THE UNIVERSITY OF MICHIGAN COLLEGE OF LITERATURE, SCIENCE, AND THE ARTS Department of Astronomy Technical Note No. 2 ABUNDANCES OF ELEMENTS IN STARS AND NEBULAE (The Abundance of Certain Elements in the Solar Atmosphere) L. H. Aller ORA Project 03719 under contract with: AIR FORCE OFFICE OF SCIENTIFIC RESEARCH AIR RESEARCH AND DEVELOPMENT COMMAND CONTRACT NO. AF 49(638)-807 Washington, D. C. administered through: OFFICE OF RESEARCH ADMINISTRATION ANN ARBOR June 1962

PREFACE Studies of chemical compositions of stars, the sun, and gaseous nebulae are dependent on knowledge of basic atomic parameters such as collision crosssections and transition probabilitieso This report is concerned with application of the transitioh probabilities measured by C. Corliss at the National Bureau of Standards to the composition of the sun. Applications of these fvalues to stellar problems will be made in other reports in this series. I am grateful to Dr. Corliss for sending me his results in advance of publication. L. H. Aller The University of Michigan Observatory June 7, 1962 iii

TABLE OF CONTENTS Page LIST OF ILLUSTRATIONS vii ABSTRACT ix I. INTRODUCTION 1 II. THEORY 2 II. FACTORS INFLUENCING THE DETERMINATION OF ELEMENTAL ABUNDANCES IN THE SOLAR ATMOSPHERE 4 IV. THE ROLE OF TRANSITION PROBABILITIES IN ABUNDANCE DETERMINATION V. SOLAR ABUNDANCES OF PARTICULAR ELEMENTS 7 REFERENCES 11 v

LIST OF ILLUSTRATIONS Table Page I Curve of Growth Data for Various Atoms 13 II Summary of Abundance Determinations 15 Figure 1 The theoretical curve of growth for iron. 16 2 Curve of growth for sodium. 17 3 Curve of growth for potassium. 17 4 Curve of growth for scandium. 18 5 Curve of growth for copper. 19 6 Curve of growth for zinc. 19 7 Curve of growth for gallium. 20 8 Curve of growth for germanium.20 9 Curve of growth for neutral strontium 21 10 Curve of growth for ionized strontium. 22 11 Curve of growth for neutral yttrium. 22 12 Curve of growth for strontium. 22 13 Curve of growth for niobium (columbium). 23 14, Curve of growth for molybdenum. 25 15 Curve of growth for ruthenium. 24 16 Curve of growth for rhodium. 24 17 Curve of growth for palladium. 25 18 Curve of growth for silver. 25 vii

LIST OF ILLUSTRATIONS (Concluded) Figure Page 19 Curve of growth for tin. 26 20 Curve of growth for antimony. 26 21 Curve of growth for ionized barium. 27 22 Curve of growth for ytterbium. 27 23 Comparison of solar abundances with "standard" abundances prepared for solar system. 28 viii

ABSTRACT The solar abundances of 21 elements (Na, Sc, Cu, Zn, Ga, Ge, Sr, Y' Zr, Nb3 Mo, Ru, Rh, Pd, Ag, Cd, In, Sn, Sb, Ba, and Yb) are revised with the aid of new t- -values obtained by C. Corliss of the National Bureau of Standards. Some implications of the new measurements are discussed and it is emphasized that a comprehensive reconsideration of solar abundances should be undertaken, in which ions as well as neutral atoms and a new model solar atmosphere are usedo ix

I, I1NTRODUCTION One of the persistent problems of interest to geochemists and others concerned with the chemistry of meteorites and planets has been the original composition of the solar systems. It is generally accepted that the surface layers of the sun provide the best sample now available. The influence of thermal diffusion in modifying the chemical composition of the solar atmosphere may have been appreciable over long intervals of time, but its exact amount is difficult to assess until we know more about the structure of the sub-photospheric layers of the solar envelope. In any event, the loss of heavier elements from the upper layers depends smoothly on the atomic number in the sense that the heavier the atom, the greater the amount of depletion. Analysis of the chemical composition of the sun's atmosphere requires that we extract from the intensity (equivalent width) of a spectral line the abundance of the element acting to produce ito Now the intensity of an absorption line in the solar spectrum depends on a number of factorsp only one of which is the abundance (Pee, for example Refso 2,354,5,6, and 7). These other factors include (a) the transition probability of the line in question or Ladenburg f:; (b) the sources of line broadening (Doppler widening, natural damping, and collisional damping); and (c) the number of atoms in the right stage of excitation and ionization. (The latter depends on the stratification of the solar atmosphere; i.e., it is determined by the dependence of temperature and density with depth.) The equivalent width of a line (defined as the energy removed from the continuous spectrum expressed in equivalent angstroms thereof), can, be measured for weak as well as for strong lines in the solar spectrum. A comprehensive collection of reliable data has been given by workers at the Utrecht Observatory. The infrared region of the solar spectrum has been surveyed by Mohler9 and by Mohler et al. 1 In addition, many individual observers have made measurements of particular lines in the solar spectrum. 1

II. THEORY In Pecker's theory,11 the equivalent width of a line at a particular point on the solar disk is given by 12,13l14 +oo A M = gx'(x) g ) (1 e-hv/kT) (1) At the center of the disk, = cos = 1 (2) Here M involves the abundance of the element in question, the factor gf%, and certain known numerical factors. In the factor gfX, g is the statistical weight.of the lower level of the transition in question, f is the Ladenburg "f" or "oscillator strength," and \, of course, is the wavelength of the line. We define x = log T (5) where To is the optical depth at the -selected standard wavelength. The factor Z(x) allows for the variation of ionization and thermal excitation with depth15 $ is Pecker's saturation function. It expresses the wellknown phenomenon that a strong line is formed higher in the star's atmosphere than is a weaker line. The "weighting function" g%(x) depends only on the structure of the atmosphere. It expresses the fact that for a weak line, different layers contribute different amounts to the line intensity. The last factor allows for the atimulated emissions (negative absorptions), The intensity of a weak line is given by +00oo x = /.MZ(x) gx(x) (1 - e-/k) dx = ML (4) -00 since $ = 1. Let us define log C = log gfX + Ax@o + L, + constf (5) where we set Go = 5040 = O for the sune The constant is known~ The curve of growth consists of a plot of log Wx/k against log C. It is clear that this relationship will differ not only for a given element in different stages of ion2

ization, but for different lines of different excitation potential or for lines of the same excitation potential which fall in different regions of the spectrum* The latter effects are generally small, however. Equivalent widths of weak or medium intensity lines as observed at the center of the sun's disk are connected to the abundance of the element in question by the relation ogW N log = log C + log (6) NH For any given line, we know AX; the difference between the excitation and ionization potentials of the line in question; the n-value from laboratory measurements; and Lx from the structure of the solar atmospheres The latter also influences the shape of the curve of growths For each element we plot log W%/\ against log C and fit these empirical plots to the proper theoretical curveso For the neutral lines of most metals we can use the iron curve, although for these elements for which the data are good, special curves can be calculated. A systematic study of the abundances of elements in the sun was undertaken by Goldberg, Muller, and Aller1l7 using an empirical solar model atmosphere and r-values derived from various sources. The observational data were taken mostly from the Utrecht8 and Michigan1 works but additional observations were also secured at the McMath-Hulbert Observatoryo 5

IIIo FACTORS INFLUENCING THE DETERMINATION OF ELEMENTAL ABUNDANCES IN THE SOLAR ATMOSPHERE Determinations of the composition of the sun may be improved in 3 ways: (1) Consideration of the formulation of each spectral line under its own particular circumstances. (2) Improvements in the model solar atmosphere. (3) Improvements in the transition probabilities, For the abundant metals it is nearly always possible to find enough lines which are unaffected by blends for one to carry out an analysis based on unblended lines. Usually the strongest lines of an abundant metal are affected by collisional damping; hence one must use either lines of moderate strength or weak lines. The rarer metals, on the other hand, are represented (if at all) by weak lines. These weak lines are sometimes totally blended with other lines-in which event they are useless for abundance studies. In other existences they fall on the edges or "wings' of stronger lines, in which event one should consider first the problem of the formation of the stronger line and then that of the weak line of the element of interest. In the ultraviolet part of the spectrum, the wings of strong or moderately strong lines overlap so thoroughly that one never sees the "continuum" that would have been defined by the negative hydrogen ion, which is the principal source of continuous absorption throughout most of the solar spectrum. Hence the calculation for each line has to be undertaken as a separate problem.l8 The computations attain a fairly high level of complexity but are being undertaken for a few particularly important elements in the solar atmosphere-lead, berylluim, and lithium.19 Likewise, the question of the model atmosphere influences the results. Some lines, such as these of sodium and potassium, are formed almost entirely in the shallowest layers. Hence it is of the utmost importance to know the temperature and pressures in these strata very accurately. Other lines such as those of oxygen, all formed primarily in very deep layers-hence the model atmosphere must be accurate there also. These problems are being investigated by Mutschlecner (particularly for shallow layers) and by Mugglestone for the deeper layers. We shall not consider these general problems here but shall limit ourselves to the third question, that of the transition probabilities, 4

IV. THE ROLE OF TRANSITION PROBABILITIES IN ABUNDANCE DETERMINATION Uncertainties in the {-values enter directly into stellar abundance determinations$ and indirectly insofar as they are employed to fix the level of ionization of a stellar atmosphere by comparing lines arising from two different stages of ionization. In the comprehensive Michigan study of solar abundancesl7Y transition probabilities were derived from many sources -both observational and experimental. Since this investigation was published. C. H. Corliss at the Bureau of Standards has obtained relative -values for 24,000 lines of 70 elements between k2000A and k9000A.20 By calibrating this scale with known, absolute fvalues, Corliss has been able to put it on an absolute basis. The technique he used was the following: An arc was struck between copper electrodes to which a single element was added in the ratio of one atom of the element to 1000 atoms of copper. Hence the operating condition of the arc is not influenced by the trace substance. Corliss then photographed the resultant spectrum in the usual way, calibrating the plates with the aid of standard lamps. In this fashion he could obtain a set of arc line intensities corresponding to uniform conditions of temperature and pressure21 Next he used lines for which relative Dvalues had been obtained in order to get the temperature of the arc, 5100~K + 110~Ko He then compared intensities of lines of 11 elements which appeared in two stages of ionization and for which relative 2f-values were known. In this way it is possible to find the relative numbers of neutral atoms and ions and then, with the aid of the Saha ionizatioh equation, to deduce the electron density in the arc. Corliss thus found an electron density of 2.4 x 1014 Om-3 With both temperature and electron density known, it is possible to determine the relative numbers of ions and neutral atoms for each of the 70 elements in the arc* It is necessary to make a small correction for the thermal diffusion of atoms from the arc stream. Thus, when the ionization in the arc (ieo, the relative numbers of atoms and ions) is known, one can calculate the absolute f -values for lines of all 70 elements, provided absolute 9f-values for some of them are known, Of course) the arc must have a unique excitation temperature; ioeo. the relative populations of the levels must be given by Boltzmann's law. Corliss checked this question carefully and came to the conclusion that the Boltzmann equation was sufficiently well satisfied. Note that a small error in the temperature would cause the tf-values for high-level lines to be systematically in error compared with tf-values for low-level lines' A similar method has been employed by C. W. Allen and his associates22 for a more limited series of elements. The copper arc metnoa is a relative raTner;nan a runcamental method. and. is still dependent on absolute methods for the basic calibration, Therefore, the accuracy of the procedure is ultimately limited by the accuracy of the basic 5

-F-values which can be obtained experimentally with the aid of (a) lifetime measurements of excited levels, (b) the atomic-beam methodQ (c) the electric furnacep (d) anomalous dispersion, and (e) the luminous shock tube* Transition probabilties can also be calculated by theory; relative P-values are usually easier to get than absolutef-lvalues,. In some instances7 it is possible to derive working gf-values from measurements of stellar spectral lines, We shall describe this procedure in a separate report. 6

V. SOLAR ABUNDANCES OF PARTICULAR ELEMENTS The great advantage of Corliss' -data is that they constitute a set of transition probabilities measured in a uniform way. If the line strength are systematically in error, they are likely to be so in a manner which is probably roughly comparable for different atoms. An error in the arc temperature, for example, would introduce an error depending only on the excitation of the levels involved.,Accordingly, we have-re-examined the solar abundances of a number of elements, Most.of the elements of the iron group are being studied by Miss Edith Mller, who is taking advantage of the great number of solar lines of these elements to investigate not only abundances but also such questions as deviations from local thermodynamic equilibrium. She is also examining the influence of changes in the model atmosphere, Paul Mutschlecner is studying lead, beryllium, and lithium —all of which are of interest in connection with nucleogenesis problems. In the present work we have simply derived new values of.log;C with the Corliss -values and constructed curves of growth with the log W/X-values previously obtained. Individual curves of growth have been obtained for only a. few elements; hence the FeI curve of growth17 has been employed for most of the heavier metals, Table I gives the basic data for curve-of-growth plots. Successive columns give for each element the wavelength, the multiplet number, log gf, the abscissa of the curve-of-growth log C, and log W/k. (W/k is the equivalent width divided by the wavelength). Table II summarizes the abundance determinations, Column 3 gives the value of log N obtained in the present work; column 4 gives previously published results17 based on the same solar line intensities. Transition probabilities are available for only a few lines of sodium and potassium. The sodium abundance is based on multiplets (5) and (6) alone. Notice that the scatter is rather large (Fig. 2); hence the determination is rather shaky. The derived abundance( is an order of magnitude smaller than the old determination* The potassium abundance is based on only 3 lines. Two of them belonging to multiplet (3), k4044 and k4047, give a potassium abundance log N =53.61 on the scale log N (hydrogen) = 12, whilst k7699 of multiplet (1) gives log N = 3.66. Thus the NBS f-values do not define as good a curve of growth as did the 7

previously employed data. We are inclined to accept the value given by k7699. Figure 3 shows the curve of growth fitted to both multiplets (1) and (3), It should be compared with Fig. 10 of Ref. 170 The problem of potassium needs further investigation. Figure 4 shows the results for neutral scandium, and the caption to Fig. 4 lists the multiplets used. Hence AX is the distance of the energy level below ionization level (expressed in electron volts), the wavelength region denotes the part of the spectrum in which the lines appear, and the multiplet number gives the number of the multiplet according to the Revised Multiplet Tableo23 For scandium we use the average iron curve of growth (Ref. 17, po 63), We derive the same abundance for scandium as had been found from the c-sum rule and from use of the Bates-Damgaard.Tables.24 The plot for copper is given in Fig. 5. The scatter appears to be somewhat larger than with the older-F-values; on the other hand, use of the Corliss f-values seems to remove the apparent discrepancy between the solarl7and SuessUrey25 abundance. The theoretical curve is the FeI curve of growth. For ZnI (Fig. 6), k6362 (multiplet 6) and the lines of multiplet (2) give discordant results. The adopted zinc abundance is somewhat smaller than that obtained earlier, but is very uncertain. More work needs to be done on this element~ The abundance of gallium (Fig. 7) is raised and that of germanium (Fig. 8) is lowered as a consequence of revising the +-values. Notice that the results are in any event uncertain because of the nature of the solar data, The single gallium line is blended and has an uncertain intensity. The GeI k4226 line is seriously blended with the strong CaI k4226074 line. Clearly, one should treat these lines from the point of view of the blended line theoryo18 The spectrum of SrI (Fig. 9) is characterized by one strong line, x4607, which falls on the transition region of the curve of growth, and a number of weak lines which fall on the linear part of the curve of growths Two lines,.4876 and \5486, fall some distance to the left of the straight line defined by the others; possibly they are affected by blends. The final abundance de-w rived from SrI is somewhat larger than that found previously, whereas the abundance found from X4161 SrII (Fig. 10) is slightly smaller. Hence the discrepancy between the strontium abundance found from SrI and SrII is increased. This element needs further investigation. The abundance of yttrium previously derivedl7 was very uncertain because of the unsatisfactory status of the f-valueso Most of the lines are weak and fall on the straight part of the curve of growth, Even with the Corliss fvalues, the scatter remains very bad This element is an example of that group whose abundance can be derived much better from the ionic lines than from the lines of the neutral element (see Fig. ll)o 8

The lines of neutral zirconium and of neutral niobium likewise all fall on the straight part of the curve of growth. The abundances derived with the aid of the NBS f-values are somewhat more than twice those previously obtainedl7 with approximate estimates of the-F-values (see Figs. 12 and 13), Molybdenum is represented in our analysis by three lines of MoI which fall on the straight-line part of the curve of growth (see Fig. 14). Rhodium and particularly ruthenium show a considerable scatter, probably as a consequence of blends and difficulties in the measurement of these extremely weak lines (see Figs. 15 and 16). For all of these elements Nb, Mo, Ru, and Rh, use of the Corliss -values results in an increase of the abundance determined, On the other hands the palladium abundance is hardly changed by the new -values (see Fig, 17)+ The scatter is less severe than for ruthenium, although for all these elements the weakness of the lines makes it impossible to obtain really reliable abundances The abundance of silver is raised by nearly an order of magnitude when the new-F-values are employed (see Figs 18)* The new value is much more closely in. line with the Suess-Urey compilation. Cadmium and indium are represented by single lines that are affected by blends or by an uncertain position of the continuum. The new abundances are only slightly increased, Neutral tin is represented by two rather weak lines which yield an uncertain abundance about three times greater than that previously published (see Fig. 19). On the other hand the abundance of antimony is decreased by more than an order of magnitude (see Fig. 20). It must be remarked however, that the Sb abundance depends on extremely weak and uncertain solar lines. The curve of growth for ionized barium is shown in Fig. 31. The dotted curve gives the theoretical relation according to the calculations by Elste, The same general features are shown in Fig. 21 as in Fig. 48 of Ref. 13 but the derived abundance is increased by about a factor of 2,5. The lanthanide rare earth ytterbium (see Fig. 22) would appear to have an abundance considerably in excess of that expected for this group of elements,26 Note that this estimate depends on a single line, which may be affected by a blend with an unknown contributor~ Now that the necessary - -values are available, it will be important to measure the abundances of other lanthanide elements, In Fig. 23 we cojmpare the new solar abundances with a previously suggested abundance scale for the solar system27 which was based partly on meteoritic data, the older solar data,17 a semi-empirical compilation by Suess and Urey,25 and a semi-theoretical treatment by Cameron.28 In some instances the discrepancy between the solar and "standard" abundances is decreased; in other instances it is increased, For most elements we are tempted to conclude that a definite improvement in the abundances has been achieved' 9

We want to emphasize that in this particular investigation we have not attempted to improve the basic solar model atmosphere, nor generally to make use of lines of ionized metals. Availability of the National Bureau of Standards transitions probabilities has made it possible to use many more lines than could be used before. Hence a comprehensive new treatment of this problem would entail the following steps: (1) Construction of a new model for the solar atmosphere, in which the best available data is used to improve the temperature distribution in both the shallowest and deepest layers. Such a model has been devised by Paul Mutschlecner.19 (2) With the aid of the new model one would calculate curves of growth and other required data not only for the lines that have been used in the previous work but also for a host of lines arising from different levels and from ions for which f-values are now available. (3) At least for the few available lines of the rarer elements, one would treat the problem of their formation on an individual basis, taking into account blends, etc. 10

REFERENCES 1. S. Chapman and L, H. Aller, Astrophysical Journal, Vol. 132, 1960, p. 461. 2. L. H. Aller, Atmospheres of the Sun and Stars, Ronald, New York, 1953. 3. V. A. Ambarzumian (ed.), Theoretical Astrophysics, J. B. Sykes (tr.), Pergamon, London, 1958. 4. J. L. Greenstein (ed.), Stellar Atmospheres (Compendium of Stars and Stellar Systems, Vol. 6), University of Chicago Press, 1960. 5. J. C. Pecker and E. Schatzman, Astrophysique Generale, Masson, Paris, 19 9. 6. A. Unsold, Physik der Sternatmospharen, Springer, Berlin, 1955. 7. R. Woolley and D. W. Stibbs, Outer Layers of a Star, Oxford University Press, 1953. 8. "Preliminary Photometric Catalogue of Fraunhofer Lines \3164-X8770," Astronomical Researches of the Utrecht Observatory, Vol. 15, 1960. 9. O. C. Mohler, A Table of Solar Spectrum Wavelengths, 11984A to 25578A, University of Michigan Press, 1955. 10. 0. C. Mohler, A. K. Pierce, R. McMath, and L. G. Goldberg, Photometric Atlas of Near-Infrared Solar Spectrum x8465 to k25242, University of Michigan Press, 1950. 11. J. C. Pecker, Annales d'Astrophysique, Vol. 14, 1951, p. 383. 12. See Ref. 4, p. 156. 13. L. H. Aller, G. Elste, and J. Jugaku, Astrophysical Journal Supplement, Vol. 3, 1957, p. 1. 14. L. G. Goldberg and A. K. Pierce, Handbtieh der Physik, Vol. 52, 1958. 15. See Ref. 4, p. 172, Eq. (41). 16. See Ref. 4, p. 173, Eq. (53). 17. L. G. Goldberg, E. A. Mfller, and L. H. Aller, Astrophysical Journal Supplement, Vol. 5, 1960, p. 1. 11

18. L. H. Aller in Physics and Chemistry of the Earth, Vol. 4, Pergamon, London, 1961, p. 1. 19. P. Mutschlecner, Unpublished Thesis, The University of Michigan, 1962. 20. C. H. Corliss, National Bureau of Standards Monograph (in press, 1962). 21. W. F. Meggers, C. H. Corliss, and B. F. Scribner, Tables of Spectral-Line Intensities, National Bureau of Standards Monograph No. 32, 1961. 22. C. W. Allen and A. S. Asaad, Monthly Notices of the Royal Astronomical Society, Vol. 117, 1957, P. 36; C. W. Allen, ibid, Vol. 117, 1957, P. 622; and Vol. 121, 1960, p. 299. 23. C. E. Moore, Multiplet Table of Astrophysical Interest, Princeton Observatory Contributions, No. 20, 1945. 24. D. R. Bates and A. Damgaard, Philosophical Transactions of the Royal Society (London), sers. A, Vol. 242, 1949, p. 101. 25. H. Suess and H. C. Urey, Reviews of Modern Physics, Vol. 28, 1956, P. 53. 26. R. A. Schmitt, A. W. Mosen, C. S.' Suffredini, J. E. Lasch, R. A. Sharp, and D. A. Olehy, Nature, Vol. 186, 1960, p. 863. 27. L. H. Aller, Abundance of the Elements, Interscience, New York, 1961. 28. A. G. Cameron, Astrophysical Journal, Vol. 129, 1959, p. 656; and Vol. 131, 1960, p. 521. 12

TABLE I CURVE OF GROWTH DATA FOR VARIOUS ATOMS Mult.. M N M. log. logC log. MN. log f log C log No. No. Sodium NaI Zinc ZnI 6160.76 5 -0.50 1.65 -5.15 4810.54 2 0.86 4.32 -4.76 6154.24 5 -0.80 1.6 -5.36 4722.16 0.69 4.20 -4.88 5282.65 6 -0.37 1.80 -4.74 4680.14 0.28 3.80 -5.05 5688.22 6 -0.07 2.09 -4.67 6362.36 6 0.44 2.32 -5.53 5153.42 6 -0.82 1.36 -5.33 Gallium Gal Potassium KI 4172.05 1 -0.27 4.60 -5.08 7698.98 1 -0.16 3.23 -4.70 4044.15 3 -0.63 2.77 -5.57 Germanium GeI 4047.19 3 -0.93 2.47 -6.01 4226.567 4 -0.08 4.19 -5.37;S~candi~umn SdcI Strontium SrI 4054.57 6 -0.18 4.13 -5.49 4607.34 2 -0.57 3.77 -5.05 3996.61 7 -0.16 4.15 -5.56 707007 -0.18 2.37 -6.62 5020.40 7 +0.39 4.70 -4.95 4020.40 7 +0.159 4.70 -4.95 6878.32 3 -0.28 2.31 -6.84 4023.69 7 +0.41 4.70 -4.91 - 4047.81 7 -0.48 3.81 -5.57 4 6 3911.83 8 +0.52 4.80 -4.85 462.5 4 -0.25 2.42 -6.5 5671.84. 12 +0.35.28 -5.68 48712.51 4 -0.20 2.42 -6.51 5708.67 12 -0.54 2.38 -6.28 4812.89 5 +0.07 2.4 -51 5686.85 12 +0.26.20 -5.91 44722.29 5 -0.13 2.49 -6.37 j717.31 12 -0.48 2.46 57179.31 12 -0.48 2.46 -6.5 640o8.48. 8 +0.50 2.4366 -6.64 5724.10 12 -0.60 2.3376 -6. 5081.59 13 +0.61 3.53 -5.80 5085.71 115 +0.157 15.50 -5.86 45521.79 9 +0.05 2.24 -6.70 5085.50 13 +0.30 3.23 -5.67 Strontium 5086.94 13 +0.14 3.08 -6.235. 4743.83 14 +0.39 3.31 -5.83 4161.80 -0.28 4.25 -5.28 5520.52 15 +0.44 2.96 -5.90 5514.22 15 +0.36 2.90 -6.44 Yttrium YI 5482.00 16 +0.50 3.02 -6.26 6191.75 2 -1.56 2.76 -6.32 5484.65 16 +0.33 2.87 -6.34 6222.61 2 -2.18 2.14 -6.75 5356.09 17 +0.37 2.89 -6.43 4128.31 5 +0.06 4.27 -5.88 5349.29 17 +0.18 2.72 -6.43 4142.85 5 -0.09 4.18 -5.78 4709.33 22 +0.25 2.35 -6.67 4102.39 7 +0.11 4.30 -6.00 3620.97 8 +0.09 4.27 -5.78 Copper CuI 5630.10 12 -0.43 2.55 -6.15 5105.55 2 -1.70 3.45 -4.79 4781.02 13 -0.80 2.14 -6.57 5700.50 2 -2.16 2.74 -5.61 4819.65 13 -0.77 2.21 -6.24 5782.10 2 -1.57 2.93.33 -5.2562 79335.16 6 +0.09 2.91 -5.56 4477.45 14 -0.55 2.42 -5.90 8092.64 6 +0.41 3.20 -5.33 4487.52 14 -0.28 2.70 -5.87 5218.21 +0.99 3.82 -5.04 4475.72 14 -0.48 2.46 -6.51 5220.09 +0.19 3.02 -5.57 4505.93 14 -0.04 2.92 -6.48 13

TABLE I (Concluded) Mult. Mult. log log C log Mu log f log C log w No. No. Zirconium ZrI Palladium PdI 4815.64 45 -0.69 +2.97 -6.36 5516.95 1 +0.07 +5.43 -5.57 4772.32 -0.56 5.08 -6.07 5404.59 2 +0.30 +5.80 -5.06 4710.08 -0.16 3.41 -6.04 3460.74 2 -0.22 +5.28 -5.54 4687.81 -0.01 3.51 -5.81 3609.56 2 +0.25 +5.61 -5.44 4739.45 -0.32 3.29 -5.99 3372.97 3 -0.21 +5.14 -5.68 4805.89 -1.04 2.54 -6.57 3342.72 3 +0.08 +5.56 -5.23 4241.71 45 -0.20 3.39 -6.23 3553.10 9 +0.54 +5.41 -5.95 4282.22 -0.56 3.02 -6.23 3433.46 11 +0.24 +5.10 -5.76 4072.20 46 +0.11 3.66 -5.89 4027.25 -0.23 3.37 -6.16 Silver AgI 403o.o3 -o.56 3.07 -6.49 403.03 -o.56 307 -6.49 53382.90 1 -0.89 +5.66 -5.39 Niobium (Columbium) NbI 3380.68 1 -0.83 +6.01 -5.12 Nioiu - -0-2 +3.8 5 8273.48 2 +0.49 +3.06 -5.92 )100.92 1 -0.20 +3.85 -5.84 4157.12 -0.72 3.39 -6.32 Cadmium CdI 4139.73 -0.53 3.45 -6.22 41968.1 -0.56.55 5 -6.14 3261.05 1 -2.84 +4.00 -6.34 3713.10 3 -0.25 3.70 -6.27 3697.87 -().65 3.37 -6.17 Indium InI 4511.34 1 -0.18 +4.52 -6.40 Molybdenum MoI 5506.51 4 -0.05 +3.83 -6. in Sn 5533.04 -0.23 3.65 -6.14 3801.03 2 -0.37 +4.04 -6.10 5570.40 -0.56 3.32 -6.20 3330.62 2 -0.49 +3.89 -5.92 Ruthenium RuI Antimony SbI 5799.535 1 -0.21 +4.66 -5.68 3332.55 2 0.54 4.91 -6.73 5798.91 1 -0.06 4.66 -5.54 3267.54 2 0.26 4.89 -6.61 5742.28 2 -0.19 4.35 -5.87 3029.83 2 O.16 4.76 -6.18 5498.95 4 +0.05 4.89 -5.13 3435.75 4 +0.11 4.83 -5.63 Barium BaII 3589.22 4 +0.25 4.73 -6.55 659 -445 4554.o4 1 -0.55 6.59 -4.45 3301.58 4 -1.17 3.65 -5.84 6. ~ 301.58 4 -1.17 3.65 -5.84 4934.08 1 -1.17 5.92 -4.49 4554.54 5 +0.15 4.25 -6.o6 6117 2 117 55- 4 6 3984.84 9 -0.25 3.63 -6.305 67 4709.51 14 -0.20 3.58 -6.20 585.69 2 -1 47 01 6496.92 2 -1.03 5.31 -4.84 4524.95 3 -0.76 3.78 -5.18 Rhodium RhI 4130.66 4 +0.66 4.93 -5.09 3692.36 1 -0.24 +4.82 -5.75 3891.78 4 +0.52 4.35 -5.19 3507.31 2 -0.33 4.40 -6.15 3434.90 -0.14 4.90 -5.66 Ytterbium YbI 3396.83 3 -0.28 4.76 -5.91 3987:97 2 -0.42 4 43 - 3470.64 +0.02 4.64 -63997 2 -42 4.4 -535 3585.11 -0.29 4.58 -6.20 5462.08 +0.02 4.74 -6.26 14

TABLE II SUMMARY OF ABUNDANCE DETERMINATIONS Log N Lo_ N Element Ion NS Ref. 17 Element Ion NS Ref. 17 Sodium NaI 5.44: 6 30 Molybdenum MoI 2.30 1.90 Potassium KI 4.66: 4.70 Ruthenium RuI 1.82 1.43 Scandium ScI 2.80 2.82 Rhodium RhI 1.37 0.78 Copper CuI 3.50 5.04 Palladium PdI 1.27 1 21 Zinc ZnI 3.52 4.40 Silver AgI 1.04 O.14 Gallium Gal 2.51 2.36 Cadmium CdI 1.66 1.46 Germanium Gel 2.49 3.29 Indium InI 1.28 1.16 Strontium SrI 3.10 2.53 Tin SnI 2.05 1.54 SrII 2.56 2.60 Antimony SbI 0.42 1.94 Strontium-Adopted 2.70 2.60 Barium BaII 2.50 2.10 Yttrium YI 3.20 2.25: Ytterbium YbI 2.28 1.53 Zirconium ZrI 2.65 2.23 Niobium NbI 2.350 1.95 (Columbium): indicates a highly uncertain result. 15

-4.0 Fe I t -5.0 log W -6.0 log C - Fig. 1. The theoretical curve of growth for iron. (Abscissae are log C; the zero point of the scale depends on the abundance.) 16

No I Nal / * log W. / xJ9 OlX =5.5 -- II Fig. 2. Curve'of growth for sodium. x = Multiplet5 * = Multiplet6 K I A7698 I I -5.0 - Mo/ Multip/le!t f/N log. -5.5Mu/tip/e! (3) -6.0 - 2.0 2.5 3.0 3.5 log C - Fig. 3. Curve of growth of potassium. 17

-4.5 Sc I /0 -5.5 / 0 o 0 0 0 o o -6.0 - / -6.5- X o -7.02.5 3.0 3. I4.0 45 2.0 2.5 3.0 3.5 4.0 4.5 5.0 log C - Fig. 4. Curve of growth for scandium. Symbol AX Wavelength Multiplet Symbol AX _ __Region ft Number 3 6.7 4000 6, 7, 8 o 5.3 4700-5700 12, 13, 14 o 4.8 5300-5600 15, 16, 17 x 4.4 4709 22 18

Cu log WX -5.0 3.0 3.5 4.0 log CFig. 5. Curve' of growth for copper. AX -5.5.5 5105, 5700, 578 4.0 log C -—' Fig. 6. Curve of growth for zinc. 6.8030, 47220, 482o x 5.9 795, 8095 * 5. 5220 -'2.5 3.0 3.5 4.0 4.5 Fig. 6. Curve of growth for zinc. * 4680, 4722, 481o o 6562 19

Gao -5.0 log WX -5.5 / I 4.0 4.5 I og C -- Fig. 7. Curve of growth for gallium. Gel -5.0 log -5.5 4.0 4.5 log C Fig. 8. Curve of growth for germanium. 20

Sri log /A 4607 3.5 4.0 log C —SrI -5.0 +; J log, +/ -6.5 + + + * X x -7.0........... 2.0 2.5 3.0 log C - Fig. 9. Curve for growth for neutral strontium. (The portions of the curve for the strong and weak lines are shown separately.) 21

Sr iI.-5.0 4.0 4.5 logC - Fig. 10. Curve of growth for ionized strontium. I' I I YI -6.5 o 0 -6.0 0 o -6.5 -o_ \- o -7.0 2.0 2.5 3.0 3.5 4.0 4.5 log C Fig. 11. Curve of growth for neutral yttrium.. I I I / NbI 4 0 -6.0 0 logWX 0 + + -6.5 * / 2.5 3.0 3.5 log C — Fig. 12. Curve of growth for strontium. 22

NbI I -6.0 log-i -6.5 35 3.5 4.0 log C - Fig. 13. Curve of growth for niobium (columbium). MOaI MoI / -6.0 WX log w / 1 3.5 log C - Fig. 14. Curve of growth for molybdenum. 25

-5.0 I RuX I -5.5 logX / x 0 x I I 3.5 4.0 4.5 5.0 log C -- Fig. 15. Curve of growth for ruthenium. AX Multiplet * 7.4 (1), (2), (4) x 6.6o (5), (9) o 6.37 (14) RhI o / / t / log WX ) -6.0 o + -6.5 I 4.5 5.0 log C - Fig. 16. Curve of growth for rhodium. 24

-5.0 logeX I o -5.5 0 + -6.0 5.0 5.5 6.0 log C - Fig. 17. Curve of growth for palladium. - I ^ I I Ag / -5.0 -5.5 / 5.0 5.5 6.0 6.5 log C -- Fig. 18. Curve of growth for silver. 25

SnI log / x -6.0 4.0 log C -- Fig. 19. Curve of growth for tin. I Sb I t-6.5 log-wx10/ _ X 5.0 log C —* Fig. 20. Curve of growth for antimony. 26

"1-'1 -I-I —-'II.I -—.-l-.. Baff -4.5 / x x -5. 0 -5.5. I I I I I I 4.0 4.5 5.0 5.5 6.0 6.5 log C - Fig. 21. Curve of growth for ionized barium. YbI log19 -5.5.I 4.5 log C - Fig. 22. Curve of growth for ytterbium. 27

7.0- Fe Ca 6.0 - 5-00 log N K,I 4.0 5 R) y 3.0 Sc \Ge Zn Sc Ba ~Ga o 2.0. on (A CO 0 — ~ 10- 20 30 40 50 60 70 _n ATOMIC NUMBER Fig. 23. Comparison of solar abundances with "standard" abundances prepared for solar systems.

