. (B)

Sc VIII
Beckman, Thesis, Upsala (1937). λ165—λ494, 15 lines, class. 13. (B)
Kruger and Phillips, Phys. Rev. 52, 97 (1937). λ164—λ375, class. 15 lines. (B)

Sc IX
Beckman, Thesis, Upsala (1937). λ119—λ538, class. 17 lines. (B)
Kruger and Phillips, Phys. Rev. 52, 97 (1937). λ318—λ427, class. 9 lines. (B)

Sc X
Beckman, Thesis, Upsala (1937). λ76—λ628, class. 26 lines. (B)
Parker and Phillips, Phys. Rev. 57, 140 (1940). λ135—λ343, class. 12 lines. (B)

Sc. XI          I.P. 248.58, 248.61 ¶
Beckman, Thesis, Upsala (1937). λ62—λ169, class. 30 lines. (A)
Edlén, Zeits. f. Physik 100, 621 (1936). λ95—λ169, class. 10 lines. (B)

Sc XII          I.P. 683.4¶
Edlén and Tyrén, Zeits. f. Physik 101, 206 (1936). λ27—λ31, class. 5 lines. (B)

Titanium 22.

Ti I (X)         I.P. 6.81 (BG)
Ti II          I.P. 13.6
Russell, Astrophys. J. 66, 283 (1927). λ1906—λ1914, class. 6 lines. (C)

Ti III          I. P. 27.6
Russell and Lang, Astrophys. J. 66, 13 (1927). λ1002—λ1927; class. 55 lines. (C)

Ti IV          I.P. 43.06
Russell and Lang, Astrophys. J. 66, 13 (1927). λ424—λ1469; class. 11 lines. (C)

Ti V          I.P. 99.7, 99.4
Kruger and Weissberg, Phys. Rev. 48, 659 (1935). λ225, λ229, class. 2 lines. (B)
Kruger, Weissberg and Phillips, Phys. Rev. 51*, 1090 (1937). λ163, λ164, class. 2 lines. (B)

Ti VI
I.P. 119
Edlén, Zeits. f. Physik 104, 407 (1937). \(\lambda 182—\lambda 202\), class. 10 lines. (B)

Ti VII
I.P. 140.1
Kruger and Pattin, Phys. Rev. 52, 621 (1937). \(\lambda 500—\lambda 522\), class. 5 lines. (B)
* Edlén, Zeits. f. Physik 104*, 188 (1938). \(\lambda 164—\lambda 179\), class. 19 lines. (B)

Ti VIII
Kruger and Pattin, Phys. Rev. 52, 621 (1937). \(\lambda 150—\lambda 162\), class. 15 lines. (B)

Ti X
Edlén, Zeits. f. Physik 103, 536 (1936). \(\lambda 101\), \(\lambda 102\), class. 2 lines. (B)

Ti XI
I.P. (264)
Edlén, Zeits. f. Physik 103, 536 (1936). \(\lambda 72—\lambda 127\), class. 14 lines. (A)

Ti XII
I.P. 290.08
Edlén, Zeits. f. Physik 100, 621 (1936). \(\lambda 61—\lambda 117\), class. 16 lines. (A)

Ti XIII
I.P. 784.6
Edlén and Tyrén, Zeits. f. Physik 101, 206 (1936), \(\lambda 23—\lambda 27\), class. 5 lines. (B)

Vanadium 23.

V Main
Meggers and Moore, Not published. 1939. \(\lambda 1248—\lambda 2000\), about 600 lines V, II, III, IV.

V I
I.P. 6.72 (M)
Moore, Phys. Rev. 55, 710 (1939). \(\lambda 1873—\lambda 2000\), 43 lines, class. 30. (C)

V II
I.P. 14.1 (Estimate)
Meggers, J. Research Nat. Bur. Stand. 25, 83 (1940). \(\lambda 1313—\lambda 2000\), 185 lines, class. 171. (A).

V III
I.P. 29.6
White, Phys. Rev. 33, 672 (1929). \(\lambda 1118—\lambda 1829\), class. 74 lines. (C)

V IV
I.P. 48.3
Ekefors, Zeits. f. Physik 71, 53 (1931). \(\lambda 675—\lambda 680\), 4 lines (B)
* White, Phys. Rev. 33*, 538 (1929), \(\lambda 675—\lambda 1999\), class. 57 lines. (C)

V V
I.P. 64.891 (KW²)
Ekefors, Zeits. f. Physik 71, 53 (1931). \(\lambda 286—\lambda 485\), 5 lines. (B)
Gibbs and White, Phys. Rev. 33, 157 (1929). \(\lambda 287—\lambda 1717\), class. 11 lines. (C)

V VI
I.P. 128.4
Kruger and Weissberg, Phys. Rev. 48, 659 (1935). \(\lambda 128—\lambda 183\), class. 4 lines. (B)

V VII
I.P. 150
Edlén, Zeits. f. Physik 104, 407 (1937). \(\lambda 148—\lambda 165\), class. 11 lines. (B)

V VIII
I.P. 172.8
Edlén, Zeits. f. Physik 104, 188 (1937). \(\lambda 135—\lambda 148\), class. 19 lines. (B)

V IX
Kruger and Pattin, Phys. Rev. 52, 621 (1937). λ125—λ127, classification. 6 lines. (B)

V XI
Edlén, Zeits. f. Physik 103, 536 (1936). λ87, λ88, classification. 2 lines. (B)

V XII         I.P. (307)
Edlén, Zeits. f. Physik 103, 536 (1936). λ61—λ107, classification. 15 lines. (A)

V XIII         I.P. 334.69
Edlén, Zeits. f. Physik 100, 621 (1936). λ53—λ100, classification. 15 lines. (A)

V XIV         I.P. 892.8
Edlén and Tyrén, Zeits. f. Physik 101, 206 (1936). λ21—λ24, classification. 4 lines. (B)

Chromium 24.

Cr Main
Bloch and Bloch, J. de phys. et rad. (6) 6, 105 (1925). λ1506—λ1926, 182 lines. (C)

Cr I (X)        I.P. 6.74 (BG)
Cr II         I.P. 16.6 (BG)
Kiess, unpublished. 1939. λ1971—λ2000, 19 lines. (B)

Cr III         I.P. 31 (B)
Bowen, Phys. Rev. 52, 1153 (1937). λ921—λ2000, classification. 89 lines. (A)
White, Phys. Rev. 33, 914 (1929). λ1196—λ1736, classification. 35 lines. (C)

Cr IV         I.P. 50.4
White, Phys. Rev. 33, 672 (1929). λ617—λ1990, classification. 65 lines. (C)
Bowen, Phys. Rev. 52, 1153 (1937). λ574—λ1968, classification. 62 lines. (C)

Cr V         I.P. 72.8
White, Phys. Rev. 33, 535 (1929). λ433—λ1820, classification. 55 lines. (C)

Cr VI         I.P. 90.17
Kruger and Weissberg, Phys. Rev. 52, 314 (1937). λ209, λ211, classification. 2 lines. (B)
Gibbs and White, Phys. Rev. 33, 157 (1929). λ335—λ338, classification. 3 lines. (C)

Cr VII         I.P. 160.4
Kruger and Weissberg, Phys. Rev. 48, 659 (1935). λ104—λ149, classification. 4 lines. (B)

Cr. VIII        I.P. 184
Edlén, Zeits. f. Physik 104, 407 (1937). λ124—λ136, classification. 10 lines. (B)

Cr IX         I.P. 208.6
Edlén, Zeits. f. Physik 104, 188 (1937). λ117—λ124, classification. 11 lines. (B)

Cr XII
Edlén, Zeits. f. Physik 103, 536 (1936). λ76, classification. 2 lines. (B)

Cr XIII        I.P. (353)
Edlén, Zeits. f. Physik 103, 536 (1936). λ54—λ92, classification. 14 lines. (A)

Cr XIV        I.P. 382.37
Edlén, Zeits. f. Physik 100, 621 (1935). λ46—λ86, classification. 16 lines. (A)

Cr XV I.P. 1008.1
Edlén and Tyrén, Zeits. f. Physik 101, 206 (1936). λ18—λ21, class. 4 lines. (B)
Tyrén, Zeits. f. Physik 111, 314 (1938). λ15—λ21, class. 9 lines. (A)

Manganese 25.