THE UNIVERSITY OF MICHIGAN COLLEGE OF LITERATURE, SCIENCE, AND THE ARTS Department of Astronomy Technical Note No. 3 ABUNDANCES OF ELEMENTS IN STARS AND NEBULAE (Spectral Line Strengths from Astrophysical Data) L. H. Aller ORA Project 03719 under contract with: AIR FORCE OFFICE OF SCIENTIFIC RESEARCH AIR RESEARCH AND DEVELOPMENT COMMAND CONTRACT NO. AF 49(638) -8o0 WASHINGTON, D. C. administered through: OFFICE OF RESEARCH ADMINISTRATION ANN ARBOR July 1962

PREFACE In order to carry out studies of the shemical. coMpositions of stars, good oscillator strength must be availablev Although such data are available for lines of neutral atoms, no comparable data are available for FtII, CrII and MnII in spectral regions of astrophysical interest,. It is the aim of the present study to partly fill this need with line strengths obtained from stellar spectral line data. It is hoped that these?fX-values may be useful in laboratory as well as in astrophysical applications. Thari.s are due to Charles Corliss of the National Bureau of Standards for supplying us witht4 -data for many atoms in advance of publication. I am also grateful to Prof. Lochte Holtgreven for sending us the Fell and Till?f-measurements secured in his laboratory at Kiel. Assistance in the reduction and assessment of data was supplied by J. Bergey, WA Dent, J. Dickel, J. Ehman, J. Kirk, H. Graboske, D. Gray, G. Withbroe, Miss. C. Parks and Miss. Nancy Houk. Helpful discussions with Dr. Ke. 0. Wright of Dominion Astrophysical Observator3 are also gratefully acknowledged. L. H. Aller zTe University of Michigan Observatoz June 18, 1962 iii

TABLE OF CONTENTS Page LIST OF ILLUSTRATIONS vii ABSTRACT ix I. INTRODUCTION 1 II. EMPIRICAL ASTROPHYSICAL TRANSITION PROBABILITIES 4 III. EMPIRICAL -VALUES FOR FeII 9 IV. ESTIMATES OF TRANSITION PROBABILITIES FOR Fell 12 V. TRANSITION PROBABILITIES FOR TilI, CrII, and MnII 14 A. Ionized Titanium 14 B. Ionized Chromium 15 C. Ionized Manganese 15 REFERENCES 17 v

LIST OF ILLUSTRATIONS Table Page I Emnpirical Log af-Values for Singly Ionized Iron, FeII 19 II Empirical Log ~fX-Values for Doubly Ionized Iron, FeIII, Derived From the Spectrum of y Pegasi 20 III Transition Probabilities for Ionized Titanium, TiII 21 IV Data for Calibration of Log ~fx for Ionized Chromium, CrII 23 V Transition Probabilities for Ionized Chromium, CrII 23 VI Data Pertaining to Transitions of Ionized Manganese, MnII 24 Figure 1 Spectral scan of q Carinae. 25 2 Tracing of a portion of the spectrum of I Carinae. 26 5 Determination off k valtes from the curve of growth. 28 4 Ionized iron line-strength solar data, compared with a Cygni line-strength data. 28 5 Calibration of solar log fX FeII data by comparison with laboratory log f x data. 29 6 Calibration of Groth's log /f's for FeII by comparison with laboratory data. 29 vii

ABSTRACT An attempt is made to compile empirical log( %F)'s for lines of metallic ions of FeII, FeIIn, CrII, Till, and MnII. Here is the Landenburg f, or oscillator strength, y is the statistical weight of the lower level, and X is the wavelength of the line. The curve of growth relates the equivalent width of a spectral line to the number of atoms capable of absorbing it and to the f-value of the transition involved. If the electron pressure and temperature in the stellar atmosphere are known, relative logipX%'s can be determined. Furthermore, if laboratory absolute t-'s are available for a few of these lines, then relative $f2's can all be converted to absolute kX's. The advantages and limitations of these "astrophysical"' Ps are described, and it is emphasized that in order to secure adequate line strength data, one must (a) secure good line-intensity data (b) interpret these with the aid of a model atmosphere, taking into account effects of the stratification in the atmosphere. The e X-values herein tabulated may be useful in analyzing the spectra of incandescent gases in laboratory or experimental sources of unknown temperature and pressure. ix

I. INTRODUCTION Quantitative interpretation of the spectra of stars and nebulae or of the properties of an incandescent laboratory gas requires a knowledge of the transition probabilities of the observed lines. The transition probability of a line may be expressed in various ways, some of which are more convenient than others, for a particular problem, If we are concerned with emission line speCtra it is most appropriate to work with the Einstein coefficient'for spontaneous emission, 1l2,3 viz., wn' 832 c 2 v n ___' = 7 ^ mne (1) A. rin 111 a n Here n is the upper level of the transition involved,; nv ~j w.ie lower level. The symbols 7t, c, m, and c have their usual significance; rn andn, are the statistical weights of the upper and lower levels respectively,, The frequency of the transition n-n' is v, and,nn is the Landenburg' y or oscillator strength. In working with absorption lines, as in the interpretation of stellar spectra, it is easiest to think in terms of the oscillator strength, since j av dV= me,n (2) where,v is the coefficient of absorption and the integration is carried out over the line. Since the other Einstein coefficients can be calculated immediately if one of them is known, the essential problem is to obtain either by theory or by experiment, the A-value of the 4-value. In addition to these quantities, we frequently define a number S. called the strength of a transition by means of the expression 1 6 l44 v3 A(aJ;a'J') 2J + 1 Sl(a ) (3) where J denotes the total quantum number of the upper level, and J' that of the lower level. Here Oa and cY' are the designations of the upper and lowter terms respectively, i.e., the quantum numbers nSLs and n'$'L'S', The total strength S1 contains two factors 1

S1 = S(CJ;alJ') a2(nC;n'1') y4) where S depends only on the angular factors in the wave function. For a given type of coupling (e.g., LS or jj) it can be tabulated once and for all. 45 The a2 factor depends on the radial quantum integral involving the radial wave functions for both the upper and the lower level. For certain atoms and ions for which good wave functions are available, a can be computed to adequate accuracy. Inserting numerical values, we can write Eq. (1) in the form6 nn = 1.5 x 10-8 K (5) where the wavelength \ is expressed in microns. Here = 2J+1 and = 2J'+1 denote the statistical weights of the upper and lower levels respectively. Similarly, Eq. (3) can be written in the form S, (6) wnere AA aenoroes mne wavelength expressed in angstrom units. The spectral lines observed in the stars arelegion and absolute - or Avalues have been measured for only a few of them. The experimental techniques pre difficult, but by utilizing absolute measurements for a few lines of a few elements to establish the electron pressure and temperature in an incandescent gas, it is possible to convert relative +-values to tolerably good absolute fvalues. Thus, quite recently Corliss7 has obtained De-values for a large number of lines observed in a copper arc, to the electrodes of which small quantities of the metal under consideratiiQn ad been added. This study has added a vast wealth of data which is badly needed for astrophysical investigations. Unfortunately, however, even this valuable collection of -determinations does not supply enough data for certain metallic ions which are of great astrophysical importance. We refer in particular to the ions FeII, Nill, CrII, MnII, and'Till, which characterize the spectra of stars of spectral classes F, A; and late B whose surface temperatures lie between about 7500~K and 153,000K. Of these elements, iron would appear to be the most important. In almost every incandescent astrophysical source it appears, as FeI and Fell in the solar photospheric spectrum, and as FeIII in the spectra of hotter stars such as the B2V star y Pegasi. Emission lines of iron, both permitted and forbidden have been observed in the solar chromosphere and corona-nearly all stages from Fell to 2

[FeXIV]. Certain stars such as XX Ophiuchi in its emission-line phase, r (Carinae, and RR Telescopii show strong lines of FeII and [Fell] (see Figs. 1 and 2). Other stars with extended envelopes and high levels of excitation show lines of [FeII!], [FeV], [FeVII], and even [FeX]! These stars include BF Cygni, CI Cygni, Z Andromedae, and T Coronae Borealis.8 Only hydrogen ranks with iron in spectroscopic importance. Transition probabilities for the forbidden lines of iron have been calculated theoretically. In particular, in recent years one may mention the efforts of Garstang9 and the program recently initiated by S. Czyzak at Wright Patterson Air Force Base. In stellar spectra and in the envelopes of the nova-like stars q Carinae and XX Ophiuchi we are particularly concerned with the permitted lines of FelI. Even with the great advances already made in the theory of atomic spectra, particularly by Racah10 and with the possibilities opened up by modern computing techniques, reliable theoretical calculations of transition probabilities for FeII appear intractable. Similar remarks apply to the lines of NiII and CrII. Experimental methods, notably the wall-stabilized are and eventually the luminous shock tube, promise to give us considerable help, but at the present time, data for only a relatively few lines of Fell and Till are available, We shall see that by using the Corliss t-values, the above mentioned FeII fvalues obtained by Roderll and a few other measurements in connection with equivalent widths of metallic ion lines measured in the sun and other stars, we can assemble a body of f -values for lines of CrII, Till, FeII, etc., that may be useful for the interpretation of the spectra of hotter stars, Such astrophysical ~f-values may be used for comparisons of stellar atmospheres, abundance determination, etc., but they should not be taken as definitive. Efforts should be continued to obtain good laboratory and theoretical transition probabilities for all ions of astrophysical interest. 3

II. EMPIRICAL ASTROPHYSICAL TRANSITION PROBABILITIES Attempts have been made to derive relativelf-values from both emission and Absorption stellar spectra. For example, W. Petrie attempted to get empirical f -values from the spectrum of the sollar Chromosphere observed by the Harvard eclipse expedition.12 Some years laterl3 the present writer and his associates attempted to get relative f-.values for FeI from the rich emission line spectrum of the "iron star," XX Ophiuchi. The energy emitted per unit volume in a monochromatic line of frequency y in an incandescent gas is proportional to the number of atoms in the upper level Nj multiplied by the Einstein A coefficient for the transition and the energy per quantum, hv, viz., E Nj Ahv (7) If an excitation temperature can be found so that the relative populations of the excited levels can be represented by Boltzmann's equation with a unique temperature, TD, then relative A-values can be found in principle from measurements of the emission line intensities. In practice, A or <f-value determinations by this method are beset by a number of severe difficulties. Since the emitting column is optically thick, self-absorption will occur and the intensity of the observed bright line will no longer be strictly proportional to the number of emitting atoms per unit volume. Second,the radiating volume may consist of strata of several different temperatures and densities. Third, within a given radiating volume there may exist substantial departures from local thermodynamic equilibrium, so that Boltzmann's formula does not represent the relative populations of the exicted levels. All of these difficulties seem to exist both in the solar chromosphere and in the envelopes of stars such as i Carinae or XX Ophiuchi. Hence the emission lines are probably useless for.f-values determinations, except perhaps for relative strengths within a multiplet. Absorption line intensities appear to offer much better possibilities. Extensive measurements of relative sf -values were made a number of years ago by K. 0. Wright14 and by Barbara Bell.15 Since then, many other determinations have been made in which not only solar but also stellar data were used. The studies yield relative?4-values unless the structure of the stellar atmosphere and its composition are sufficiently well known to permit one to obtain absolute4 -values, 4

Consider, first-of all, the simplest model for a stellar atmosphere-a photosphere which radiates a continuous spectrum overlaid by a "reversing layer" which produces absorption by discrete spectral lines. For this SchusterSchwarzschild Model, Menzel,l6 Unsold,2 Wrubel,l7 and others have calculated theoretical curves of growth. The ordinate of the curve of growth is log W/x c/v where W is the equivalent width expressed (as is \) in Angstroms, c is the velocity of light, and v is the most probable velocity of the absorbing atoms. If there is no turbulence this will be the most probable gas kinetic velocity, but if microturbulence is present, then 2 = 2kT + 2 (8) M where e is the most probable velocity of the turbulent elements, M is the mass of the atom, and T is the gas kinetic temperature. The abscissa of the curve of growth (in Menzel's notation) is log Xo = Nr- l og N + log - GXr,s - log v Ur(T) + log (9) where Nr = number of atoms above the photosphere in the r-th stage of ionization. For neutral atoms r = 0. Xrs = is the excitation potential (in electron volts) of the level (r,s) of statistical weight 0r, Q = 5040/T where T is the excitation temperature. v is defined by Eq. (8). Ur(T) is the partition function for the atom in the rth state of ionization. The constant log /ce2/mc = -1.826. If, as is usually true, the coefficient of continuous absorption varies with wavelength, we must replace log Nr by log Nr K(Xo)/ (x) where r(Xo) is the continuous absorption coefficient at wavelength %o. Hence Nr refers to the number of atoms above the photosDhere at \o Let us suppose that the spectrum of a given star is analyzed by curve-ofgrowth techniques using lines whose J-values are known.l4 15 Then one obtains 5

the velocity of v and a good enough estimate of the temperature T to compute Ur(T). If we are willing to suppose that the same excitation temperature, T, holds for the ions whose ~ -values are unknown as for the atoms whose? -values are known, we may use the curve of growth to derive empirical ~-values. The equivalent width, W1, of a line Xi is measured in the spectrum of the star. We then calculate log W/x c/v and enter the curve of growth (see Fig. 3), to read off the value of log X0. From this value one can then obtain the quantity log Nr+log4 X2. It is clear that we cannot obtain the absolute value of log At unless log Nr is known. Sometimes we can make a good estimate of log Nr if the temperature and electron pressure are known. For example, we possess tolerably good estimates of the mean temperature and electron pressure in the solar atmosphere; T = 5700~K, Pe a 11 dynes/cm2 are reasonable values to use. These estimates are obtained for those elements which are observed in two stages of ionization and for which?f-values are known. Suppose that we have -values for a neutral atom, e.g., FeI, and wish to determine p;-values for the ion Fell. We can determine the excitation temperature from the FeI lines in the following way. First we sort out the lines in groups according to their excitation potentials. For each line we have the value of log W/x c/v and we read the value of log Xc from the curve of growth. Hence for each group of lines of a given value of lower excitation potential X, we can write from Eq. (9) Y1 = log Xo - log 4X = log Cr - X (10) Similarly, for lines of lower excitation potential X2, Y2 = log Xo - log X = log Cr - X2 (11) Hence by plotting Y against X we can determine 0 = 5040/Ttx and log Cr. From log Cr we can compute log Nr, since v is known and Ur(Tex) can be calculated. With the electron pressure and temperature known, the ionization level of the metal can be found, i.e., we get log Nr+l/Nr. For the ionic lines the process is repeated, but we now know Nr+l, 9, v, U(T), etc., so we can derive log gX for the transitions of interest. It is clear that any uncertainity in the abundance of the element involved, in the theory, or in the P. and T adopted will enter directly into the absolute %y's. Hence such empirical %F's are likely to be useful in comparing the spectra of similar stars, since the effects of atmospheric structure and inadequacies in the theory are likely to enter in the same direction in each instance. However, they might be less useful in the interpretation of emission spectra observed in the laboratory. 6

Essentially the same procedure is employed if we use the Milne-Eddington approximation, except that in place of log Xo as abscissa, one uses log ro where. = Nr, so /K; (12) Nr s is the number of atoms per gram of stellar material in the level (r,s) capable of absorbing the line in question, ao is essentially the coefficient of monochromatic absorption at the center of the line, and K is the coefficient of continuous absorption (defined per gram of material). The form of the curve of growth is slightly different; Wrubell8 has published a number of such curves. The conventional curve-of-growth approach assumes that the same curve applies to both neutral atoms and ions. In a more detailed treatment we take into account the stratification of temperature and density in the stellar atmosphere. Then each atom in each stage of ionization and excitation has its own curve of growth. Usually it is possible to use a single curve for atoms in a given stage of ionization. For example, most of the neutral metals in the sun follow rather well the curve of growth for neutral iron. The ionized metals probably require a different curve of growth. In the version of the theory proposed by Pecker19 and applied by Aller, Elste, and Jugaku,20 the abscissa of the curve of growth is log C = log rs fX + AXr,s G0 + log L* + log K (13) where K is a numerical constant that depends on the mean molecular weight of the stellar material, o: = 5040/To (where To may be taken as the mean temperature of the stellar atmosphere), and AXr, = - Xr s (neutral atoms) - Xr s (ions) where I is the ionization potential. Now Lax is defined by an integral which involves the weighting function for the stellar atmosphere and the distribution of absorbing atoms as a function of depth. 7

The ordinate of the curve of growth is log W/X. For any given line, L*t can be calculated, AXr s GO is readily computed, K is known, and log C is read from the curve of growth. Hence logF X can be determined directly. The resultant value is subject to uncertainties caused by inadequacies in the model atmosphere and by inaccuracies in the' -values. A procedure similar to that which we have described was employed by Groth21 to get f-values for FeXll Instead of the Pecker theory, however, he used Unsolds curve of growth, which is similar in form to Menzel's. The empirical method is useful for getting relative -values for many ions, Then, if an absolute calibration for a few lines of a given ion is known, these relatives'-values can be converted to absolute S-values. This program has been carried out for Or I, Fell, and Till, but for Mnwi it was necessary to use entirely empirical ff's. 8

III. EMPIRICAL ~f-VALUES FOR Fell We are primarily interested in obtaining pf-values for lines of ions which appear in stars of spectral classes F, A, and late B. Among available sources of line-intensity data I have chosen the following: K. 0. Wright's compilation of data for the solar spectrum.14 The Utrecht catalogue of equivalent widths in the solar spectrum.22 W. Buscombe's measurements of equivalent widths in the A stars, a Cygni (Deneb) and y Geminorum.23 Unpublished measurements in the spectrum of a Cygni secured by the writer at the Lick Observatory in 1937. In addition to these data we have Groth's determination21 of log -values based on his study of the spectrum of a Cygni. With the aid of a model atmosphere for this star and Unsold's curve of growth, he attempted to derive absolute line strengths. Although the solar spectrum exhibits numerous lines of FeII, many of which are prominent in the spectra of hotter stars, here they are blended with neutral metal lines. Furthermore, although most iron atoms in the sun are singly ionized, the observable FeII lines arise from high levels for which the Boltzmann factor is unfavorable. Since the sun is often used as a standard for comparison with other stars, it seems worthwhile to see what information can be obtained from solar FeII lines. The spectra of A stars would appear to offer better possibilities for determinations of'k-values.. The ionic lines are still strong but blending is much less severe. The chief difficulty is that ionic lines tend to fall on the straight-line part of the curve of growth; hence a small error in the measured intensity can produce a large error in log ~f. Buscombe measured equivalent widths in the spectra of the dwarf y Geminorum and the supergiant a Cygni using coude plates secured with the 100-inch telescope at Mt. Wilson. He used Wrubel's curves of growth to obtain log 7O's from which log fXs's- could be calculated when the excitation temperature and electron pressure were determined. The Lick data consisted of equivalent widths of lines measured on Mills spectrograph plates obtained by G. F. Paddock. They covered only the region?4300-X4630, but many plates of excellent quality were available. Groth used Mt. Wilson coude data. The first step was to reduce all log \f \-data to Groth's system and to the 9

system of the solar line-strengths. In order to do this, one must know the excitation temperature, which is not necessarily the same as the ionization temperature. In fact, Gex 0.-75 for y Geminorum- corresponding to a temperature somewhat lower than the ionization temperature for this star. We assumed Gex _ 1.04 for the FeII lines in the sun-a value derived from the Fel data. It would have been better to have analyzed the solar Fell data with the aid of a model atmosphere and properly calculated contribution functions. Figure 4 shows a plot of log(Qf4)) -[a Cygni - Lick data] vs. log(,X). tsolar data - due mostly to Wright]. Here X is given in Angstroms. The zero points have yet to be adjusted. We prepared a similar plot of log(eK)oQ vs. log(~ X)Groth as well as corresponding diagrams comparing the Buscombe data: with those of Groth and;with solar data. In this fashion, we could reduce all our log fgX's to the same system, i.e., to the same zero point-either that of Groth or the solar f-system for which we adopted an iron abundance and a mean electron pressure and:temper.ature for the solar atmosphere. Although the solar-based and Deneb-based (Groth model) relative f-values agreed reasonably will with one arother, there occurred appreciable differences in the corresponding absolute -values. Hence the adopted atmospheric parameters and iron abundance for one star or the other (or both) must be a fault* Hence it is most urgent to seek absolute 9-measurements for FeII. The extensive Bureau of Standards.f-value determinations7 do not help much here, as little data are available for the spectral regions accessible to observation. Fortunately, some absolute oscillator strengths have oeen measured by Roderl working with a wall-stabilized arc in the laboratory of Lochte-Holtgreven in Kiel. The measured transition probabilities were obtained for lines belonging to multiplets:(27), 37, 38, 42, and 49 of Fell, but the absolute fvalues were obtained by calibrating these with respect to just one-multiplet of FeI, multiplet 41. The 4F-values for this multiplet were taken from the work of King24 as calibrated by Bell, Davis, King, and Routly.25 In Figs. 5 and 6 we plot logg#X as measured by Roder at Kiel against the log f X-values obtained from the sun and from a Cygni (Groth). In this way we find the reduction factors necessary to express all our log fX-values on the same absolute scale, Table I gives the results: we have expressed X in centimeters. The columns headed B, A, and G refer to the log(afX) values found from the spectrum of a Cygni with data obtained by Buscombe, the present writer, and by Groth. Buscombe's y Geminorum data supply the data for the next column; the solar data are in the next column; and the last column gives the adopted mean value. It would probably be just as satisfactory to use Groth's.i-values. Several obvious improvements can be made. Additional stars should be ob10

served; we may mention Traving's investigation of Sirius, wherein accurate equivalent widths were measured. A reliable model atmosphere can be constructed since the temperature and surface gravity of this star are accurately known. Several other late B and A stars could be similarly investigated. 11

IV. ESTIMATES OF TRANSITION PROBABILITIES FOR FeIII Lines of FeIII are often visible in the spectra of B stars. In the later subdivisions they are sometimes seen with lines of FeII-or, in the case of HR6870, are present along with lines of both FeII and FeI. These lines could be very useful in fixing the excitation tempurature of a stellar atmosphere. In order to get empirical logyfX's we shall use line-intensity measurements for y Pegasi.26 Jugaku, Sargent, and Greenstein27 suggest that the excitation level of the spectrum of this star can be represented by T = 18,000~K, log PE = 0.234. We shall use Wrubel's curve of growth19 for B1/Bo = 4/3. His curves were calculated for the Milne-Eddington approximation and the assumption B = B + B (15) where T is the optical depth in the continuum in the neighborhood of the line. The ordinate of the theoretical curve of growth is We W c ( log = log (16) % v ^ M t and the abscissa is log - = log - log v - log U(T) Nr mca + logf\ - GXrs - log nk(T,P.) (17) The turbulent velocity Et, is 4.5 km sec-1 according to Struve and Mrs. BohmVitense.28 We interpolate the coefficient of continuous absorption from Allen's29 tables for log Pe = 2.34, 9 = 5040/T = 0.278. Putting in numerical values, log v = 5.70 log U(FeIII) = 1*73, log K varies from 0.04 at \4000 to 0.225 at X5000. The abundance of iron in y Pegasi is not known a priori, so we have assumed it to be the same as in the sun; each gram of stellar material is presumed to contain 1.42 x 1018 iron atoms. By application of the ionization equation we find N(FeIIl)/N(Fetotal) - 0.83. 12

Details of this calculation are given in Table II. Successive columns give the wavelength of the line, the multiplet number according to the Revised Multiplet Table,30 the excitation potential X (in e.v.), the equivalent width of the lines in mA (B indicates a blend), W/X x 10, log W/x c/v (of Eq. 16), log KX, and log rQ. The last two columns give the quantities: -OX + log ekfx and log f X (calculated for 0 = 0.278). Notice the marked increase of logg with excitation potential. The probable explanation of this phenomenon is as follows: In the atmosphere of y Pegasi the high excitation lines arise in the deeper layers where the Boltzmann correction -OX is smaller, whilst the low excitation lines originate in the cooler, shallower layers where OX is larger. The curve of growth procedure uses the same value of 0 for all levels; hence a systematic error in logoff is introduced. When we compare two stars, the same approximation is employed in both, and the errors tend to cancel out. Hence the numbers in the last column are in error, if they are interpreted strictly as log (X values. 13

V. TRANSITION PROBABILITIES FOR Till, CrII, AND MnII A. IONIZED TITANIUM Experimental transition probabilities for Till in the astrophysically important range from 3000A to 8000A has been measured by R. B. King and A. S. King,31 by J. B. Tatum32 by K. H. Wobig,3 and by C. Corliss.7 For many lines not measured by these authors we can use empirical Xk's from the solar curve of growth, calibrating them with the laboratory f-values. Both Tatum and Wobig calibrated the tf-values from the sum rules, whereas Corliss interpreted his relative -values with aid of the measured temperature and electron density in the arc. Since$ -values are available for both TiI and Till we can solve for the ionization equilibrium in the sun. Assuming Tion = 57000K, we get log PE 1.12. This (Pc,T) combination can then be used to get rough estimates of absolute logAf X-values for other ions, such as NiII or MnII, for which no other data are available. For getting the solar log;PX-values, we employed Menzel's curve of growth as revised by Wright, and read off log Xo values corresponding to the observed log W/X's. The lines whose FX-values were known permitted calculation of the quantity Y = log X - log X = log Cr - 9X (18) Then, by plotting Y for each multiplet against X, the excitation potential of the lower term, we obtained log Cr and Or. Two independent solutions in which different weights were used for different lines gave gex = 1,03 and 1.12, corresponding to excitation temperatures of 4500~K and 4900~K respectively. Table III gives the final results. The first three columns give the wavelength of the line, the multiplet number, and the excitation potential. The next three columns give log af as measured by Corliss, Tatum, and Wobig. The seventh column gives F = () x 106 (19) noting important blends, and the last column gives the finally adopted value of logr -X. Where laboratory data are available they are used; where they are not we have resorted to empirical solar log SXk's. 14

B. IONIZED CHROMIUM Ionized chromium represents a more difficult problem. Our treatment is similar to that for FeII in that we employ data from both the sufh and a Cygni, including Groth's empirical log g4-values for Deneb. Absolute -f-values are available for only a few CrII lines in the astrophysically important region X3800-X5000A. The data for the calibration are given in Table IV. The columns headed "Theory and Laboratory" consist of relative strengths for multiplet 44 (CrII) computed for LS coupling calibrated with Corliss' measurement for X4558.68. For other lines, laboratory4 -determinations are available. Table V gives the finally adopted log f-X-values for CrII, based on both solar and a Cygni equivalent widths and calibrated with the aid of the data in Table IVo Since the absolute calibration depends on very few lines, additional measurements of f -values for CrII are urgently needed. A plot of log 4X-values obtained from solar line intensities against corresponding quantities obtained by Groth from his study of a Cygni show a considerable scatter, probably partly due to blends. These blends are likely to be much more serious in the sun then in the hotter star, whose spectrum is much less crowded with lines and consists mostly of lines of ions. Accordingly, we have given greater weight to Groth's data in forming the final means in Table V. C. IONIZED MANGANESE The A-star 53 Tauri is characterized by strong lines of Manganese, particularly those of MnII, The stronger lines of MnI are also observed in this star; the p-values for the resonance lines have been measured by a number of experimenters. Determination of log afX for MnII lines from solar of stellar curves of growth are beset with a number of severe difficulties. In most A-stars the MnII lines are few and weak; lines of this ion appear in the sun (cf. Table VI). In Table VI the first three columns give the lines, multiplet numbers, and excitation potentials. The fourth column gives the intensity of the line divided by the wavelength, i.e., W/X,multiplied by 106 These lines are all extremely weak, with Rowland intensities of 0, -1, and -2. We have selected only those lines that are believed to be reasonably free of blends. In any event, the intensities are subject to large, random, observational errors. This cause alone would disaualify them from consideration as reliable soureces of line-strength data. Column 5 gives the quantity log S/Zs (where s denotes the strength of the line, and Zs the total strength of the whole multiplets)o These quantities are calculated for LS coupling. Unfortunately MnII does not follow LS coupling, so theoretical calculations of even relative f-values will not be possible unless 15

an intermediate coupling theory along with other refinements are used. Multiplet (2) represents an intercombination, a5D - zP, while multiplets (7) violate the LS coupling scheme because the initial and final terms arise from different parents, i.eo, a4F and a4P of MnIII. We would expect the — values for multiplet (2) to be smaller than those for multiplets (6), (7), and (8) —a prediction that seems to be substantiated by the data.'he last coliumn gives logqF, —X as derived from the solar MnII lines. We assumed a solar log:P and T, as described above. The MgI lines were used to fix the constants in the abscissa of the curve of growth. The ionization equilibrium and adopted 3M abundance permits one to calculate log %Xf-~X. Notice that most of the available lines have a rather high excitation potential; only the levels of multiplet (2) lie within 2 volts of the ground level. It is,,unlikely that the same value of ~0 applies to both these levels since the 5.37volt lines must originate, on the average, in deeper layers than the 1.8-volt lines. We conclude theiqrfolie that the solar data give us no help in establishing usable log )'-values for MnI; it would be better to invert the procedure and obtain log i,%-values from the MhII lines in 535 T'uri and other Muanganiese stars, using preferably a model atmosphere technique. In the meantime, laboratory investigations should be pushed further in an effort to obtain absolute QF-values for at least a few lines of MnIZ. Perhaps the luminous shock tube can yield satisfactory results. 16

REFERENCES 1. E. Condon and G. Shortley, Theory of Atomic Spectra, Cambridge University Press, 1935. 2. A Unsold, Physik der Sternatmospharen, Springer-Verlag, Berlin, 1955. 3. L. H. Aller, Atmospheres of the Sun and Stars, (Chapter 5), Ronald, New York, 1953. 4. L. Goldberg, Astrophysical Journal, Vol. 82, 1935, P. 1; and Vol. 84, 1936, p. 11. 5. H. E. White, Introduction to Atomic Spectra, McGraw-Hill, New York, 1934, p. 439. 6. D. R. Bates and A. Damgaard, Philosophical Transactions of the Royal Society of London, Series A, Vol. 242, 1949, p. 101. 7. C. Corliss, National Bureau of Standards (in press). 8. See e.g., J. Sahade, "Composite and Combination Spectra," and D. B. McLaughlin, "Spectra of Novae" in Stars and Stellar Systems, ed. by J. L. Greenstein, University of Chicago Press, 1960. 9. R. Garstang, Monthly Notices of the Royal Astronomical Society (in press). 10. G. Racah, Physical Review, Vol. 61, 1942, p. 186; Vol. 62, 1942, p. 438; Vol. 63, 1943, p. 367; Vol. 76, 1949, p. 1352. 11. O. Roder, Zeitschrift fur Astrophysik (in press). 12. W. Petrie, Unpublished Thesis, Harvard University, 1942. 13. L. H. Aller in Proceedings of the Indiana National Science Foundation: Conference on Stellar Atmospheres, ed. by M. Wrubel, 1954. 14. K. 0. Wright, Publications of the Dominion Astrophysical Observatory, Vol. 8, 1948, p. 1. 15. B. Bell, Special Report 25, AMC Contract W19-122ac-17, Harvard University Observatory, 1949. 16. D. H. Menzel, Astrophysical Journal, Vol. 84, 1936, p. 462. 17

17. M. Wrubel, Astrophysical Journal, Vol. 119, 1954, p. 51. 18. M. Wrubel, Astrophysical Journal, Vol. 109, 1949, pa 66'Vol. 111, 1950, p. 157. (See also the curve of growth for pure absorption quoted in Greenstein (Ref0 8), p* 199) 19. J. C. Pecker, Annales d'Astrophysique, Vol. 14, 1951, p. 3835. 20. L. H. Aller, G. Elste, and J. Jugaku, Astrophysical Journal Supplement, Vol. 3, No. 25, 1957, p. 1. 21. H. G. Groth, Zeitschrift fur Astrophysik, Vol. 51, 1961, p. 206. 22. Utrecht Observatory Publications, Vol. 12, 1960. 23. W. Buscombe, Astrophysical Journal, Vol. 114, 1951, p. 73, 24. R. B. King and A. S. King, Astrophysical Journal, Vol. 87, 1938, p. 24. 25. G. D. Bell, M. H. Davis, R. King, and P. M. Routly, Astrophysical Journal,. Vol. 127, 1958, p. 775. 26. L. H. Aller and J. Jugaku, Astrophysical Journal Supplement, Vol. 4, 1959, p. 109; Astrophysical Journal, Vol. 127, 1958, p. 125. 27. J. Jugaku, W. L. Sargent, and J. L. Greenstein, Astrophysical Journal, Vol. 134, 1961, p. 783. 28. 0. Struve and E. Bohm-Vitense, Astrophysical Journal, Vol. 123, 1956, p. 288. 29. C. W. Allen, Astrophysical Quantities, Athlone Press, London, 1955. 30. C. Moore, "Revised Multiplet Table, " Princeton University Observatory Contribution No. 20, 1945. 31. R. B. King and A. S. King, Astrophysical Journal, Vol. 94, 1941, p. 27; Vol. 95, 1942, p. 78. 32. J. B. Tatum, Monthly Notices of the Royal Astronomical Society, Vol. 112, 1961, p. 311. 33. K. H. Wobig, Zeitschrift fiir Astrophysik (in press). 18

TABLE I EMPIRICAL LOG.%F-VALUES FOR SINGLY IONIZED IRON FeII Deneb 7 Adopted k ___ — B A G Geminorum Sun Mean 3762.94 -5.99 -5-99 3781 -7.22 -7.25 -7.23 3783.39 -7.51 -7.24 -7.50 3824.95 -7.31 -6.58 -7.30 3827.07 -6.95 -6.43 -6.92 3845.18 -6.52 -6.52 3864 -7.03 -7.03 3935.96 -5.37 -5.65 -4.88 -5.60 3938.31 -7.95 -7.82 -7.90 3938.95 -5.27 -5.86 -5.36 -5.31 -5.80 3960.90 -4.82 3975 -6.48 -5.52 -6.45 4024.55 -6.19 5.85 -6.00: 4044 -6.66 -6.66 4122.64 -6.88 -7.57 -7.90 -7.57 4128.74 -7.85 -7.31 -7.82 4173.48 -6.74 -6.76 -6.54 -6.75 4177.69 -7.75 -6.88 -7.75 4178.86 -6.73 -6.45 -6.91 -6.67 4233.18 -5.77 -6.10 -6.43 -6.05 -6.04 4258 -7.53 4273.33 -7.28 -7.55 -7.36 -7.45 4296.56 -6.99 -7.24 -7.15 -7.00 -6.91 -7.03 4103.19 -6.53 -6.91 -6.70 -6.83 -6.57 -6.65 4351.76 -6.22 -6.41 -6.17 -6.25 4369 -7.66 4385.39 -6.73 -6.79 -6.67 -6.83 -6.59 -6.66 41416.81 -6.80 -6.92 -6.74 -6.89 -6.86 -6.80 4473 -7.63 4489.17 -6.97 -7.03 -7.04 -6.97 -7.33 -7.05 4491.42 -6.86 -6.91 -6.93 -7.02 -7.00 -6.90 4508.28 -6.53 -6.39 -6.52 -6.74 -6.77 -6.60 4515.33 -6.61 -6.71 -6.62 -6.72 -6.84 -6.67 4520.20 -6.56 -6.83 -6.68 -6.98 -6.91 -6.73 4522.60 -6.17 -6.42 -6.30 -6.44 -6.23 -6.25 4541.55 -6.89 -7.08 -7.11 -7.06 -6.54 -7.01 4555.89 -6.47 -6.52 -6.45 -6.48 -6.81 -6.52 4576.30 -6.98 -7.08 -7.02 -6.99 -7.28 -7.08 4582.85 -7.21 -7.25 -7.06 -7.44 -7.26 4583.82 -6.02 -5.93 -6.05 -6.35 -6.08 -6.06 4629.34 -6.50 -6.65 -6.54 -6.64 -6.55 4635.33 -5.00 -5.64 -5.05 -5.60: 19

TABLE II EMPIRICAL LOG X-VALUTES FOR DOUBLY IONIZED IRON, FeIII, DERIVED FROM THE SPECTRUM OF 7 PEGASI W Wo -X — x1.0( lo —-- -g — - g - X lMultiplet X W llx 1 og +lo log fF*o 4571.58 4 8.21 5B 1,1 -1.17 0.11 -0.98 10.68 -7.39 4382.56 8.21 10 2.3 -0.87 0.11 -0.67 93.7 -7.08 4352.58 8.21 11 2.6 -0.82 0.11 -0.62 9.33 -7.04 4415 55 8.21 30 6.8 -0.40 0.12 -0.12 8.82 -6.53 o 4431.03 8.21 22B 5.0 -0.53 0.12 -0.29 8.88 -6.69 4022 38 45 11.53 11 2.7 -0.79 o 04 -0.59 93.5 -6.14 4039.21 11.53 10 2.5 -0.83 0.05 -0.62 9-38 -6.17 4122.02 118 20.51 14 354 -0.70 0.06 -0.49 9.23 -3.53 4164.92 20.54 31B 7.4 -03.6 0.07 -o0.6 8.79 -5.09 4137.71 20.52 24 5.8 -0.46 0.07 -0.21 8.94 -5.24 *See text