Mn Main
Bloch and Bloch, J. de phys. et rad. (6) 6, 154 (1925). λ1465—λ1869, 178 lines. (C)

Mn I (X) I.P. 7.41 (BG)
Mn II I.P. 15.56
Curtis, Phys. Rev. 53, 474 (1938). λ953—λ1960, class. 432 lines. (A)

Mn III I.P. 34.4
Kruger and Gilroy, Phys. Rev. 48, 720 (1935). λ892—λ895, class. 3 lines, corrections to Gilroy (1931). (B)
Gilroy, Phys. Rev. 38*, 2217 (1931). λ915—λ1998, class. 64 lines. (C)

Mn IV I.P. 52 (B)
Bowen, Phys. Rev. 52, 1153 (1937). λ540—λ1973, class. 156 lines. (A)
White, Phys. Rev. 33, 914 (1929). λ1742—λ1790, class. 12 lines. (C)

Mn V I.P. 75.7
Bowen, Phys. Rev. 47, 924 (1935). λ382—λ458, class. 74 lines. (B)
White, Phys. Rev. 33*, 672 (1929). λ405—λ1621, class. 49 lines. (C)

Mn VI
Cady, Phys. Rev. 43, 322 (1933). λ307—λ330, class. 30 lines. (B)

Mn VII I.P. 118.677
Kruger and Weissberg, Phys. Rev. 52, 314 (1937). λ111—λ468, class. 21 lines. (A)

Mn VIII I.P. 195.5
Kruger, Weissberg and Phillips, Phys. Rev. 51, 1090 (1937). λ122, λ125, class. 2 lines. (B)
Kruger and Weissberg, Phys. Rev. 48, 659 (1935). λ122, λ125, class. 2 lines. Approximate wavelengths.

Mn IX I.P. 221
Edlén, Zeits. f. Physik 104, 407 (1937). λ105—λ115, class. 9 lines. (B)

Mn X I.P. 247.2
Edlén, Zeits. f. Physik 104, 188 (1937). λ100—λ105, class. 11 lines. (B)

Mn XIII
Edlén, Zeits. f. Physik 103, 536 (1936). λ67, class. 2 lines. (B)

Mn XIV I.P. (402)
Edlén, Zeits. f. Physik 103, 536 (1937). λ57—λ80, class. 13 lines. (A)

Mn XV I.P. 433.14
Edlén, Zeits. f. Physik 100, 621 (1936). λ45—λ75, class. 12 lines. (A)

Mn XVI I.P. 1130.5
Tyrén, Zeits. f. Physik 111, 314 (1938). λ13—λ19, class. 9 lines. (A)

Absorption

Paul, Phys. Rev. 52, 923 (1937). λ1085—λ1923, 57 lines.

Iron 26.

Fe Basic

Bloch and Bloch, Comptes Rendus 197, 679 (1933). λ 365—λ 1149, 171 lines. (B)
Bloch and Bloch, J. de phys. et rad. (6) 6, 105 (1925). λ 1505—λ 1896, 286 lines. (C)
Bloch and Bloch, Ann. de physique (10) 6, 409 (1926). λ 1877—λ 2000, 279 lines, indicated as weak. (C)

Fe I  I.P. 7.83 (BG)
Green, Phys. Rev. 55, 1209 (1939). λ 1934—λ 1974, 29 lines. (B)
Bloch and Bloch, Ann. de physique (10) 6, 409 (1926). λ 1934—λ 1999, 44 lines. (C)

Fe II  I.P. 16.16 (M)
Green, Phys. Rev. 55, 1209 (1939), λ 897—λ 2000, class. 273 lines. (B)
Bloch and Bloch, Ann. de physique (10) 6, 409 (1926). λ 1877—λ 1999, 22 lines. (C)
Dobbie, Proc. Roy. Soc. A151, 703 (1935), Annals of the Solar Physics Observatory (Cambridge), Vol. 5, Part I (1938). Terms.
It is known that Edlén has significant unpublished material on Fe II.

Fe III  I.P. 30.48
Bowen, Phys. Rev. 52, 1153 (1937). λ 1123—λ 1926, class. 12 lines. (B)
Green, Phys. Rev. 55, 1209 (1939). λ 860—λ 2000, class. 22 lines. (B)
Green, unpublished 1939. λ 1550—λ 1997, 202 lines. (B)
Bloch and Bloch, Ann. de physique (10) 6, 409 (1926). 1877—2000. 213 lines. (C)
* Edlén, unpublished 1939; see Astrophys. J. 90*, 378 (1939). λ 727—λ 2000, class. 500 lines. (A)

Fe IV  I.P. 56.8
Kruger and Gilroy, Phys. Rev. 48, 720 (1935). λ 526—λ 527, class. 3 lines. Corrections by Gilroy. (1931). (B)
* Gilroy, Phys. Rev. 38*, 2217 (1931). λ 575—λ 1826, class. 70 lines. (C)

Fe V.
Bowen, Phys. Rev. 52, 1153 (1937). λ 364—λ 1554, class. 145 lines. (A)
White, Phys. Rev. 33, 914 (1929). λ 1431—λ 1469, class. 9 lines. (C)

Fe VI
Bowen, Phys. Rev. 47, 924 (1935). λ 276—λ 319, class. 101 lines. (A)

Fe VII
Cady, Phys. Rev. 43, 322 (1933). λ 231—λ 246, class. 33 lines. (B)
Bowen and Edlén, Nature 143, 374 (1939). Only terms; wavelengths not published. Reanalysis by Cady.

Fe VIII  I.P. 150.427
Kruger and Weissberg, Phys. Rev. 52, 314 (1937). λ 93—λ 371, class. 15 lines. (A)

Fe IX  I.P. 233.5
Kruger, Weissberg and Phillips, Phys. Rev. 51, 1090, class. 2 lines. (B)

Fe X  I.P. 261
Edlén, Zeit. f. Physik 104, 407 (1937). λ 94—λ 98, class. 8 lines. (B)

Fe XI                I.P. 288.9
Edlén, Zeits. f. Physik 104, 188 (1937). λ86—λ91, class. 12 lines. (B)

Fe XIV
Edlén, Zeits. f. Physik 103, 536 (1936). λ59—λ60, class. 2 lines. (B)

Fe XV                I.P. (454)
Edlén, Zeits. f. Physik 103, 536 (1936). λ50—λ71, class. 12 lines. (A)

Fe XVI               I.P. 487.01
Edlén, Zeits. f. Physik 100, 621 (1935). λ39—λ97, class. 13 lines. (A)

Fe XVII              I.P. 1259.7
Tyrén, Zeits. f. Physik 111, 314 (1938). λ12—λ17, class. 9 lines. (A)

Cobalt 27.

Co Principal
Bloch and Bloch, J. de phys. et rad. (6) 6, 105 (1925). λ1425—λ1882, 342 lines. (C)
Moore, unpublished 1939, λ1836—λ2000. Co I, II, III.

Co I                I.P. 7.84
Russell, King and Moore. Phys. Rev. 58, 407 (1940). λ1814—λ2000, 144 lines, class. 79. (C)

Co II               I.P. 16.9 (M)
Findlay, Phys. Rev. 36, 5 (1930). λ1941—λ1998, class. 6 lines. (B)

Co V                I.P. 83.1
Kruger and Gilroy, Phys. Rev. 48, 720 (1935). λ356, class. 3 lines, corrections to Gilroy (1931). (B)
Gilroy, Phys. Rev. 38, 2217 (1931). λ413—λ1489, class. 57 lines. (C)

Co VI
Bowen, Phys. Rev. 53, 889 (1933). λ266—λ307, class. 103 lines. (B)
Phillips and Kruger, Phys. Rev. 54, 839 (1933). Another classification is proposed.