TABLE III TRANSITION PROBABILITIES FOR IONIZED TITANIUM, TiII log F Solar and Multiplet X C W T (Eq. 19) Blend Stellar 3679.71 "15 1.57 24.7 -4.70 3706.24 73 1.56 -0.05 37.2 -4.48 3741.65 72 1.57 0.31 44.6 -4.12 3757.68 72 1.56 -0.08 46.8 -4.51 3759.29 13 0.60 0.34 115 -4.09 3761.32 13 0.57 0.24 16 -4.19 3761.87 107 2.58 0.45 -3.97 3813.39 12 0.60 38 -5.25 3814.59 12 0.57 33.6 CN -5.41 3882.30 40 1.11 30.9 CN -4.93 3900.54 34 1.13 -0.02 -0.08 41 -4.43 3913.48 34 1.11 -0.06 -0.20 39.4 -4.47 3932.02 34 1.13 -1.20 -1.53 34.7 -5.41 3982.01 11 0.57 20.7 -6.28 3987.60 11 0.60 14.8 -6.99 4025.14 11 0.60 -1.69 20.9 -6.08 4028.35 87 1.88 -0.42 -0.81 22.3 -4.81 4029.68 87 1.88 26.8 -4.28 4053.80 87 1.88 16.0 -5.41 4161.53 21 1.08 23.1 Fe -5.58 4163.66 105 2.56 o.45 25.7 -3.93 4171.92 105 2.59 0.22 21.6 -4.16 4174.10 105 2.59 9.8 -6.76 4184.32 21 1.08 18.2 -6.02 4287.89 20 21.0 -5.72 4290.23 41 1.16 -0.79 -0.80 -0.68 27.3 -5.02 4294.10 20 1.08 -0.79 -5.16 4300.06 41 1.18 -0.55 -0.47 -0.40 38.6 -4.92 4301.53 41 1.16 -0.95 -1.25 -1.17 34.4 -5.32 430-.88 41 1.16 -1.15 -1.00 -4.99 4312.88 41 1.18 -0.94 -1.14 -1.14 35.5 CH -5.30 4314.99 41 1.16 -1.15 -0.96 28.0 -5.02 4320.99 41 1.16 16.4 -6.17 4330.24 94 2.04 9.7 -5.94 4337.93 20 1.08 -0.78 -0.97 -0.89 24 -5.14 4350.84 94 2.05 14 - 5.46 4367.66 104 2.38 -0.82 22.4 CH -4.oo 21

TABLE III (Concluded) log F Solar and X Multiplet X C W T (Eq. 19) Blend Stellar 4386.88 104 2.59 -0.96 13.4 -5.08 4394.07 51 1.22 18 -5.87 4395.04 19 1.08 -0.46 -0.47 -0.30 30.7 -4.82 4395.85 41 1.24 15.0 -6.34 4399-78 51 1.23 -1.06 -1.11 -1.20 26.4 -5.42 4409.25 61 1.24 6.8 -6.90 4409.51 61 1.23 7.9 -6.93 4411.90 61 1.22 12.0 -6.65 4417.72 40 1.22 -1.17 -1.15 22.4 -5.46 4418.34 51 1.23 15.8 -6.19 4421.94 93 2.05 11.5 -6.88 4443.80 19 1.o8 -0.65 -o.64 -0.64 30.8 -5.00 4444.54 1.11 12.4 -6.69 4450.48 1.08 -1.46 -1.46 17.8 -5.68 4456.64 115 3.11 15.9 -4.10 4464.45 40 15.2 -6.33 4468.50 31 1.13 29.3 -4.99 )469.13 11.0 -6.74 4488.33 115 3.11 0.28 -0.44 -0.46 10.7 -4.07 4501.28 31 1.11 -0.68 -0.70 -0.79 29.1 -5.03 4518.34 18 10.0 -6.84 4529.49 1.56 13.0 -6.11 4533.97 50 -0.50 -0.46 24 Co -5.84 4544.02 60 7.7 -6.86 4470.85 40 12.5 -6.58 4549.64 38 -0.16 -0.08 -0.14 32.5 -4.64 4563.77 50 1.22 -0.87 -0.72 -0.84 28.0 -5.21 4568.33 40 1.22 6.1 -7.17 4571.96 82 1.56 -0.18 -0.19 -0.24 29.7 -4.52 4589.92 50 1.23 -1.51 -1.43 -1.81 -5.85 22

TABLE IV DATA FOR CALIBRATION OF LOG- FX FOR IONIZED CHROMIUM, CrII log 94 log fx Theory and Theory and.... _ kLaboratory Solar k Laboratory Solar 3677.69 3.37 3.58 4634.08 2.83 3.18: 3677.91 3.55 3.97 4616.63 2.43 2.94: 3712.95 3.67 3.42 4588.21 3.52 3.44: 4012.55 4.62 4.57 4558.66 3.78 3.58 4554.99 2.74 2.99: TABLE V TRANSITION PROBABILITIES FOR IONIZED CHROMIUM, CrII ~, Alog1 log7 X x: logf lo 5677.69 -4.62 4051.97 -6.25 4284.18 -5.48 3677.91 -4.44 4070.86 -5.41 4554.99 -5.15 5712.95 -4.33 4111.00 -5.87 4558.64 -4.22 3738.38 -5.67 4145.78 -3.31 4565.74 -5.63 3765.61 -5.94 4179.43 -4.35 4588.20 -4.47 4003.30 -3.80 4252.62 -5.60 4592.08 -5.20 4012.55 -3.38 4261.92 -5006 4616.55 -5.11 4038.00 -3.36 4269.27 -5.88 4618.81 -4.90 4049.13 -2.59 4275.58 -5.40 4634.07 -4.82 23

TABLE VI DATA PERTAINING TO TRANSITIONS OF IONIZED MANGANESE, MnII log _.X X Multiplet X F log S/ZS -GX 07.25 (2) 1.82 0.6 -10.69 053.7 1.80 9.8 - 9533 38.79 1.82 5.2 - 9.69 30 47 1. 84 34.42 (6) 5.35 6.3 -1.24 - 9.60 00.20 5.36 11.9 -1.35 - 9.18 43.97 53.7 11.5 -0.55 - 9.21 45.60 53.7 2.3 -1.50 - 9.09 6,.43 (7) 53.7 14.0 -0.50 - 9.00 44.27 5.35 2.1 -1.10 -10.13 36.91 (8) 53.7 5.1 -1.50 - 9.70 08.05 5.37- 16.5 -0o74 - 8.67 -9.50 5.37 2.4 -1.35 -10 07 24

HP II Hy,FeI Fe I Fe ~ I r FeIL ~~~~~~~~SLIT~~I He 9A DARK CURRENT resolution is about 9A, so that many of the FeII lines are blended. resolution is about 9A, so that many of the Fell lines are blended.

Fe If FeUllF -4452.11 445795 Fel Fell Fell 4 445.95 — ---- 4471.5 447491 —-- 448.85 4491.40 e Fel 4515.34 Fe IE -4508.28 452022?FejI 4472.92 Fell FeI/, x~~~n, I i r^~Ni 50 FellI 4485.22 4432.45 Ti52 Ti II 4443.87?T II 4501.27 R) Fig. 2. Tracing of a portion of the spectrum of Tj Carinae. This is a part of a microphotometer tracing of a coude' spectrogram of Tj Carinae secured at Mt. Stromlo Observatory. Notice the dome-shaped profiles and sharp spikes of individual emission lines, indicating complicated physical conditions in the radiating gas. (Courtesy T. Dunham)

Fe*\~~~ /\ -^FeU U FenP-Fl Fe 63 4549.5 4556.89 4583.83 452Z63 Fe IH 4576.33 Fe 3 Cr IL 4529.39 t 4558.66 Fe. Hf Cr I' \I|45805| r 4588.22 A Fe I —-L [FeI 4533.00 4541.52 Mol 4571.10 r U c 4581.2 Fig. 2 concluded.

+1 ol1> 0 - /n / -I II5~~~~~~~II I =- 0 +1 1-2. +3 log Xo Fig. 3. Determination of ~-F values from the curve of growth. For each line whose equivalent width W is known, one calculates log W/x c/v and reads log Xo from the curve of growth. 3.0 "..O 2.0 i/LI I 1.0 2.0 3.0 log (g f X) sun Fig. 4. Ionized iron line-strength solar data, compared with a Cygni line-strength data. 28

-7.0/ - 0 655. -6.0 o~I, I log(gfX) SUN Fig. 5. Calibration of solar log qf% FeII data by comparison witn laboratory log rf\ data. -2.5 o 0 2.0 / -3 /. / - 1.0 - -I.50 -2.0 -25 -3.0 log (gf)) Fig. 6. Calibration of Groth's log O's for FeIl by comparison with laboratory data. 29

UNIVERSITY OF MICHIGAN 3 9015 02499 5535II 3 9015 02499 5535

THE UNIVERSITY OF MICHIGAN COLLEGE OF LITERATURE, SCIENCE, AND THE ARTS Department of Astronomy Technical Note No. 4 ABUNDANCES OF ELEMENTS IN STARS AND NEBULAE (Determination of Ratio of Gas Pressure to Electron Pressure in Stellar Atmospheres) C. Cowley A. Cowley L. Aller ORA Project 03719 under contract with: AIR FORCE OFFICE OF SCIENTIFIC RESEARCH AIR RESEARCH AND DEVELOPMENT COMMAND CONTRACT NO. AF 49(638)-807 WASHINGTON, D.C. administered through: OFFICE OF RESEARCH ADMINISTRATION ANN ARBOR July 1962

I

ABSTRACT The ratio of gas pressure to electron pressure has been computed over a range of temperatures by an approximate method and also by a more exact IBM 709 program. The approximate method shows large errors at low temperatures and high gas pressures. Tables I through III give the values of log Pg and log Pe computed by the 709 as well as the relative abundances to which these values refer. iii

I. INTRODUCTION A knowledge of the relationship between the gas and electron pressure (Pg and Pe) is necessary before one can begin to construct a model stellar atmosphere. All sources of opacity, whether discrete or continuous, depend upon the number of atoms or particles in the line of sight which are capable of absorbing a quantum of the frequency considered. For any given over-all chemical composition, the availability of a particle for the absorption of a photon of frequency will depend upon the "state" of ionization and excitation of the particle considered. Thus in general, only a fraction of the particles will be able to produce an absorption (or emission) at a given frequency, and this fraction may be determined with the aid of the Boltzmann and Saha, or ionization equations. The purpose of the program described below is to find the relationship between the gas and electron pressures for a given chemical composition. So long as we are not concerned with the states of excitation among the various degrees of ionization, this may be done with the Saha equation alone. The Saha equation furnishes the ratio of the number of atoms which are i+l-fold ionized to the number which are i-fold ionized in terms of the partition functions of the ith and (i+l)th states of ionization and the ionization potentials Xi,i+l. In addition it is necessary to specify the electron pressure and the temperature. Stellar models are obtained by an integration of the equation of hydrostatic equilibrium which depends upon the gas pressure and not upon the electron pressure. The procedure generally followed, therefore, is to prepare a graphical relationship between the gas and electron pressures for the entire anticipated ranges of gas pressure and temperatureo The Saha equation is used in the form Ni+l 1 (2?)/ T5/2 e ~T5/2 e:kT Ni Pe h3 where the symbols have their usual meanings.* The ratio is then determined by the computer program described below. *L.H. Aller, Atmospheres of the Sun and Stars, Ronald Press,, N. Y., 1953. 1

II. METHOD OF SOLUTION BY AN IBM 709 COMPUTER PROGRAM Ionization potentials, partition functions, abundances temperatures, and electron pressures are read into the computer as data. The machine solves the Saha equation in non-logarithmic form for each element to give ratios of the number of first-ionized to neutral atoms. This ratio (RAT) for each element is divided by one plus the same ratio to yield the number of firstionized to first-ionized-plus-neutral atoms. The resulting quantity is then multiplied by the respective abundances and summed over all elements included in the program. The result is divided by the sum of the abundances, and from this quantity the ratio of the gas to electron pressure is determined by adding unity and dividing by the quantity. The Fortran statements follow: 1 DIMENSION PE(14), CHI(28), PRRT(28), T(15),ABND(28) 2 READ INPUT TAPE 7,3, (CHI(M), M=l,28) 5 FORMAT ( 6E12.3) 4 READ INPUT TAPE 7,5, (PRRT(M), M=1,28) 5 FORMAT ( 6E12.3) 6 READ INPUT TAPE 7,7, (T(N),N=1,15) 7 FORMAT ( 5F8.1) 8 READ INPUT TAPE 7,9,(PE(I),I=1,14) 9 FORMAT ( 8E9.1) 31 READ INPUT TAPE 7,32, (ABND(M),M=1,28) 32 FORMAT ( 6E12.3) 10 N=1 12 COF=3.3319E-01 13 BOLTEV = 8.616E-05 33 I=l 14 CONTINUE 34 DO 20 M=1,28 16 X=-CHI(M) /(BOLTEV*T(N)) 35 DIMENSION RAT(28),EXI(28) 17 RAT(M) =COF*T(N)**2.5*PRRT(M) *EXP (X)/PE(I) 36 EXI(M) =RAT(M)/(RAT(M) +1.0) 20 CONTINUE 37 SEAB=O.O 38 DO 24 M=1,28 39 SEAB=SEAB+( EXI(M) *ABND(M)) 24 CONTINUE 40 SAB=O.O 41 DO 29 M=l,28 42 SAB=SAB+ABND(M) 29 CONTINUE 2

43 EEE=SEAB/SAB 44 PGPE=(1. +EEE) /EEE 45 WRITE OUTPUT TAPE 6,46, PGPE,N,I 46 FORMAT(7H PGPE=1PE11.3,4H N=I2,4H I=I2) 47 I=I+1 48 IF(14-I)49,14,14 49 N=N+1 50 IF(15-N) 51,33,33 51 CONTINUE 30 END 5

III. AN APPROXIMATE SOLUTION FOR THE RATIO Pe to Pg The original computations of the Pc, Pg relations were carried out in 1959 and 1961 by hand calculations. In order to make these computations manageable, the elements of similar ionization potential were grouped together. For example, hydrogen (I.P. = 13.54), nitrogen (I.P. = 14.49) and oxygen (I.P. = 13.56) were handled together using a mean ionization potential of 13.54. Thus the solution of the ionization equation gave in this case the approximate ratio of ionized to neutral particles of H, N, and 0 combined. Five groups were considered: Group I: Helium, Neon Mean Ionization Potential = 24.48 Group II: Hydrogen, Nitrogen, Oxygen Mean Ionization Potential = 13.54 Group III: Carbon Mean Ionization Potential = 11.20 Group IV: Silicon, Iron, Magnesium, Nickel (other metals) Mean Ionization Potential = 7.9 Group V: Calcium, Aluminum, Sodium, Potassium Mean Ionization Potential = 5.8 The resulting log P., log Pg values for the various values of 9 (where ~ = 5040/T) are displayed in the Figs. 1-3 by dotted lines. By use of The University of Michigan's IBM 709 computer we were able to solve the ionization, equation exactly for each individual element, and then to determine extremely accurate log PE, log Pg relations for the same compositions which had been handled by the approximate method described above. The results of these calculations are shown in Figs. 1-3 by the solid lines and are also tabulated for each abundance. Apparently the approximate method fails most seriously at the lowest temperature and for high values of the gas pressure. 4

TABLE I SOLAR-LIKE ABUNDANICE* (IBM 709 Solution) log P log Pl log PE Q=.50 9=.60 Q=.70 @=.75 9=.80.=.85 6=.90 9=.95 -2.0 -.61.07 -1.5 -.22 +.36 1.07 -1.0 -.46 -.02. +.63 1.34 2.04 -.5 -.17 -.16 - 05. + +.91 1.62 2.33 2.98 0 +.33 +,34 +.65 1.19 1.88 2.60 3.28 3.84 +.5 +.83.87 1.48 2.15 2.87 3.57 4.19 4.58 1.0 1.34 1.45 2.41 3.12 3.84 4.51 4.99 5.26 1.5 1.84 2.14 3.38 4.11 4.81 5.38 5.72 5.94 2.0 2.37 2.96 4.36 5.08 5.72 6.16 6.43 6.66 2.5 2.94 3.89 5.355- 6.03 6.56 O50 3.62 4.86 6,32 log Pg.=1 10 =1. 1 9=1.2 9=1.4 9=1.6 9=1.8 9=2.0 -2.0.78 1.92 2.22 2.65 3.27 3.64 4.20 -1.5 1.75 2,60 2,78 3.36 3.89 4.33 5.06 -1.0 2.65 3.20 3 39 4,08 4.53 5.09 5,95. 5 346 3.79 4.06 4.76 5.22 5.95 6.88 0 4.14 4.42 4.76 5.41 5.97 6,85 7.85 +.5 4.78 5,10 5.50 6.09 6.82 1.0! 5.44 5.83 6 22 6.83 15 6.14 6.58 *As used by L, H, Aller in Stellar Atmospheres, ed. by J. L. Greenstein, University of Chicago Press, 1960, p. 234. Relative Number of Atoms H = 1,000,000. Ne = 2,000. Misc. = 2. He = 140,00oo Si = 40, "Ca = 2.4 C = 250. Fe = 6. Al = 1.6 N = 410. Mg = 19. Na = 2.0 0 = 1,o00. Ni = 0.6 K = 0.09 5

TABLE II ABUNDANCE A (IBM 709 Solution) llog Pg log e G=.53 =.4. G=.5 G=.6 G=. 7 G=.8 G=.9 -53.5 -3.0 -2.07 -2.5 -1.89 -1.23 -2.0 -1.6 -1.45 -1.44 -1.28 -.34 -1.5 -1.03 -.95 -.93 -.56 +.55 -1.0 -.70 -.70 -.48 -.45 -.39 +.28 1.41.5 -.20 -.19 +.04 +.06 +.22 1.19 2.28 0 +.30 +.33.55 +.57.94 2.08 3.13 +.5 +.80 +.83 1.06 1.11 1.79 2.98 3.89 1.0 +1.30 +1.44 1. 56 1.71 2.69 3.88 4.54 1.5 +1.80 2.00 2.07 2.43 3.61 4.72 5.17 2.0 2.31 2.54 2.60 3.27 4.54 5.46 5.82 2.5 2.82 3.05 3.20 4,18 5.45 6.14 6.53 3.0 3.35 3.56 3.90 5.12 6.29 6.83 7.30 r~I s t~~~logPg Q=1.0 =1.2 9=1.4 0 =1.7 9=2.0 =2.53 9=2.6 -.5 j-1.86 -.18 +.12 +.93 1.48 2.22 3.44 -3.0 -.99 +.48.69 1.61 2.12 3.08 4538 -2.5 -.16 1.05 1.51 2.24 2.80 5.97 5.36 -2.0 +.67 1.59 1.97 2.86 3558 4.88 6.35 -1.5 1.52 2.15 2.69 3.50 4.45 5.85 7435 -1.0 2.30 2.74 3.44 4.17 5.35 6.83 8.34 -.5 2.98 3.39 4.15 4.92 6.29 7.83 9035 0 3.58 4.09 4.82 5.77 7.26 8.82 10.34 +.5 4.17 4.85 5.50 6.69 8.25 9.82 11.35 1.0 4.80 5.60 6.23 7.63 9.24 1.5 5.48 6.32 7.05 8.62 2.0 6.23 7.05 7.95 9.60 2.5 7.02 350 7.79 Relative Number of Atoms He = 160,000. C = 1,000. (Other metals) 2. Ne = 500. Si = 32. Ca = 1.6 H = 1000000. Fe = 4 Al = 1.6 N = 125. Mg = 25. Na = 2.0 0 = 800. Ni = 0.9 K = 0.07 6

TABLE III ABUNDANCE B (IBM 709 Solution) _____log P _____________log Pg lpog ~ Q=.53 =.4 G=.5 Q=.6 G=.7 e=.8 =. 9 -3.5 -3.0 -1.42 -2.5 -1.24 -.66 -2.0 -1.51 -.83 -.81 -.63 +.07 -1.5 -.81 -.52 -.29 +.09.76 -1.0 -.70 -.69 -.09 +.19 +.26,85 1.48 -.5 -.20 -.18 +.57.69.87 1.60 2.28 0 +.30 + 0 +.5 1,14 1.21 1.59 2.32 5.06 +.5.80.94 1.67.1.75 2.36 5.10 5.77 +1.0 1.50 1.61 2.19 2.36 3.13 3.91 4.39 +1.5 1.80 2.34 2.71 3.08 5.90 4.68 5.00 +2.0 2.31 3.02 3.25 3.86 4.72 5.35 5.64 +2.5 2.83 363 35.85 4.66 5.54 5.99 6.35 +53.0 5.58 4.18 4.55 5.47 6.28 6.66 7.12 log Pg.......... 9=1.0 9=1.2 G=1.4 9=1.7 9=2.0 0=2.3 0=2.6 -3.5 -1.45 -.35 -.06.75 1.30 2.04 3.25 -5.0 -.79 +.50 +.51 1.42 1.95 2.90 4.19 -2.5 -.12.87 1.13 2.06 2.62 3.79 5.18 -2.0 +.62 1.41 1.79 2.68 5.40 4.70 6.16 -1.5 1.43 1.96 2.51 3.32 4.26 5.67 7.17 -1.0 2,17 2.56 35.25 35.99 5.17 6.65 8.16 -.5 2,82 3.21 35.96 4.74 6.11 7.65 9.17 0 5.40 5,91 4.64 5.59 7.08 8.64 10.16 +.5 3.99 4.67 5.352 6.50 8.07 9.64 11.16 +1.0 4.61 5.42 6.05 7.45 9.06 1.5 5.30 6.14 6.87 8.45 2.0 6.05 6.87 7.77 9.42 2.5 6.83 5.0 7.61 Relative Number of Atoms He = 16o0000. C = 1,000. (Other metals) 2. Ne = 500. Si = = 32. Ca = 1.6 H = 10,000. Fe = 4. Al = 1.6 N = 125. Mg = 25. Na = 2.0 0 = 800. Ni = 00.9 K = 0.07 7

LOG Pe vs. LOG Pg +3_____________________~~ @0s.50 +3~~~~~~~~~~~~/.^ ^ ^~.70 Q~~~~~~~~~' ~ ~ ~ ~ ~ i +2- / ^^ ^^^ ^^ ^^.^~=.85 +0 a,=140,000. Si 40. Co = 2.4 0* 0 co 0~ -J cc~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~r - I 0 +1 +2 +3 +4 +5 +6 LOG Pg FIG. I RELATIVE ABUNDANCE H=1,000,000. Ne= 2,000. Misc.= 2. HeI-140,000. Si = 40. Ca= 2.4 CZ 250. Fe= 6. AI = 1.6 N= 410. Mg= 19. No= 2.0 0= 1,300. Ni 0.6 K= 0.09

ABUNDANCE"A"HYDROGEN 0.1 NORMAL =0o.3 G0.5 E@ 0.6 80.7 ~=0.8~=0.9 ~ 1.0 +3 +2 +1 0 (91~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 0 -2 -2 0 -I 0 +1 +2 +3 +4 +5 +6 +7 +8 +9 LOG Pe FIG. 2 ABUNDANCE "A" He= 160,000. C= 1,000. (other metals) 2. Ne= 500. Si= 32. Ca= 1.6 H= 100,000. Fe= 4. Al= 1.6 N:= 125. Mg: 25. Na= 2.0 0= 800. Ni= 0.9 K= 0.07

ABUNDANCE "B" HYDROGEN 0.01 NORMAL 8=0.3 8=0.5 ~:0.6 ~=0.7 +3 aO LOGo Pgo +10 ~ ~ ~ ~~~~e 50 i 2 a. Q~~~~~~~Oi 0 0 -2 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 -I 0 0 +1+2 +3 +4 +5 +6 +7+8+ LOGIO Pg FIG. 3 ABUNDANCE "B" He= 160,000. C: 1,1000. (other metals)'2. Ne= 500. Si= 32. Co= 1.6 H= 10,000. Fe= 4. Al= 1.6 N= 125. Mg= 25. Na 2.0 0= 800. Ni= 00.9 K= 0.07

UNIVERSITY OF MICHIGAN 3 901 115 02693 87491111 3 9015 02693 8749

THE UNIVER S I T Y OF MI C H I G A N COLLEGE OF LITERATURE, SCIENCE, AND THE,ARTS Department of Astronomy Technical Note No,.5 ABUNDANCES OF ELEMENTS IN STARS AND NEBULAE (The Manganese Star 53.Tauri) L. H. Aller The University of Michigan Observatory and W. P. Bidelman Lick Observatory, The University of California ORA Project 03719 under contract withs AIR FORCE OFFICE OF SCIENTIFIC RESEARCH AIR RESEARCH AND DEVELOPMENT COMMAND CONTRACT NO. AF 49(638)-807 WASHINGTON, D.C. administered through: OFFICE OF RESEARCH ADMINISTRATION ANN ARBOR July 1962

PREFACE In the study of stellar chemical compositions, our attention has been focussed mainly on hydrogen-deficient objects but hydrogen deficiency is often correlated with abundance anomalies of other kinds. Conversely, if a star showing abundance anomalies does not show a hydrogen deficiency this fact is likewise of interest, since it implies that the abnormal abundance must have been produced by processes that did not at the same time seriously affect the hydrogen concentration. Granted that hydrogen is much more abundant than the metals, most nuclear processes invoked to explain abnormal abundances require mixing of surface layers with hydrogen-deficient cores-or processes in the atmospheric layers which themselves might be expected to affect seriously the concentration of hydrogen in these layers. Abundance anomalies of the type described here differ from one star to another. We are indebted to Mrs. Anne Cowley for her care in making the microphotometer tracings and for her other assistance with the redictions. Thanks are due to Miss C. Parks, and Messrs. John Dickel, Jerry Ehman and Harold Graboske for their help in the reductions. We are grateful to Dr. J. Jugaku and Dr. W. Sargent for communicating to us their spectral energy results in advance of publication. L. H. Aller The University of Michigan Observatory July 23, 1962 iii

TABLE OF CONTENTS Page LIST OF ILLUSTRATIONS vii ABSTRACT ix I. INTRODUCTION: THE OBSERVATIONAL DATA 1 II. CURVE OF GROWTH 3 III. THE ATMOSPHERIC ELECTRON PRESSURE 6 IV. DETERMINATION OF THE IONIZATION TEMPERATURE 9 V. RESULTS OF THE CURVES OF GROWTH ANALYSIS 10 A. Carbon 10 B. Magnesium 10 C. Silicon 10 D. Calcium 11 E. Titanium 11 F. Chromium 12 G. Manganese 12 H. Iron 13 I. Gallium 14 J. Strontium 14 K. Yttrium 14 L. Zirconium 15 REFERENCES 17 v

LIST OF ILLUSTRATIONS Table Page I Equivalent Widths in the Spectrum of 53 Tauri 18 II Summary of Abundances in 53 Tauri 25 Figure 1 Tracings of a portion of the spectrum of 53 Tauri. 26 2 Determination of the number of neutral hydrogen atoms in the second level. 28 3 Curve of growth for titanium. 29 4 Curve of growth for ionized calcium and ionized chromium. 30 5 Curve of growth for neutral and ionized manganese. 31 6 Curve of growth for neutral and ionized iron. 32 7 Curve of growth for CII, MgII, SiII, GaI, SrII, and YII. 33 8 Curve of growth for ZrIIo 34 9 Determination of the concentration of ionized zirconium atoms 35 Plates 1 - 7 36-42 vii

I

ABSTRACT The chemical composition of the Manganese star 53 Tauri was studied with the aid of data secured with the coude spectrograph of the 120-inch reflector at the Lick Observatory. Transition probabilities required for the problem are obtained from laboratory sources whenever possible, but some are obtained by the calibration of stellar data. The temperature of the star, deduced from ionization equilibrium, is near 11,000~K-a value in good agreement with that found from spectral energy scans by Jugaku and Sargent. The electron pressure in the atmosphere, deduced from the total intensities of hydrogen lines, appears to lie in the neighborhood of 500 dynes cm-2. Several striking abundance anomalies are found; they cannot be explained by deviations from local thermodynamic equilibrium. Manganese seems to be the most abundant metal in the star, but gallium appears to show the greatest enhancement of abundance over the "normal" value. Strontium, yttrium, and zirconium are also enhanced in abundance, whilst magnesium, and calcium seems to be depleted.. ix

I. INTRODUCTION:- THE OBSERVATIONAL DATA Spectral class A includes objects with a variety of peculiarities. Metallic line stars and magnetic stars have attracted the most attention. Of these objects perhaps the most exotic are the magnetic stars whose spectral peculiarities are linked with changes in magnetic fields. Less intensively studied have been the so-called "silicon," "strontium," and "manganese" stars. As an example of a manganese star we have selected for studys 53 Tauri [a = 4n 16m 29s (1950); 8 = + 21~ 1' 23" (1950)]. This star is also designated as Boss PGC 997, GC 5210, BD + 20~ 733, and HD 27295. Its magnitude is 5.39, Draper spectral class B8, and Mto Wilson spectral class AlS; it is a sharp-lined star. The most noticable characteristic of 53 Tauri is the great strength of the, lines of manganese, particularly those of MnII. Excess manganese abundance is sometimes associated with.other abundance anomalies as noted, for example, by Sargent aid Jugaku in'kappa Cancri.1 Other stars of peculiar composition, e.g., 3 Centauri, which has been studied by Sargent, Jugaku, and Greenstein,2 may be associated with these manganese stars. Some manganese stars also show strong chromium lines, but 53 Tauiri shows the strongest manganese line spectrum. Plates 1-7 reproduce the spectrum of 53 Tauri from about X35650 to X4670 as photographed with the 120-inch reflector coude' spectrograph at the Lick Laboratory. The strong, broad hydrogen lines are characteristic of spectral classes A or late B. Note the sharpness of the metallic lineso Only the "K" line of CaII shows a distinct fuzziness-due to collision broadening and radiation damping coupled with a large number of absorbing atoms. The richness of the MnII spectrum is particularly striking; many lines are not even listed in the Revised Multiplet Table.5 There are large numbers of lines of other ions as well, e.g., CrII, ZrII, and TiII. The FeII lines which so often dominate the enhanced metallic spectrum of an A-type star are present but are generally weaker than the MnII lines. The CII pair at 4267 and 3919, 3920 are present but weak; k3856,62; k4l28,4131 of SiII are strong. The Gal k4172.06 line is unmistakably present and measurable-indicating, as- we shall see, a remarkably high abundance for this metal (see Fig. 1). Table I gives line intensities -and other data relevant to an analysis of the spectrum of 53 Tauri. The first column gives the measured wavelength of the line. The second column gives the identification; the number in parentheses indicates the number of the multiplet in the Revised Multiplet Table.3 For many of the MnII lines no classification is available. Other MnII lines identified and classified by Laura Iglesias at Catalan's laboratory in Madrid are designated by (igl). The third column gives the equivalent width of the line in angstrons, tabulated to the nearest milliangstrom. The fourth column, denoted "Eff. 1

Cnt gives the amount by which the bakground continuum at the position of the line is affected by wings of strong lines, particularly those of hydrogen; thus a "l.0" means that there is no blending and a number such as 087, means that the continuum is depressed by 13%, i.e., the intensity in the hydrogen line is 0.87 of the continuum intensity at the point where the line is formed. The fifth column gives the quantity log W/x c/v which is used as the ordinate of the curve of growth. Here c is the velocity of light and v is the most probable speed of the atom, viz., = 2kT + 2 (1) where M is the mass of the atom, T is the gas kinetic temperature, k is the Boltzmann constant, and Et is the most probable "turbulent" velocity. The sharpness of the lines and the character of the curve of growth both suggest that turbulent velocities are small. Accordingly, for the purpose of a first reconnaissance we have assumed that 0t = 0 and have taken the gas kinetic temperature as 12,000~K. The sixth column gives log -fx, where < is the statistical weight of the lower level of the transitipn involved and f is the Landenburg f or oscillator strength. The seventh column gives the excitation potential of the lower level, and the last column gives a quantity essential for the curve of growth analysis. The oscillator strengths have been taken from various sources as will be described in the following sections. For the metallic lines we have used mostly the data obtained by Corliss at the National Bureau of Standards,5 supplemented for lines of several ions-particularly FeII-by data extracted from stellar curves of growth. 2

II. THE CURVE OF GROWTH The chemical analysis of a star may be carried out to several different degrees of precision, depending on the character of the input data, the accuracy of the observed line intensity, the reliability of the f -values, and how much we know about the star's surface gravity and effective temperature. In a first reconnaissance it is wise to adopt the curve of growth procedure which permits one to obtain a quick survey of the problem-the atmospheric pressure and temperature and a rough idea of the compositions. Later, when adequate observational material of sufficient high accuracy has been accumulated and the temperature and surface gravity of the star has been reasonably well evaluated, a model atmosphere approach may be justified. We should note, however, that the time and effort involved in a model atmosphere type of analysis is several times greater than that entailed in a curve of growth study. Several theoretical curves of growth- have been calculated by various workers. Among these we may mention especially those of Unsold,6 Menzel,7 and Wrubel,8 which have used by many workers in this field. Unsold's and Menzel's formulations were proposed originally for, the Schuster-Schwarzschild model, in which it is supposed that the continuous spectrum is produced wholly in underlying strata, whilst the line spectrum is entirely produced in a thin overlying layer. The physical model is thus a highly simplified one. Wrubel has published curves both for this Schuster-Schwarzschild model and for the Milne-Eddington model, in which it is supposed that the ratio of line to continuous absorption coefficient is independent of optical depth. That is, the line and continuous spectra are both formed in exactly the same layers. A further assumption made in the equations used by Wrubel is that the source function depends linearly on the optical depth in the continuum: %B (iT) = Bo + B1 Ty (2) where Bv is the Planckian function ) 2hv 1 (T) — 2 ehV/kT - 1 and the temperature T(TV) is a known function of the optical depth TV; dTyv = Ky P dh (4) 5

where KV is the coefficient of continuous absorption, p is the density, and dh is the element of geometrical height. The Wrubel curve of growth depends not only on the ratio of the damping constant to the Doppler width but also on the value of Bo/B1. Hence different curves will apply to different regions of the spectrum. If one wishes to use the same curve over an extended range, it is necessary to apply to the abscissa small corrections that depend not only on the wavelength but also on the intensity (equivalent width) of the line. In a first reconnaissance it is doubtful that the extra effort involved is really justified; at this stage of the analysis one may just as well use the Unsold or Menzel formulation, The more sophisticated model atmosphere treatment yields different curves of growth for different stages of ionization,9 a refinement which is hardly possible at this stage of the analysis. All of these curves of growth are based on the assumption that the lines are formed by the mechanism of pure scattering (Milne's monochromatic radiative equilibrium) If we suppose that the lines are formed by the mechanism of pure absorption, the curves are modified9 and the variation of Bo/B! has to be taken into account quite carefully.' The "scattering" curves of growth seem adequate for the first treatment of the problem. We have chosen for this analysis a curve of growth published by K. 0O WrightlO for lines of iron in the sun. It is essentially Menzel's curve of growth with small empirical corrections. The ordinate is log W/x c/v, and the abscissa is Log Xo given by:9 log X =log N- exX + log N + c 3() (5) Here 5040 9ex 54 0 (6) Tex where Tex is the excitation temperature of the atmosphere. X is the excitation potential of the lower level of the line (tabulated in the seventh column of Table I)o Na is the number of atoms "above the photosphere" at X4200A; and c () = log - log v - log u(T) mc (7) + log K(%O)/K(X). Here v is the most probable velocity, 2kT/M, u(T) is the partition function, and the first term is a numerical constant. The last term allows for the changing 4

"depth of the photosphere" as one goes from one spectral region to another. In regions where the coefficient of continuous absorption K(x) is large the atmosphere is more opaque; hence fewer atoms lie "above the photosphere." In more transparent regions K(x) is smaller and the depth of the photosphere is greater. We have chosen \o = 4200 as the standard wavelength. 5

III, THE ATMOSPHERIC ELECTRON PRESSURE Certain essential step in the curve of growth analysis must be carried out by successive approximation. We do not know the temperature of the star, nor the electron pressure in its atmosphere a priori. From the spectral type and level of excitation in the atmosphere we know that the effective temperature of the star must lie between about 10,000~K and about 13,000~K. As noted in Section IV, this range is further narrowed by spectrophotometric data secured by Sargent and Jugaku. We finally adopted T = 11,200~K. The electron pressure may be assessed in a number of ways. If the temperature were known accurately, one could determine the electron pressure from the ionization equilibrium.* Another method'is to use the hydrogen lines, The confluence of the Balmer series gives one estimate of log N. by the Inglis-Teller formula;ll another guess may be made from the total intensities of the hydrogen lines. The Inglis-Teller relation between the ion (or electron) density and the last resolvable member of the Balmer series is: log N = 23.26 - 7.5 log Nm (8) From visual inspection of the plate, Nm ~ 17, whence log N = 14i04, and log PC = 223 if T = 11,0000K. Still another method is to adapt a procedure suggested by Unsold12 to employ the total intensities of the hydrogen lines Hy and H5. The depth in the line is presumed to be given by Minnaert's empirical formula R = 1 (9) XY Rc where Rc is the central depth of the lineo In 53 Tauri, Rc= 0.84 for Hy, 0.78 for HS, and 0.76 for He; x~ is the absorption coefficient in the line, viz., *In theory, one could get both T and Pt by comparing two elements with appreciably different ionization potentials in two successive stages of ionization. In practice, however, this procedure does not work out as the relative numbers of atoms are rarely determined with the accuracy required. 6

xv = a(Al) no2 (1) where no2 is the number of hydrogen atoms in the second level above the photosphere and a(An) is the line absorption coefficient which can be written in the form13 CE.3/2 a(AN) = - [1 + (Ne,T) \ A4] (11) (Ax)/2 where Eo is the average microscopic field produced by the surrounding ions at the radiating hydrogen atom. That is, E = 2.60 N/3 (12) 0 1 Ni is the number of ions per cm3, e is the charge on the electron, and C is a constant that has been tabulated by Traving for the higher members of the Balmer series. Then, approximately, d(AX) w; = Rdx = ~ AX;/2.1 (13) CE3/2 N02 [1 + ~NAX] Rc wnire R(NeT) is a correction factor to the Holtsmark equation that has been tabulated by Kolb, Griem, and Shen.l2 For various estimates of Ni = NE = P./kT and an adopted value of no2, we calculate WX and interpolate to obtain NE. Thus in order to apply the method one must have some estimate of n2, which we obtain by a procedure proposed by Unsold.ll If the hydrogen lines were formed in an optically thin layer we would have that -W ~ 0.886 x 10-12 \2F nO2 (14) For the lower members of'the Balmer series, this equation will certainly fail because the thin-layer approximation is far from valid, while for the higher members of the series, overlapping wings will cause the continuum to be estimated too low and Wx will be too weak. Unsold's procedure is to plot no2 against the quantum number n and extrapolate to obtain a limiting value. In this fashion a value of log n02 = 16.55 is obtained (see Fig. 2). 7

observed equivalent line width log PC Hy 8.9k 2.73 HS 11.4 2.77 He 11.9 2.80: Adopted value log PC = 2.75 The value derived from HE is approximate and is not used in the final average. Note that the value of log PC obtained in this way is only approximate. We assume that the hydrogen-line profiles can be represented as though they were formed in an atmosphere at constant pressure and temperature. If we take the above-derived values of no2, PE, and the assumed T = 11,000~K we find a total quantity of hydrogen above the photosphere that is about an order of magnitude smaller than the quantity reckoned from the continuous absorption coefficient on the assumption that the photosphere lies at an optical depth T=0.5. If a larger value of n2 is adopted, the derived log Pc is reduced and the required ionization temperature is slightly lowered, but the transparency of the atmosphere is increased. Hence the discrepancy remains. If 53 Tauri lies on the main sequence, the predicted log Pc is about 2.56 for an optical depth T=0.40 at %4235. It appears that the difficulties can be resolved only by detailed line profile computations based on ad appropriate model atmosphere. We have retained log PC = 2.75 in our subsequent calculations. Perhaps a smaller value, log P. = 2.50, would have been better, but then the ionization equilibrium would have required a lower ionization temperature. Furthermore the spectrophotometric data appear to favor the higher temperature, T' 11,000~K. 8

IV. DETERMINATION OF THE IONIZATION TEMPERATURE Basically, our procedure was to estimate a rough temperature from the spectral class. Then, from the broadening of the hydrogen lines according to the Kolb-Griem theory, the electron pressure could be assessed. The next step was to improve the temperature estimate by using the ionization equilibrium of elements that appear in two stages of ionization, eg., FeI and FeII, Sil and Sill, and MnI and MnII. From the curve of growth analysis we derived log N(I) and log N(II) for each of these elements. One may compare Nr s with N r-ls to obtain a combined excitation and ionization temperature. Such a situation occurs when lines arising from low levels in a neutral atom are compared with lines arising from high levels in the corresponding ion. The ionization or ionization-excitation temperatures derived in this way for Mn, Fe, and Si all agreed'in giving a temperature near 11,200~K, 9 = 0.45. This value was used in subsequent calculations. Later we learned that' Sargent and Jugaku had measured the energy distribution in the spectrum of 53 Tauri. They found that the energy distribution is very close to that of a Leonis; hence our ionization temperature of 11,000 seems reasonable. A more elaborate procedure would entail representing this energy distribution by means of a model atmosphere, which could then be employed to predict spectral line intensities by the Pecker theory or some more elaborate procedure. The hydrogen line profiles could be.represented by the model atmosphere and the KolbGriem theory to fix the proper value of the surface gravity, and to check the run of temperature and density in the adopted model. 9

V. RESULTS OF THE CURVE OF GROWTH ANALYSIS We shall now present results for the individual elements whose abundances have been determined by the curve of growth method, Throughout, we have assumed (ex = Gion - 0,450 A. CARBON Multiplets (4) and (6) are represented in the spectrum of 53 Tauri, The fvalues are calculated by the Bates-Damgaardl4 method, which seems to yield reliable results for light atoms of this type. We obtain for the number of atoms above the photosphere log N(CII) = 19.44 after allowing for the small contribution of neutral carbon we find log N(C) = 19.45 B. MAGNESIUM MgI is represented by very weak lines 3829, 53838, whose intensities are uncertain; MgII is represented by the strong 4481 pair. With our adopted 0 = 0.45, we obtain log N(MgI) = 12.00 log N(MgII) = 16.60 The corresponding ionization temperature would be implausibly high; therefore we have adopted the abundance from the MgII lines which yield log N(Mg) = 17.19 The corresponding value from MgI would be 16.40; evidently further study of this element is needed, possibly by means of a model atmosphere approach. C. SILICON SiI is represented by 23905, from whose intensity we derive a concentration log N(SiI) = 13591 Ionized silicon is represented by lines of multiplets (1) and (3). The — values were taken from the Kiel work; we obtain 10

multiplet log N-OX OX log N(SiII) (1) 14.12 5.08 17.20 (5) 12.74 4.41 17.15 mean log N(SiII) = 17.19 The N(SiII)/N(SiI) ratio yields an ionization temperature near 11,000~K, which is in good agreement with that obtained from other ratios. The finally adopted abundance of silicon is log N(Si) = 1722: see however the note on p. 16. D. CALCIUM We observe only the lines of ionized calcium. multiplet log N-OX OX log N(CaII) (1) 13.83 0.00 13.83 (3) 12.36 1.40 13.80 The K line gives a result which is in excellent accord with that found from other, weaker lines of ionized calcium. Most of the calcium is doubly ionized. log N(Ca) = 15.08 E. TITANIUM Numerous lines of TiII are observed in the spectrum of 5. Tauri. They arise from levels with excitation potentials between 0.59 ev and 5.11 ev. For each excitation potential X, we plot log W/\ c/v against C3 + log Aft and fit each plot of points to the theoretical curve in order to obtain log N - OX. number of number of X lines log N-OX X lines log N-OX 0.599 9 14.76 1.57 8 14.12 1.08 11 14.56 1.88 2 13.84 1.11 7 14.56 2.04 2 14.74 1.16 9 15.04 2.05 1 14.44 1.23 10 15.14 2.58 5 13.44 2.11 2 13.0: From a plot of log N-OX against X we obtain log N(TiII) by fitting a line with a slope G = 0.45. The points derived from a small number of lines show a consiberable scatter. The adopted concentration of TiII is 11

log N(TiII) = 15.24 Most of the titanium atoms are doubly ionized. No lines of TiI are observed. After correction for ionization equilibrium the abundance of titanium is found to be log N(Ti) = 15.95 The curve of growth for titanium is given in Fig. 3. F. CHROMIUM This metal is represented by lines of CrII. The F-values (taken from data given in Technical Report No. 3 of this series) are obtained from stellar line intensities calibrated with the aid of Croliss p-value measurements secured at the National Bureau of Standards. The scatter from some of the highlevel lines may be at least partially attributed to inaccuracies in these empirical+ -values. number of number of X lines log N-9X X lines log N-9X 2.69 3 12.82 4.04 7 13.59 3.09 3 14.09 5.50 2 11.85 3.81 1 12.82 5.64 1 11.42 3.84 5 13.87 6.46 4 11.32 From a plot of log N - GX against X represented by a line of slope 9 = 0.45 there results: log N(CrII) = 14,68 Allwoing for the effects of ionization, We find log N(Cr) = 14.91 This value is somewhat uncertain because'of the scatter of the log N - GX plot. The curve of growth for ionized calcium and ionized chromium is given in Fig. 4. G, MANGANESE This appears to be the. most abundant metal in the star. Lines of both MnI and MnI are present. Laboratory transition probabilities are available for the lines of MnI but unfortunately very little -'-value data are available for the lines of ionized manganese. For this ion we have had to use empirical solar f-values which suffer from two very grave defects: (1) since the lines are all very weak 12

and may be affected by blends, the so-determined line strengths are highly uncertain; and (2) there may be a zero-point error due to a wrong estimate of the ionized manganese concentration in the sun-or an error in the excitation temperature. For MnI we obtain the following results: number of X lines log N-GX 0.00 3 13.57 2.13 11 12.04 log N(MnI) = 13.35 while we find that the concentration of the MnII ion may best be found from the lines arising from the levels at 1.82 ev, viz., log N(MnII) = 16.75 There are also a number of MnII lines arising from levels at 5.36 ev. These gave a ridiculously small abundance of MnII. Presumably the empirical t4-values were at fault; probably too small on excitation temperature had been assumed for the sun. Adopting these concentrations of MnI and MnII we can solve for the ionization equilibrium. The resultant ionization temperature lies near that found from silicon and from iron. The finally adopted abundance of manganese is based mostly on the MnI data. It is log N = 1730. The curve of growth for manganese is given in Fig. 5. H. IRON There are three lines of FeI from multiplets (20) and (41) at X = 0.86 eV, and (43) at X = 1.48 eV. They yield an FeI concentration of log N(FeI) = 12.90. The FeII data are more extensive. Using the empirical log c\k values obtained from solar and stellar data we find number of X lines log N-OX 2.66 6 15..24 2.81 8 15.48 5.90 2 14.20: whence we find log N(FeII) = 16.62 13

From the ratio log N(FeII)/N(FeI) we obtain an ionization temperature of 11,200~K, which is in good agreement with that' found from manganese and silicon: log N(Fe) = 16.75 The curve of growth for iron is given in Fig. 6. I. GALLIUM Perhaps the most remarkable feature of the spectrum of 53 Tauri is the presence of neutral gallium.only one line is suitable for measurement and since it is perturbed by a neighboring, stronger, line there is some uncertainity. We get log N(GaI) < 11.65 while after correction for ionization effects we find log N(Ga) < 15.3 The curves of growth for GaI, as well as for CII, MgII, SiII, SrII, and YII, are given in Fig. 7. J. STRONTIUM Strontium is represented by the X4077,4215 lines of SrII, from which we obtain log N(SrII) = 12.98 most of the strontium is doubly ionized so we get log N(Si) = 14.62 K. YTTRIUM From four lines of ionized YII for which f-values have been measured by Corliss we find log N(YII) = 12.89 log N(Y) = 14.16 14

L. ZIRCONIUM Zirconium is represented by a fair number of lines of ZrII for all of which N.B.S. laboratory log v —values are available. number of X lines log N-OX 0.52 4 13.91 0.72 11 13.82 1.22 3 13.68 The curve of growth for ZrII is given in Fig. 8; the extrapolation to log N(ZrII) is given in Fig. 9. We obtain log N(ZrII) = 14.16. Correction for the ionization of this metal leads to a final value log N(Zr) = 14.62 Table II summarizes the abundances in 53 Tauri, as compared with a so-called "normal" scale of abundances derived for the solar system.15 The scale is nor-malized for hydrogen on the assumption that the proportion of this element is the same in the sun and in 53 Tauri. We could equally well have used iron; doing so would not have changed the essential conclusions. Carbon appears to be more abundant in 53 Tauri than in the sun by a factor of slightly more than two:, but this result could be modified by a slight change in the temperature. If the temperature is slightly higher, the abundance difference is eradicated. Magnesium and siliconmay-be about;, four:;or.five times less abundant than in the sun, and calcium is about thirty times less abundant. If the ionization temperature is higher, this difference could be reduced somewhat but probably it could not be eliminated, at least for calcium. Titanium appears to be about four times more abundant in 53 Tauri than in the sun, but the scatter of the individual determinations is large. Chromium is about eight times less abundant in 53 Tauri. Here again the scatter of individual determinations is fairly large, but not large enough to account for the difference. An appreciable depletion of chromium must exist. Manganese is nearly sixty times as abundant as in the sun; in fact it appears to be more abundant than any other metal. It is three times as abundant as iron, which may be slightly less abundant in 53 Tauri than in the sun. Gallium presents a special problem. The presence of lines of GaI in a star as hot as 53 Tauri indicates that this element must be very abundant-probably at least a hundred times as abundant as in the sun. Strontium yttrium, and 15

zirconium are appreciably more abundant than in the sun, by factors of thirty, twenty, and fifty respectively, It is difficult to explain these abundance anomalies in terms of nucleogenesis patterns. Elements of the iron group are presumably produced by equilibrium reactions. Then iron is always the most abundant, and it is hard to see how manganese can be so prominent. Before pursuing this matter further it will be well to obtain abundance data for additional elements, Sargent and Jugaku find oxygen to be normal but it would be valuable to obtain data on nitrogen and sodium also. Since the preceding was written Wallace Sargent has informed us of his and Searle's conclusion that the abundance of silicon in 553 Tauri is probably not much depleted since the "equivalent width of 4130 of Sill is entirely normal." Magnesium may be deficient but a factor of five may be too much. With respect to iron, silicon and magnesium are depleted by factors between two and three according to the results presented here. The Sill lines have fairly high excitation potentials. A slight error in the excitation temperature of ionized silicon could remove the discordance with Sargent's conclusion; possibly application of model atmosphere methods will remove the discrepancy entirely. 16

REFERENCES 1. W. Sargent and J. Jugaku, Astrophysical Journal, (in press, 1962). 2. W. Sargent, J. Jugaku, and J. L. Greenstein, Astrophysical Journal, Vol. 154, 1961, p. 7853 5. C. E. Moore, Revised Multiplet Table, Princeton University Contribution Contribution No. 22, 1945. 4. L. Iglesias, Journal of the Optical Society of America, Vol. 46, 1956, p. 449; Vol. 47, 1957, p. 852. 5. C. Corliss, National Bureau of Standards Monograph (in press). 6. A. Unsold, Physik der Sternatmospharen, Springer-Verlag, Berlin, 1955. 7. D. H. Menzel, Astrophysical Journal, Vol. 84, 1957, p. 462. 8. M. Wrubel, Astrophysical Journal, Vol. 109, 1949, p. 66; Vol. 111, 1950, p. 157. 9. See e.g., Chapter 4 of Stellar Atmospheres, Compendium on Stars and Stellar Systems, Vol. 6, ed. by J. L. Greenstein, University of Chicago Press, 1960. 10. K. 0. Wright, Astrophysical Journal, Vol. 99, p. 249, 1944. 11. D. R. Inglis and E. Teller, Astrophysical Journal, Vol. 90, 1959, p. 439. 12. A. Unsold, Zeitschrift fur Astrophysik, Vol. 21, 1941, p. 37. 13. A. C. Kolb, H. Griem, and K. Y. Shen, Physical Review, Vol. 116, 1949, p.4. 14. D. R. Bates and A. Damgaard, Philosophical Transactions, Royal Society of London, Ser. A, No. 242, 1949, p. 101 15. L. H. Aller, Abundances of the Elements, Interscience, New York, 1961, p. 192. 17

TABLE I EQUIVALENT WIDTHS IN THE SPECTRUM OF 53 TAURI 2'.D I. D.Con log W/x c/v log x C 3+log W Cont. 3 C 3677.69 CrII(12).026 1.0 +.037 -4.624 2.69 -12.97 3677.91 CrII(12).030 1.0 +.090 -4.444 2.69 -12.79 3679.71 TiII(75).016 1.0 -.206 -5.70 1.57 -14.67 3685.07 MnII(8).025 1.0 NA 3685.22 TiII(14).074 1.0 +.468 -4. 083 0.59 -13.06 3706.03 CaII(3).035 0.92 +.109 -4.651 3.11 -12.33 3706,24 TiII(73).027 0.95 +.034 -4.481 1.56 -13.47 3706.89 MnII(8).013 0.97 -.258 -3.96 5.37 -12.00 3708.05 MnII(8).012 1.0 -.311 -2.93 5.37 -10.97 3712.95 CrII(12).033 0.84 +.139 -4.330 2.69 -12.71 3715.40 CrII(145) o016 0.97 -.175 4.91 3729.50 MnII(8).066 0.o 987 +.447 -4.33 5.37 -12 38 3736.93 CaII(3).052 0.76 +.269 -4.537 3.14 -12.22 3738.38 CrII(20).022 0.87 -.038 -5.67 3.09 -14.05 3741.65 TiII(72).070 1.0 +.438 -4.117 1.57 -13.11 MnII 37453.8 M38 n.052 1.0 NA 3.09 (no class) 3745.93 ZrII(112).017 0.75 -.030 -3.616 1.75 -12.53 3748.02 TiII(107).061 0.93 +.381 2.59 5754.59 CrII(20).026 0.82 +.018 3.09 3755.21 (l.023 o.84 NA (no class) 3757.68 TiII(72).063 1.0 +.392 -4.505 1.56 -13o50 3759.29 TiII(13) o100 1.0 +.601 -4.085 0.60 -13.08 3761.32 TiII(13).095 1.0 +.570 -4.185 0.57 -13.18 3761.87 TiII(107).044 1.0 +.232 -5.975 2.58 -12.97 3763574 MII.068 0.900 NA (no class) 3765.61 CrII(20).009 0.78 -.432 -5.94 3 09 -13.53 3774.35 YII(7).013 0.64 -.147 -4.583 0.13 -12.92 3776.05 TiII(72).039 1.0 *+.183 -5.183 1.57 -14.20 3778.32 MnII.043..90 NA (no class) 3783.89 MnII (igl).016 1.0 NA 37863.5 TiII(12).016, 0.99 -.212 0.60 3788.71 YII(7).010. 0.87 -.277 -4.762 0.10 -13.12 3806.71 MnI(6).022 0.86 -.038 -3.45 2.11 -11.74 3812.25 MnII (igl).4 1.0 NA 3812.53 MnII (igl).012 1.0 NA 3813.39 TiII(12).037 loO +.156 -5.25 0.60 -14.27 5814.59 TiII(12).051 1.0 -.706 -5.41 0.57 -14.43 18

TABLE I (Continued) TI. D. w EfC. log W/x c/v log F X C +log f _. _____________ Cont.3,.. 86 -.... 3820.42 FeI(20).015 0.97 -.202 -.99 0.86 -12.88 3823.51 MnI(6).017 0.94 -.160 -3.61 2.13 -11.91 3823.87 MnI(6).011 0.93 -.328 -4.32 2.15 -12.62 3825.02 MnII (igl).022 0.89 NA 3829.37 MgI().013 0.69 -.446 -4.04 2.70 -11.49 3832.94 YII(7).020 0.51 +.009 -4.976 0.18 -13.35 3844.17 MnII (igl).053 0.82 NA 3848.24 MgII(5).011 0.94 NA 3848.62 MnII (igl).013 0.96 NA 3849.57 NiII(11).018 0.98 -.115 4.o01 3850038 MnlI(5).013 1.0 NA 3853.67 SiII(l).062 1,0 +.255 -6.11 6.83 -14o04 3856,01 SilI(1).110 1.0 +.506 -5.16 6.83 -13.10 3862.59 SiII(l).089 1.0 +.415 -5.38 6.83 -13.31 5863.44 MnII (igl) 032 1.0 NA 3865 60 CrII(167),029 0.029 +. 66 5.30 3879.00 MnII (igl).028 0.80 NA 3882.30 TiII(34).029 0.67 +.036 -4.93 1.11 -13.98 3897.60 MnII (igl).036 0,76 NA 5898.09 MnII (igl) 041 0.78 NA 5900.55 TiII(34).064 0.67 +.380 -4.429 1.13 -13.48 3902.40 MnII (igl).010 0.91 NA 3905.48 SiI(3).023 0.96 -.174 -5.118 1.90 -13.37 3905,.67 CrII(167).027 0.96 +.031 5.31 3913.48 TiII(34).094 1.0 +.548 -4.467 1.11 -13.52 3915.97 ZrII(17).019 1.0 -.011 -5.297.52 -14.27 3917.33 MnII (igl).039 1.0 NA' 3918.95 CII(4).010 1.0 -.722 -4.67 16.26 -12.80 3920.66 CIT (4).022 1.0 -.686 -4.98 16.26 -13.11 3926.13 MnII (igl) 015 1.0 NA 3926.45 MnI(44).031 1.0 +.089 -3.91 3.83 -12.24 3930.99 MnII (igl).16 1.0 NA 3932.01 TiII(34).047 1.0 +.244 - 5.605 1.13 -14.67 3934.79 ZrII(43).0098 1.0 -.296 -5.515 0.71 -14.30 3938.98 FeII(190).0051 1.0 -.686 -5.87 5.89 -14.81 3941.23 MnII (igl).024 1.0 NA 3943.60 MII (igl),.019 1.0 NA 3950,37 YII(6).010 0.98 -.294 -55113 0.10 -13552 3952.43 MnaI (igl) -.009 0.94 NA 3958.22 ZrII(16).018 0.820 -.040 -4.832 0.52 -15.82 3967.95 MnlI (igl).009 0.40 NA 3968.46 CaIl (none).12 0-370 +.613 -5.371 0.00 -13.13 19

TABLE I (Continued) X Iw.D. W Eff log W/x c/v log x X C +log "~-~.D. W Cont.3 3979.53 CrI(183).017~ 0.76 -.190 5.65 3982.01 TiII(11).014 o.840 -.275 -6.28 0.57 -15.36 3986.60 MnII (igl).013 0.91 NA 3987.60 TiII(11).013 0.92 -.330 -6.99 0.60 -16.07 3991 16 ZrII(30).027 0.96 + 132 -4.699 0.75 -13.69 3995.30 MnII (igl).015 1.0 NA 3998.97 ZrII(16).020 1.0 +.014 -4.898 0.56 -13589 4000.04 MnI (igl).026 1.0 NA 4003530 CrII (194).021 1.0 -.089 -3.80 6.46 -12.27 4012.40 TiII(11).052 1.0 +.282 -5.877 0.57 -14.96 4012.55 CrII(183).014 1.0 -.260 -3.38 5.64 -11.85 4018.07 MnI(5).011 1.0 -.370 -4.05 2.11 -12.40 3838.28 MgI(3).020 1.0 -.256 -3.49 2.18 -10.95 5933.67 CaII(1) 1.0 +1.073 -3.075 0.00 -10.80 4018.07 MnI(5).016 1.0 -.197 -4.046 2.105 -12.41 4025.15 TiII(11).035 1.0 +.102 -6.085 0.60 -15.18 4028,35 TiII(87).051, 1.0 +.272 -4.815 1.88 -13.91 4029.70 ZrII(41).014 1.0 -.139 -5.045 0.71 -14.05 4030.77 MnI(2).042 1.0 +.215 -4.875 0.00 -13.24 4053307 MnI(2).042 1.'0 +.210 -5.035 0.00 -13540 4034.52 MnI(3).026 1.0 +.001 -5.275 0.00 -13.64 4035.73 MnI(5) 016 1.0 -.197 -4.025 2.13 -12.39 4038.00 CrII(194).022 1.0 -.071 -336 6.46 11.83 4041.39 MnI(5).034 1. o +.116 -3.466 2.11 -11.83 4045.6o0 ZrII(3).018 1.0 -.049 -5.055 0.71 -14.o06 404.5.83 FeI(43).013 1.0 -. 06 -3.673 1.48 -12.62 4048.71 ZrII(4).035 1.0 +.138 -4.14 2.15 -12.50 Mnl(5) 4049.13 CrII(193).16 1.0 -.207 -2.59 6.46 -11.06 4050.32 ZrII(43).007 1.0 -.460 -5.353 0.71 -14.37 4051.97 CrII(19) o018 1.0 -.173 -6.25 3.09 -14.73 4053083 TiII(87).052 1.0 +.272 -4.28 1.88 -13538 4054.09 CrII(19).011 1.0 -.390 3.09 4055056 MnI(5).023 1.0 -.059 -3.922 2.13 -12.29 4058.94 MnI(5).010 1.0 -.429 -4.132 2.17 -12.50 4o64.36 TiII(106) 013 1.0 -.331 2.59 40670o6 NiII(ll).009' 1.0 -.419 4.01 4070.86 CrII(193).01 1.0 -.429 -3.41 6.46 -11.89 4075.46 SiII (none).020 1.0 -.265 4076.78 SiII (none).016 1.0 -.363 4077.71 piII(1).059 1,0 +.277 -5.170 0.00 -12.76 4079.24 MnI(5).014 1.0 -.256 -4.290 2,18 -12.66 20

TABLE I (Continued) AI.D. W Ct log W/x c/v ogC+logX X C3+1 f Cont. 4081.45 MnII (igl).039 1.0 NA 4083.65 MnII(2) o014 0.942 -.277 -4.149 2.15 -12.52 4085.41 MnII (igl).022 1.0 NA 4110.62 MnII (igl).023 0.764 NA 3.094111.00 CrII(1826).018 0.79 -.167 -5.87 -14.36 412808 SiII) 0.12 1.0 +.655 -4.16 1.77 -12.16 MnII(2) 4129.12 SiII(1).012 1.0 -.479 4130.87 SiII(3).095 1.0 +.413 -3.98 9.80 -11.98 4136.91 MnII (igl).060 1.0 NA 4140.47 MnII (igl) o016 1.0 NA 4145.78 CrII(162).023 1.0 -.067 -3.31 5.30 -11.81 4149.22 ZrII(41).026 1.0 +.102 -4.512 0.80 -13.55 ZrII(42) 4150.98 ZrI().012 1.0 -.357 -5.402 5.31 -13590 CrII(163) 4156.24 ZrII(29).011 1.0 -.274 -5.231 0.71 -14.27 4158.27 MnII (igl).012 1.0 NA 4161,19 ZrII(42) 018 1.0 -.051 -5.091 0.71 -14.13 4161,53 TiII(21).028 1.0 -.011 -5.58 1.08 -14.70 4163.65 TiII(105) 078 1.0 +.441 -3.931 2.58 -13505 4171.05 MnII (igl).016 1.0 NA 4171.53 MnII (igl) o014 1.0 NA 4171092 TiII(105).064 1.0 +.355 -4.160 2.59 -13.28 4172.06 GaI(1).016 1.0-.156 -4.650 o010 -12.51 4173.45 FeII(27).010 1.0 -.442 2.57 4175.52 TiII(21) o.033 1.0 +.o66 1o08 4174.10 TiII(105).019 1.0 -.182 -6.76 2.59 -15.88 4174.34 MnII(2).037 1.0 NA MnII (igi) 4177.524).027 1.0 NA YII(14) 4178.89 FeII(28).025 1.0 -.022 -6.45 2.57 -15.45 4179.45 CrII(26).022 1.0 -. 098 -4 35 3.81 -12.86 4184.29 TiII(21).011 1.0 -.430 -6.02 1.08 -15.15 4184,47 MnII (igl).020 1.0 NA 4200.28 MnII (igl).027 1.0 NA 4205.39 MnII(2) 051 1.0 +.278 -7.40 1.80 -15.58 4206.39 II(7).078 -l.o +.464 -3.26 5.37 -11.44 4207.22 - MnII(2).007 1.0 -.553 -8. 74 1.82 -16.92 4209.01 ZrII(41).009 1. 0 -.348 -4i916 0.71 -13597 4211.89 ZrII(15) o 0096 1.0 -.55336 -5586 0.52 -14.64 4215.54 SrII(1).024 1.0 +.o 048 - 5.565 0.0 0 -12.98 21

TABLE I (Continued) ^I<ID.W CEnt ff log W/x c/v log FX X C +log f Cont.o3 9 4233.19 FeII(27) 046 1o0 +.232 -6.11 2.57 -15.13 4235.33 MnI(23).015 1.0 -.260 -3.794 2.88 -12.21 4237.85 MnII(7) 014 1.0 NA 4238,81 MnII(2).029 1.0 +.032 -7.74 1.82 -15.93 4239.21 MnII(7).0o 4 1.0 NA 4240. 39 MnII (igl).019 1.0 NA 4242.96 MnII(7) 016 1.0 NA 4244.27 MnII(7).031 1.0 +. 060 -4.41 5.35 -12.60 4247.97 MnII (igl).032 1.0 NA 4251.74 MnII (igl).035 1.0 NA 4252.62 CrII(31) o016 1.0 -.234 -5.60 384 -14.12 4252.99 MnII(7).052 1.0 NA 4253~16 MnII (igl).019 1.0 -.158 -3.65 5.36 -11.84 4259.20 MnII(7).065 1.,0 NA 4260.49 MnII(2).012 1.0 NA 4261.92 CrII(31).o040 1.0 +.153 -5.06 3.85 -13.59 4266.98 CII(6) o014 1l.o -.620 -4.09 17.97 -12.29 4267.26 CII(6) 016 l0-.o564 -3.63 17.97 -11.83 4269.27 CrII(31).014 1. -.290 -5.88 3.84 -14.41 4275 58 CrII(31).050 1.0 +.030 -5.40 3.84 -13,93 4275.89 MnII (igl).010 1.0 NA 4278.61 MnI(6).024 1.0 NA 4284.18 CrII(31).030 1.0.032 -5.48 3.84 -14.01 4284,44 MnlIl(6).031 1.0 +.062 -3.88 5.35 -12.09 4287,89 TiII(20) o041 l.o +.143 -5.72 1.08 -14.88 4288.07 MnII (igl).024 1.0 NA 4289.60 MnI(6) 003 1.0 NA 4290.23 TiII(41) o083 1.0 +.456 -5.158 1.16 -14.31 4292.23 MnlI(6).065, 1.0 NA 4294.11 TiII(20).072 1.0 +.5390 -5.157 1.08 -14o31 4300.o6 TilI(91).075 1.0 +.410 -4,917 1.18 -14.07 4300.27 MnII(6).024 1.0 -.058 -3.44 5.36 -11.65 4301.95 TiII(41).061 1.0 + 321 -5.316 1.16 -14,47 4303 518 FeII(27).020 1.0 -.128 -6.70 2.69 -15.73 4307,89 TiII(41).056 1.0 +.284 -4.986 1.16 -14.14 4308.18 MnII (igl).027 1.0 NA 4312 88 TiII( 41).063'10 +.333 -5.305 1.18 -14.46 4414.99 TiiI(41) o064 1,o +.336 -5.02 1.16 -14.18 4351682.0I (18 1.0 -.223 NA Ti i(4)4) 4317.77 InII (igl).009 1.0 NA 4320.99 TiII(41) o.02 1,0 +.032 -6.17 1.16 -15.33 22

TABLE I (Continued) I.D. W Cn. low/ / lolog f X C3+log 4325.07 MnII(6).018 0.97 NA 4330.24 TiI(94).021 o0864 -.173 -5.94 2.o4 -15.10 4330.70 TiII(41).020 0.86 -.173 1.18 4337.93 TiII(20).032 0.48 +.041 -5-143 1.08 -14.30 4344.01 MnII(6).050 0.53 +.059 -3.47 5.37 -11.68 4344.27 TiII(20).011 0.55 -.422 -5.952 1.08 -15.11 4345.64 MnII(6).014 0.64 -.286 -4.35 5.37 -12.56 4346.40 MnII (igl).016 0.71 -.248 NA 4348.42 MnII (igl).055 0.78 +.097 NA 4350.81 TiII(94).013 0.86 -.355 -5.46 2.05 -14.63 4351.77 FeII(27).017 0.88 -.220 -6.17 2.69 -15.21 4356.63 MnII (igl).039 1.0 +.148 NA 4359,75 ZrlI(79).011 1.0 -.302 -4.761 1.23 -13.85 4363.27 MnII (igl).025 1.0 -.042 NA 4365.24 MnI (igl).029 1.0' +.025 NA 4567.66 TiII(104).o41 1.0 +.135 -4 o00 2.58 -13.17 4370.99 ZrII(79).018 1.0 -.084 -5.130 1.20 -14.22 4377.77 MnIT (igl).021 1.0 -..116 NA 4383.60 FeI(41).006 1.0 -.649 -4.01 1.48 -13.04 4384.64 MnII(10).017 1.0 -.217 NA 4385,41 FeII(27).012 1.0 -. 380 -6.67 2.59 -15.72 4386.88 TiII(104).034 1.0 +.055 -5.08 2.59 -14.25 4390 59 MnTI(10).031 1.0 +. 48 NA 4391.03 TiII( 41).020 1.0 -.166 4393.39 MnII (igl).024 1.0 -.075 NA.4594.07 TiII(51).035 1.0 +.062 -5.87 1.22 -15.04 4395.04 TiII(19).092 1.0 +.487 -4.817 1.08 -13.99 4395.85 TiII(61).036 1 0 +.082 -5.34 1.24 -14.51 4399.78 TiII(51).058 1.0 +.290 -5.417 1.23 -14.59 4403 52 MnII (igl).081 1.0 +.461 NA 4409.51 TiII(61).016 1.0 -.281 -6.93 1.23 -16.11 4411.08 TiII(115).09 1.0 +.111 3.08 4411.90 TiII(61).008 1.0 -.586 -6.65 1.24 -15.83 4416.80 FeII(27).011 1.0 -.412 -6.75 2.77 -15.80 4417.72 TiII(40).057 1.0 +.279 -5.485 1.16 -14.66 4418.34 TiII(51).029 1.0 -.022 -6.19 1.23 -15.37 4421.94 TiII(93).021, 1.0 -.163 -5.88 2.05 -15.06 4434.06 MnII (igl).035 1.0 NA 4441.73 TiII(40).016 1.0 -.277 1.18 4441.99 MnII (igl).009 1.0 NA 4443.02 ZrII(88).067 1.0 +.486 -4.842 1.48 -13.95 4443.80 TiII(19).088 1.0 +.464 -5.002 1.08 -14.19 23

TABLE I (Continued) \ ZTI.D. W Cnt. log W/x c/v log,pX X C3+logofX,Cont. 3 444454 TiII(31).021 1.0 -.155 -6.69 1,11 -15.87 4450.48 TiII(19).047 1.0 +.194 -5.882 1.08 -15.07 4456,62 TiII(115) 009 1.0 - 541 -410 3.11 -13.29 4462.03 MnI(28) 016 1.0 -.238 -3.67 3.06 -12.13 4464.45 TiII(40).042 1,0 +.137 -6.33 1,16 -15.52 4468.51 TiII(31).095 1,0 ++494 -4.93 1.13 -14.12 4469.13 TiII(18),009 1.0 +.485 -6.74 1.o8 -15.93 4470.85 TiII(40).019 1.0 -.200 1.16 4478.63 MnII (igl) o056 1.0 +.293 NA 4481.14 MgII(4).15 1.0 +.556 -3.87 8.83 -11.51 4481.35 MgII(4).14 1.0 +.514 -3.64 8.83 -11.28 4488.33 TiII(115).059 1.0 +.284 -4.068 3.11 -13.26 4489.19 FeII(37).013 1.0 -.347 -7.04 2.82 -16.11 4491.40 FeTI(37).017 1.0, -.215 -6.93 2.84 -16.oo 4496.98 Zrii(40).007 1.0 -.488 -5.477 0.71 -14.59 4501.28 TiII(31).092 i.0 +.477 -5.027 1.11 -14.23 450318 MnII (igl).018 1.O NA 4508,28 FeII(37).027 1.0 -.022 -6.52 2.84 -15.60 4515.33 FeII(37).012 1.0 -.390 2.83 4518.34 TiII(18).012 1.0 -.409 -6.84 1.08 -16.04 4518 94 MnII (igl). 037 1.0 NA 4519.23 MnII (igl).014 1.0 NA 4520.22 FeIT(37).023 1.0 -.102 -6.69 2.79 -15.77 4522.62 FeII(38).040 1.0 +.149 -6.30 2.83 -15.38 4525 32 MnIIIT (igl).021 1.0 NA 4529.48 TiII(82).039 1.0 +.100 -6.11 1.56 -15.31 4533.95 TiII(50).093 1.0 +.480 -4.844 1.23 -14.04 4544.o1 TiII(60).009 1.0 -.556 -6.86 1.24 -16.07 4549.49 FeII(38).028 1.0 -.008 2.82 4549.62 TiII(82).090 1.O +.465 -4.502 1.58 -13.71 4554.99 CrII(44).035 1.0 +.072 -5.15 4.05 -13.74 4555.89 FeII(37).015 1.0 -.294 -6.45 2.82 -15.53 4558.64 CrlI(44).061 1.o +.314 -4.221 4.06 -12.81 4563.77 TiI(50).084 1.0 +.430 -5.211 1.22 -14.42 4565.74 CrII(39).018 1.0 -.213 -5.63 4.02 -14.22 4568.33 TiII(60).007 1.0 -.661 -7.17 1.22 -16.38 4571.96 TiII(82) o084 1.0 +.433 -4.520 1.56 -13.73 4583.82 FeII(38).035 1.0 +.089 -6.05 2.79 -15.14 4588.20 CrTI(44).062 1.0 +.315 -4.47 4.05 -13.07 4589.95 TiII(50).028 1.0 -.054 -5.848 1.23 -15.06 4591.08 CrII(44).030 1.0 4616.65 CrIl(44).038 1.0 +.096 -3.11 4.05 -11.71 24

TABLE I (Concluded) Pf f X I.D. W C log W/X c/v loggfX X C3+log aX Conit. 4618.81 CrII(44).053 1.0 +.242 -4.90.4.o6 -13.50 4629.32 FeII(37).024 1.0 -.090 -6.54 2.79 -15.63 4634.07 CrII(44).043 1.0 +.151 -4.82 4.05 -13.42 4635.34 FeII(186).019 1.0 -.186 -5 64 5.93 -14.74 4657.22~ TiII(59).022 1.0 -.159 1.24 4727.85 MnII(5).050 1.0 NA 5.35 4730.39 MnII(5).044 1.0 NA 5.35 4738.31 MnII(5).049 1.0 NA 5.36 4755.71 MnII(5) o084 1.0 NA 5.37 4283.78 MnII(6).028 1.0 NA..5.55 TABLE II SUMMARY OF ABUNDANCES IN 53 TAURI log N log Na Elementlog A log N 53 Tau "normal" lg H 22.42 22.42 C 19.45 19.02 i+0.43 Mg* 17.19(?) 17.82 -0.63 Si* 17.22(?) 17.92 -0.80 Ca 15.08 16,61 -1-55 Ti 15.95 15o31 +o.64 Cr 14.91 15,80 -o.89 Mn 17.50 15.54 +1.76 Fe 16 75 17.00 -0.25 Ga <1-53 12.87 <2,4 Sr 14 62 15312 o150 Y 14o16 12,87 1.29 Zr 14062 12 92 1.70 *see note on p 16. 25

9a _ 4024.52 Fe - 4025. 15 Ti I yf (qJ X ^-5 —r- 4028. 32 TIt i-J. Hn 0 - 4029.70 Zr ] i - 4030. 77 Mnl 0 0 r-b h- 4033.07 Mn' 0 -4034.52 MnI -4035.73 Mnr - 4038.00 Crl

I I n MM _- -.- F-. F. I I1 11 1 1 1 1 1u Fig. 1 (Concluded) ro tU

4 60 o0 16.0 0 / z 0 0 15.0 I I I I I l I I I I I I I I I 5 6 7 8 9 10 II 12 13 14 15 Hr H8 He PRINCIPAL QUANTUM NUMBER — Fig. 2. Determination of the number of neutral hydrogen atoms in the second level. Following Unsold's procedure, we plot No2, calculated from Eq. 14 against the principal quantum number, n, and extrapolate to obtain an asymptotic calue of N02. 28

I ox ^^+a x ~ t | X +Symbol X | O XPX 1.1eV 0 -A + X 1 -I 0 1 +2 +3 Log X - nearly the se excitation potential are plotted against + o and 0 J or xi TITANIUM,+ + Symbol X + /0 0.59 eV nearly/~~~~~~~ ~ t+ 1p.2 eV m^ * 1.57eV v 2.58 eV -I 0 +1 +2 +3 Log X, Fig. 3. Curve of growth for titanium. Lines arising from levels of very nearly the same excitation potential are plotted against Cg + log 8kT\ and each group is fitted separately to the theoretical curve of growth. The ordinate is log W/% c/v, the abscissa is log Xo.