Co VIII
Cady, Phys. Rev. 43, 322 (1933). λ181—λ193, class. 27 lines. See above under Fe VII (Bowen and Edlén). (B)

Co XI               I.P. 304
Edlén, Zeits. f. Physik 104, 407 (1937). λ81—λ85, class. 6 lines. (B)

Co XV
Edlén, Zeits. f. Physik 103, 536 (1936). λ53, class. 2 lines. (B)

Co XVI              I.P. (510)
Edlén, Zeits. f. Physik 103, 536 (1936). λ47—λ62, class. 9 lines. (A)

Co XVII             I.P. 544.08
Edlén, Zeits. f. Physik 100, 621 (1935). λ41—λ59, class. 10 lines. (A)

Co XVIII            I.P. 1396
Tyrén, Zeits. f. Physik 111, 314 (1938). λ12—λ15, class. 7 lines. (A)

Nickel 28.

Ni Principal
Bloch and Bloch, J. de phys. et rad. (6) 6, 105 (1925). \(\lambda 1370—\lambda 1859\), 330 spark lines. (C)

Ni I
I.P. 7.606
Russell, Phys. Rev. 34, 821 (1929). \(\lambda 1964—\lambda 2000\), class. 7 lines. (C)

Ni II
I.P. 18.4 (M)
Shenstone, Phys. Rev. 30, 255 (1927). \(\lambda 1812—\lambda 1996\), class. 8 lines. (C)
Lang, Phys. Rev. 31, 773 (1928). \(\lambda 1703—\lambda 1952\), class. 13 lines. (C)
Lang, Phys. Rev. 33, 547 (1929). \(\lambda 1253—\lambda 1537\), class. 26 lines. (C)
Menzies, Proc. Roy. Soc. A122, 134 (1929). \(\lambda 1164—\lambda 1965\), class. 111 lines. (C)
Yamanouchi, Proc. Phys. Math. Soc. Japan 20, 242 (1938). Terms (the journal was not available).

Ni VI
Gilroy, Phys. Rev. 38, 2217 (1931), \(\lambda 844—\lambda 1192\), class. 25 lines. (C)
Kruger and Gilroy, Phys. Rev. 48, 720 (1935). \(\lambda 260—\lambda 261\), class. 3 lines. (B)

Ni VII
Phillips and Kruger, Phys. Rev. 54, 839 (1938). \(\lambda 205—\lambda 229\), class. 92 lines. (B)

Ni IX
Cady, Phys. Rev. 43, 322 (1933). \(\lambda 146—\lambda 155\), class. 20 lines. See the verification in Fe VII (Bowen and Edlén). (B)

Ni XVIII
I.P. 604.1
Edlén, Zeits. f. Physik 100, 621 (1936). \(\lambda 44—\lambda 53\), class. 4 lines. (B)

Copper 29.

Cu Principal.
Bloch, Bloch and Farineau, J. de phys. et rad. (7) 3, 437 (1932). \(\lambda 386—\lambda 1377\), 427 lines. (B)
Bloch and Bloch, Comptes Rendus 197, 132 (1933). \(\lambda 240—\lambda 383\), 86 lines. (B)
Kruger, Phys. Rev. 44, 826 (1933). \(\lambda 109—\lambda 476\), 440 lines, possibly Cu VI—IX. (B)
Bloch and Bloch, J. de phys. et rad. (6) 6, 154 (1925). \(\lambda 1359—\lambda 1931\), 177 lines. (C)
Selwyn, Proc. Phys. Soc. 41, 392 (1929). \(\lambda 1685—\lambda 1825\) 13 lines. (C)

Cu I
I.P. 7.68 (BG)
Shenstone, Phys. Rev. 34, 1623 (1929). Classification of Selwyn’s lines. (C)

Cu II
I.P. 20.18
Shenstone, Phil. Trans. Roy. Soc. A751, 195 (1936). \(\lambda 676—\lambda 2000\), class. 476 lines. (A)
McLennan and Quinlan, Phil. Mag. (7) 14, 823 (1932). \(\lambda 1979—\lambda 1990\), 2 lines. (C)

Cu III
Bloch and Bloch, Comptes Rendus 200, 2017 (1935). Classification.

Cu VII
Kruger and Gilroy, Phys. Rev. 48, 720 (1935). λ200—λ201, class. 3 lines. (B)

Cu XIX
Edlén, Zeits. f. Physik 100, 621 (1936), λ47, class. 2 lines. (B)

Zinc 30.

Zn Principal
Bloch and Bloch, Comptes Rendus 201, 137 (1935). λ425—λ479, 24 lines, included in Ann. de physique (1936). (B)
Bloch and Bloch, Ann. de physique (11) 5, 325 (1936). λ227—λ1981, 490 lines, 155 indicated as weak. (B and C)
Selwyn, Proc. Phys. Soc. 41, 392 (1929). λ1671—λ1972, 11 lines. (C)

Zn I
I.P. 9.36 (BG)
Bloch and Bloch, Ann. de physique (11) 5, 325 (1936). λ1590, 1 line. (C)
Hetzler, Boremann and Burns, Phys. Rev. 48, 656 (1935), 2 calculated lines, λ1403, λ1458.

Zn II
I.P. 17.89 (BG)
Lang, Proc. Nat. Acad. Sci. 15, 414 (1929). λ984—λ1930, class. 11 lines. (C)
Takahashi, Ann. d. Physik (5) 3, 27 (1929). λ834—λ1969, 46 lines, class. 30. (C)
Bloch and Bloch, Ann. de physique (11) 5, 325 (1936). λ1361—λ1928, 4 lines. (C)

Zn III
I.P. 39.5 (KS)
Mazumder, Ind. J. Phys. 10, 171 (1936). λ498—λ1975, class. 226 lines. (C)
Bloch and Bloch, Ann. de physique (11) 5, 325 (1936). λ678—λ1981, 54 lines. (B and C).

Zn IV
Bloch and Bloch, Ann. de physique (11) 5, 325 (1936). λ467—λ1900, 107 lines, class. 32. (B and C)

Absorption

Beutler and Guggenheimer, Zeits. f. Physik 87, 176 (1933). λ714—λ1109, class. 24 lines.

Gallium 31.

Ga Principal
Lang, Phys. Rev. 30, 762 (1927). λ1455—λ1538, 12 lines. (C)

Ga I (L) I.P. 5.97 (BG)
Ga II
I.P. 20.43
Sawyer and Lang, Phys. Rev. 34, 712 (1929). λ829—λ1846, class. 38 lines. (C)

Ga III
I.P. 30.6 (BG)
Lang, Phys. Rev. 30, 762 (1927). λ632—λ1535, class. 9 lines. (C)

Ga IV
I.P. 63.8 (KS)
Mack, Laporte and Lang, Phys. Rev. 31, 748 (1928). λ422—λ1466, class. 39 lines. (C)

Germanium 32.

Ge I
I.P. 7.89 (M)
Gartlein, Phys. Rev. 31, 782 (1928). λ 1874—λ 2000, 27 lines, class. 21. (C)
K. R. Rao, Proc. Roy. Soc. A124, 465 (1929). λ 1639—λ 1999, 77 lines, class. 64. (C)

Ge II
I.P. 16 (M)
Lang, Phys. Rev. 34, 697 (1929). λ 999—λ 1649, class. 18 lines. (C)

Ge III
I.P. 34.07
Lang, Phys. Rev. 34, 697 (1929). λ 543—λ 1978, class. 48 lines. (C)

Ge IV
I.P. 45.5 (BG)
Lang, Phys. Rev. 34, 697 (1929). λ 440—λ 1648, class. 19 lines. (C)

Ge V
I.P. 93.0
Kruger and Shoupp, Phys. Rev. 46, 124 (1934). λ 295—λ 305, class. 3 lines. (B)
Mack, Laporte and Lang, Phys. Rev. 31, 748 (1928). λ 943—λ 1222, class. 33 lines. (C)

Arsenic 33.

As Principal
Queney, J. de phys. et rad. (6) 10, 448 (1929). λ 711—λ 2000, 238 lines, class. 16. As IV, V. (C)

As I
I.P. 10.5
A. S. Rao, Proc. Phys. Soc. 44, 594 (1932). λ 1313—λ 1945, 161 lines, class. 126. (C)

As II
I.P. 20.1
A. S. Rao, Proc. Phys. Soc. 44, 343 (1932). λ 932—λ 1769, 46 lines, included in Ind. J. Phys. (1933). (C)
A. S. Rao, Ind. J. Phys. 7, 561 (1933). λ 803—λ 1769, class. 68 lines. (C)

As III
I.P. 28.0 (BG)
K. R. Rao, Proc. Phys. Soc. 43, 68 (1931). λ 604—λ 1274, class. 19 lines. (B)
Pattabhiramiah and A. S. Rao, Ind. J. Phys. 3, 437 (1929). λ 866—λ 1750, class. 12 lines. (C)

As IV
I.P. 49.9
K. R. Rao, Proc. Roy. Soc. A134, 604 (1932). λ 530—λ 1481, class. 45 lines. (B and C).
Queney, J. de phys. et rad. (6) 10, 448 (1929). λ 742—λ 981, class. 12 lines. (C)

As V
I.P. 62.5 (BG)
Queney, J. de phys. et rad. (6) 10, 448 (1929). λ 715—λ 1030, class. 4 lines. (C)
Sawyer and Humphreys, Phys. Rev. 32, 583 (1928). λ 601—λ 1057, class. 9 lines. (C)

As VI
I.P. 126.9
Kruger and Shoupp, Phys. Rev. 46, 124 (1934). λ 219—λ 233, class. 4 lines. (B)
Pattabhiramiah and A. S. Rao, Zeits. f. Physik 53, 587 (1929). λ 812—λ 1016, class. 29 lines. (C)

Selenium 34.

Se Main

K. R. Rao and Murti, Proc. Roy. Soc. A145, 694 (1934). λ 561—λ 861, 44 lines. (B)

Goudet, J. de phys. et rad. (7) 6, 433 (1935). λ 332—λ 1294. 500 lines, 238 indicated as weak. (B)

Lacroute, J. de phys. et rad. (6) 9, 180 (1928). λ 1234—λ 1994, 68 lines. (C)

Se I                                                                     I.P. 9.70*

Ruedy and Gibbs, Phys. Rev. 46, 880 (1934). λ 1314—λ 1995, 135 lines, class. 108. (A)

K. R. Rao and Murti, Proc. Roy. Soc. A145, 694 (1934). λ 1413—λ 1995, 47 lines, class. 27. (B)

* Gibbs and Ruedy, Phys. Rev. 40, 204 (1932). λ 1607—λ 1961, 10 lines with approximate wavelengths, included in the work of Ruedy and Gibbs, Phys. Rev. 1934.

Se II                                                                    I.P. 21.3, 21.6*

Martin, Phys. Rev. 48, 938 (1935) λ 695—λ 1657, class., 127 lines. (A)

* Krishnamurty and K. R. Rao, Proc. Roy. Soc. A149, 56 (1935). λ 746—λ 1657, class. 50 lines. (B)

Goudet, J. de phys. et rad. (7) 6, 433 (1935). λ 746—λ 1294, 38 lines. (B)

Bloch and Bloch, J. de phys. et rad. (7) 6, 441 (1935). Classification.

Se III                                                                   I.P. 33.93*

Badami and K. R. Rao, Proc. Roy. Soc. A140, 387 (1933). λ 685—λ 1126, class. 56 lines. (B)

* K. R. Rao and Murti, Proc. Roy. Soc. A145, 681 (1934). λ 518—λ 1993, 88 lines, class. 30. (B)

Goudet, J. de phys. et rad. (7) 6, 433 (1935). λ 518—λ 1207, 113 lines. (B)

Bloch and Bloch, J. de phys. et rad. (7) 6, 441 (1935). Classification.

Se IV                                                                    I.P. 47.72

K. R. Rao and Badami, Proc. Roy. Soc. A131, 154 (1931). λ 636—λ 1967, class. 24 lines. (B)

Goudet, J. de phys. et rad. (7) 6, 433 (1935). λ 636—λ 1167, 18 lines. (B)

Se V                                                                     I.P. 72.8

K. R. Rao and Badami, Proc. Roy. Soc. A131, 154 (1931). λ 506—λ 1150, class. 22 lines. (B and C)

K. R. Rao and Murti, Proc. Roy. Soc. A145, 694 (1934). λ 675, class. 1 line. (B)

Goudet, J. de phys. et rad. (7) 6, 433 (1935). λ 505—λ 1228, 20 lines. (B),

Se VI                                                                    I.P. 81.4*

Goudet, J. de phys. et rad. (7) 6, 433 (1935). λ 453—λ 887, 7 lines. (B)

* Sawyer and Humphreys, Phys. Rev. 32, 583 (1928). λ 453—λ 887, class. 7 lines. (B)

Se VII                                                                   I.P. 165.5*

K. R. Rao and Murti, Proc. Roy. Soc. A145, 694 (1934). λ 760—λ 819, class. 4 lines. (B)

* Kruger and Shoupp, Phys. Rev. 46, 124 (1934). λ 171—λ 181, class. 4 lines. (B)

Goudet, J. de phys. et rad. (7) 6, 433 (1935). λ 561—λ 861, 42 lines. (B)

Bromine 35.

Br Main

Vaudet, Comptes Rendus 185, 1270 (1927). λ 1302—λ 1943, 101 lines. (C),

Lacroute, Ann. de physique (II) 3, 5 (1935). λ 646—λ 1994, 135 lines, 122 indicated as weak. (C)

Br I  I.P. 11.80 (BG), 12.2
Lacroute, Ann. de physique (II) 3, 5 (1935). λ 1385—λ 1730, 23 lines. (C)
* de Bruin and Kiess, Science 69*, 360 (1929). Classification.

Br II  I.P. 19.1 (B)
Lacroute, Ann. de physique (II) 3, 5 (1935). λ 851—λ 1981, 50 lines. (C)
Bloch, Bloch and Lacroute, Comptes Rendus 199, 41 (1934). Terms.

Br III  I.P. 35.7
K. R. Rao and Krishnamurti, Proc. Roy. Soc. A161, 38 (1937). λ 666—λ 818, class. 15 lines. (B)
Lacroute, Ann. de physique (II) 3, 5 (1935). λ 649—λ 1994, 49 lines. (C)

Br IV  I.P. 50
A. S. Rao and Krishnamurti, Proc. Phys. Soc. 46, 531 (1934). λ 538—λ 736, class. 39 lines. (B)

Br V
A. S. Rao and K. R. Rao, Proc. Phys. Soc. 46, 163 (1934). λ 482—λ 856, class. 12 lines. (B)

Br VI
A. S. Rao and K. R. Rao, Proc. Phys. Soc. 46, 163 (1934). λ 499—λ 940, class. 14 lines. (B)

Br VII
A. S. Rao and K. R. Rao, Proc. Phys. Soc. 46, 163 (1934). λ 502—λ 780, class. 35 lines. (B)

Br VIII  I.P. 208.8
Kruger and Shoupp, Phys. Rev. 46, 124 (1934). λ 138, λ 140, class. 2 lines. (B)

Krypton 36.

Kr I  I.P. 13.94
Boyce, Phys. Rev. 47, 718 (1935). λ 945—λ 1235, class. 10 lines. (A)

Kr II  I.P. 24.47
Boyce, Phys. Rev. 47, 718 (1935). λ 554—λ 964, class. 82 lines. (A)

Kr III  I.P. 36.8
Boyce, Phys. Rev. 47, 718 (1935). λ 516—λ 1923, class. 138 lines. (A)

Kr IV
Boyce, Phys. Rev. 47, 718 (1935). λ 805—λ 842, class. 3 lines. (B)

Absorption

Beutler, Zeits. f. Physik 83, 177 (1935). λ 850—λ 1003, class. 29 lines.

Rubidium 37.

Rb Fundamental
Ricard and Valiancogne, Comptes Rendus 207, 1093 (1938). λ 1046—λ 2000, 92 lines. (C)

Rb I (L)
I.P. 4.16 (BG)

Rb II
I.P. 27.3

Laporte, Miller and Sawyer, Phys. Rev. 38, 843 (1931). λ 697—λ 741, class. 3 lines. (C)

Rb III
Tomboulian, Phys. Rev. 54, 350 (1938). λ 482—λ 815, class. 30 lines. (B)

Absorption

Beutler, Zeits. f. Physik 91, 131 (1934). λ 595—λ 810, class. 39 lines.

Strontium 38.

Sr I (L)
I.P. 5.67 (BG)

Sr II
I.P. 10.98 (BG)

Saunders, Schneider and Buckingham, Proc. Nat. Acad. Sci. 30, 291 (1934). λ 1483—λ 1996, class. 18 lines. (A)

Sr IV
Tomboulian, Phys. Rev. 54, 350 (1938). λ 358—λ 710, class. 29 lines. (B)

Yttrium 39.