^~~ ~ ~~~ I I + ~ ~ y ca +1 0 0 0 0 -a -I - CrI * 2.69 eV o 3.09eV X 3.84eV 0 4.05eV V 6.46eV -2 -I 0 +1 +2 23 Log XFig. 4. Curve of growth for ionized calcium ana ionized chromium. We plot log W/x c/v against log X0.

+1 0 x x t o f~~r J MnI * X= OeV 0 X = 2.13eV MnIE X X = 1.82eV -2 -I 0 + I +2 +3 Log X,Fig. 5. Curve of growth for neutral and ionized manganese. Data for the 5.8 eV lines are not included.

4T~ x 0 l FeI o - I -FeIET X 2.66 eV /A 2.81 eV -I 0 l1 +2 Log XoFig. 6. Curve of growth for neutral and ionized iron.

o o CI -j Silt go + Mg I ~ * Sr 3I X Y 3 o Ga I ~~-lI~ ~':~ O. f ~+ I -1 0 -ILog X — *Fig. 7. Curve of growth for CII, MgII, Sill, Gal, SrII, and YII. 33

+lo I I I I 0 /^ + 1.22eV Log XZ -X Fig. 8. Curve of growth for ZrII.

-------— Zr-II z 0 13.0 _ 0 I 2 Fig. 9. Determination of the concentration of ionized zirconium atoms. We plot log N - OX = log C3 + log?fX against X. The slope of the straight line corresponds -to 9 = 0.45. Its intercept with the X = 0 axis gives log N(ZrII). 35

9i 3763.74 Mnll 3765.61 Cr I 3 715.40 C 3770.63 -3721.94I3. 3774.35 Y II 3776.05 Till 3677.69 CrH 3778.32 M nn 3677.91 CCrr 3778.61 -3729.49 Mnll 3679.71 Ti - 3730.07 Mn It -3783. 89'M nI__ 3734.37 H 3685.07 Mn l 3786.35 Ti 3685.2 Ti 3736.93 Cof 3788.71 Y]l - 3738.38CrII 3740 28 3741.65 Til 3743.38 Mnl[ 3745.93 Zrll 3797.9 H 3748.0 2 T i IL 3750.15 H -3703.86 H - 3754.59 Crt1 -3755.21 MnrL 3706.03 Call 3806.71 Mn TI -3706.24 Ti TT -3757.68 Ti -3706.87M 3706.87 MnII -3759.29 Till -3708.87 Mn1L -3812:.53 MnTI -3761.87 TiTl 3711.97 H -3813.39 Ti D 3712.95 Cr I' -3763.74 MnI[ _), 0 -3814.59 Ti -_ CM~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~- 1

3865.60 CrI — 3814.59 TilT 3917.33 MnIr 3918.97 CIL 3920.66 C D- 3820.42 Fel 3823.51 MnI 3823.87 MnI 3926.13 Mnlr 3926.45 Mn I 3879.00 Mnll - 3829.37 MgI 3930.99 Mn-_ 3882.30 Ti II 3932.01 Till - 38 32.94 Y!I 3933.67 CaII 3934.79 ZrlT -3835.39 H ItJlI * l t 3889.05 H l l l *-3838.28 Mgl 3938.98 Fell 3941.23 MnT 3943.60 Mn T 3844.17 Mnlr - 3897.60 Mnll 3848.24 Mg 3 3898.09 Mn3l 3848.62 Mnll 3900.55 Ti r 3849.57 Nill 3954.37 y l - 3850.38 MgI 3902.40 Mnll 3952.43 Mnl' 3853.67 Si Il 3905.48 Si 1 39 05.67 Cr TI 3856.01 Si ll 3958.22 Zr la 3913.48 TiAO - 3862.59 SilT 3863.44 MnllI 3865.60 Cr23917.33 Mnll

- 4067.06 Ni Tr - 3,967.95 MnIr - 4018.07 MnI -4070.86 Crlr - 3970.07 H -4075.46 Sill -4076.78 Sil - 4024.52 Fell - 4077.71 SrII - 4025,15 Ti I 4079.24 MnI - 4028.32 Ti:I - 4081.45 MnIl 3979.53 Crtr - 4029.70 ZrIL - 4030.77, Mnl -4083.65 Mn! - 4030.77 Mnl - 3982.07 TiII -4085.41 Mn II - 4033.07 MnI - 4034.52 MnI - 4035.73 MnI - 3986.60 Mn r - 3987.60 Ti If - 4038.00 CrII - 3991.16 Zr Ir - 4041;39 Mn I 0 C - 3995.30 MnU' - 4045.60 Zr I' 4045.83 Fe I - 3998,9'7 ZrIT - 4048.71 ZrTll MnI - 4101.74 H - 4049,13' CrIl - 4000.04 Mn I - 4050.32 ZrII - 4051.97 Cr - 4003.30 CrlI - 4053.83 Til] 4054. 09 Cr II - 4055.56 Mn I - 4110.62 Mnll -4111.00 6CrM I - 4058.94 MnI - 4012.40 Ti Ir 4012.55 Cr 1 - 4064.36 Ti 1I - 4067.06 Ni II - 4018.07 MnI 0 o ) CM

- 4171.05 Mnll - 4171.53 MnIT = 4171.92 Ti l4172.06 Ga I - 4173,52 TiIT - 4174.10 Til3 - 4174.34 Mnll -- 4177:.52- Mn I - 4233.19 Fe3 YI N - 4178.89 Fe I - 4179.45 Cr11 - 4128.08 SiII Mn II - 4235.33 MnI * Ikii- - 4129.12 Si Ii - 4130.88 Sill - 4237.85 Mn'r -4238.81 MnU - 4239.21 MnU- 4184.29 Till - 4184..47 Mn'II - 4240.39 Mnll - 4242.57 Mn IL - 4242.96 Mn I 413691 Mn - 4244.27 Mn 71 4140.47 Mn]7 - 4247.94 Mn 11 - 4251.74 Mg IL - 4252.62 Cr -r 4145.78 CrE C = 4252.99 Mnl' 4253.06 Mn l - 4-_00.28 Mnl 1 -- 4149,22 Zrl' - 4150.98 Zr'r - 4259.20 Mn II - 4260.49 Mn lT - 4205.37 Mn'll - 4206.39 MnlT' - 4261.92 CrT 4207.22 Mn 4207.22 Mnll — 4156.24 ZrU - 4209.01 Zr Tr - 4158.27 MnI - 4266.98 Cr - 4211.89 ZrII - 4267, 26 C 3r -4161.19 Zr r - 4161.53 Ti II - 4269.27 Cr*T - 4215.54 Srl - 4163.65 Ti l 03 | -- e 04 N'

o1t -4326.66 =4275.58 CrMI 4275.89 MnIr 4377.77 Mnll 4379.72 Blend -4330.24- 4278.61 MnII - 4330.24 TiIr - 4330.70 Ti fl - 4281.96 - 428 2.49 4383.60 Fe I4282.49 4384.64 MgTL 4283.78 =428 4. 18 Cr t -4385.41 Fell 4284.144 Mnl - 4386.88 TiT -- 4337.93 TiII 4287.89 Till 4288.07 MnIL -- 43 40,47 H - 4390.59 Mg-3 -4391.09 Ti I - 4290.23 TIIl - 4292.23 MnTT -4393.39 MnI - 434401 MnII - 4394.07 Ti ll -4344.27 Ti U -4395. 04 Ti - -4294.11 Till -- 4345.64 Mn'rr -4395.85 TillI -4346.40 MnlL - 4348.42 Mn ll 4399.78 Till -4350.81 Till -4300.06 Till - ~4351.77 Fe~ - 4300.27 Mnll OI -4301.95 Ti3Ir 4403.52 MnIl - 4303.18 Fe II{ - 4356.63 Mnll - 4307.89 Til - 4308.18 Mnll 4409.51 Till -4359.75 Zr4 1 -4411.08 Tilr -4111.90 TiIr - 4363.27 Mnll -4312.88 Till -4365.24 MnJI - 4314.99 Till'-4416.80 Fell'MnTT - 4367.66 Ti - 4316.82Ti - 4417. 72 Til Ti 4418.32 TiII - 4317.77 MnIr - 4318.55 - 4370.99 Zrll - 4320.99 Tirr 4221.94 Ti 4320.99 Ti -4374.87 yrr Till -43 25.0 7 Mnrr.4." 4377. 77 MnTT -~ iB - 4326.66 -- 01 0) -4 01,la -=.

-4529.48 Till -4478.63 MnhI - 4481.14 Mg I -4533.95 Till - 4481.35 Mg4434.06 Mn - 4434.06 MnTI' - 4488.33 Till -4489.19 FeII - 4441.73 Ti.ll 4441.99 Mn l - 449.1.40 FelE -4443.02 Zr l' - 4443.80 Ti II - 4544.01 Till — 4444.54 Ti II 4549.49 Fell - 4496.98 Zrl 4549.62 Till - 4450.48 Till -4451.56 - 4500.58 Mn ll' - 4501.28 Ti l. 4'554.99 Crll - 4503.18 MnI' l'r 4555.89 Fel 9 - 4456.62 Till 4558.64 Crll -4508.28 Fell - 4462.03 MnI - 4563.77 Ti ll - 4464.45 Till - 4565.74 CrIT -4515.33 TilI - 4568.'33 Till - 4468.51 Till - 4469.13 Ti IL - 4518.34 TiIr = 4518.94 Mnll -47.6Tl4519.23 Mnll - 4470.85 Till - 4571.96 Til'r - 4520.22 Fel - 4522.62 Fell -4525.32 MnTr - 4478.63 Mn ll -4529.48 Till 0 _4 <8)F 0 o

-4583.82 Fenl 4634.07 Cr I.4635.34 Felr 4588.20 Cr11 4589.95 Till -4591.08 Cr I, O( a 4657.22 Till c 2. 4616.65 CrIl 4618.81 Cr I 4629.32 Fell a 4634.07 CrlI ro N I' 0

UNIVERSITY OF MICHIGAN 3 9015 02499 5600

THE UN I V E R S I T Y OF MI C HI GAN COLLEGE OF LITERATURE, SCIENCE, AND THE ARTS Department of Astronomy Final Report ABUNDANCES OF ELEMENTS IN STARS ANDJ,NEBULAE L. H. Aller ORA Project 03719 under contract with: AIR FORCE OFFICE OF SCIENTIFIC RESEARCH AIR RESEARCH AND- DEVELOPMENT COMMAND CONTRACT NO. AF 49(638)-807 WASHINGTON, D.C. administered through: OFFICE OF RESEARCH ADMINISTRATION ANN ARBOR September 1962

ACKNOWLEDGMENTS The investigations described in this summary and in the preceding technical reports could not have been undertaken without the generous help of many individuals and institutions. The study of the spectrum of NGC 7009 and two other planetary nebulae was made possible by a cooperative arrangement with the Mto Wilson and Palomar Observatories. I am indebted to Director I. S. Bowen for'many helpful discussions in connection with this work. All of the work carried on in the southern hemisphere, viz., (1) spectrophotometry of gaseous nebulae in the Magellanic Clouds, (2) spectrophotometry of clusters, and (3) studies of y Velorum and HD 96446, were made possibly by the award of a Senior Postdoctoral Fellowship at the Austrialian National University for 1960-61, and by generous-cooperation and assistance from the Australian National University. In particular I would like to cite help received from Director BartLJ. Bok, Prof. T. Dunham, Jr., Dr. S.C.B. Gascoigne, Dr. A. Hogg, Dr. H. Gollnow, Dr. A. Przybylski, and members of the technical staffs of the Mt. Stromlo Observatory and Mt. Bingar Field station. These astronomers generously gave up time from their own investigations and programs Ito permit us to carry out our spectrophotometric and spectroscopic programs. D. J. Faulkner supplied valuable data after my return to the U.S.A. Thanks are due to Mrs. Anne Cowley, who supervised much of the routine work and carried out a number of the investigations, to Mrs. H. Dickel for her thorough discussion of the spectrophotometric data for the Large Magellanic Cloud, to Mrs. Katherine Grey, who reduced the data for NGC 7009, to Mr. Herbert Rood, who verified the reductions of the star clusters and y Velorum, and to Tom Toon, Eric Leighton, and Tom Jones, who did much of the routine work. Individual acknowledgments are also included in the technical reports. iii

TABLE OF CONTENTS Page LIST OF TABLES vii LIST OF FIGURES ix SECTION. I. SUMMARY OF THE YEAR'S ACTIVITIES 1 SECTION II. ENERGY DISTRIBUTION IN GLOBULAR STAR CLUSTERS 7 A. Introduction 9 B. Data for Southern Globular Clusters 11 1. Properties of Some Southern Globular Clusters 11 2. Distance to the Globular Clusters of the Magellanic Clouds 12 3. Summary 17 C. Measurements of the Energy Distributions in Southern Globular Clusters' 17 Section II: Tables, Figures, and References 23 SECTION III. SPECTROPHOTOMETRY OF THE WOLF-RAYET STAR OF 72 VELORUM 39 Section III: Tables, Figures, and References 45 SECTION IV. THE HYDROGEN/HELIUM RATIO IN HD 96446 55 Section IV: Tables, Figures, and References 61 SECTION V. THE EMISSION NEBULOSITIES IN THE LARGE MAGELLANIC CLOUD 71 Section V: Tables, Figures, and References 79 SECTION VI. EMISSION NEBULOSITIES IN THE SMALL MAGELLANIC CLOUD 95 Section VI: Tables, Figures, and, References 99 SECTION VII. PHOTOGRAPHIC REGION OF THE SPECTRUM OF NGC 7009 107 Section VII: Tables, Figures, and References 113 SECTION VIII. REPORT ON RESEARCH CARRIED OUT AT THE UNIVERSITY OF MICHIGAN AND SUPPORTED.BY THE U.S.A.F. UNDER CONTRACT NO. AF 49(638)-807 139 v

LIST OF TABLES Table Page 2-1 Properties of Some Southern Globular Clusters 25 2-2 Southern Hemisphere Globular Cluster Scans 26 2-3 Globular Cluster Magnitudes Relative to l/\o = 1.80 27 2-4 Energy Distribution Equivalents for Globular Clusters 28 3-1 A, Values of 72 Velorum 47 5-2 -ml/1 Values of 72 Velorum 48 5-3 Ij/Io Values of 72 Velorum 49 4-1 Coude/ Observations of HD 96446 63 4-2 Ax Values of HD 96446 64 4-3 Determination of N,2H for Hydrogen 65 4-4: Determination of Nr for. Helium 66 r s 5-1 Scanner Slots and Apertures Used on Emission Nebulosities in the Large Magellanic Cloud 81 5-2 Data for the Emission Nebulosities in the Large Magellanic Cloud 82 5-3 Relative Intensities for Hp = 10 85 5-4 Electron Densities and Ionic Concentrations for Te = 10,0000K 84 5-5 Electron Densities and Ionic Concentrations for Te = 15,000~K 85 6-1 Relative Intensities of the Nebular Lines Observed in the Small Magellanic Cloud 101 vii

LIST OF TABLES (Concluded) Table Page 6-2 Electron Densities in the Small Magellanic Cloud 102 7-1 Details Concerning Spectrograms Secured With the Coude Spectrograph on the 100-Inch Telescope 115 7-2 Photographic Region of the Spectrum of NGC 7009 116 viii

LIST OF FIGURES Figure Page.2-1 Schematic diagram of the Michigan spectrophotometer. 29 2-2 The Michigan spectrophotometer mounted on the 50-inch telescope at Mt. Stromlo. 30 2-3 Scan of 47 Tucanae (Aug. 25, 1960). 31 2-4 Scan of NGC 1866 (Oct. 27, 1960). 31 2-5 Energy distributions for NGC 362, X Centauri, NGC 2808, and 47 Tucanae. 32 2-6 Energy distributions for 47 Tucanae, NGC 6752, and NGC 419. 33 2-7 Energy distributions for NGC 1866, NGC 1783, NGC 1831, and NGC 1978. 34 2-8 Energy distribution for NGC 330. 35 5-1 Spectral scan of 72 Velorum (Oct. 12-13, 1960, Mt. Bingar). 50 5-2 Energy distribution for 72 Velorum. 51 4-1 Energy scan of HD 96446 (Nov. 10-11, 1960, Mt. Bingar). 67 4-2 Energy distribution for HD 96446. 68 5-1 The Large Magellanic Cloud. 86 5-2 Emission nebulosities in the Large Magellanic Cloud near 30 Doradus* 87 5-3 The emission nebulosity Henize 144 in the Large Magellanic Cloud. 88 5-4 The emission nebula near S Doradus, 89 5-5 Emission nebulosities Henize 11 in the Large Magellanic Cloud. 90 ix

LIST OF FIGURES (Concluded) Figure Page 5-6 Scans of typical emission nebulosities in the Large Magellanic Cloud. 91 6-1 The Small Magnellanic Cloud. 103 7-1 Spectrum of NGC 7009. 125 7-2 Microphotometric tracing of the spectrum of NGC 7009. 129 7-3 Spectral scan of NGC 7009. 155 x

SECTION 1 SUMMARY OF THE YEAR'S ACTIVITIES

I

The program on the "Abundances of the Elements in Stars and Nebulae" carried out from August 1961 through August 1962 in the Astronomy Department of The University of Michigan has involved both theoretical and observational studies. Some of these are reported in previous technical reports and others are described here in Sections II-VIII. Our investigations of the gaseous nebulae are described in-Sections V and VI, written by Helene Ro Dickel, Do J. Faulkner, and L. H. Aller, and in Section VII. Our work this year has been entirely concerned with observational problems. A series of photoelectric spectrophotometric measurements of emission nebulosities in both the Small Magellanic Cloud and the Large Magellanic Cloud were made in Australia by Faulkner and Aller. They have been analyzed and discussed by Mrs. Dickel; the principal results may be summarized briefly as follows. The nebulosities appear to be similar to those found in our own galaxy although a number are larger and more massive than the best known galactic diffuse nebulae in our own neighborhood. Since the distance of the clouds is known, we can convert the measured surface brightnesses (expressed in conventional c.g.s. units) into emissions per unit volume and derive electron densities, provided the electron temperature is known or can be inferred. These mean densities turn out to be rather low compared with those found in objects like the Orion nebula, but when effects of filamentary structure are taken into account, the actual densities are probably more nearly comparable. The intensities of the strongest forbidden lines of [OII], [OIII], and [NeIII] have been measured and from these the ionic concentrations are found. The ratios of O+/H, O++/H, and Ne++/H are comparable with those found in emission nebulosities in our own galaxy. In all gaseous nebulae it is difficult to find the actual abundance of an element such as 0 or Ne from the abundance of only a few of its ions, and these difficulties are particularly severe for diffuse nebulae in which only a few lines can be observed. From results obtained for a large number of nebulosities, it is concluded that the ratio of oxygen to hydrogen and of neon to hydrogen is about the same in the Large Magellanic Cloud and in the galaxy. A similar situation probably holds for the Small Magellanic Cloud but the diffuse nebulae are fewer and generally fainter, so that a definitive study is more difficult. Spectroscopic observations of the brighter diffuse nebulae in both clouds are urgently needed. In Section VII we describe in detail the spectrum of the planetary nebula NGC 7009 and present lists of lines, their identifications, and rough intensities. The purpose of this investigation is to provide material for theoretical studies, which are being carried out in collaboration with Stanley Czyzak of the Fundamental Physics Laboratory at Wright-Patterson Air Force Base. Stellar investigations have involved both theoretical and observational studies. During the fall of 1961 Dr. Donald Mugglestone spent several months with us studying basic problems in the determination of abundances in stellar spectra; his own summary of his work is presented in Section VIII. 5

An urgent necessity for improving stellar and solar abundances is improved +-values. We have employed, (-values determined by Corliss at the National Bureau of Standards to obtain improved abundances for a number of metals in the sun. The results were described in Technical Report No. 2. In addition we have attempted to obtain ff-values for a number of ions by combining laboratory and stellar data (Technical Report No. 3). These F -values were then employed in a curve-of-growth analysis for a remarkable star rich in manganese observed by William P. Bidelman at Lick Observatory (Technical Report No. 5). Among purely theoretical investigations, mention must be made of the work of Anne and Charles Cowley, who developed a machine routine for computing the relation between gas and electron pressure for stellar atmospheres of arbitrary composition (Technical Report No. 4). The advantage of this technique is that one may allow for the effect produced by individual atoms that are easily ionized. One of the objectives of our work was an investigation of properties of hydrogen deficient stars. In detailed theoretical study made of a pure helium atmosphere (Technical Report No. 1) it was demonstrated that conventional mathematical methods capable of yielding good results for a pure hydrogen atmosphere failed when applied to one of pure helium. The reason for this discordance is not known,' and a separate study will have to be devoted to it. Prominent among hydrogen deficient stars are those of the Wolf-Rayet type. In Section III we describe some results obtained for the brightest of these objects, 72 Velorum, a star of the "carbon" type. Accurate spectrophotometry must provide one of the bases for assessment of these objects. This star appears to be remarkably blue; its energy distribution corresponds to a temperature of 32,000~K, indicating that there exists little if any space reddening. The remarkable hydrogen deficient star HD 96446 appears to be one of the brightest absorption-line objects, but with the data we were able to secure it has been possible to determine only the hydrogen/helium ratio. The ratios of other elements cannot be determined until better spectrograms are obtained. In extending composition studies to remote stellar systems, one can no longer work with individual stars and must investigate the composite spectra of many of them. By observing the spectra of clusters and galaxies, one may observe certain lines and molecular bands that are sensitive to luminosity, temperature, and the metal/hydrogen ratio. If one can also measure the energy distribution in the spectra of such clusters or galaxies, it may be possible to determine the metal/hydrogen ratios. With this hope in mind.we present in Section II the results of an extended study of the energy distributions in globular clusters, both in our own galaxy and in the Magellanic Clouds, Comparison of the energy distributions which are expressed in terms 4

of spectral types of standard stars with estimated spectral classes for the clusters provides some clue to metal deficiencies, but it should be emphasized that quantitative results -are possible only when color-magnitude diagrams and luminosity functions are available for the stars under consideration and when energy distributions in stars of various degrees of metal deficiency are known. 5

I i

SECTION II ENERGY DISTRIBUTION IN GLOBULAR STAR CLUSTERS by L. H. Aller and D. J. Faulkner

A. INTRODUCTION Some of the most engaging opportunities for the study of stellar evolution and differences in stellar composition are provided by star clusters. Differences in apparent brightness correspond to differences in intrinsic brightness, whereas observed differences in color correspond to true differences in color. Furthermore the stars of a cluster (with the possible exception of the very faintest ones) all have the same age. Some clusters of the "open" or "galactic" type are sufficiently close to the sun to permit detailed studies of individual stars. In more remote clusters such as those of the globular type, spectra of individual stars cannot be investigated except with the very- largest telescopes. Even then, it is possible to study only the very brightest stars in the very nearest clusters. For most clusters we are limited to the following observational data: (1) Magnitudes and colors of individual stars; (2) Integrated magnitude and colors of the cluster; 3() Integrated spectrum of the cluster; (4) Radial velocity of the cluster; and (5) Energy distribution in the integrated light of the cluster, Most attention has been paid to the construction of color-magnitude arrays for individual clusters, as these have given much useful information that can be interpreted in terms of theories of stellar evolution. Theoretical color-magnitude arrays can be calculated; from a comparison of observed and computed arrays it is then possible to estimate the age of the cluster and the metal/hydrogen ratio. The chemical composition of a star or at least the metal/hydrogen ratio determines not only the relation between its color and effective temperature but also its evolutionary track in the color-magnitude [Russell-Hertzsprung] diagram. For example a metal deficient star of about 12 or 1.5 solar masses becomes brighter in the late stages of its evolution than does one of normal stellar composition. By comparing theoretical evolutionary tracks computed for different metal/hydrogen ratios with the observed color-magnitude arrays it is possible to estimate abnormalities in the metal/hydrogen ratios. The integrated magnitude of a cluster can be measured with a relatively small telescope, and it is also possible to measure colors with a relatively long baseline in wavelength. -Such studies are very valuable in that they 9

provide convenient checks on the assumed stellar content and chemical composition of a cluster. Spectrograms secured with a slit spectrograph yield not only spectral classifications but also radial velocities, which are necessary for assessing the kinematics of these objects. The present investigation is concerned with the energy distributions in globular clusters. In a sense energy-distribution measurements are comparable with multi-color photometry. The chief advantage is that narrower band passes may be used and the entire spectrum traced, whereas in multi-color photometry one is limited to effective wavelengths determined by the filter and a rather broad band-pass. On the other band, broad band-pass photometry often permits one to cover a broader spectral range, and to work much faster. Since the time required for a single observation is much shorter, the observer is less at the mercy of the sky transparency. Hence a greater accuracy can be obtained. We want to emphasize that energy distribution measurements such as those we have secured with purities of 20A, 40A, or sometimes worse should be supplemented by slit spectrograms. A judicious combination of spectrograms of moderate dispersion and energy scans should provide valuable boundary conditions for testing hypotheses of the compositions of stars contained in star clusters. Ideally, one could proceed as follows. From color-magnitude arrays we synthesize the spectrum and energy distribution in a star cluster. For each interval of magnitude we know the number of stars and the distribution of these stars with respect to color. The only adjustable parameter is the chemical composition. In our ownm galaxy we can find stars of different chemical composition and can determine their energy distributions, colors, and spectral characteristics. For each assumed ratio of metals to hydrogen, one can construct what the integrated spectrum and energy distribution of a star cluster would look like-given its color-magnitude array and luminosity function. Thus, the problem is a well determined one, The metal/hydrogen ratio in a cluster can be found provided we know: (1) The color-magnitude -array for a star's contributing to the integrated light of the cluster; (2) The luminosity function for all stars contributing appreciably to the integrated light of the cluster; (3) The spectral class and the intensities of strategic luminosity-dependent and composition-dependent lines; 1C

(4) The energy distribution in the integrated spectrum or the energy flux from the cluster at certain pre-selected wavelengths, with a band-pass (5) The energy distributions in stars of different temperatures, intrinsic luminosities, and chemical compositions. Unfortunately, this program cannot yet be carried out for a single cluster. For only a few clusters (see Part B of this Section) do we have detailed color-magnitude arrays and luminosity functions. The spectroscopic observations have almost all been carried out with low dispersions and give only fragmentary data on line ratios needed for luminosity classifications and chemical compositions. On the other hand, energy distributions have been measured for only a limited number of stars, most of which are of normal chemical composition. Energy distributions have been measured for only a few stars of abnormal chemical composition! In the present study we have attempted to supply data on the energy distributions in several southern globular clusters. We have included several objects in the Large Magellanic Cloud, LMC, and one object, IC 419, in the Small Magellanic Cloud, SMC. Part B is based on a compilation by Herbert Rood and gives a brief summary of some of the salient features of the clusters observed, while Part C describes our observational techniques, the results obtained therewith, and some conclusions that may be drawn from our study. B. DATA FOR SOUTHERN GLOBULAR CLUSTERS* 1. PROPERTIES OF SOME SOUTHERN GLOBULAR-CLUSTERS (TABLE 2-1) Table 2-1 gives data for the southern globular clusters contained in our program. The number in brackets [ ] denotes the reference from which the data were obtained. The superscript after brackets denotes the number of the relevant footnote. Column 1 NGC Number. 2 Right ascension and declination for 1950o 3 Galactic latitude and longitude for 1900 (based on the old coordinates of the pole). L*Prepared by H. Rood. 11

Column 4 Reference to place where color-magnitude diagram is found. 5 Reference to place where luminosity function is found. 6 Spectral type as estimated by CH/Hy intensity ratio, Hy strength, strength of FeI from X4250 to X4400. Spectral types of [30] are Harvard types. 7 Photographic magnitude P or mpg. 8 Color index P-V or mpg-mpv. 9 Distance modulus corrected for interstellar absorption and reduced to a value of the absolute magnitude of the RR Lyrae stars of MRR = +0.5. Number in parenthesis is probable error. 10 Distance in KPC calculated from the eq. log(dist) = 0.2(cor. mod.+5)-3 = 0.2(cor. mod.-l0). 11 Measured radial velocity in km/sec. Number in parenthesis is probable error. 12 Mass of cluster in solar masses, 13 Concentration class on the system of Shapley and Sawyer, Harvard Bull. No. 849(1927). 14 Apparent angular diameter. Refs. [18] and [29] give estimates from images on plates. Ref. [15] gives estimate from star counts. Ref. [28] gives photoelectric diameter, the diameter within which is found 1/2 the light = effective diameter. Ref. [31] gives 2 x mean distance of RR Lyrae stars from cluster center = mean diameter, 2. DISTANCE TO THE GLOBULAI CLUSTERS OF THE MAGELLANIC CLOUDS a. The Distance Modulus Thackeray (IAU Symp. 7, Co-ordination of Galactic Research p. 81, 1959) states that the "uncertainty regarding location in depth within the clouds adds a random error of about +0.13 magnitudes.") As we shall see later, this is much smaller than the uncertainties in distances to the clouds. Furthermore, if we assume that the clouds' globular clusters are of the order of or less than a typical dimension of each cloud away from the cloud center, then the true distance modulus to each cloud equals the true distance modulus of its clusters to well within the total uncertainty of the distance modulus. 12

Color-magnitude diagrams have been obtained for NGC 1783, NGC 1978, NGC 330, and NGC 419. To construct'a color-absolute magnitude diagram, the true distance modulus and absorption must be known. Their values as used in connection with observed color-magnitude diagrams are given below. NGC 1783: Sandage and Eggen [17] take the true modulus to the LMC = true modulus of NGC 1783 as 18.7. They assume no reddening even though the redness of some stars may be explained if EB_V = +o'm6, because some stars are observed near the theoretical blue limit of B-V = -0.4 inside and nearby the cluster. NGC 1978: Hodge [23] takes the distance modulus of LMC = distance modulus of NGC 1978 as 18.7. He does not consider reddening. NGC 330 and NGC 419: Arp [24] and [26] uses as distance modulus of SMC = distance modulus of NGC 330 = distance modulus of NGC 419 = 19.2. He does not consider reddening. IAU Transactions for 1961, p. 296, gives 19.0+0.5 magnitudes as probably the best present value of the distance modulus of the clouds. Since absorption is probably close to zero (as far as can be concluded from present data), and the LMC and SMC have the same modulus (within the accuracy of present data), we will use as the distance modulus to every cloud cluster: m - M = 19.0 + 0.5 or a distance of 63(+19, -13)KPC. b. The Individual Clusters We now consider data for the individual clusters that have been observed. (1) 47 Tucanae.-The distance of 47 Tucanae has been studied in recent years by Feast and Thackeray [8], by Wildey [3], and by Kinman [12]. From studies of foreign stars Feast and Thackeray estimated the absorption to be 0.2 magnitudes; they adopted the absolute magnitudes of the RR Lyrae stars as +0.5 and obtained a corrected distance-modulus of 13.2, corresponding to a distance of 4.4 kiloparsecs. Wildey concluded that if the absolute visual magnitude of the stars in the horizontal branch of the color-magnitude array is 0.0, then m-M, the distance modulus, = 14.1. If the blue end of the horizontal branch is as faint or fainter than the horizontal branch of NGC 6356, taken as M = 0.9, then m-M < 13.2. From a discussion of all available data and methods of distance determination, Kinman found m-M = 13.8~0.4 for MRR = 0. Wildey constructed a colormagnitude diagram from 500 stars to a limiting magnitude B = 15.4, V = 14.6 13

(Johnson-Morgan magnitudes being denoted here by U = ultraviolet, B -= blue, and V = visual). He finds this giant branch to be fainter than the giant branch in most globular clusters (which are hydrogen deficient) and''concludes that 47 Tucanae has a metal/hydrogen ratio more nearly comparable with that' of the sun than with that of extreme population type II. Gascoigne and Burr [7] measured photoelectrically the surface brightness, integrated magnitudes, and color of 47 Tucanae. They determined an integrated magnitude P = 4.7+0.05 and a color P-V = 0.67. They found 47 Tucanae to be redder at the center than at the edge. Kinman [5] finds an integrated spectral type of G3, and Feast and Thackeray [8] have determined the spectral types of 32 bright stars by comparing them with standards on the MK system, They find that the red giant stars in 47 Tucanae are much closer to "normal" population type I giants than to the extreme population II globular cluster stars, and conclude that the metal/hydrogen ratio is not less than about 0.25 normal. The brightest star in the cluster, MV = -3.0, has a normal B8III spectrum. Thackeray [9] and. Thackeray and Wesselink [13] note that 47 Tucanae is unique in that it is the only globular clusteri in the galaxy known to contain M stars. Kinman [5] studied the spectra of some bright giant K stars in 47 Tucanae and found them to be "normal." Feast and Thackeray [8] found no rotation in 47 Tucanae from the radial velocities of red giant stars, although they did find some indication that late-type giants may be losing mass since their radial velocities became more negative with advancing spectral class from G8 to M2. If this correlation is taken into account, the virial theorem gives an upper limit to the mass for the cluster as 250,000 Msun., Then M/L < 0.3 solar units, which is much less than for elliptical galaxies. Data on variable stars are given by Mrs. Hogg [16]. See also Feast, Thackeray, and Wesselink [13]o (2) u Centauri = NGC 5139o -From a discussion of all available data, Kinman [12] gets a distance-modulus m-M = 13.5~0.7. The apparent diameter of this very large cluster is given.by Mrs. Hogg [1] as 65.41 whereas Lindsay [15] from star counts finds a diameter of 101'. The cluster is noticeably elliptical. Belserene [4] has obtained a color-magnitude diagram in B and V to a limiting magnitude of B = 19.5 and has also obtained the luminosity function from star counts, A stellar density 25,000 times that near the sun is estimated for the center of-the cluster: From the CH/Hy ratio, Kinman [5] obtained an integrated spectral type F7. Morgan classified the spectrum by comparing intensity ratios and intensities with a sequence of standard dwarfs on the MK system, viz,, 14