Y Ground
McLennan and Liggett, Trans. Roy. Soc. Canada 20, 372 (1926). 1788, 1 line. (C)

Y I (X)
I.P. 6.5 (BG)

Y II (X)
I.P. 12.3 (BG)

Russell and Meggers, Bur. Stand. J. Research 2, 733 (1929). Terms.

Y III
I.P. 20.4 (BG)

Bowen and Millikan, Phys. Rev. 28, 923 (1026). λ 989, λ 996 — class. 2 lines. (C)
Russell and Meggers, Bur. Stand. J. Research 2, 733 (1929). Terms.

Y V
I.P. 76.5

Paul and Rense, Phys. Rev. 56, 1110 (1939). λ 314—λ 629, class. 41 lines. (B)

Zirconium 40.

Zr I (X)
I.P. 6.92 (BG)

Zr II
I.P. 13.97

Kiess and Kiess, Bur. Stand. J. Research 5, 1205 (1930). λ 1744—λ 1999, class. 21 lines. (A).

Zr III
I.P. 24.00

Kiess and Lang, Bur. Stand. J. Research 5, 305 (1930). λ 756—λ 1990, class. 61 lines. (B)

Zr IV
I.P. 33.83

Kiess and Lang, Bur. Stand. J. Research 5, 305 (1930). λ 629—λ 1608, class. 12 lines. (B)

Zr VI
I.P. 98.4

Paul and Rense, Phys. Rev. 56, 1110 (1939). λ 237—λ 568, class. 46 lines. (B)

Columbium 41.

Cb Fundamental
McLennan and Liggett, Trans. Roy. Soc. Canada 20, 377 (1926). λ 1590—λ 1983, 71 lines. (C)

Cb I (O)
Cb III          I. P. 24.2 (B)
Eliason, Phys. Rev. 43, 745 (1933). λ 1423—λ 1600, class. 26 lines. (B)
Cb IV          I.P. 38.1
Lang, Zeeman Verh. (The Hague, 1935), p. 44. λ 542—λ 1978, class. 95 lines. (A)

Cb V          I. P. 49.3 (B)
Trawick, Phys. Rev. 46, 63 (1934). λ 465—λ 1877, class. 12 lines. (B)

Molybdenum 42.

MO I (X)        I. P. 7.35 (BG)
MO IV
Eliason, Phys. Rev. 43, 745 (1933). λ 856—λ 1995, class. 41 lines. (B)

MO V
Trawick, Phys. Rev. 48, 223 (1935). λ 410—λ 1849, class. 91 lines. (A)

MO VI
Trawick, Phys. Rev. 46, 63 (1934). λ 373—λ 1576, class. 12 lines. (B)

Ruthenium 44 (O).

Ru I (X)        I. P. 7.7 (M)

Rhodium 45 (O).

Rh I (X)        I. P. 7.7 (BG)

Palladium 46.

Pd I          I. P. 8.3 (BG)
Shenstone, Phys. Rev. 36, 669 (1930). λ 1946—λ 1993, 9 lines, class. 7. (C)

Pd II          I. P. 19.8
Shenstone, Phys. Rev. 32, 30 (1928). λ 1212—λ 2000, class. 53 lines. (C)

Silver 47.

Ag Fundamental
Bloch, Bloch and Farineau, J. de phys. et rad. (7) 3, 437 (1932). λ 260—λ 1321, 727 lines. (B and C)
Bloch and Bloch, J. de phys. et rad. (6) 6, 157 (1925). λ 1389—λ 1890, 298 lines. (C)
Selwyn, Proc. Phys. Soc. 41, 392 (1929). λ 1709—λ 1995, 3 lines. (C)

Ag I          I. P. 7.54 (BG)
Shenstone, Phys. Rev. 57, 894 (1940). λ 1507—λ 1850, class. 14 lines. (A)

Ag II          I. P. 21.9 (BG), 21.4
Shenstone, Phys. Rev. 31, 317 (1928). λ 933—λ 1994, class. 2 lines. (C)
Menzies, Proc. Roy. Soc. A122, 134 (1929). λ 1107—λ 1196, class. 3 lines. (C)
* Gilbert, Phys. Rev. 47*, 847 (1935). λ 729—λ 1999, class. 111 lines. (C)

Ag III                 I. P. 39.5
Gilbert, Phys. Rev. 48, 338 (1935). λ 710—λ 2000, class. 193 lines. (C)

Absorption

Paul, Phys. Rev. 52, 923 (1937). λ 1032—λ 1893, 21 lines.

Cadmium 48.

Cd Principal
Bloch and Bloch, Comptes Rendus 201, 137 (1935). λ 493—λ 547, 20 lines, included in Ann. de physique. 1936. (B)
Bloch and Bloch, Ann. d. physique (11) 5, 325 (1936). λ 212—λ 1996, 906 lines, 125 indicated as weak. (B and C)
Selwyn, Proc. Phys. Soc. 41, 392 (1929). λ 1669—λ 1820, 3 lines. (C)

Cd I (Q)             I. P. 8.96 (BG)
Cd II               I. P. 16.84
Takahashi, Ann. d. Physik (5) 3, 27 (1929). λ 711—λ 1995, 96 lines, class. 51. (C)

Cd III               I. P. 38.0 (KS)
Bloch and Bloch, Ann. de physique (11) 5, 325 (1936). λ 677—λ 1944, 33 lines. (B and C)
Gibbs and White, Phys. Rev. 31, 776 (1928). λ 677—λ 1943, class. 35 lines. (C)
McLennan, McLay and Crawford, Trans. Roy. Soc. Canada 22, 45 (1928). λ 1470—λ 1943, class. 31 lines. (C)

Cd IV
Bloch and Bloch, Ann. de physique (11) 5, 325 (1936). λ 495—λ 1930, 124 lines, class. 32. (B and C)

Absorption

Beutler, Zeits. f. Physik 87, 19 (1933). λ 683—λ 1023, 27 lines.

Indium 49.

In I (L)             I. P. 5.76 (BG)

In II               I. P. 18.79
Lang and Sawyer, Zeits. f. Physik 71, 453 (1931). λ 681—λ 1977, class. 76 lines. (A)

In III              I. P. 27.9 (BG)
Lang, Proc. Nat. Acad. Sci. 13, 341 (1927). λ 686—λ 1749, class. 11 lines. (C)
Lang, Proc. Nat. Acad. Sci. 15, 414 (1929). λ 882—λ 1744, class. 5 lines. (C)

In IV.              I. P. 57.8 (KS)
Gibbs and White, Phys. Rev. 31*, 776 (1928). λ 472—λ 1726, class. 36 lines. (C);

Tin 50.

Sn I               I. P. 7.297
Meggers, J. Research Nat. Bur. Stand. 24, 153 (1940). λ 1697—λ 2000, 80 lines, class. 56. (A)

Sn II
I. P. 14.56
McCaugwik and Sawyer, Phys. Rev. 54, 71 (1938). λ888—λ1900, classified.
32 lines. (A)

Sn III
I. P. 30.5 (BG), 30
Green and Loring, Phys. Rev. 30, 675 (1927). λ1161—λ1818, classified.
9 lines. (C)
Gibbs and Vieweg, Phys. Rev. 34, 400 (1929). λ744—λ1992, classified. 35 lines. (C)

Sn IV
I. P. 39.4 (BG)
Lang, Proc. Nat. Acad. Sci. 13, 341 (1927). λ956—λ1438, classified. 9 lines. (C)
Lang, Proc. Nat. Acad. Sci. 15, 414 (1929). λ500—λ710, classified. 8 lines. (C)

Sn V
I. P. 80.7 (KS)
Gibbs and White, Proc. Nat. Acad. Sci. 14, 345 (1938). λ356—λ1535, classified.
34 lines. (C)

Antimony 51.

Sb Basic
Bloch and Bloch, J. de phys. et rad. (7) 8, 217 (1937). λ263—λ1997,
448 lines, 188 indicated as weak. (B and C)

Sb I
I. P. 8.35
Bloch and Bloch, J. de phys. et rad. (7) 8, 217 (1937). λ1699—λ1971,
6 lines. (C)

Sb II
I. P. 18
Lang and Vestine, Phys. Rev. 42, 233 (1932). λ691—λ1991, 99 lines,
classified. 60. (B)
Krishnamurty, Ind. J. Phys. 10, 83 (1936). λ876—λ1875, classified. 20 lines. (C)
Bloch and Bloch, J. de phys. et rad. (7) 8, 217 (1937). λ900—λ991,
87 lines. (C)

Sb III
I. P. 24.7
Bloch and Bloch, J. de phys. et rad. (7) 8, 217 (1937). λ691—λ1945,
37 lines. (B and C)
Pattabhiramiah and A. S. Rao, Ind. J. Phys. 3, 437 (1929). λ1814,
λ1840, classified. 2 lines. (C)
* Lang, Phys. Rev. 35*, 445 (1930). λ691—λ1947, classified. 35 lines. (C)

Sb IV
I. P. 44 (BG), 42
Bloch and Bloch, J. de phys. et rad. (7) 8, 217 (1937). λ517—λ1915,
45 lines. (B and C).
* Green and Lang, Proc. Nat. Acad. Sci. 14, 706 (1928). λ805—λ1514,
classified. 19 lines. (C)
Gibbs and Vieweg, Phys. Rev. 34, 400 (1929). λ456—λ1667, classified.
35 lines. (C)
Badami, Proc. Phys. Soc. 43*, 538 (1931). λ1358—λ1915, classified. 5 lines. (C)

Sb V
I. P. (BG)
Bloch and Bloch, J. de phys. et rad. (7) 8, 217 (1937). λ831—λ1906,
12 lines. (B and C)
Lang, Proc. Nat. Acad. Sci. 13, 341 (1927). λ746—λ1525, classified. 9 lines. (C)
Badami, Proc. Phys. Soc. 43, 538 (1931). Terms.

Sb VI
I. P. 107.1
Kruger and Shoupp, Phys. Rev. 46, 124 (1934). λ279—λ293, classified 4 lines. (B)
Bloch and Bloch, J. de phys. et rad. (7) 8, 217 (1937); Comptes Rendus 204, 424 (1937). λ280—λ1338, classified 30 lines. (B and C)
Schoepfle, Phys. Rev. 43, 742 (1933). λ883—λ1332, classified 31 lines. (C)

Tellurium 52.

Te Fundamental.
Bloch and Bloch, J. de phys. et rad. (7) 6, 441 (1935). λ157—λ1312, 521 lines, 87 indicated as weak. (B)
Lacroute, J. de phys. et rad. (6) 9, 180 (1928). λ1256—λ1998, 93 lines. (C)
Bloch and Bloch, Comptes Rendus 208, 336 (1939). λ110—λ332, 108 lines. (B)

Te I
I. P. 8.96 (M)
Krishnamurty, Ind. J. Phys. 10, 365 (1936). λ1555—λ1995, 24 lines, classified 22. (C)
Bartelt, Zeits. f. Physik 88, 522 (1934). Classification.

Te III
I. P. 30.5
Bloch and Bloch, J. de phys. et rad. (7) 6, 441 (1935). λ840—λ1145, 34 lines. (B)
* Krishnamurty, Proc. Roy. Soc. A151, 178 (1935). λ839—λ1145, classified 37 lines. (B)
Krishnamurty and K. R. Rao, Proc. Roy. Soc. A158*, 562 (1937). λ612—λ1805, classified 27 lines. (B)

Te IV
I. P. 37.7
K. R. Rao, Proc. Roy. Soc. A133, 220 (1931). λ749—λ1459, classified 17 lines. (B)
Bloch and Bloch, de phys. et rad. (7) 6, 441 (1935). λ749—λ1197, 15 lines. (B)

Te V.
I. P. 60.0 (BG)
Bloch and Bloch, J. de phys. et rad. (7) 6, 441 (1935) λ359—λ1282, 24 lines, classified 5. (B)
Gibbs and Vieweg, Phys. Rev. 34, 400 (1929). λ603—λ1550, classified 23 lines. (C)

Te VI
I. P. 72.0
K. R. Rao, Proc. Roy. Soc. A133, 220 (1931). λ540—λ1314, classified 10 lines. (B)
Bloch and Bloch, J. de phys. et rad. (7) 6, 441 (1935). λ242—λ1071, 13 lines, classified 6. (B)
Lang, Proc. Nat. Acad. Sci. 13, 341 (1927). λ954, λ1078, classified 2 lines. (C)

Te VII
I. P. 136.5
Kruger and Shoupp, Phys. Rev. 46, 124 (1934). λ227—λ224, classified 6 lines. (B)
Bloch and Bloch, J. de phys. et rad. (7) 6, 411 (1935). λ804—λ1123, 12 lines. (B)
Bloch and Bloch, J. de phys. et rad. (7) 8, 217 (1937); Comptes Rendus 204, 424 (1937). λ228—λ1189, classified 24 lines. (B)
Schoepfle, Phys. Rev. 43, 742 (1933). λ784—λ1124, classified 24 lines. (C)

Iodine 53.

J Main
Bloch, Bloch and Felici, J. de phys. et rad. (7) 8, 355 (1937). λ 190—λ 1010 — 438 lines, 290 indicated as weak. (B)
McLeod, Phys. Rev. 49, 804 (1936). λ 809—λ 1876, 223 lines. (B and C).
Lacroute, Ann. de physique (II) 3, 5 (1935). λ 766—λ 1999, 406 lines, 265 indicated as weak. (C)

J I  I. P. 10.2 (M)
Lacroute, Ann. de physique (II) 3, 5 (1935). λ 1493—λ 1876, 15 lines. (C)
Deb, Proc. Roy. Soc. A139, 380 (1933). Classification.

J II  I. P. 19.4
Bloch, Bloch and Felici, J. de phys. et rad. (7) 8, 355 (1937). λ 480—λ 995, 139 lines, class. 13. (B)
Kalia, Ind. J. Phys. 9, 179 (1934). λ 1275—λ 1982, 105 lines. (C)
Lacroute, Ann. de physique (II) 3, 5 (1935). λ 766—λ 1999, 148 lines, class. 42. (C)
Murakawa, Zeits. f. Physik 109, 162 (1938). Classification.

J III
Bloch, Bloch and Felici, J. de phys. et rad. (7) 8, 355 (1937). λ 436—λ 1004, 93 lines. (B)
Lacroute, Ann. de physique (II) 3, 5 (1935). λ 767—λ 1999, 102 lines. (C)

J IV and higher
Bloch, Bloch and Felici, J. de phys. et rad. (7) 8, 355 (1937). λ 483—λ 953, 45 lines. (B)

J VIII  I. P. 169.1
Kruger and Shoupp, Phys. Rev. 46, 124 (1934). λ 190—λ 202, class. 4 lines. (B)
Bloch, Bloch and Felici, J. de phys. et rad. (7) 8, 355 (1937). λ 190, λ 194 — class. 2 lines. (B)

Absorption

McLeod, Phys. Rev. 45, 802 (1934). λ 1420—λ 1830, 8 lines.

Xenon 54.

Xe I  I. P. 12.078
Boyce, Phys. Rev. 49, 730 (1936). λ 1192—λ 1469, class. 4 lines. (A)
Abbink and Dorgelo, Zeits. f. Physik 47, 221 (1928). λ 1027—λ 1469, 15 lines. (C)

Xe II  I. P. 21.1
Boyce, Phys. Rev. 49, 730 (1936). λ 740—λ 1244, class. 20 lines. (B). For checking and performing analyses see: Humphreys, J. Research Nat. Bur. Stand. 22, 19 (1939).
Déjardin, Ann. de physique (10) 13, 83 (1930). λ 1881—λ 1997, 26 lines. (C)

Xe III  I. P. 32.0
Boyce, Phys. Rev. 49, 730 (1936). λ 627—λ 1978, class. 128 lines. (A)

Xe IV
Déjardin, Ann. de physique (10) 13, 82 (1930). λ 1881—λ 1999, 49 lines. (C)

Absorption

Beutler, Zeits. f. Physik 93, 172 (1935). λ926—λ996, classified. 18 lines.

Cesium 55.