CH/Hy gives F7 (in agreement with Kinman); Hy intensity gives a "hydrogen" type F8; and Fel intensity gives a "metallic line type" FO. Since the metallic lines are weak, the metals appear to be deficient. In Morgan's system of eight groups I-VIII in order of increasing metallic abundance, c Centauri falls in group II. Since the pioneer work of Popper, the weakness of metallic lines in individual stars has been confirmed (Kinman [5 ]). Gascoigne and Burr [7] obtained the integrated mangitude, brightness distribution, and color for X Centauri. They find P = 4.25; and P-V = 0.68. The extensive literature on the variable stars up to 1959 has been summarized by Mrs. Hogg [16]. See also Belserene [14]. (3) NGC 6572.-Kinman studied the spectra of some giant stars in this cluster and found that if they were red giants, their metallic lines were weakened but not very much weakened. c. Clusters of the Large' Magellanic Cloud In discussing the four clusters of the Large Magellanic Cloud we refer to Section 2a, "The Distance Modulus." (1) NGC 1783.-Sandage and Eggen [17] obtained a color-magnitude diagram to V = 17. Three features indicate that NGC 1783 is different from "normal" clusters in the galaxy: (a) The giant branch is well populated up to B-V = 2.4, whereas all clusters in our galaxy terminate at B-V about 1.6. No giant stars in clusters in the galaxy are as-red as B-V = +2.4. (b) The slope bV/3(B-V) = 1.25 is gentler than that of the giant branches in normal clusters. (c) There is no drop towards a vertical subgiant sequence beginning at (B-V)o = +0.9. The giant. sequence remains linear to the limit of the observationsThe diagrams for NGC 419 and for,NGC 1978 as obtained by Arp and Hodge are very similar to that of-NGC 1783, the giant-branch slopes being almost the same. The spectra of the stars in NGC 6356 imply a high metal abundance. The color indices of some stars in NGC 1783 are as red as B-V = 2.4 and are not limited by TiO bands at B-V = 1.7 as are stars with high metal abundance. 15

Why then are the NGC 6356 and NGC 1783 giant-branch slopes the same? Gascoigne [28] gives the photoelectric effective diameter of the cluster as 1.2. This is the diameter within which is concentrated half the light of the cluster. (2) NGC 1831. -Hodge [20] describes the color-magnitude diagram as similar to that for Mil. NGC 1831 is similar in appearnce to globular clusters but is much younger. It differs from galactic clusters by the enormous quantity of stars and the location of the giants in the color-magnitude diagram. They are bluer and brighter than in the open clusters. Hodge [21] describes the cluster as "large and concentrated," Gascoigne derives a photoelectric effective diameter of 1.0', The magnitudes of the half-dozen brightest measurable stars are roughly mpg = 18.0+0.5. (3) NGC 1866.-Preliminary results (cfo Arp, Sandage, and Thackeray [22]) indicate that the color-magnitude diagram for this cluster resembles that of Mll. Thackeray [9] regards it as the prototype of early-type globulars. It contains many 3-day Cepheids, is rich in bright blue stars, and has many red ones. Hodge [21] found more thani 2000 stars from mB = 16.6 to mB = 19.0 in the brightest part of the luminosity function. (4) NGC 1978. —This cluster is described by Hodge [23] as an unusual eccentric ellipse. He has determined a color-magnitude diagram which has a heavily populated giant branch, whose gentle slope implies a relatively high metal abundance. It is a rich cluster with some remarkable variable stars. See also Hodge [23], Hogg [16], and Thackeray and Wesselink [27]. d. Clusters of the Small Magellanic Cloud In the Small Magellanic Cloud we studied two clusters, NGC 330 (which is not a globular cluster) and NGC 419. (1) NGC 330.-For this cluster the right ascension and declination for 1950 is: a = O - 54.m9, = -72 ~4] Arp [24] obtained a color-magnitude diagram to V = 19. It differed considerably from that of h and X Persei, the red supergiants being bluer and more abundant, and the!Hertzsprung gap narrower. The brightest stars are comparable in luminosity with their counterparts in h and X Persei. Arp suggests that the differences in the color-magnitude diagrams are due to composition differences; but since the spectra of the bright SMC stars indicate no significant metal deficiency (Feast and Thackeray, 1958-Observatory 28, 156), the-composition differences may suffice to affect the internal structure but not the spectrum of the star, Arp found that the main sequence was narrow 16

and well populated up to MV = -4.5, above which occurred A and F supergiants between -4.5 and -5.5 and some red' supergiants at -5. The integrated magnitude is P = 9.0 (de Vaucouleurs [25]). (2) NGC 419.-Arp [26] found the color-magnitude diagram for this globular cluster to resemble that of the clusters in our galaxy. The reddest stars in NGC 419, however, have B-V = +2.0 as compared with B-V = 1.6 in similar clusters in our galaxy. The color-magnitude diagram which Arp measured to V = 19.4 and B = 20.1 has no horizontal branch. Kron [6] observed the luminosity distribution in NGC 419 by measuring it with apertures of different size. The distribution of luminosity with distance from the center strikingly resembles that of My in our galaxy. In other words, globular star clusters tend to be built on the same models in our own galaxy and in the Magellanic Clouds. 3. SUMMARY In summary, then, the published data suggest that the color-magnitude arrays for clusters in the Magellanic Clouds imply that there are composition differences between the galaxy and these stellar systems. On the other hand, the appearance of the spectra of individual stars as observed particularly by Thackeray and his associates suggest that these differences are not large. Here, then is the paradox; can colors or energy distributions in the integrated light of Magellanic Cloud star clusters throw any light on it? Unfortunately, we cannot yet get a clear-cut answer, but we are convinced that energy distribution measurements can help to solve this problem, although such measurements alone certainly do not suffice. C. MEASUREMENTS OF THE ENERGY DISTRIBUTIONS IN SOUTHERN GLOBULAR CLUSTERS in order to observe the spectral energy distributions in globular clusters we employed the photoelectric spectrophotometer built at The University of Michigan by W. Liller with aid of/a grant from the Rackham Foundation. The device was built originally-to study particularly gaseous nebulae (see Section V), but has also been used for many other problems-stellar energy distributions, comets, etc. Since the instrument has been described in detail by Liller [33] we restrict ourselves here to a brief description of its principles and operation. Figure 2-1 is a schematic diagram of the spectrophotometer. Since the instrument was originally designed for an f/5 system and has to be used in Australia on telescopes of such aperture ratios as f/15 or f/18, it was necessary to install an auxiliary optical system in front of the entrance slit17

more properly called an entrance slot. We also provided a guiding diagonal consisting of an unsilvered, thick glass plate which was placed in the beam. The entrance slot can be varied depending on the purity of the spectrum one wishes to obtain for an extended object. Slot A gives a purity of 6A, B one of 15A, C one of 25A, and D one of 59A. The length of the slot can be limited by a circular aperture which can be varied from 0.5 to 4.0 mm (corresponding to 11" to 88" at the focus of a 74-inch reflector). The light then falls on the Newtonian flat and passes to the collimator mirror and thence to the grating, which is rotated by a motor. With a 60 cycle current the rate of scan is 30A/min at slow speed, 90A/min at medium speed, and 270A/min at high speed. The spectrum formed by the grating is reflected by a second Newtonian flat through the exit slots and a Fabry lens, and thence to the photocell, The exit slots have purities of 2A, 9A, 19A, and 47A. Thus a combination of two B-slots (B-B) gives a purity of 24A, C-C gives a purity 44A, and D-D gives a purity of 106A-provided the cluster under observation fills the entrance slot. Figure 2-2 shows the instrument mounted on the 50-inch telescope at Mt. Stromlo. At the extreme right is the control panel. The guiding eyepiece is fitted with illuminated cross-wires and provision for removing the glass plate. An eyepiece can be inserted in the beam directly behind the entrance aperture and slot so that the observer can verify what the cell "sees." The four knobs controlling the sizes of the aperture and widths of the slots are plainly visible. Above and to the left is seen the wavelength recorder. The grating can be unlocked from the motor and turned by hand when necessary. The cold box which contains the photocell hangs at the bottom of the instrument; from its base dangles the signal lead. The photocell current is further amplified with a GR amplifier and the resultant potentiometer deflections are displayed on a Brown recorder. Let hk denote the deflection on the recording paper as normalized to some selected amplification. Let the true flux from a star or cluster in the wavelength range X to X+d\ at the top of the earth's atmosphere be denoted by Fi. Let Sk denote the combined effects of the sensitivity of the photocell and the transmission of the optics of the scanner plus that of the telescope. Then -kksec z hk = SXFe e where kk is the coefficient of atmospheric extinction, and z is the zenith distance. A priori, the atmospheric extinction is not known; neither is the sensitivity function, So. We determine these quantities in a straightforward way 18

by observing a star whose energy distribution is known, and if necessary by taking another star which can be observed in the course of the night at widely different altitudes. In the northern hemisphere, where fundamental energy distributions have been carried out for a number of stars, one could often choose a single star to serve both purposes. In the southern hemisphere, however, it was necessary to pick a star near the celestial equator whose energy distribution had been measured; we used these stars to determine secondary standards of energy distribution for use in the southern hemisphere. This problem is being treated in a separate investigation. In praxis, it turned out that we determined the atmospheric extinction each night by observing one of a relatively -small number of southern stars that were to serve eventually as southern standards. These stars were in turn compared with northern standard stars. Hence it was possible to determine both Sk and the atmospheric extinction. Of course, one need not determine -S explicitly. He can compare the cluster or nebula directly with a standard star-taking into account the effects of differential atmospheric extinction. We observed the clusters as close to the meridian as possible so as to minimize effects of atmospheric extinction. Table 2-2 summarizes the individual observations. The first column gives the designation of the cluster and its position. The second gives the date of observation; all observations were secured between August 23, 1960, and March 18, 1961, so the year is not entered. The third column gives the instrument employed. Bin denotes the 26-inch reflector at the Mt. Stromlo field station on Mt. Bingar, near Yenda, N.S.W. The 50-inch reflector at Mt. Stromlo was originally the "Great Melbourne Telescope" installed at the now-defunct Melbourne Observatory a century ago. It was purchased by R.v.d.r, Woolley and mounted on Mt. Stromlo. The 74-inch ireflector on Mt. Stromlo was made by Grubb-Parsons. Column 4 gives the scan speed-fast, medium, or slow. With the 50-cycle current, the slow speed was 25A/mm, the medium was 75A/mm, and the fast was 225A/mm. The slots and apertures are denoted in the fifth column. For example, C4-C4 means entrance slot C, aperture 4 (3.0-mm diameter) and exit slot C, aperture 4 (5.0-mm diameter). Entrance aperture 5 has a diameter of 4.0 mm; exit aperture 5 has a 5-mm diameter with a yellow filter. The spectral range covered is listed in the last column, Figure 2-3 shows a scan of the spectrum of 47 Tucanae, as obtained with a medium speed on August 25 with the 50-inch reflector. The spectral resolution is 24A. The yellow filter is introduced to cut out overlapping orders. Tracings were made both in the direction of increasing wavelength and in the reverse direction in order to secure a check on the guiding and the atmospheric transparency. These tracings were then superposed to obtain a final "mean" trace. The discrepancies are larger than we would like-probably they are at least partly due to the drifting of individual stars off from and into the slot. For this reason it is advantageous to use a small telescope in working on a large extended cluster such as C Centauri. 19

Figure 2-4 shows a scan of NGC 1866, as observed with the 74-inch reflector on Mt. Stromloo Notice the prominent hydrogen lines, characteristic of a spectrum near A5, and the atmospheric line at 5577, which serves as a valuable fiducial mark. Here the spectral resolution is 44A. The final results are summarized in Table 2-3 and in Figs. 2-5, 2-6, and 2-7. We measured the scans at the wavelenths chosen by Code [34] and also at many other positions of the spectrum, particularly in the neighborhood of the G band (X~ 4308), \4227, the "H" and "K" lines X3968 and 3933, and also numerous points in the ultraviolet. We compared the clusters individually with stars observed by Code. The primary entries in Table 2-3 give the brightness F(l/%) of the globular cluster in magnitudes relative to the brightness at l/Xo = 1.80. One magnitude corresponds to 4 db. The secondary entries for each cluster give an estimate of the error involved. The size of the error increases at the limits of the scan because of the decreased sensitivity of the photocell, and decreased in the ultraviolet because of the increased uncertainty in the atmospheric transmission. In Figs. 2-5, 2-6, and 2-7 the brightness is plotted in magnitudes against 1/X (in reciprocal microns as abscissa). Vertical dotted lines indicate the order of magnitude of the errors involved. They are estimated uncertainties, not errors calculated by statistical theory. We deemed it best to retain the "wiggles" in the individual curves. Some of them are almost certainly spurious, but others, such as those corresponding to the "G-Band" and particularly to the H and K lines, seem to be unambiguously established. We must emphasize, however, that these fine features of the spectra can be observed only with slit spectrograms; the resolution of the scanner is too low and the noise level too high to permit delineation of many spectral characteristics that are useful for assessing absolute magnitude effects, metal deficiencies, etc. in Fig. 2-5 we compare NGC 362, c Centauri, NGC 2808, and 47 Tucanae. Of these clusters, 47 Tucanae is probably the most "metal rich." There is a sharp drop in the continuum in the neighborhood of the H and K lines because of the many overlapping metallic lines in this spectral region. The spectral distribution fits nr Hercules G4III for 1/A < 2.6, but is brighter than this star in the ultraviolet (l/\ > 2.6) by about 0.3 magnitudes. If we fit the energy distribution of 47 Tucanae to that of 31 Comae GOIII, we find that a good fit can be obtained for 1/\ > 2.3. That is, 47 Tucanae is redder than GOIII. It probably corresponds to about G2III. The cluster NGC 2808 appears to have about the same energy distribution as 47 Tucanae; i.e., it corresponds to about G2III, being redder than a G4dwarfo It appears to be more metal deficient than 47 Tucanae but is certainly not an extreme type II population object such as a subdwarf. 20

The energy distribution in w Centauri falls between that of a G4V star such as 51 Pegasi, and that of a G2V star such as 16 Cyg A. It appears to be metal deficient in comparison with 47 Tucanae. We experienced some difficulty in measuring the ultraviolet energy distribution in NGC 362 (1/A > 2.7) when the highest resolution 24A was sought, so we have disregarded these measurements in drawing the final curve. The cluster appears to be definitely metal deficient; its energy distribution fits that of a G4V star to within the estimated error of the measurements, although it may be slightly redder. It is much bluer than a G4III star. We estimate its energy distribution as corresponding to that of a G5V star. In Fig. 2-6 we compare 47 Tucanae with NGC 6752 and NGC 419, a globular cluster in the Small Magellanic Cloud. NGC 6752 fits 51 Pegasi G4V over nearly the entire range covered. Because of its faintness, NGC 419 was observed only in the blue. In the interval 2.0 < 1/\ < 2.6 its energy distribution fits that of 51 Pegasi G4V but it has strong H and K lines and a strong G band. The weakness of the cluster in the ultraviolet may partly arise from observational error, but part of it may arise from the influence of the extremely red stars noted by Arp [26]. Further observations of this cluster are essential, since the present fragmentary data would suggest that the cluster is abnormally rich in metals. Figure 2-7 shows the data for the clusters in the Large Magellanic Cloud. NGC 1866 and NGC 1831 were observed with higher resolution, i.e., 44A, than the other two clusters, NGC 1783 and NGC 1978, which could be measured only with a resolution of 106A. The energy distribution in NGC 1866 resembles that of a late A-type star. In the interval 1.8 < 1/ < 2.6 the slopes of the energy curve corresponds to a spectral class a little later than A5III, although the curve shows a drop in the extreme red. The Balmer jump corresponds to that of B Tri spectral class A5III. The strong Balmer lines are easily visible upon the tracing and show clearly that the spectral class is definitely A, in agreement with Thackeray's result (cf. remarks to Table 2-1). The energy distribution in the cluster NGC 1831, which was also observed with the 74-inch reflector, shows a good fit to 110 Hercule's F6V in the interval 1.7 < 1/\ < 2.5, but the Balmer jump seems to be too large and the ultraviolet portion of the spectrum appears to be relatively too faint by about 0.5 magnitude in the region 2.65 < 1/\ < 2.9. Here again we must emphasize that the ultraviolet data are uncertain and that further observations should be obtained. The' Balmer lines still seem to be relatively strong. The clusters NGC 1785 and NGC 1978 are fainter than NGC 1866 and NGC 1831. Unfortunately we-were unable to observe them with the large telescope. Not only is the spectral purity lower but the influence of sky background is greater, particularly in the ultraviolet. The energy distribution in NGC 1783 appears to resemble that of 51 Pegasi G4V for 1/\ < 2.65. In the ultraviolet the errors appear to be large because of effects of the sky. The star 21

is bluer than v Hercules G4III except in the ultraviolet. The energy distribution in NGC 1978 seems to fit that of a G2V star to within the observational error. Note that what we have derived are energy distribution equivalents, not spectral classes. The results are summarized in Table 2-4. Among globular clusters associated with our galactic system, the spectral classes tend to be earlier than the energy distribution equivalents-which is exactly the effect we would anticipate for stars that were deficient in metals! The exception is 47 Tucanae, which everyone agrees has a normal or nearly normal metal/hydrogen ratio. Because accurate spectral classes are not now available for the clusters in the Magellanic Clouds, similar comparisons cannot be made. The fragmentary data do suggest that perhaps the metal/hydrogen ratio tends to be more nearly "normal" in the globular clusters in the clouds than in those of our own galaxy. Note that the energy distribution equivalent of NGC 1783 is later than the assigned spectral class, whereas the opposite is true for NGC 1831. Possibly the spectral classes are themselves in error; the energy distributions in NGC 419, NGC 1978, and NGC 1783 are very similar, as are their colormagnitude arrays. No emphasis can be laid on the amount of the quantitative difference between the energy distribution equivalents and the spectral classes. In each instance one deals with the composite spectra of a host of stars. A stellar distribution ranging from FO to K5 may give a spectral class resembling G2, but the energy distribution may differ from that of a G2 star. The present investigation suggests that spectral scans of star clusters may be helpful in getting an estimate of the metal/hydrogen ratio. Before the data can be interpreted quantitatively, it will be necessary to supplement it with spectroscopic observations (for line intensities), color-magnitude arrays, and luminosity functions. There is some question whether spectral scans of the type we have obtained are the most useful kind of observations to be secured with a spectral scanner. When quantitative studies of the composite spectra of clusters are carried out, it will be possible to select certain spectral features for special attention. An extension of the Strtmgren photometric technique to globular clusters is clearly suggested, but the actual spectral intervals must be chosen with care. Furthermore, attention must be paid to getting spectral energy distributions for a variety of stars of different hydrogen/ metal ratios and surface temperatures, We secured spectral scans of NGC 330 in the SMC. This object is an open cluster containing many stars of early spectral type; Fig. 2-8 shows its energy distribution. Although the Balmer jump corresponds to that of a B5 star, the energy distribution and the strength of the hydrogen lines correspond to a later spectral class. 22

SECTION II: TABLES, FIGURES, AND REFERENCES

TABLE 2-1 PROPERTIES OF SOME SOUTHERN GLOBULAR CLUSTERS (Compiled by Herbert Rood) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (15) (14) NGC a(1950) (900)b C-m 1-f Spetral P-m P-V Mod. Dist. Vr Mass Conc. Diam. h m ~'C-r I -CH/Hy Hy FelI (47 Tue) 104 00 21.9-72 21 272 -45 G3 G3 G3 4.7 o.67 15.5(.4) 4.6 - 24(3) <6x105 III 44 [1] [1] [1] [1] [3] [5] [5] [5] [7]1 [7] [12]2 [5] [8] [1] [1] coCent 5139 13 23.8-47 03 277 +15 F7 F8 FO 4.25 0.68 13.0(.7) 4.0 +230(7) VIII 101 65.4 10.4 [1] [1] [ [ 1[1] [4] [4] [10] [10] [10] [7] [7] [12] [5 [12] [15] [1] [31] 362 01 00.6-71 07 268 -47 F8 GO F7 8.0 15.0(.7) 10 +221(7) III 17.7 4.0 [1] [1] [1] [1] [5] [5] [5] [1] [12] [5] [1] [1] [31] 2808 09 10.9-64 39 250 -11 F8 F7 F7 7.8 14.0(.8) 6.3 +101(7) I 18.8 [1] [1] [1] [1], [5] [5]- [5] [1] [12] [5] [1] [1] 6752 19 06.4-60 04 304 -27 F6 F7 F5 7.2 13.5(.8) 5.0 - 39(5) VI 41.9 [1] [1] [1] [1] [5 ] [5] [5] [1] [12] [5] [1] [1 ] LMC 1783 4 59.0-66 03 [17] F5 10.55 +0.48 VI VI 2.1 1.2 1.4 [28] [19] [19] [7 []7 [18] [29] [18] [28] [29] V1<)5~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ "Iarge LMC 1831 5 6.1-64 59 [20] G5 10.65 +0.18 V Concentrated" 1.0 1.5 [28] [19] [19 ] [29] [21] [28] [29] "Large LMC 1866 5 13.5-65 31 F8 9.38 +0.07 IV Concentrated" 0.9 2.2 [50 ] [19] [19 ]5 [29] [21] [28] [29] LMC 1978 5 28.0-66 16 [235] G 10.59 +0.64 VI V 1.9 1.0 1.0 [28] [19] [19] [-]7 [-]7 [29] [18] [18] [28] [29] SMC 419 1 7.1-73 6 [26] G 10.62 0.54 0.8 1.4 [28] [19 ]3 [19 ]5 [-]7 [-]7 [28] [28 ]6 1. Kron [6] gives a plot of relative magnitude vs. the 6 colors of Stebbin and Whitford. 2. Feast and Thackeray [8] obtain corrected modulus = 13.2. Wildey [35] obtains corrected modulus = 14.1 if Mv = 0.0 for the horizontal branch. He gets corrected modulus < 13.2 if the blue end of horizontal branch is as faint or fainter than the horizontal branch of NGC 6356. 3. Kron [6] gives a plot of relative magnitude vs. the 6 colors of Stebbin and Whitford. 4. In [28] it is stated that Thackeray gets Type A3 from a slit spectrum. 5. In [28] it is stated in Note (3) under Table XV that Irwin (private communication) has measured photoelectric colors: For NGC 1866 P-V = 0.01; for NGC 419, P-V = 0.55. 6. In [28] Gascoigne lists the diam. 1:4 from Harvard plates, given "somewhere" in the literature. 7. See Section 2a, "The Distance Modulus."

TABLE 2-2 SOUTHERN HEMISPHERE GLOBULAR CLUSTER SCANS Name Scan Date Instrument Scan Speed Slots Spectral Range c) Centauri March 14-15 Bin Slow B 3570A-4800A a = 13h23m8 ( 0 March 15-16 Bin Medium C4-C4 3400A-5900A 8 = 47~03' ( ) March 11(?) Bin Medium C4-C4 3400A-5900A March 11-12 Bin Fast D4-D4 3300A-6100A 47 TUC Oct. 12-13 Bin Medium B5-B4 3400A-6000A Aug. 25-26 50" Medium C5-C4 3400A-5000A a = oh21m9 (1950) Aug. 25-26 50" Medium B5-B5 3600A-5900A 8 = 72021' Aug. 24-25 50" Fast C5-C4 3400A-5900A Aug. 25-26 50" Fast B 5000A-5868A Aug. 26-27 50" Fast C5-C4 3400A-5900A NGC 362 Aug. 26-27 50" Medium B5-B4 3600A-5700A a = 01hOO006 Aug. 24-25 50" Fast C5-C4 3500A-6100A 8 = 71~07' (195 Aug. 25-26 50" Fast B5-B4 3500A-6000A Nov. 9-10 Bin Medium C4-C4 3500A-6000A NGC 2808 a = 09hldm9.(195o March 16-17 Bin Medium C4-C4 3800A-6000A 8 = 64039' ( March 16-17 Bin Fast D4-D5 3400A-6000A NGC 6752 a = 19ho6.m4 Oct. 16-17 Bin Fast D5-D4 3400A-6000A = -600o 4,(1950) Oct. 16-17 Bin Fast C5-C4 3400A-4900A iMC-NGC 1831 Nov. 14-15 74' Medium C5-C4 3800A-4950A a8 5h-64Q (1950), Nov. 14-15 74' Fast D5-D4 3400A-5900A = -649 J Jan. 14-15 Bin Fast D4-D4 3400A-4900A LMC-NGC 1783 a = 4h59s 0 1 Sept. 20-21 50" Fast D5-D4 3500A-5900A 8 = -66 3' (1950) Jan. 13-14-14-15 Bin Fast D4-D4 3500A-4900A J (Composite) LMC-NGC 1866 a 5hl3m5 Oct. 27-28 74" Medium C4-C5 3500A-5800A a 6 ~ 1* (1950)4 r Jan. 13-14 Bin Fast D4-D4 3400A-5900A 8 _ -05 J Oct. 27-28 74" Fast D4-D5 3400A-5900A LMC-NGC 1978 Jan. 17-18 50" Fast D5-D4 3400A-6000A a = 5h28m0 (1950) Nov. 23-24 50" Fast D5-D4 3400A-5900A 8 = -66 ~16' 9 Jan. 9,13-14,14-15 Bin Fast D4-D4 3400A-4900A j - (Composite) SMC-NGC 419 Sept. 19 50" Medium C4-C4 3600A-4900A a h7l (= 1950 Sept. 19 50" Fast D4-D4 3400A-5000A 8= - 73~6' Aug. 28-29,30-31 50" Fast D4-D4 3400A-4900A (Composite) SMC-NGC 330 Nov. 22-23 50" Medium C5-C4 3400A-5000A a _7241, (1950 8 h54i o (1950)0 Nov. 22-23 50" Fast D5-D4 3400A-5900A 8 = 74 j,Aug. 31* 50" Fast C3-C4 3900A-5000A S26= SMC-"Kron's Blue Cluster" a = oh55m8 (1950)l Nov. 23-24 50" Fast D4-Df4 3400A-6000A 8 = -72032' J Nov. 23-24 50" Fast D3-D4 3400A-4900A *Very low weight; scale uncertain; sky uncertain. 26

TABLE 2-3 GLOBULAR CLUSTER MAGNITUDES RELATIVE TO 1/\o = 1.80 1/* = 2.94 2.74 2.59 2.48 2.39 2.18 1.975 1.80 1.72 w Centauri 2.12 1.56 1.19.83.79.43.21 0.0 -.25 Est. Error.40.10.10.05.05.10.10.10.40 47 TUC 2.25 1.90 1.75 1.15 1.09.53.22 0.0 -.10 Est. Error.25.15.08.05.05.04.05.12.15 NGC 362 1.88 1.65 1.34 1.03.86.48.28 0.0 -.15 Est. Error.50.50.20'.10.05.04.10.15.25 NGC 2808 2.18 1.96 1.63 1.33 1.11.71.38 0.0 -.10 Est. Error.07.10.15.05.08.15.20.30 LMC-NGC 1831 2.17 1.78.97.49.46.26.14 0.0 -.o6 Est. Error.15-.07.05.10.05.07 LMC-NGC 1866 1.75 1.21.40..13.07 -.12 0.0.18 Est. Error.40.20.05.07.05.10.20.15.25 LMC-NGC 1783 2.41 1.80 1.26.76.54.46.26 0.0 -.07 Est. Error.80.20.15.10.10.10.10 NGC 6752 1.92 1.46 1.21.88.78.46.22 0.0 -.o6 Est. Error.40.15.10.10.05.05.10.05.10 LMC-NGC 1978 2.04 1.63 1.33.99.78.47.25 0.0 -.10 Est. Error.20.20.10.20.10.10.15.10.13 SMC-NGC 330.34.28 -.13 11 -.10 -.o 07 -.03 0.0.16 Est. Error.20.05.10.07.05.05.10.15.20 *1/\ is given in reciprocal microns. 27

TABLE 2-4 ENERGY DISTRIBUTION EQUIVALENTS FOR GLOBULAR CLUSTERS Energy Distribution Spectral Class Name NGC ~Name__ _ NGC_ Equivalent (See Table 2-1) 47 Tucanae 104 G2III G3 362 G4V F8 2808 G2III F7 w.Cehtauri 5139 G4V F8 6752 G4V F6 SMC Cluster 419 G4V G LMC Clusters 1783 G4V F5 1831 F6V G5 1866 A8III A 1978- G2V G 28

GUIDING COLD BOX DIAGONAL GRATING PHOTO MULT1 PL1ER \ CORRECTING r\ 1 I " \ \ ^ ~^ ~ - l I ENS-FA R LENS TE LES C~OPE ~ —— ^ *- \< ^. S L I \ I I, I~FI ro / / /~ \1 v \ \ \ CORRPRECTIALID OF NT \OI \ Fig. 2 -1. m dg o \ ta \ I l. \ \ ^\ COL IMATOR Fig. BOLOID 2F c dteMican COLLIMATOR Fig. 2-1. Schematic diagram of the Michigan spectrophotometer.

0 Fig. 2-2. The Michigan spectrophotometer mounted on the 50-inch telescope at Mt. Stromlo.

e X |.47 TUCANAE |.~~..''' ~~~~~~~AUG. 25 1960 Yellow Filter G Band H K Fig. 2-3. -Scan of 47 Tucanae (Aug. 25, 1960). Yellow ^^ NGC 1866 Fiter Oclober 27, 1960 HH X4340 X4101 X4861 )5577 Night sky / I \ SKY A^\ _______ BACKGROUND 5800 57 56 55 54 53 52 51 50 4900 Fig. 2-4. Scan of NGC 1866 (Oct. 27, 1960).

BRIGHTNESS IN MAGNITUDES 00 -^ 4 c ().) cn cn ( J N N _ 0 ( 0 b0 o 0 b 0 n bLn<~ ^ ~' ~i0 0 -q ~3 c) c+d-~ --------- co ~3/'^ c~ t3________________. *____________L___________^___________* r____________

0.0 0.5 -' 1.0 1.5 2.0 2.5 r' I i u3.5- -'^t3 () C 4.0 4.5 -K Z T 4.0 7 49 _ NGC 41952 4.9 4 5.5!.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 Fig. 2-6. Energy distributions for 47 Tucanae, GC 6752 ad NGC 419. ~~~5.5~~3 55

hcj J. Oq IV BRIGHTNESS IN MAGNITUDES,* 0o <D D O0:X) - -4.- 03 0 (. t. C s N N - 0o b u i 0 b 0 i.n 0 01 0 0 0 0b C. l 0 H) 00 ot 03 H NI C h) 0 t4 —' C O

I I I I-I-I I I I -^^ —-^-I' NGC 330 i t t,,ni - 0.5 I/X-.I I I I I I I I I I 0 1.7 2.0 2.3 2.5 2.9 Fig. 2-8. Energy distribution for NGC 33550.