Cs I (L)  I. P. 3.87 (BG)
Cs II  I. P. 23.4
Laporte, Miller and Sawyer, Phys. Rev. 39, 458 (1932). λ612—λ927,
 8 lines, included in Olthoff a. Sawyer, Phys. Rev. 42, (C)
Olthoff and Sawyer, Phys. Rev. 42, 766 (1932). λ607—λ927, classified.
 9 lines. (C)
Ricard, Givord and George, Comptes Rendus 205, 1229 (1937). λ1179—
 λ1501, classified. 3 lines. (C)

Cs III  I. P. 35 (B)
Fitzgerald and Sawyer, Phys. Rev. 46, 576 (1934). λ530—λ878, classified.
 17 lines. (C)

Absorption

Beutler and Guggenheimer, Zeits. f. Physik 88, 25 (1934). λ640—
 λ1008, 128 lines.

Barium 56.

Ba I (L)  I. P. 5.19 (BG)
Ba II  I. P. 96 (BG)
Saunders, Schneider and Buckingham, Proc. Nat. Acad. Sci. 20,
 291 (1934). λ1398—λ1986, classified. 26 lines. (A)

Ba III  (O) I. P. 35.5 (B)
Ba IV
Fitzgerald and Sawyer, Phys. Rev. 46, 576 (1934). λ570—λ740, classified.
 13 lines. (C)

Lanthanum 57.

La Main
McLennan and Liggett, Trans. Roy. Soc. Canada 20, 377 (1926). λ1699—
 λ1858, 3 lines. (C)
La I (L)  I. P. 5.59 (RM)
La II (X)  I. P. 11.38 (RM)
La III  I. P. 19.1 (RM)
Lang, Can. J Research A13, 1 (1935). λ1082—λ1462, classified. 5 lines. (B)
Lang, Can. J. Research A14, 43 (1936). Corrections to Can. J. Research
 (1935).

Cerium 58.

Ce I (L)  I. P. 6.5 (estimate)
Ce II (X)  I. P. 12.3 (estimate)
Ce III  I. P. 20
Russell, King and Lang, Phys. Rev. 52, 456 (1937). λ1680—λ1987, classified.
 31 lines. (B)

Ce IV  I. P. 33.3
Lang, Can. J. Research A13, 1 (1935). λ900—λ1881, classified. 11 lines. (B)
Lang, Can. J. Research A14, 127 (1936). λ447—λ1937, classified. 22 lines. (B)

J. K. BOYS

Lang, Phys. Rev. 49, 552 (1936). λ742—λ1915, class 4 lines, included in Can. J. Research 14 (1935). (B)
Badami, Proc. Phys. Soc. 43, 53 (1931). λ1335—λ1950, class 2 lines. (A)

Praseodymium 59.

Pr Principal
McLennan and Liggett, Trans. Roy. Soc. Canada 20, 377 (1926). λ1533—λ1961, 3 lines. (C)

Pr I (L)
I. P. 5.7 (estimate)

Neodymium 60.

Nd Principal
McLennan and Liggett, Trans. Roy. Soc. Canada 20, 377 (1926). λ1626, 1 line. (C)

Nd I (L)
5.7 (estimate)

Samarium 62.

Sm I (L)
I. P. 5.64 (M*)

Europium 63.

Eu I (L)
I. P. 5.64 (B)

Eu II (X)
I. P. 11.4 (M)

Gadolinium 64.

Gd I (L)
I. P. 6.6 (estimate)

Terbium 65.

Tb I (L)
I. P. 6.7 (estimate)

Dysprosium 66.

Dy I (L)
I. P. 6.8 (estimate)

Holmium 67 (O)

Erbium 68 (O)

Thulium 69 (O)

Ytterbium 70.

Yb I (L)
I. P. 6.23 (M)

Yb II (X)
I. P. 12.05 (M)

Lutetium 71 (O)

Hafnium 72.

Hf I (O) Hf II
I. P. 14.8

Meggers and Scribner, Bur. Stand. J. Research 13, 625 (1934). λ1623—λ1993, class 20 lines. (B)

Tantalum 73 (O)

Tungsten 74.

W Fundamental.
Bloch and Bloch, J. de phys. et rad. 6, 105 (1925). λ1453—λ1878, 409 lines. (C)
Bayen, Comptes Rendus 180, 57 (195). λ1873—λ1999, 75 lines. (C)

W I (O)
W II 8.1 (M)
Laun, J. Research Nat. Bur. Stand. 21, 207 (1938). λ1961—λ2000, class. 4 lines. (B)

Rhenium 75.

Re I (O) I. P. 7.85 (M)

Osmium 76.

Os I (O) I. P. 8.7 (M)

Iridium 77 (O)

Platinum 78.

Pt Fundamental.
Bloch and Bloch, J. de phys. et rad. 6, 154 (1925). λ1330—λ1912, 461 lines. (C)

Pt I I. P. 8.9
Livingood, Phys. Rev. 34, 185 (1929). λ1929—λ1996, 11 lines, class. 4 (C)

Pt II I. P. 18.47
Shenstone, Phil. Trans. Roy. Soc. A237, 453 (1938). λ976—λ1998, class. 352 lines. (A)

Gold 79.

Au Fundamental.
Bloch and Bloch, J. de phys. et rad. 6, 154 (1925). λ1342—λ1930, 386 lines. (C)
McLennan and Liggett, Trans. Roy. Soc. Canada 20, 377 (1926). λ1553—λ1997, 100 lines. (C)
Selwyn, Proc. Phys. Soc. 41, 392 (1929). λ1647—λ1992, 36 lines, 22 indicated as weak. (C)
Bloch and Bloch and Farineau, J. de phys. et rad. (7) 3, 437 (1932). λ296—λ1342, 510 lines. (B and C).

Au I I. P. 9.20
Selwyn, Proc. Phys. Soc. 41, 392 (1929). λ1647—1952, class. 7 lines. (C)
* McLennan and McLay, Proc. Roy. Soc. A134*, 35 (1931). λ1624, λ1587, class. 2 lines, and classification of published materials. (C)

Au II I. P. 20.0 (B)
McLennan and McLay, Trans. Roy. Soc. Canada 22, 103 (1928). λ1398—λ1925, class. 31 lines. (C)
Selwyn, Proc. Phys. Soc. 41, 392 (1929). λ1674—λ1925, class. 15 lines. (C)
Sawyer and Thomson, Phys. Rev. 38, 2293 (1931). λ1167—λ1362, class. 3 lines. (C)
B. V. R. Rao, Proc. Roy. Soc. A142, 118 (1933). Classification.
Mack and Fromer, Phys. Rev. 48, 357 (1935). Classification.

Mercury 80.

Hg Principal

Bloch and Bloch, Ann. de physique (II) 6, 561 (1936). \(\lambda 223\)—\(\lambda 1628\), 1397 lines, 294 indicated as weak (B and C).
Déjardin, Ann. de physique (10) 8, 424 (1927). \(\lambda 1860\)—\(\lambda 2000\), 69 lines indicated as weak. (C)
Subbaraya, J. Mysore University 7, 100 (1934). Hg II (There was no journal available.)

Hg I

I. P. 10.38 (BG)
Beutler, Zeits. f. Physik 84, 289 (1933). \(\lambda 1204\)—\(\lambda 1435\), class. 17 lines in emission and absorption. (B)
Selwyn, Proc. Phys. Soc. 41, 392 (1929). \(\lambda 1775\)—\(\lambda 1849\), class. 3 lines. (C)

Hg II

I. P. 18.672
Paschen, Preuss. Akad. Wiss. Berlin Ber. 32, 536 (1928). \(\lambda 893\)—\(\lambda 1988\), 50 lines, class. 27. (B)
Bloch and Bloch, Ann. de physique (II) 6, 561 (1936). \(\lambda 893\)—\(\lambda 1728\), 80 lines. (B and C)
Déjardin, Ann. de physique (10) 8, 424 (1927). \(\lambda 1860\)—\(\lambda 2000\), 46 lines. (C)
Naude, Ann. d. Physik (5) 3, 1 (1929). \(\lambda 939\)—\(\lambda 1996\), 55 lines, class. 31. (C)
McLennan, McLay and Crawford, Proc. Roy. Soc. A134, 41 (1931). \(\lambda 1459\)—\(\lambda 1981\), class. 4 lines and classification of published materials. (C)
Venkatasachar and Subbaraya, Zeits. f. Physik 73, 413 (1932). Classification.
Subbaraya, Zeits. f. Physik 73, 541 (1932). Classification.