REFERENCES 1. Helen Sawyer Hogg, Handbuch der Physik Lll1, 204-206 (1959). 2. J.L.E. Dreyer, New General Catalogue of Nebulae and Clusters of Stars (1888). 3. R. L. Wildey, Ap. J. 133, 430 (1961). 4. E. P. Belserene, A. J. 64, 58 (1959). 5. T. D. Kinman, M. N. 119, 538 (1959). 6. G. E. Kron, P.A.S.P. 73, 202 (1961). 7. S.C.B. Gascoigne and E. J. Burr, M.N.R.A.S. 116, 570 (1956). 8. M. W. Feast and A. D. Thackeray, M.N 120, 463 (1960). 9. A. D. Thackeray, A.J. 64, 437 (1959). 10. W. W. Morgan, P.A.S.P. 68, 509 (1956). 11. W. W. Morgan, A.J. 64, 432 (1959). 12. T. D. Kinman, M.N.A.S.S.A. XVII, No. 3 (1958) = Radcliffe Observatory Reprint No. 9. 13. M. W. Feast, A. D. Thackeray, and Wesselink, M.N. 120, 64 (1960). 14. E. P. Belserene, A.J. 66, 38 (1961)A. 15. E. M. Linsay, Irish Astronomical Journal 2, pp. 145-146 (1953). 16. Helen Sawyer Hogg, Handbuch der'Physik Llll, 182 (1959), David Dunlap Publ. 1, No. 20 (1939) and 2, No. 2 (1955). 17. A. Sandage and 0. Eggen, M.N. 121, 232 (1960). 18. P. W. Hodge, Ap. J. 131, 351 (1960). 19. S.C.B. Gascoigne and G. E. Kron, P.A.S.P. 64, 196 (1952). 20. P. W. Hodge, Nature 186, 622 (1960). 37

REFERENCES (Concluded) 21. P. W. Hodge, Ap. J. 133, 413 (1961). 22. H. C. Arp, A. Sandage, A. D. Thackeray, Mt. Wilson and Palomar Report 1956-57, P. 51. 23. P. W. Hodge, Ap. J. 132, 346 (1960). 24. H. C. Arp, A.J. 64, 254 (1959). 25. G. de Vaucouleurs, P.A.S.P. 71, 202 (1959). 26. H. C. Arp, A.J. 63, 273 (1958). 27. A. D. Thackeray and A. J. Wesselink, Nature 171, 693 (1953), M.N.A.S.S.A. 13, 99 (1954). 28. S.C.B. Gascoigne, Suppl. to the Australian Journal of Science, Vol. 17, No. 3, Dec. 1954, pp. 27-28. 29. H. Shapley and H. B.'Sawyer, Harvard Bull. 852, 22 (1927). 30. Annie J. Cannon, Harvard Bull. 868, 1 (1929). 31. S. van Den Bergh, Z. Ap. 41, 61 (1956). 32. P. N. Kholopov, Var, Stars l1, No. 3, 202 (1958), H. Wilkens, Circ. La Plata Obs. No. 16 (1960). 33. W. Liller, Publ. Astron. Soc. Pac. 69, 511 (1957). 34. A. Code, Stellar Atmospheres, ed, V. L. Greenstein, Univ. of Chiago Press (1960). 38

SECTION III SPECTROPHOTOMETRY OF THE WOLF-RAYET STAR OF 72 VELORUM by L. H. Aller, D. J. Faulkner, and Herbert Rood

I

Wolf-Rayet stars are among the most perplexing of the hydrogen deficient stars in the sky. They are high temperature objects with broad emission; lines of helium, carbon, nitrogen, and oxygen, and often with sharp absorption lines on the violet edge of the emission structure. A number of these objects occur in the constellation of Cygnus, where they have been intensively studied by numerous investigators. The brightest Wolf-Rayet star in the sky is the more luminous component of the binary y Velorum (visual magnitude = 2.22). The fainter component some 40" away, y Velorum = HD 68243, spectrum B2IV (visual magnitude = 4.57), appears to be a normal star, whose absolute magnitude is about -3.1. Hence Y2 Velorum must have an absolute magnitude MV _. - 5.6. Henry V. Smith [1] suggusts that Y2 Velorum and Zeta Puppis, a star of somewhat similar temperature and absolute magnitude, are not only apparently close together in the sky but actually close together in space. They may be associated with an association of bright B stars, and the two of them together probably excite the gaseous nebula discovered by Gum [2]. The true dimensions of this nebula are greater than 650 light years, making it the largest such object known in our galactic system. The spectrum of Y2 Velorum is unusual in that its spectrum is composite and variable. Superposed on the wide lines of a carbon Wolf-Rayet star specral class WC7 are broad hydrogen and helium absorption lines characteristic of an early 0 star, which Smith classifies as 07. Many of the strong emission lines have strong central absorptions. Remarkable spectral variations involving both emission and absorption features were noted by Perrine [3] and studied intensively by Smith in March 1953, who notes that periods of activity are ephemeral and infrequent. Shells of material appear to be ejected from the star from time to time. Other examples of Wolf-Rayet stars with variable spectra are known, e.g., HD 45166. A spectrophotometric study of both components of this system was undertaken by us at Bingar and at Stromlo. Spectral scans of the fainter component were difficult to -secure and most of our effort was concentrated on measuring the energy distribution in the brighter component. Scans were obtained on the nightb of Oct. 12-13 and Nov. 5-6, 1960 at Bingar and Jan. 28-29, Jan. 30-31, and Jan. 31, 1961 at Stromlo. Figure 3-1 shows a scan obtained with medium speed. Different gain settings of the amplifier were required. If after reducing each scan to a common amplifier gain,, the wavelength (abscissa) is measured in Angstroms and the height of the continuum (ordinate) is measured in arbitrary units, then the quantity Ax (called "Equivalent Angstroms") may be defined for each observed spectral line on the tracing by area of the line at A A = -- o height of the continuum at X 41

Ap is thus tho —width (in Angstroms) of a rectangle, the height of which is equal to the height of the continuum at \ and the area of which is equal to the area of the line at X. We may use A\ values to derive relative line intensities as soon as the energy distribution in the star is known. Table 3-1 gives Ax values for the observed lines. Column 1 gives the wavelength of each line. Column 2 gives its identification (when known). The next 10 column give, for each night, the average Ak value obtained from the scans of that night and the uncertainty estimate, which is taken as the mean difference between the average Ak value and the Ak value for each tracing of the night. Where no uncertainty estimate is listed, there was only scan. The last column gives'<X the average A; value for all the nights. The color temperature is obtained by the following procedure. For a given wavelength, the apparent magnitude of the continuum ml/X (corresponding to intensity per unit wave-number) is obtained from the formula ml/X = -(2.51oghk + kksecz + S\)' + constant where hk is the continuum height at. measured in arbitrary units, kk is the atmospheric absorption at the zenith (in magnitudes) at the place of observation, z is the zenith distance at the time of observation, and S. is the relative sensitivity of the telescope-photocell combination in magnitudes(per unit wave number) obtained by comparing the tracings of standard stars with the known intensity distributions given by Code (see [34] of Section II). The constant depends on the units in which h is measured. Since only differences in magnitudes at various wavelengths are used to obtain the color temperature, we need not know the value of the constant and the magnitudes given in Table 3-2 are obtained by arbitrarily setting the constant equal to zero. Table 3-2 gives -ml/X at spectral line wavelengths for the five nights of observation, the estimated uncertainty, and the average value of -ml/X for the five nights. Plots of -ml/X vs. 1/\ were constructed for each of the five nights and superimposed to obtain the final linear -ml/A vs. 1/X plot given in Fig. 3-2. With use of this graph, the color temperature may be easily obtained, from the Planckian formula logB(l/l) - logB(l/%2) = log (lhc/kX2-l) - logs% (1hc/kTl l 1) where B(l/Xl)/B(l/?2) is the intensity ratio for wave numbers l/Al and 1/X2, h = Plank's constant, k =,Boltzmann's constant, c = speed of light, and T is the color temperature. Thus the color temperature is found to be 32,000~K. Although the intensities of absorption lines may be expressed conveniently in equivalent angstroms, X emission line intensities so described can be compared with one another only if the energy distribution in the continuous spectrum is known, That is, to compare A(lX) with another line A(X2) 42

we must know F( X)/F(2)). We may express the relative intensities of all emission lines in terms of a hypothetical emission line of equivalent width A(X) = 1 at X5000. Thus the intensity IX of any emission line of equivalent width Ax is given by IX 5000 5 (hc/5kTxlO5 h 5000 il (1, -1) 1 = ---- (1h.-.T) AX Table 3-3 gives I/Io at the spectral line wavelengths for the five nights of observation, the estimated uncertainties, and the average value of I Io for each night. "Apparent" excitation temperatures may be obtained by comparing intensities of emission lines arising from two suitable levels of a given ion. That is, using the above values of I/Io, the excitation temperatures for various combinations of lines are computed from Boltzmann's formula W1 = e -(X1.- X2)/kT ^1A1v1 -/Io A2= ev2 I2"/IO hD2A2V2 where Xi and X2 are the excitation potentials for the lines of wave numbers l/X1 and 1/^, Al and A2 are Einstein probability coefficients and w and w2 are statistical weights. For the lines used to obtain excitation temperatures, excitation potentials were obtained from Ref. [4],Table 7 and SlAAlv were obtained from Ref. [4], Table 6. Excitation temperatures derived from various line combinations are given below: CIII 4069/4650 23,000~K CIII 3609/4650 22,000~K CIV 3934/5805 85,000~K CIV 4441/5805 110,000~K CIV 3411/3562 36,000~K CV 4124/5590 1,900,000~K Abundances of various elements cannot/be determined for this star until a more complete study of its spectrum is made. 43

I

SECTION III: TABLES, FIGURES, AND REFERENCES

TABLE 3-1 AX VALUES OF 72 VELORUM (AX = Area of Line at X/Intensity of Continuum at x) () Ident. Oct. 12-13, 1960 Nov. 5-6, 1960 Jan. 28-29, 1961 Jan. 30-31, 1961 Jan. 1, 1961 31 Ah () E.U.* (I) AX (L) E.U.* (X) AX (A) E.U.* (IL) AN (I) E.U.* (I) A, (X) E.U. ) 3411 OIV 2.98 (0.30) 2.88 3.54 (0.51) 2.96 (0.42) 3.94 3.26 3500 0.670.67 3560 OIII,OIV,HeI 1.20 (0.06) 1.49 1.20 (0.10) 1.64 (0.04) 1.33 1.37 3607 CIII 1.65 (0.14) 2.181.93 (0.48) 1.54 (0.20) 1.591.78 3687 2.55 (0.21) 3.10 2.72 4.41 3.89 3.33 3722 CIV 3.50 (0.34) 4.43 3.88 (0.70) 5.53 (0.31) 4.20 3.91 3758-3769 OnI 1.61 (0.05) 2.53 2.81 (0.45) 3.30 (0.07) 2.322.51 3811 OVI 0.98 (0.16) o.69 1.70 (0.02) 1.40 1.01 1.16 3835 OVI 1.00 (0.35) 1.36 (0.27) 1.051.14 3889 Hel, CIII.96 (0.06) 2.50 5.22'(0.42) 5.34 (0.76) 5.454.49 3934 CIV5.82 (0.6) 2.605.78 (0.37) 5.78 (0.13) 6.465.29 3965 HeII,OIII 0.81 0.810.81 4025 Hel,HeII 1.37 1.921.65 4069 CIII 6.32 (0.68) 5.95 6.84 (0.81) 7.48 (0.94) 7.456.81 41007 CIII 41231 OV 2.91 (0.67) 2.84 3.01 (0.59) 2.80 (0.44) 2.50 2.81 4157 CIII 1.44 (0.13) 1.17 1.56 (0.32) 1.23 (0.33) 1.111.30 4189 CIV 1.54 (0.08) 1.61 1.48 (0.25) 2.12 (0.61) 2.321.81 4229 CIII 4235 Hell 0.88 (0.11) 1.26 1.08 (0.20) 1.28 (0.43) 1.13 1.13 4267 CII 2.74 (0.26) 3.22 3.43 (0.35) 4.25 (0.70) 3.04 3.34 4330 CIII,HeII 3.64 (0.24) 3.22 4.48 (0.29) 4.11 3.62 3.81 4375 CIII,HeII 3.10 (0.23) 1.66 2.68 (0.36) 2.01 (0.01) 1.752.24 4441 CIV 4.38 (0.23) 4.00 4.60 (0.45) 4.48 (0.20) 3.574.21 4471 HeI 1.04 (0.18) 1.24 2.68 (0.52) 0.99 (0.61)1.49 4513 CIII 1.02 (0.02) 0.72 1.95 0.84 (0.03)1.13 4544 Hell 1.52 (0.03) 0.78 1.60 1.32 (0.36) 1.31 4650 CIV,CIII 61.8 (5.8) 57.8 79.4 (5.4) 67.2 76.068.4 4686 Hell 16.4 (2.4) 15.8 19.8 (0.3) 19.4 23.919.1 4786 CIV 1.47 (0.06) 1.94 2.50 (0.52) 2.712.16 4861 Hell 1.88 (0.18) 4.34 3.35 2.80 4.53 3.38 4923 HeI 1.37 (0.04) 1.59 1.571.51 5019 CIV,HeI 2.50 (0.41) 2.28 4.85 4.00 4.88 3.70 5060 0.85 0.870.86 5132 CIII 6.61 (0.82) 3.16 4.80 3.46 6.40 o4.89 5248-5257 CIII 1.74 (0.53) 0.73 0.76 0.56 1.531.06 5411 Hell 2.79 (0.37) 1.76 4.36 3.08 (0.31) 3.51 3.10 5470 CIV 4.14 (0.21) 3.09 3.90 3.45 (0.25) 2.92 3.50 5592 OV,OIII 2.20 1.70 2.252.05 5695 CIII 51.2 (3.3) 60.2 55.4 54.4 55.054.8 5805 CIV 40.0 (0.5) 35.4 36.7 39.9 44.6 39.3 5876 HeI 8.56 (0.87) 7.51 10.1 13.9 14.410.9 *Estimated Uncertainty Average % Error = Z Estimated Error/ZAX (with errors estimated) = 41.48/476.62 = 8.7%.

TABLE 3-2 -ml/% VALUES OF 72 VELORUM (-ml/X = Appareit Magnitude of the Continuum at X) Oct. 12-13, 1960 Nov. 5-6, 1960 Jan. 28-29, 1961 Jan. 30-31, 1961 Jan. 31, 1961 (A) -/ h' -mlA, E. U.E.U ml E.U.* -ml/ E. U. -ml/ - E. U. * -U/A 3411 5.99 (0.07) 6.48 5.77 (0.02) 5.89 (0.06) 5.70 5.97 3500 6.59 6.59 3560 5.95 (0.06) 6.60 5.67 (0.02) 5.79 (0.04) 5.61 5.92 3607 5.96 (0.07) 6.42 5.66 (0.03) 5.76 (0.03) 5.59 5.88 3687 5.92 (0.02) 6.53 5.56 5.73 5.53 5.85 3722 5.90 (0.04) 6.49 5.59 (0.04) 5.68 (0.02) 5.52 5.84 3758-3769 5.86 (0.03) 6.48 5.52 (0.06) 5.63 (0.03) 5.47 5.79 3811 5.85 (0.02) 6.52 5.52 (0.05) 5.65 5.47 5.80 3835 5;85 (0.02) 5.59 (0.05) 5.45 5.62 3889 5.86' (0.07) 6.42 -5.48 (0.05) 5.58 (0.04) 5.41 5.75 3934 5.79 (0.01) 6.43 5.49 -(0.05) 5.58 (0.04) 5.41 5.74 3965 5.54 5.42 5.48 4025 5.54 5.395.46 4069 5.75 (0o.o) 6.37 5.50 (0.07) 5.56 (0.05) 5.40 5.72 41007-41231 5.75 (0.o0) 6.31 5.49 (0.06) 5.56 (0.04) 5.40 5.70 4157 5.73 (o.o1) 6.28 5.43 (0.05) 5.54 (0.04) 5.37 5.67 f- 4189 5.73 (0.02) 6.26 5.44 (0.06) 5.53' (0.04) 5.37 5.67 4229-4235 5.70 (0.02) 6.23 5.42 (0.05) 5.50 (0.05) 5.34 5.64 4267 5.69 (0.02) 6.24 5.41 (0.06) 5.51 (0.04) 5.32 5.63 4330 5.65 (0.03) 6.22 5.39 (0.05) 5.52 5.32 5.62 4375 5.63 (0.00) 6.20 5.36 (0.05) 5.49 (0.03) 5.30 5.60 4441 5.65 (0.02) 6.19 5.35 (0.05) 5.48 (0.04) 5.31 5.60 4471 5.63 (0.02) 6.20 5.35 (0.06) 5.47 (0.03) 5.315.59 4513 5.64 (0.02) 6.21 5.41 5.47 (0.04) 5.68 4544 5.62 (0.02) 6.20 5.39 5.44 (0.04) 5.66 4650 5.66 (0.05) 6.27 5.35 (0.07) 5.465.22 5.59 4686 5.65 (0.04) 6.22 5.32 (0.08) 5.41 5.23 5.57 4786 5.62 (0.01) 6.17 5.26 (0.03) 5.34 5.60 4861 5.66 (0.01) 6.22 5.11 5.28 4.89 5.43 4923 5.69 (0.01) 6.26 5.36 5.76 5019 5.61 (0.02) 6.13 5.14 5.28 4.80 5.39 5060 5.48 5.155.31 5132 5.46 (0.00) 6.04 5.12 5.23 4.98 5.37 5248-5257 5.44 (o.oo) 5.97 4.85 5.15 4.76 5.23 5411 5.35 (0.00) 5.91 4.93 5.10 (0.02) 4.82 5.22 5470 5.38 (0.01) 5.92 4.92 5.11 (0.02) 4.83 5.23 5592 4.91 5.11 4.82 4.94 5695 5.44 (0.02) 5.99 5.00 5.18 4.88 5.30 5805 5.32 (0.03) 6.01 4.82 4.98 4.665.16 5871 5.25 (0.16) 5.78 4.59 4.68 4.47 4.95 *Estimated Uncertainty.

TABLE 3-3 IV/Io VALUES OF 72 VELORUM (Ix/Io = Intensity of Emission Line of Equivalent Width AX) x, ( R) Oct. 12-13, 1960 Nov. 5-6, 1960 Jan. 28-29, 1961 Jan. 30-31, 1961 Jan. 31, 1961,,/IIo E.U.* IT/Io E.U.* Ih/Io E.U.* IUI;IO E. U. E.U.* I"IT 3411 10.87 (1.10) 10.49 12.89 (1.86) 10.78 (1.53) 14.33 11.87 3500 2.122.12 3560 3.77 (0.19) 4.67 3.77 (0.31) 5.15 (013) 4.17 4.31 3607 5.00 (0.42) 6.61 5.85 (1.46) 4.66 (0.61) 4.82 5.39 3687 7.27 (0.60) 8.84 7.75 12.6 11.1 9.51 3722 9.65 (0.94) 12.2 10.7 (1.93) 9.74 (0.86) 11.6 10.78 3758-3769 4.30 (0.13) 6.76 8.71 (1.40).04 (0.19) 6.207.00 38 112.53 (0.41) 1.78 4.38 (0.05) 3.61 2.61 2.98 3835 2.51 3.41 (0.68) 2.64 2.85 3889 9.46'(0.14) 5.98 12.5 (1.01) 12.7 (2.52) 13.0 10.73 39341 13.4 (1.45). 98 13.3 (0.85) 13.3 (0.30) 14.9 12.18 3965 1.81 1.81 1.81 4025 2.88 4.04 3.46 4069 12.7 (1.37) 12.0 13.7 (2.00) 15.0 (1.88) 15.0 13.68 41007-41231 5.65 (1.30) 5.50 5.84 (1.15) 5.43 (0.85) 4.85 5.45 4157 2.69 (0.24) 2.18 2.92 (0.60) 2.30 (0.62) 2.08 2.43 4189 2.80 (0.15) 2.93 2.6) (0.45) 2.86 (0.82) 4.22.10 \o0 4229-4235 1.54 (0.19) 2.20 1.89 (0.35) 2.24 (0.75) 1.98 1.97 4267 4.68 (0.44) 5.50 5.86 (0.60) 7.26 (1.19) 5.20 5.70 4330 5.97 (0.39) 5.29 7.36 (0.48) 6.75 5.946.26 4375 4.90 (0.36) 2.62 4.24 (0.57) 3.18 (0.02) 2.76 3.54 4441 6.57 (0.34) 6.00 6.90 (0.68) 6.73 (0.03) 5.356.31 4471 1.54 (o.27) 1.84 3.96 (0.77) 1.46 (0.90) 2.20 4513 1.46 (0.03) 1.03 2.79 1.20 (0:04) 1.62 4544 2.13 (0.04) 1.09 2.24 1.85 (0.51) 1.83 4650 80.4 (13.00) 75.2 103.0 (7.00) 87.4 98.8 89.0 4686 20.6 (3.0) 19.9 24.9 (0.38) 24.4 30.1 24.0 4786 1.85 (0.08) 2.27 2.93 (0.61) 3.172.56 4861 2.07 (0.20) 4.78 3.68 3.14 4.97.73 4923 1.45 (0.04) 1.69 1.661.60 5019 2.45 (0.40) 2.24 4.75 3.92 4.78.63 506 0 0.8.2 0.81 5132 5.95 (0.74) 2.84 4.32 3.12 5.76 4.40 5248-5257 1.44 (0.44) 0.61 0.63 0.46 1.27 0.88 5411 2.12 (0.28) 1.34 3.32 2.34 (o.24) 2.66 2.36 5470 3.02 (0.15) 2.26 2.85 2.52 (0.18) 2.86 2.70 5592 1.50 1.16 1.53 1.40 5695 32.6 (2.10) 38.5 34.1 34.8 35.2 35.0 5805 24.0 (0.3) 21.2 22.0 24.0 26.8 23.6 5871 4.80 (0.49) 4.21 5.67 7.77 8.06 6.10 *Estimated Uncertainty

ON~S - Gain= Zero Mag. oIvt r-cm l? C Gain= I Magnitude Gain 1 Magnitude CD A w CD ^ ^ S'^^/\^\A C~~~~~~~~~~~~~~a~~~~~~~~Gain 2 Magnitudes CD ID N^F Wv. ^ \ 10 foy(D~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~o n^-'^ 2 10iu (O C\I~~~~~~~~~~~~~\l Fig.5-1 pc a s o 7 l (O 1 1 19 M i0 co ~ a c CD 0( a ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~~~~A m AD re, a V A F 3 Se a c o Vl (c 1 3 9 M B0 0 Ca (0 Fig. 5-1. Spectral. scan of' 72 Veloruin (Oct. 12-15, 1960, Mt. Bingar).

— 0 -I-1.50 m ( 1/,.) |0.o~ / 7 | ~ MAGNITUDES _ 0 -T=1.32,00 I/k- I.....I — I......... —I o —.-o 10.50 1.50 2.00 2.50 3.00 Fig. 3-2. Energy distribution for 72 Velorum.

REFERENCES 1. Smith, Henry, J. Thesis, Harvard, 1955. 2. Gum, C. S. Observatory 72, 151, 1952. 3. Perrine, C. D. Ap. J. 47, 52, 1918. 4. Aller, L. H, Ap. J. 97, 1355 1943. 53

SECTION IV THE HYDROGEN/HELIUM RATIO IN HD 96446 by Anne Cowley and L. H. Aller

I

One of the brightest of the hydrogen deficient stars is HD 96446 (a = llizhl8m. =59024t; 1900), whosespectrum has been described by the Jascheks [1]. Its broad lines of hydrogen and helium suggest that the star is on or below the main sequene [2]. About 400 lines have been measured in the spectral range XA3350-4850 arising from H, He, CII, CIII, NII, NIII, OII, OIII, NeII, SiIII, SiIV, P1, SII, ClII, ArIl, KlT, Call, FeIII, FeIV, NIII, and possibly KrII. (he identification of the latter is uncertain becuase the strongest lines of this ion are not present. Spectral scans of this star were secured by D. J. Faulkner and L. H, Aller at Mt. Bingar on the nights of Nov. 9-10 and Nov. 10-11, 1960, Jan. 1415, 1961, and Feb. 24-25, 1962, by Faulkner, In Fig. 4-1 we reproduce a typical, medium-speed spectral scan. In addition to the strong lines of hydrogen Hp, Hy7, HS, etc., there also appear prominent helium lines: 4471, 4388, 4143, 4121, 4026, 4009. The Balmer jump is not'prominent. Figure 4-2 shows the reduced energy distribution; the monochromatic magnitude of the star is plotted against 1/?. In order to make a detailed study of the spectrum T, Dunham and L. Aller secured a number of spectrograms with the Coudd spectrograph on the 74-inch reflector at Mt. Stomlo. Observing conditions were extremely unfavorable at the time, so that although a number of lines could be measured on the plates it Was not possible to carry out a complete abundance analysis for this star.. We selected the five plates which were most suitable for measurements of equivalent widths and traced them with the microdensitometer at The University of Michigan Observatory. Photometric calibration had been provided by two spectra of a wedge slit illuminated by a lamp and photographed above and below the stellar spectrum simultaneously with it. Unfortunately the two wedges did not always give the same deflection-log I curve, indicating that one of the wedges had been poorly illuminated; in some instances the second wedge image was partially missing. For this reason we used only one of the wedges to determine the deflectionlog I curves. We also measured equivalent widths of the stronger lines on the photoelectric scans,. Table 4-1 gives the list of Coude' observations of eD 96446, Table 4-2 gives the equivalent widths (A) determined from Coude plates and from! photoelectric spectral scans. Column 1 gives the measured wavelength; Column 2 gives the identification; Column.3 gives the average equivalent width determined from plates Cd 19, Cd 119, Cd 124, Cd 134, and Cd 144; and Column 4 gives the equivalent widths of the stronger lines as obtained from the spectrum scans. In this preliminary reconnaissance we follow the method proposed by UnsUld [3] to get the helium/hydrogen ratio. If one assumes that a Balmer line formed in an onticallv thin layer, its equivalent width will be related 57

to the number of atoms "above the photosphere" in the second level, No 2H, by Ax = 0.886 x 1iO12 HXNO 2H This approximation fails for the earlier members of the series but it becomes more nearly valid the greater the principal quantum number, until the lines begin to overlap. Hence for each Balmer line of principal quantum number, n, one may plot AA(n) against N0,2H and extrapolate to an asymptotic value of NO 2H. Table 4-3 summarizes the relevant data. The principal quantum number, n, and the wavelength X are given in the first two columns. The third column gives the equivalent width AX in cm; the fourth column gives the i-value; and the last column gives log NO-2H. The asymptotic value is log N0 2H = 15.95. A similar procedure may be carried-out for helium, considering each of the various series, 23P-n3D, 21P-nlD, 23P-n3S, and 21D-nlS, separately (see Table 4-4). In this way we obtain the number of atoms Nr.s above each cm2 of the photosphere in the rth excitation level of the neutral (s = 0) stage of ionization. We may refer all helium atoms to the 25P level, for which we find log N(23P)H = 16.16. The number of ionized helium atoms is related to the number of helium atoms in the 25P level by log He 50u0r+ log He = [IHeX23p)] +- logT-0o48 + logu(He) log N(23p) T He~2-'P) g(23p) A similar expression holds for hydrogen. Then since at the temperature T 18,000~K and electron pressure - 103 dynes cm2 for the atomsphere of HD 96446, both hydrogen and helium are essentially all ionized, and we can write log NHe(total) - log NHe+ log -NH(total) = log NHW or NH N002 5040 4 l og N- = log - T x 0.12 - log 4 - log He58 N(2) T 9 58

since IH - X2 =.39 ev, IHe X23p = 3.51 ev, ul = 1, u(He+) - 2 g2 = 8, g(23P) = 9, Hence if we assume T 17,000~K we find NH H 0.38 log - 0.42 or 0.38 NHe He The uncertainty in the temperature introduces a much smaller uncertainty in the final result than does that in the extrapolation procedure, Although HD 96446 is a hydrogen deficient star, it is not by any means an extreme example of this class. Further studies of this interesting star will require improved observational data. 59

SECTION IV: TABLES, FIGURES, AND REFERENCES

TABLE 4-1 JOUDE OBSERVATIONS OF HD 96446 Plate Number Date Emulsion 19 Feb. 25, 1961 103a-0 119 May 26, 1961 IIa-O 124 May 28, 1961 IIa-O baked 134 Feb. 6, 1961 IIa-O baked 144 June 6, 1961 IIa-O 63

TABLE 4-2 Ak VALUES OF HD 96446..../ ~.............Average From A Element Coude Average Scans 3634.37 HeI 2.50 3705.14 HeI. 3770.65 H" 3797.92 H 2.71 3819.61 HeI 2.98 2.73 3835.39 H 4.20 3.26 3867.48 HeI.64 3871.82 HeI 1.78 1.62 3888.65 HeI 85.50 3889.05 H 3918.98 CII.09 3920.68 CII.10 3926.53 HeI 2.20 1.37 3933.66 CaII.30 3935091 HeI.24 3964.73 HeI.53 3968.47 CaII.15 3970.07 H 4.88 4.231 5995.00 NII.07 4009.27 HeI 2.60 1.96 4026.19 HeI 3.71 3.03 4069.77 OII.09 4072.16 OII o6 4075.87 oI o.08 4101.74 H 4.65 5.01 4120.84 HeI.82 1.16 4143.76 HeI 2.63 2.18 4169.23 OII.39 4267.17 CII.16 4340.47 H 4.69 4.61 4387~.93 HeI 2.53 2.35 4437.55 HeI.34 4469.92 HeI 4471.48 HeI 390 367 4481.23 MgII.16 4713014 HeI.57.62 4861.33 H 3.17 4921.93 HeI 1.62 64

TABLE 4-3 DETERMINATION OF N0 2H FOR HYDROGEN n. Ak (cm) F log N0,2H 4 4861.33 3.17 x 108 1.19 x 10-1 14.104 5 4340.47 4.65 x 10-8 4.47 x 10-2 14.795 6 4101.74 4.83 x 10-8 2.21 x 10-2 15.166 7 3970.06 4.66 x 10-8 1.27 x 10-2 15.420 8 3889.00 Combined with He 8.04 x 10-3 9 3835.38 3.73 x 10-8 5.43 x 10-3 15.722 10 5797.92 2.71 x 10-8 3.85 x 10-' 15.740 Extrapolated Value...................... 15.95 65

TABLE 4-4 DETERMINATION OF N FOR HELIUM rs n X A.(cm) + log NH n x Ax (cm) flog H 23P-n3D 21P-nlD 4 4471.48 3.67 x 10-8.1231* 14.226 4 4921.93 1.62 x 10-8.1205* 13.797 5 4026.19 3.37 x 10-8.04686* 14.700 5 4387.93 2.35 x 10-8.04293* 14.506 6 3819.61 2.86 x 10-8.0260 14.930 6 4143.76 2.40 x 10-8.0199 14.899 7 3705.14 3.34 x 10-8.0152 15.257 7 4009.27 2.28 x 10-8.0112 15.155 8 5364.37 2.50 x 10-8.00853* 15.344 8 3926.53 1.78 x 10-8.00737* 15.242 9 3871.82 1.70 x 10-8.00471 15.434 23P-n3S 21D-nlS 4 4713.14.62 x 10-8.01075* 14.467 5 4120.84.99 x 10-8.003** 15.341 5 4437.55.34 x 10-8.00320 14.785 6 3867.48.64. x 1-8.00183 15.421 6 4169.23.39 x 10-8.00156* 15.210 *Given by Trefftz, et al. [4]. **Given by Allen [5].

HD 96446 Nov 10-Il 1960 CM 6z CD cu h I X OD O\ I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~I CD o,~~~~~~~~~~~~~~~~~~~~~~~~r Yellow FilterS K 6000 5500 5000 4500 4000 3500 Fig. 4-1. Energy scan of ID 96446 (Nov. 10-11, 1960, Mt. Bingar).

-1.0Ia l I I I I I I I -I I I I I 01.0 O 0 0 0 Fig. 4-2. Energy distribution for HD 96446. Fig. 4-2. Energy distribution for ED 96446.

REFERENCES 1. Jaschek, M. and Jaschek, C. Publ. Astron. Soc. Pac. 71, 465, 1959. 2. Aller, L. H., Buscombe, W. and Dunham, T. Astron. Journ. 67, 110, 1962. 3. Unsold, A., Zeits fur Astrophysik, 21, 22, 1941. 4. Trefftz, E., Schliiter, Detmar, and Jbrgens. Zeits f. Ap. 44, 1, 1957. 5. Allen, C. W. Astrophysical Quantities, Athlone, London, 1960. 69

SECTION V THE EMISSION NEBULOSITIES IN THE LARGE MAGELLANIC CLOUD by H. R. Dickel, L. H. Aller, and D. J. Faulkner

The Magellanic Clouds are the nearest of all external galaxies. They contain many luminous stars, star clusters, emission nebulae, and large quantities of neutral hydrogen, as revealed by the 21-cm surveys. Some years ago Henize [1] published a catalogue of Ha emission stars and nebulae in the Magellanic Clouds. He employed a red-corrected 10-inch camera with a 15~ objective prism, together with red-sensitive (Kodak 103aE) photographic emulsion and a red plexiglass filter so as to observe the emission nebulosities in the light of the red hydrogen line Ha (see Fig. 5-1). This catalogue gives the right ascension and declination of each nebulous patch for the Epoch 1950, the rectangular coordinates X and Y, the diameters x and y in seconds of arc, and other data including identifications with NGC objects. Other, less extensive discussions of emission nebulosities in the Large Cloud have been given by various observers. For example, Shopley and Miss Wilson [2] discussed 22 bright NGC objects in the Large Cloud, all of which are brighter than the Orion Nebula. The most spectacular object is 30 Doradus, the Tarantula Nebula for which Johnson [3] finds a mass 5,Q00,000 that of the sun (exclusive of its stellar content!) thus making it by far the most massive gaseous nebula known in our group of galaxies. With a diameter of the order of 500 parsecs, it mostly exceeds in size any other gaseous nebula. A detailed study ofj this object is being undertaken by D. J. Faulkner. Photometric measurements of the emission nebulosities in the Large Magellanic-Cloud are not numerous' Doherty, Henize, and Aller [4] microphotometered both widened and unwidened objective prism photographs covering a narrow wavelength range near Ha to obtain cross-sectional "intensity profiles" for certain nebulosities in the Large Cloud. These intensities were converted to surface brightnesses by tracing the widened spectra of stars of known magnitude and color and using the method of Ambarzumian [5]. Peak intensities for each scan across a nebulosity are given in ergs cm-2 sec-1 steradian-1. Many of the emission nebulosities in the LMC are filamentary in structure. The classic example is 30 Doradus but other objects such as Henize 144. 159, and 160 show a pronounced fine structure. (See Figs. 5-2, 5-3, 5-4, and 5-5.) This inhomogeneity must be taken into account when we convert surface brightness into electron densities, Densities in the filaments may appreciably exceed those found from averages taken over the volume occupied by the nebula. Spectrophotometry of the emission nebulosities in the LMC was undertaken mostly at Mt. Bingar with the Michigan Photoelectric Scanner. Thanks to a generous allotment of time for this program, we were able to observe nearly all of the brighter nebulosities. Extension of the program to fainter nebulosities would have required a much larger telescope then was available to us. Throughout, we used the D'slots which sufficed to admit enough light to the photocell to resolve most of the lines of interest. Table 5-1 gives the data on the slot sizes including the dimensions in seconds of arc at the 73

focus on the slot and the area in steradians. The observational procedure was to center the nebulosity in the slot, using the periscope immediately behind the entrance aperture. Then the spectrum was scanned from about 5200A to 3400A, the direction of the grating drive reversed, and the scan repeated. By superposing direct and reverse scans, the reliability of the tracing and spectral features displayed could be assessed. In Fig. 5-6 we compare direct and reverse scans for two nebulosities of somewhat different levels of excitation. The [OIII] \4949 and X5007 lines are much stronger in NGC 2040, indicating that it has a higher level of excitation. The problem of nebulae spectrophotometry can be resolved into two aspects: (1) the surface brightness in one of the monochromatic emissions, e. g., HM; and (2) the relative intensities of the principal nebular lines. The second problem is relatively simple; one need know only the atmospheric extinction and sensitivity function of the photocell and the transmissivity of the optics. Both of these quantities can be obtained by observing proper comparison" stars, as noted previously in Sections II and III. The determination of the surface brightness is more troublesome in that one must compare the monochromatic radiation of, say, Hp with the flux from a star in a unit wavelength interval at \4861 and then determine the monochromatic magnitude of the star. Since we observed mostly stars of early spectral class whose colors differed from that of the sun, it was necessary to modify Ambarzumian's method to take this into account. If the B and V magnitudes of the comparison stars are known, we can determine their monochromatic magnitudes at \4861 since their energy distributions have been measured. Since different spectral purities are involved (the stars being always observed with the B slots, i.e., a 9A resolution) this effect must also be taken into account. The method of reduction is straightforward although it is a little involved. The principal results are summarized in Table 5-2. The first column gives the number in Henize's catalogue. The second column gives the quality of the scan of HM. For most of these nebulae, the quality was A or B, although occasionally poor deflections were obtained. Column 3 gives the slot sizes for both the entrance and exit slots. In most instances the D slots had to be used, yielding a spectral purity of 106A. In many instances the nebulosities did not fill the slot. In yet others the slot was smaller than the nebula. Column 4 gives the estimated fraction of the slot area filled by nebulosity. An entry of "1" means that the nebula filled the sloto Stencils corresponding to slot sizes as reduced to the scales of appropriate photographs of the nebulosities were used to estimate the areas actually filled by the nebulosities. Extensive notes and sketches of the nebulosities as they appeared in the slot at the time of observation also aided the making of such estimates as are entered in Column 4 Column 74

5 gives the surface brightness in Hp in units of 10-4 ergs cm 2 sec-1 ster1' (averaged over the amount of the nebula in the spot). In some instances the nebulosity consisted of a bright blob on a diffuse background. For example, in Henize 69, an intense blob filling only one-tenth of the slot gives S(HB)= 8 in our units; however, the average over the slot would be 0.8. In Column 6 we give the ratio of the sum of the intensities of the green nebular [OIII] lines to Hp. From this ratio one may estimate the "excitation class" of the nebula [6]. For Henize 77 the excitation class is estimated from -I(N+N2)/ 1(3727) to be 4.5. The larger the number in the last column the higher the level of excitation of the nebula. Table 5-3 gives the relative intensities of the lines observed in the various nebulosities on the scale Hp = 10. To each line measured on the tracing one may assign a quality raging from A (for good data) to E (for a line which is very poor). The lines listed are the green nebular lines 5007+4959 = N1+ N2 [OIII], Hp = 10, HeI x4472, Hy \4340, H \x4101, \3969 (HE+[NeIII]), k3865+ x3889 ([NeIII]+H), \3835 (rarely), and x3727 [OII]. In all of these nebulosities the intensities of the helium lines are low and the quality of the measurements is poor. If we included all the nebulosities and take a straight average we get a He/H abundance ratio of 1/7, but the uncertainty is so large that this number must be regarded as meaningless. It is best to adopt the He/H ratio from the work of Hugh Johnson [3]. We now describe briefly the utilization of these data to derive ionic abundances for hydrogen oxygen (0+ and 0++) and Ne++. The methods have been adequately described in the literature (see e.g. [6] or [7]). The density of hydrogen can be obtained from the surface brightness if one postulates a uniformly radiating sphere and chooses an electron temperature. For most estimates we chose T. = 10,000~K. For each nebulosity one estimates the radius in seconds of arc from Henize's catalogue. Assuming a distance to the LMC one can then convert this angular radius to radius in cms. For this purpose we assumed a distance of 46 kpc. Then rneb = 6.87 x 1017 n" where Gn" is the radius in seconds of arc. Then from the measured surface brightness one can calculate the electron density (assuming hydrogen is much more abundant than helium) of e.g. [8], p. 204. We use, however, the correction to the Boltzmann formula B4 (10,000~K) obtained by Burgess [9]. Similar calculations may be carried out for TE = 15,000~K. Turning to the forbidden lines of [OII] and [OIII], we may calculate 75

N(o+) _ 1(3727) Pc(NE,Tc) N(O++) I( Nl+N2) where 1(3727) is the intensity of 3727 of [OII] and I(N1+N2) is the sum of the intensities of the two green nebular lines. Here Pe(NETE) is a known function of the electron temperature, electron density, and atomic parameters involved in the ground configurations of 0+ and 0++. For details see Aller [10] and Seaton and Osterbrock [11]. Similarly [10] we can write N(O++) = B(Ne,T) I(N1+N2) I(HP) where B can be tabulated as a function of Ne and TE. Table 5-4 gives the results for an electron temperature of 10,000~K. Successive columns give the number of the nebula in Henize's catalogue, the electron density (number of electrons-per cm3), the number of 0++ ions calculated from the ration I(Nl+N2)/I(H), and the ratio N(O+)/N(O++) calculated from I(3727)/I(N1+N2). In order to see what the effects of uncertainty in the electron temperature would be, we also carried out calculations for T. = 15,0000K (see Table 5-5). The electron density is increased by about 20%. The number of doubly ionized oxygen atoms is cut in half and N(O+)/N(0++) is reduced to about three quarters of its value for 10,0000K. Similarly one may estimate the concentration of ions of Ne++ from the 3868 line. The [NeIII] lines are always relatively weak and are blended with Balmer lines. Hence only rough estimates of their intensities can be extracted from the data. To do this we plot the Balmer decrement from HM to H6 and extrapolate with the aid of the well known theory [6], [7] to get the expected intensity of \3868. The difference between the observed and calculated value is then attributed to [NeIII]. We can write [10] N(Ne++) I1(3868) N(O ++) I(Ni+N2) Numerically PNe = 4.04 for Te = 10,000 and 3.07 for TE = 15,000~K. To obtain the ratios of oxygen and neon to hydrogen we now proceed as follows. Consider first the oxygen problem. Oxygen can exist as 0, 0+, and O++; little is triply ionized. If one plots a histogram of N(O+)+N(O++)/N(H) he can get an idea of the absolute abundance of oxygen by assuming that the nebula with the largest value of this ratio would contain oxygen in only these two stages of ionization. In this way one finds the limiting value to be 76