Hg III

I. P. 34.05
Mack and Fromer, Phys. Rev. 48, 357 (1935). \(\lambda 740\)—\(\lambda 1415\), class. 14 lines. (B)
* Johns, Can. J. Research A15, 193 (1937). \(\lambda 570\)—\(\lambda 1982\), class. 248 lines. (B)
Bloch and Bloch, Ann. de physique (II) 6, 561 (1936). \(\lambda 788\)—\(\lambda 1728\). 146 lines. (B and C)
Déjardin, Ann. de physique (10) 8, 424 (1927). \(\lambda 1894\)—\(\lambda 1995\), 10 lines. (C)
Ricard, J. de phys. et rad. (7) 7*, 315 (1936). \(\lambda 1231\)—\(\lambda 1383\), class. 5 lines. (C)

Hg IV

Bloch and Bloch, Ann. de physique (11) 6, 561 (1936). \(\lambda 997\)—\(\lambda 1728\), 68 lines. (B and C)
Déjardin, Ann. de physique (10) 8, 424 (1927). \(\lambda 1881\)—\(\lambda 1998\), 13 lines. (C)
Subbaraya, Proc. Ind. Acad. Sci. A1, 39 (1934). Classification.

Absorption

Beutler, Zeits. f. Physik 84, 289 (1933). \(\lambda 1204\)—\(\lambda 1435\), class. 17 lines.
Beutler, Zeits. f. Physik 86, 710 (1933). \(\lambda 745\)—\(\lambda 1301\), class. 40 lines.

Thallium 81.

Tl Principal

Mack, Phys. Rev. 34, 17 (1929). \(\lambda 869\)—\(\lambda 1070\), 86 lines. (C)
Bloch, Bloch and Walden, J. de phys. et rad. 10, 49 (1939). \(\lambda 200\)—\(\lambda 1400\). Many lines, predominantly Tl IV and higher. (B)

Tl I (L)
I. P. 6.07 (BG)

Tl II
I. P. 20.33*
Smith, Proc. Nat. Acad. Sci. 14, 951 (1928). λ1121—λ1909, class. 20 lines (C)
Ellis and Sawyer, Phys. Rev. 49, 145 (1936). λ639—λ1909, 91 lines, class. 90. (C)

Tl III
I. P. 29.7 (BG)
McLennan, McLay and Crawford, Proc. Roy. Soc. A125, 50 (1129). λ1231—λ1661, class. 7 lines. (C)
Pattabhiramaya and A. S. Rao, Ind. J. Phys. 5, 407 (1930). Classification.

Tl IV
I. P. 50.5 (B)
Mack, Phys. Rev. 34, 17 (1929). λ1028—λ1964, class. 13 lines. (C)
K. R. Rao, Proc. Phys. Soc. 41, 361 (1939). Classification.
Mack and Fromer. Phys. Rev. 48, 357 (1935). Classification.

Absorption

Beutler and Demeter, Zeits. f. Physik 91, 202 (1934). λ1490, λ1610, class. 2 lines.
Beutler and Demeter, Zeits. f. Physik 91, 218 (1934). λ651—λ891, class. 9 lines.

Lead 82.

Pb Principal.
Arvidson, Ann. d. Physik (5) 12, 787 (1932). λ251—λ1439, 1050 lines, 532 indicated as weak. (B)

Pb I
I. P. 7.38 (BG)
Gieseler and Grotrian, Zeits. f. Physik 39, 377 (1926). λ1644—λ1972, class. 3 lines. (C)

Pb II
I. P. 14.96*
Arvidson, Ann. d. Physik (5) 12, 787 (1932). λ1110—λ1434, 5 lines. (B)
* Earls and Sawyer, Phys. Rev. 47, 115 (1935). λ840—λ1922, class. 50 lines. (C)

Pb III
I. P. 31.9 (BG)
Arvidson, Ann. d. Physik (5) 12, 787 (1932). λ515—λ1439, 45 lines. (B)
Smith, Phys. Rev. 34, 393 (1929). λ709—λ1827, class. 32 lines. (C)
Smith, Phys. Rev. 36, 1 (1030) λ961—λ1712, class. 4 lines. (C)

Pb IV
I. P. 42.11,* 42.0**
Arvidson, Ann. d. Physik (5) 12, 787 (1932). λ433, λ1435, 84 lines. (B)
* Crawford, McLay and Crooker, Proc. Roy. Soc. A158, 455 (1937). λ477—λ1982, class. 12 lines, and class. published materials. (B)
** Schoepfle, Phys. Rev. 47, 232 (1935). λ1715—λ1959, class. 4 lines and class. published materials. (C)

Pb V
I. P. 69.40*
Arvidson, Ann. d. Physik (5) 12, 787 (1932). λ254—λ1418, 219 lines. (B)
Mack and Fromer, Phys. Rev. 48, 357 (1935). Classification.
* Schoepfle, Phys. Rev. 50, 533 (1936). Classification.

Pb VI and higher
Arvidson, Ann. d. Physik (5) 12, 787 (1932). λ252—λ1433, 179 lines. (B)

Absorption

Kremenevsky, C. R. Acad. Sci U.S.S.R. 3, 251 (1935). λ 1350—λ 2000.

Bismuth 83.

Bi Basic
Arvidson, Ann. d. Physik (5) 12, 787 (1932). λ 176—λ 1487, 928 lines. 336 indicated as weak. (B)

Bi I  I. P. 7.25 (M)
Toshniwal, Phil. Mag. (7) 4, 774 (1927). λ 1902—λ 1990, 10 lines, class. 3. (C)

Bi II  I. P. 16.6 (B)
Zumstein, Phys. Rev. 38, 2214 (1931). λ 1777—λ 1902, class. 5 lines. (B)
Arvidson, Ann. d. Physik (5) 12, 787 (1932). λ 1021 λ—1487, 18 lines. (B)
Crawford and McLay, Proc. Roy. Soc. A143, 540 (1934). λ 1186—λ 1990, class. 27 lines and class. of published materials. (C)

Bi III  I. P. 25 (BG)
Arvidson, Ann. d. Physik (5) 12, 787 (1932). λ 590—λ 1455, 32 lines. (B)
Crawford and McLay, Proc. Roy. Soc. A143, 540 (1934). λ 1363—λ 1989, class. 19 lines and class. of published materials. (C)

Bi IV  I. P. 45.1 (B)
Arvidson, Ann. d. Physik (5) 12, 787 (1932). λ 353—λ 1438, 112 lines, class. 23. (B)
McLay and Crawford, Phys. Rev. 44, 986 (1933). λ 1723—λ 1925, class. 4 lines and class. of published materials. (C)

Bi V  I. P. 55.7
Arvidson, Ann. d. Physik (5) 12, 787 (1932). λ 234—λ 1394, 103 lines. (B)
Schoepfle, Phys. Rev. 47*, 232 (1935) Classification.

Bi VI  I. P. 93.97
Arvidson, Ann. d. Physik (5) 12, 787 (1932). λ 176—λ 128, 163 lines class. 8. (B)
Mack and Fromer, Phys. Rev. 48, 357 (1935). Classification.
Schoepfle, Phys. Rev. 50, 538 (1936). Classification.

Bi VII and higher
Arvidson, Ann. d. Physik (5) 12, 787 (1932). λ 265—λ 309, 38 lines. (B)

Polonium 84 (O)

Rn I (O)  I. P. 10.69 (BG).

Radon 86.

Ra I (L)  I. P. 5.25 (B)

Radium 88,

Ra II  I. P. 10.10
Rasmussen, Zeits. f. Physik 86, 24 (1933). λ 1888—λ 1976, class. 4 lines. (C)

Actinium 89 (O)

Thorium 90

Th I (O)
Th IV  I. P. 29.38
Lang, Can. J. Research 14, 43 (1935), \(\lambda 846\)—\(\lambda 1959\), class. 15 lines. (B)

Protactinium 91 (O)

Uranium 92 (O)