N(O) = 2.6 x 10-4N(H) as compared with 4 x 10-4N(H) for the Orion Nebula [12]. For neon the problem is very much more difficult since we observe neon only in one stage of ionization, viz. Ne++. Hence we must estimate N(Ne+)/ N(Ne). If we choose nebulae with about the same level of excitation as the Orion Nebula and assume that the distribution of atoms among various stages of ionization is the same for Orion and the nebulosities under consideration we can estimate the abundance of neon. The scatter in the resultant neon abundance is considerable. Most of this must arise from the distribution of atoms among various ionization stages and some of it comes from the uncertainties in the intensities which are guessed in only a very rough way. We may summarize our results as follows. -Let us assume an electron temperature of 10,000~K. The electron density ranges from 3 electrons cm-3 to 50 electrons cm-3 in the objects studied, with an average of 15. This value is considerably lower than that found for the Orion Nebula or the denser regions in 30 Doradus. The upper limit must be much too low as we have made no allowance for the effects of filamentary structure. Most nebulae fall in excitation classes 3 or 4. That is, they are comparable with planetary nebulae such as IC 418 or IC 2149 in our own galaxy [6]. For every 10,000 hydrogen atoms we would find the following figures for atoms of other types: Helium.-No reliable estimate can be obtained from the data herein presented since the helium line intensities are in all instances too uncertain. It is best to adopt Johnson's value for this ratio or that obtained by D. J. Faulkner from a photoelectric spectrophotometric study of the 30 Doradus nebula. Oxygen.-The range for N(O +O++) is from 0.9 to 2.6. We have adopted the upper limit on the assumption that all the oxygen would then be concentrated in these two ionization stages. If the electron temperature is lower than 10,000K, the ratio may be higher. Neon.-The range is from 0.10 to 1.7. For 17 of the 55 nebulosities studied, no measureable line was observed. The range and average values are adopted from the data for the 38 nebulae which were believed to show [NeIII] lines, albeit weak. Theaverage is N(Neon) = 0.5. Probably this is much too low as we can imagine that the allowance for the atoms in the lower ionization stages is inadequate. Literally interpreted, the results would suggest that neon is less abundant in the Large Magellanic Cloud than in our galaxy, but we do not believe that this is correct, 77

The conclusion drawn from this investigation is that the abundance ratio of oxygen (and probably neon as well) is about-the same in the IMC as in our galaxy. We must stress the uncertainty in this result, however, and urge that more detailed attention be paid to the brighter nebulosities, particularly 30 Doradus. Scanner observations should be supplemented by photographic spectroscopy of the fainter lines. In any event the biggest stumbling block is going to be the same as that found in the study of other gaseous nebulae, namely (1) allowing for the distribution of atoms among various stages of ionization and (2) allowing for the effects of filamentary structure. Although the first difficulty will require both difficult observations and erudite theoretical studies, we can make some progress on the second. A study of the filamentary structure of 50 Doradus has been undertaken by D. J. Faulkner, and some of the brighter nebulosities are being studied by H. R. Dickel with the aid of the isophotometer, The objective is to find to what extent our average electron densities have to be increased in the filaments and lowered throughout the bulk of the volume occupied. The data obtained in this program should be useful for another problem, namely the total amount of ionized hydrogen in the LMC. The amount of neutral hydrogen and its spatial distribution is being studied by the CSIRO radiophysics group with the large dish at Parkes. Regions of high gas density can readily be found and it will be interesting to see if most of the ionized hydrogen regions are simply Strdmgren spheres. Results obtained with more modest equipment showed that the 50 Doradus region was the focal point of a concentration of gas unequalled by any other known nebula in the local system of galaxies. 78

SECTION V: TABLES, FIGURES, AND REFERENCES

I

TABLE 5-1 SCANNER SLOTS AND APERTURE USED ON EMISSION NEBULOSITIES IN THE LARGE GEMAGELLANIC CLOUD Slot Area Bingar E-W Dimension N-S Dimension in: Slots Sec. of Arc Sec. of Arc Steradians D5 200 88 4.14 x 10-7 D4 157 88 3.26 x 10o7 D3 circular r = 57 1l83 x 10-7 DO 214 88 4.44 x 107l C5 228 38 2o02 x 10-7 C4 168 38 1l4 x 10-7 81

TABLE 5-2 DATA FOR THE EMISSION NEBULOSITIES IN THE LARGE MAGELLANIC CLOUD Henize Quality Slots Estimated Fraction S I(N+N) Excitation Catalogue of (Entrance, of the Slot Area Surface I ) I(HP) Class Number HF Exit) Filled by Nebulosity Brightness 5 B D5-D4 1.6 1.5 3 8 B D5-D4 3/8 1.7 4.1 4 9 C D5-D4 1.30 2.1 3 11AB A D5-D4 1 3.9 4.0 4 11B 4.4 4 11C 4.1 4 11CD A D5-E2 1 3.8 4.5 4 llE D5-D4 2/3.9 4.0 4 11F A D5-D4 1 1.6 2.1 3 11I C D5-D4 1/2.7 2.6 3 17 B D5-D4 1/2.6 2.0 3 23A A D5-D4 -2/3 1.1 2.7 3 30B C D5-D4 1/5 3.2.8 5 44B B D5-D4 1 1.0 1.6 3 44C B D5-D4 5/6 2.1 6.3 4 44D A D4-D4 2/3 1.6 1.3 3 44F B D4-D4 1/3 2.1 1.9 3 51A B DO-D4 2/3.8 1.2 3 51C A D5-D4 5/8 1.6 1.0 3 51D B D5-D4 2/3 1.1 1.8 3 55 A D5-D4 1 2.7 2.0 3 57A B D5-D4 3/4.8 1.6 3 59A B D5-D4 1 3.3 6.4 4 59B A DO-D4 2/3 1.3 3.2 3 4 1 (filled?).5 63 B D5-D4 1/6 (intense -) 2.9 2.0 3 68B D5-D4 (if 1, then — ).8 (if 1/10, then- ) 8.4 1.6 3 69 A- D5-D4 (if 1, then-.) 1.0 (if 1/10, then) 9.8 1.4 3 70 C D5-D4 ~1.25 1.8 3 77 1 1.5 1(3727) = 3.7 79 A D5-D4 (if 1, then.) 3.4* (if 1/4, then -) 5.9 3.9 4 79 (north) A DO-D4 1 1.9 1.5 3 83 A D5-D4 5/8 3.4 4.1 4 91 A D5-D4 2/3 2.4 2.2 3 92 C D5-D4 2/3.65 2.2 3 103 E D5-D4 1.65 105A A D5-D4 1 1.5 2.9 3 113i ABED A D4-D4 2/3 1.7 1.9 3 1l3ii E D4-D4 1/2?.4 3.4 4 119 A D5-D4 fi 1.6 1.9 3 120 A D5-D4 ~1 2.4 1.2 3 144 A D5-D4~ 1? 1.7 4.9 4 D4-D4_J 148C C D5-D4 3/4.8 2.1 3 154A A D5-D4 3/4 2.0 1.9 3 158 B D5-D4 3/4 1.5 2.8 3 159 D4-D4 5/6 1.7 2.5 3 159A B D4D41 14 1/4.4 5.8 4 D3-D4J 159C A D5-D4 1 1.7 3.2 3- 4 159D+F A D5-D4 5/6 1.7 1.9 3 160 D4-D4~ i1 3.2 4.4 4 160A C5-C4 2/3 9.0 4.1 4 160A+D D5-D4 5/6 4.5 4.4 4 160B+C A D5-D4 2/3 1.6 3.8 4 163 B D5-D4 1.5 2.0 3 164 B D5-D4 1.7 1.8 3 180 A D5-D4 5/6 1.1 2.4 3 *(1/4 for the most intense blob) 82

TABLE 5-3 RELATIVE INTENSITIES (HO - 10) Nebula NO.b [OIII] Nl+N2 H Helium 4472 Hy H8 3969 3865 3835 [OII] 3727 NO.... 5 A 15 B 10 E 1.1 E 2.1 E.9 E 1 B 15 8 A 41 B 10 C 5 D 3.4 E 1.4 E 2.5 B 20 9 C 21 C 10D 26 11AB A 40 A 10 C.7 A 4.1 B 1.5 B 1.2 B 2.8 A 16.5 N1 N2 11B 32 11.5 10 11C 32 8.6 10 11CD A 45 A 10 E.6 B 4 B- 2 C 1.4 B- 3.8 B+ 22 11E 40 10 1.7 5.3 4.9 1.1 5.5 2.3 25 11F A 21 A 10.2 C 3.8 D 2.8 E.6 E 1.1 A 30 111 C 26.5 C 10 C 4.4 C 36 17 B 19.5 B 10 D 3.8 E 1.4 B 35 23A A 27 A 10 D 3.8 D 3.5 E 1.5 E 4.7 (A) 22 30B C 8 C 10 D 2.5 D 4.4 D 4.9 E.6 E 1.6 D 10 44B B- 16 B 10 D+ 2.5 D- 2.4 E.8 E 1.6 A 19 44C A 63 B 10 E.5 B 5 D 2.1 E 1.3 C- 4.1 1.1 A 21.3 44D A 13 A 10?E.8 C 5 D 2.8 - E 1.8 E 1.2 E.6 A 37 44F A 19 B 10 D D 1.4 E 1.5 E 1.4 B 34 51A B 12 B 10 E 3.8 E 1.7 B 35 51C B 9.7 A 10 E.5 B 3.8 D- 3.1 E 1.5 E 2.5 A 37 51D B 18 B 10 B 21 55 A 20 A 10 E.2 B 3.9 C 2.25 D 1.5 D 1.6 A 32 57A B 16 B 10 E 1.1 D 4.4 E- 4.5 E- 4.1 A 51 59A A 64 B 10 C 7 B 3.8 B 2 C 15 B 2.8 A 13 59B A 32 A 10 E 1.2 B 5 C 2.8 E.9 D 2.7 A 32 63 B 20 B 10 E 1 C 3.8 D 5.6 E 3 E 2.8 B 31 66A C 14.5 68B 16 10 C 6.9 D 2.8 E 1.5 E 1.1 A 41 69 B+ 14 A- 10 B 4.3 C 4.2 D 2.1 D 2.1 A 43 70 C 18 C 10 D 4.3 E 3.8 B 40.5 77 10 3.7 79 A- 39 A 10 C 3.8 D 2.5 E.6 E 1.2 A 23 79N A 15 A 10 E.5 C 1.9 E 1.4 E 1.5 E.8 A 35 83 41 A 10.7 B 4 C 2.1 D- 1.1 D- 1.9 E.6 A 22 91 A 22 A 10 E.7 D 4 D 2.8 E 1.4 E 3.2 E 1.9 A 31 92 B 22 C 10 D 4.5 E 4.2 E 2 E 4 C 19 103 E 10 E 4.6 105A A 29 A 10 E.35 C 3.5 D 2.1 E 1.5 E 1.6 A 28 113i ABED A 19 A 10 C 3 D 1.4 E.9 E 2.1 E I A 29 11311 E 34 E 10 E? 10 D 44 119 A- 19 A 10 E.2 B 3.8 D 2.1 D 1.4 D 3.2 A 27 120 A 12 A 10.35 B 4.3 C 2.4 C 2.7 A 33 144 A 49 A 10 E.35 C 3 6 D 2.95 D 2 D 5.4 A 23 148C B 21 C 10 E 4.5 D 5.3 C 28 154A A 19 A 10 D-.95 C 3.4 C 1.7 C 1.4 D 1.3 A 21 158C A 28 B 10 E.8 C 4.8 C 3.4 D 1.4 C 2.7 D 1.4 A 25.5 159 25 10.6 3.1 3.1 1.2 3.5 31.5 159A A 58 B 10 D.8 C 4.3 C- 3.9 D 1.6 C- 3.2 E.8 A 30 159C A 32 A 10 D-.5 B 4.1 C 1.7 D.9 D+ 1.4 A 16 159D+F A 19 A 10 E.2 B 4,3 B 2.5 D 1.1 C 1.6 A 30 160 43.5 10.8 3.8 -1.8 1.5 2.8.8 14 N1 N2 160A 30 10.6 10 160A+D 43.5 10 3.6. 2.4 2 2.4 16 160B+C A 38 A 10 C 1.9 B 4.6 D+ 3.5 E 1.1 D 1.6 A 27 163 B 20 B 10. C- 4.1 E 2 E.8 B 2.2 B 25 164 A 18 B 10 E.35 C 4.4 D 6.2 1.2 B 28 180 A 24 A 10 E.95 B 6.3 B 6 C 1.8 C 3.2 A 31 Ft3z

TABLE 5-4 ELECTRON DENSITIES AND IONIC CONCENTRATIONS FOR T: = 10,000~K Nebula N(OII) NebulaI) Ne N(OIII) N(OII) No. Ne N(OIII) N(OII) No. N(III No.bua -3 40.,00 x lol 5 8.39 x 103 1.13 79A 40 5.00 x 10.68 8 19 2:55.56 79E 15.5.752.65 9 4.27 1.35 83 12 1.51.61 11AB 17 2.12.47 83A 36 4.68.61 11CD 22 3.30.54 91A+B 25 1.79 1.57 11E 12.5 1.63.70 92 9.65 1.01 11F 15.5 1.10 1.57 103 6 - ill 14.5 1.20 1.51 105A 10.6 1.10 1.06 17 12.5.78 2.00 113ABED 19 1.19 171 23A 15.5 1.37,90 113ii 4.42 1.44 c 30B 57 14.4 14 119 7.391.62 ~t UB- 18.97 1,31 120 20.80 3.02 44C 28 5.77.38 144 9 2.43.52 44D 20.87 3.10 148C 12.5.85 1.49 44F 27 1.65 2,02 154A 25 1.57 1.24 51A 29 1.12 3+33 158 36 3.32 1.05 51C 20.60 4,32 159 11.6.91 1.44 51D 7.37 1.31 159A 40 7.80.59 55 12.5.81 1,82 159C 17 1.73.54 57A 21 1.05 3.55 159D+F 17 1.66 1.71 59A 21 4.21.22 160 10 1.46.34 59B 15.5 1.62 1.13 160A 68 8.84 63 5.6.33 1.73 160A+D 34 4.98.40 63A 40 2.59 1.75 160B+C 22 2.66.79 68B 11.6, 98 60, 5.1 2,84 163 8.50 1.39 69 12.5, 104:;^, 4.72 3.37 164 8.44 1.75 70 3.7.23 2.54 180B 9.69.46

TABLE 5-5 ELECTRON DENSITIES AND IONIC CONCENTRATIONS FOR TE = 15,000~K Nebu N( ) ebula N) NNbua eN(OIII) N(OII) Ne N(OIII) Ne N(OIII) No. N(OIII) No..N(OIII) 5 10.18 x 10-3.79 79A 5 2.34 x 10-3.49 8 253 115.40 79E 18.55 1.85 9 4.5.12.97 85 15.5.72 45 11AB 20.97,32 85A 44 224.45 11CD 26.5 1.46.38 91A+B 31.81 1.12 11E 15.5 72.50 92 10.27,72 1F 1~8 50 1.12 11063 81ll 17.54, 06 105A 13.5.49.76 17 15.5 35 1. 40 l113ABED 23.53 1,21 23A 20.65.65 113ii 51.21 1.05 co 30B3 70:6.88.097 119 8.18 1.13 44B 23.44.92 120 24.37 2.14 44C 34 2.67.27 144 10 61.38 44D 25.40 2.20 148C 15.5 c38 1.04 44F 33 o77 1.42 154A 31.69.88 51A 36.53 2534 158C 43 1.50.72 51C 24.28 3004 159 14.5.40 01 51D 8.14.92 159A 49 3.47.41 55 15.5.55 1.28 159C 20.78.38 57A 25.50 2.50 159D+F 22.53 121 59A 25.69.16 16o0 11.6.62 2 59B 18.74 o79 160A 82 4o.32 63 6.5.15 1.22 160A+D 40 2.57 29 63A 51 122 122 160B+C 26.5.21:.56 68B 14,.5 120.27, 2.34 2.00 163 0 9'.22 99 69 15.5, 127.25, 2,04 2..58.164 9 18 1,24 70 4.5.10 1. 80' 18oB 10 o.32 1.03 ~~~~~~~~~~~~~~~~~~~~~;:..'... _- _

Fig. 5-1. The Large Magellanic Cloud photographed by Karl G. Henize at the Lamont-Hussey Observatory with a 10-inch telescope loaned by Mt. Wilson Observatory. A red plexiglass filter and red-sensitive emulsion were used to record the nebulosities in the red radiation of hydrogen. 86

Fig. 5-2. Emission nebulosities in the Large -Magellanic Cloud near 30 Doradus. This photograph was obtained with the 74-inch. reflector at-Mt. Stromlo on Dec. 21, 1960. The photograph shows Henize nebulosities 159 and 160, which include NGC 2077, 2078, 2079, 2080, 2083, 2084, 2085, and 2086. (ORI (red) filter, 103aE emulsion, 60 in exposure.)

Fig. 5-3. The emission nebulosity Henize 144 in the Large Magellanic Cloud photographed with the 74-inch reflector at Mt. Stromlo. This object includes NGC 1962, NGC 1965, NGC 1966, and NGC 1970. (ORI filter, 103aE emulsion, 60 min exposure.) 88

Fig. 5-4. The emission nebula near S Doradus photographed with the 74-inch reflector at Mt. Stromlo Observatory December, 1960.. (ORI filter, 103aE emulsion, 60 min exposure.) 89

"lo 0 Fig. 5-5. Emission nebulosities, Henize 11, in the Large Magellanic Cloud photographed with 74-inch reflector at Mt. Stromlo. This nebula includes IC 2116, NGC 1769, NGC 1763, IC 2115, NGC 1775, and NGC 1760. (ORI filter, 103aE emulsion, 60 min exposure.)

59B=NGC2040 Coco 0 I, - J o' 1W) 01 51C Fig. 5-6. Scans of typical emission nebulosities in the Large Magellanic Cloud. 91

I

REFERE CES 1. Henize, K. G. Ap. J. Suppl. 2, 315, 1956. 2. Shapley, H. and Wilson, H.H. Harvard Circular 271, 1924, 3. Johnson, H, Publ. Astron. Soc. Pac. 71, 425, 1959. 4. Doherty, L., Henize, K. G., and L. H. Aller. Ap. J. Suppl. 2, 345, 1956. 5. Ambarzumian, V. A. Zeits f. Astrophysik 6, 107, 1933. 6. Aller, L. H. Gaseous Nebulae, Wiley, New York, 1956. 7. Seaton, M. J. Progress in Physics. 23, 313, 1960. 8. Aller, L. H. Nuclear Transformations, Stellar Interiors and Nebulae, 1954, Ronald Press-Co. 9. Burgess. Monthly Notices Roy. Astron, Soc. 118, 488, 1958. 10. Aller, L. H. Ap. J. 120, 401, 1954. 11. Seaton, M. J. and Osterbroch, D. E. Ap. J. 126, 66, 1957. 12. Liller, W., and Aller, L. H. Ap. J. 130, 45, 1959. 93

SECTION VI EMISSION NEBULOSITIES IN THE SMALL MAGELLANIC CLOUD by H. R. Dickel, L. H. Aller, and D. J. Faulkner

The Small Magellanic Cloud (see Fig. 6-1) is considerably farther than its companion although it contains a number of bright stars and clusters. Two of the clusters, NGC 330 and NGC 419, have been described previously. The emission nebulosities are fainter and less conspicuous than those in the LMC. Shapley and Miss Wilson [1] discussed 106 nebulosities which have an average diameter of 8 or 10 Parsecs. Whitney, et al. [2] discussed 152 nebulosities in the SMC, and Lindsay found six small emission nebulosities and possibly a dozen planetaries identified on Schmidt plates secured at the Boyden Station. Henize (Ref. [1] of Section V) listed 90 sets of emission nebulosities, some of which were described as separate units, e.g., 84A, 84B, 84C, and 84D. The brightest and largest nebula is NGC 346 = Henize 66. It is comparable in size with Gum's nebula-the largest known HII region in our galaxy [3]. The level of excitation of the nebula is slightly higher than that of Orion. The hydrogen/helium ratio in this nebula [4] is found to be exactly the same as in the LMC (Ref. [3] of Section V), our galaxy [5] and M33 [6]. Scans of 12 emission nebulosities in the SMC were secured with the 50inch and 74-inch reflectors. Table 6-l.gives the relative intensities of the principal lines observed.. Column 1 lists the nebula according to the number in Henize's catalogue; Column 2 gives the surface brightness of HI in ergs cm-2 sec-1 ster-1; Column 3 gives the number of scans; and the remaining columns give the intensities for various nebular lines of the scale I(HP)=10. The quality of each line is also indicated. The most detailed study was carried out for NGC 346, for which the additional intensities were measured (see also Ref. [4]): \5007 4959 4363 3889 43.1 16',1 1.0 2 For HeI \4471, see Ref. [4]o Table 6-2 gives the electron densities for the various emission nebulosities observed in the SMC. Successive columns give the number of the nebulosity, its radius in seconds of ar, the electron density calculated both for temperature 10,000~K and 15,000~K. (The electron temperature of NGC 346 seems to be about 14,000~K.)The problem of estimating elemental abundances for the ions involves the same difficulties as those previously noted for the LMC. Fewer objects are involved, however, and [NeIII] is observedin only one nebulosity, NGC 346. The general level of the nebular line intensities and excitation indicates that the abundance ratios of 0 to H and of Ne to H are probably also about the same as for the LMC and our own' galaxy. An intensive spectroscopic study of NGC 346 ought to be carried out to observe weaker lines that may serve to fix the general level of ionization and to supply abundance data. It cannot be 97

too often emphasized that small abundance differences that may exert a marked influence on stellar structure and therefore on color-magnitude arrays, etc., are very difficult to establish from studies of gaseous nebulae. 98

SECTION VI: TABLES, FIGURES, AND REFERENCES

TABLE 6-1 RELATIVE INTENSITIES OF THE NEBULAR LINES OBSERVED IN THE SMALL MAGELLANIC CLOUD Nebula SH? k+ N1+Na2.H H Hy 3969 3865 3835 3727 12A 1.2 x 10-4 2 A A B+ C D C B+ 48 10 4.6 2.5 1.2 3.2 12.5 12B.32 1 A C C E D B 67 10 7.9 5.2 5.0 15.2 19.27 1 B+C C D E B+C 22 10 3.5 2.6 28.5 36.58 1 A B-B- D- D E? B+C 45 10 7.1 2.8 2.1 15.4 37.30 1 A B C D B 33 10 7 3.8 16 7 66 2.2 7 A B B C C B B 62.3 10 4.7 2.5 2.4 3.8 8.4 76.47 2 A C C-D C-D D C C 80.7 10 5.0 4.8 2.9 8 10.9 78.56 1 17 10 80.22 1 28.3 10 83 1.36 4 A B B C D C-D A-B 35 10 5 2.7 1.2 2.2 14.9 84 1.0 3 A A B C-D D D A 52 10 5.8 3.1 1.6 5.5? 31.9 90 1.4 1 -A C D D E D C 44 10 2.5 3.1.7 2.0 12.1 Notes: Neb 12B: 3 additional scans of N1+N2; Neb 19: 2 additional scans of N1+N2, A3727; Neb 36: 2 additional scans of 3727, 3 of N1+N2; Neb 66: 9 additional scans of N1+N2. 101

TABLE 6-2 ELECTRON DENSITIES IN THE SMALL MAGELLANIC CLOUD SMC on Ne(Te= 15 x 104 ~K) N(TE = 104 K) 12A 56 19 15 12B 56 10 8 19 75 8 6 36 75 11 9 37 150 6 5 66 100 19 16 76 100 9 7 78 156 6 5 80 88 6 5 83 90 16 13 84 80 14 12 90 87 16 13 102

Fig. 6-1. The Small Magellanic Cloud photographed by K. G. Henize with the Mt. Wilson 10-inch telescope at the Lamont-Hussey Observatory of The University of Michigan. A red plexiglas filter and red sensitive emulsion is used. 103

I

REFERENCES 1. Shapley, H. and Wilson, H. H. Harvard Observatory Circular 275, 276, 1925. 2. Whitney, C. A., Wade, C. M. and Nail, V. Proc. Nat'l. Acad. Sci. 39, 1168, 1953. 3. Johnson, H. Publ. Astron. Soc. Pac. 73, 20, 1961. 4. Aller, L. H. and Faulkner, D. J. Publ. Astron. Soc. Pac. 74, 219, 1962. 5. Mathis, J. S. Ap. J. 125, 318, 328, 1957. 6. Mathis, J. S. Ap. J. (in press). 105

y

SECTION VII PHOTOGRAPHIC REGION OF THE SPECTRUM OF NGC 7009 by L. H. Aller

From the standpoint of theories of physical processes in gaseous nebulae and nebular compositions, one of the most interesting planetary nebulae is NGC 7009, a = 21ho2m6s 5 = -11~32' (1960), sometimes called the "Saturn" nebulae because of its well-developed ansae. Its southern declination makes it a not to favorable object for study. Although NGC 7027 [1] shows a much richer spectrum than does NGC 7009 [2,3], it is a less favorable object for theoretical studies and abundance investigations. The reason appears to be that NGC 7027 is characterized by a filamentary structure which may show a great range in excitation conditions. On the other hand, the structure of NGC 7009 is relatively smooth for a planetary nebula [4] and it seems possible to allow for effects of stratification within the nebula. Whereas the central star of NGC 7027 has never been observed, that of NGC 7009 is well-defined. It seems to show a continuous spectrum with neither emissions nor absorption lines. A characteristic feature of the spectrum of NGC 7009, noted in the clas-: sical investigations of Bowen and Wyse [2] and of Wyse [3], is the great number of permitted lines of OII, CII, NII, and other ions. These arise from recombination and permit one to deduce abundances of a number of ions, e.g., i++, which cannot be studied in other ways. With the development of reliable theories of recombination and subsequent cascade by Seaton and his associate, it will be possible to use these lines to derive ionic abundances and the distribution of atoms among various ionization stages. Earlier attempts to deduce the chemical composition [5] of NGC 7009 were based on rather limited material. It is hoped that when the present investigation is extended to other spectral regions and improved intensity calibrations are obtained, the problem can be handled in a much more effective manner. In August, 1961, I was able to secure a graded sequence of spectrograms of NGC 7009 with the Coude' spectrograph of the 100-inch telescope covering the region from \3100 to k4900. The plates were calibrated photometrically with a calibrating spectrograph and were traced with the microphotometer at The University of Michigan. Table 7-1 lists the relevant data. The image rotator was used for all observations so that the nebula was held in a fixed position on the slit. In the two-night exposure, care was taken to avoid getting the light of the central star on the slit. The four plates in Fig. 7-1 show the appearance of the spectrum as photographed on plate 14767. The faint background continuum arises from the nebula. At wavelengths longer than k3665 it arises from recombinations on the third and higher levels in H; shortward of X3665 it comes mostly from recombinations on the second level of hydrogen. Most lines are intensified in the region of the bright ring but spurious sky lines denoted as L.A. —due to the light of Los Angeles-extend across the entire field. Some lines, e.g., 4227.49 [FeV] are intensified in spots in the ring; others have a more uniform appearance. 109

A microphotometric tracing of the spectrum is shown in the twelve strips of Fig. 7-2. Deflection is not strictly proportional to intensity since the characteristic curve of the plate is involved. The tracing shows, however, many of the features that are not too easily seen in the reproduction. The marked rise in intensity at the Balmer limit is clearly shown. The dotted line indicates the estimated position of the background continuum, -First we used the photographic calibration exposures secured in the calibrating spectograph to get the deflection - log I relation. This relation changed slowly with wavelength, so it was necessary to use different curves for different wavelength regions. With the aid of these curves we determined the intensities of the lines on a relative scale. It is not possible to compare lines in different regions of the spectrum because the sensitivity of the plate changes with wavelength, atmosphere extinction varies with wavelength, and the transmission of the optics changes with wavelength. The intensity of the spectrum of the background star changes with wavelength but since its temperature is known approximately, we might expect that it could be used to calibrate the wavelength-dependent effects. Unfortunately this hope is not justified since atmospheric dispersion seriously affects the energy distribution in the stellar image but does not affect the nebula at all. Photographic calibrations may be carried out by comparing the nebula with a standard star. This procedure has been used for a number of nebula by various observers, but suffers from several difficulties inherent in using photographic photometry in any fundamental way. Another alternative, which has actually been employed, is to use photoelectric photometry. Since a wide slit must be used, care must be exercised in regions of density packed lines as, for example, near the Balmer limit. In undertaking the calibrations we proceeded as follows. First, the intensities obtained from the three Coude plates were combined in one master list. The weaker lines appeared only on the two night exposure on which the strong lines were "burned out." Hence, we used the short exposure to get the intensities of the stronger lines and the intermediate exposure to tie together the strong and weak lines with the aid of lines of medium intensity. To secure definitive line intensities in this way, it will be necessary to obtain additional plates, covering the ranges of intermediate and high intensities. The photoelectric scanning techniques give only the intensities of strong lines and these must be compared with. the stronger lines observed photographically to ascertain how the effects of plate sensitivity, atmosphereic transmission, etc., are to be taken into account, Two series of photoelectric measurements were used for this investigation. One series of observations was secured by Liller and the writer at the Mt. Wilson Observatory in 1956 (see Fig. 7-3). These observations covered the lines from the Balmer limit to longer wavelengths. In order to secure a 110

calibration for the ultraviolet, Faulkner and the writer observed NGC 7009 with the Michigan scanner on the 74-inch telescope, covering the spectral range \3100-X4100. Thus it was possible to obtain a set of photoelectric intensities for all the stronger lines; and by comparing the photographic intensities with them, the effects of atmospheric transmission) plate sensitivity, etc., could be assessed. Nevertheless, the lines intensities in the ultraviolet are still subject to considerable uncertainties. We have no assurance that the atmospheric transmission remains the same from one night to another. Hence the entire calibration of intensities is to be regarded as tentative. The results are given in Table 7-2. The first column gives the wavelength measured on the Coude plate 14767, the second column is the identification taken from Miss Moore's Multiplet Table of Astrophysical Interest, and from more recent work by Bowen. The third column gives the multiplet number from Miss Moore's table and the last column gives the tentative intensity. 111

SECTION VII: TABLES, FIGURES, AND REFERENCES

TABLE 7-1 DETAILS CONCERNING SPECTROGRAMS SECURED WITH THE COUDE SPECTROGRAPH ON THE 100-INCH TELESCOPE Plate No. Date Emulsion Exposure Remarks Kodak 14763 Aug. 29, 1961 IIaO 300 min (1) 14765 Aug. 50, 1961 IIaO baked 459 min (1) 14767 Aug. 31-Sept. 1, 1961 IIaO baked 906 min 2 nights (2) (1) Centered on star. (2) Centered on bright ring, avoiding star. 115

TABLE 7-2 PHOTOGRAPHIC REGION OF THE SPECTRUM OF NGC 7009 A Identification Multiplet Tentative I 3121.71 OIII 73.8 3132.87 OIII 678.0 3187.74 HeI 78.7 3203.10 HeII 158.6 3260.98 OIII 8 7.9 3263.43 NeII 15 7.3 3265.46 0III 8 5.4 3267.3 2.7 3299.36 OIII 3 48.9 3306.60 OII 23 46.4 3305.15 OII 23 2.6 3312.30 OIII 3 108 3354.87 NeII 2 13.7 3340.74 OIII 3 134.7 3342.5 [NeII ] 8.3 H~ei L 8 3354.92 8.1 NeIl 2 3367.00 OII 52 4.4 3375.74 OII 52 3392,78 NeII 7 3397.90 NeII 36 1.5 3399.80 2.0 3405.74 OIII 15 6.0 OI11 15 3408.13 NeII 45 3.2 NeIl 45 3411.84 OIV 2.2 3414.8 OIII 15 7.1 3417.00 [NeV] 2.6 3425.57 OIV 3 3428.70 OIII 15 47.7 3430.55 OIII 15 8.3 3438 35 2.7 3490g32 OIII 13 3.3 3442.00 2.4 3444.00 0III 15 208.3 3447,59 HeI 7 5.0 3450.78 OIII 25 3464.21 1.8 3466.04 OIII 25 0.8 3468.44 1.4 3472.00 HeI 44 1.0 3478.95 HeI 43 2,2 116

TABLE 7-2 (Continued) \h ~ Identification Multiplet Tentative I 3480.80 NeII 49 3487.77 HeI 42 0.6 3494.55 OII 70 1.5 3498.70 HeI 40 2.6 3504.23 1.8 3512.51 HeI 38 2.3 3520.40 3525.38 3530.58 HeI 5.2 3554,55 HeI 5.0 3565.11 3568.62 NeII 9 4.4 3570*27 3574.29 NeII 9 0,9 3578.08 0.6 3580.05 3583003 1 2 5587.36 HeI 8.3 3597.91 5613.70 HeI 6.1 3634.35 HeI 14,1 3657.59 H 0.3 3658.81 H 0,7 3659. 28 H 1.1 3660.22 H 1.4 3661.25 H 1.9 3662.28 H 3.4 3664.48 H 6.5 3665,93 H 8.2 3667 52 H 10.5 3669.27 H 10. 9 3671.32 H 13.3 3673.61 H 14.7 3676.25 H 16 3679.25 H 17.1 3682.74 H 213. 3686.86 H 23.8 3691.49 H 28.9 3694.17 NeII 7.7 3697.11 H 35.1 3702.82 OIII 11.7 3703.82 H 41.3 3705.02 He 24.5 117

TABLE 7-2 (Continued) \ ~ Identification Multiplet Tentative I 3707.23 0111 11.4 0III 3709.67 NeII AII 2.6 3711.92 H 44.8 37135.04 NeII 8.4 3714.04 OIII 7.7 3715.15 HeII 10. 4 3721.90 H 82-2 3726.01 [OII] 258.7 3728.73 [OII] 112.9 3730.85.8 3732.60 HeI 3.1 3734.39 H 68.5 3736.62 5740.24 OII 2.2 3746.37 AII 1530 1.0 5750.14 H 85.9 5754,50 0II2 18.3 5757.33 01II 2 11.8 3759.93 OIII 2 54. 5 3762.40 SiIl 3 3.4 3766.48 NeII 1 2.2 3768.93 HeII 2.9 3770,63 H 120.0 3773.08 SiIV 5 2.4 3774.13 OIII 7.2 5777.16 Nell 1 2.6 3778.65 SIII 2.5 5784,91 Hei 1.5 5791.42 0III2 11.8 5796.36 HeIl 5 5.6 3797.93 H 151.7 3805.96 HeI 1.8 3811.08 01II 2 1.2 3813 59 HeII 4 *9 3819.66 HeI 40.0 3826.93 AII 54.8 3829.538 MgI 3 1.1 3831,74'1.4 383355.77 HeII- 6.8 3835.43 H 237'3 3838.23 HeI. 35.53 3839.83 [FeV] 3851.10 OIl 1.9 118

TABLE 7-2 (Continued) k Identification Multiplet Tentative I 3852.90 1.2 3855.90 SiII 1 3.4 3858.04 HII 4 4.9 3861.52 SiII 4.2 3866.61 1.4 3868.63 [NeII] Too Bright 3870.58 5.6 3871.73 CII 18 3.0 3882.38 OIII 12 4,0 3888.79 H He Too Bright 3891.44 [FeV] 3.8 3895.66 [FeV] 3.2 3905.83 1.5 3907.38 OII 11 1,2 5914.90 1.1 CII 4 5918.9117 1.3 17 5920o52 CII 4 1.8 3923.51 HeII 4 6.3 3926.56 HeI 58 5,8 5928.87 0.8 3933.19 [FeII] 8 0.9 3935 02 0.9 3944.87 OII 6 1.1 3947.99 CII 32 1.7 3959.94 OII 6 1.3 3960.72 0.9 3964.79 HeI 5 23.6 3967.40 [NeIII] Too Bright 5970.00 H Too Bright 3973.16 -OII 6 2.0 3976.77 0.5 3979-71 0.7 5982.96 OII 6 0.7 3986.19 o.9 3993.06 1.0 3995.04 0. 8 5997.15 [FeIII] o.6 3998.9i 1.3 4001.03 0 5 4003.62 NIII 16 2.3 4009.31 HeI 8.0 4026.15 HeI 74.6 119

TABLE 7-2 (Continued x Identification Multiplet Tentative I 4035.08 OII 51 2.2 441.41 OII 50 4041.41 9 4.0 NII 39 4043.57 1.5 4056.03 CIII 24 0.7 4060.22 OII 97 1.4 NeIl 53 4062.91 II 601.9 4068.37 [SII] 4069.96 OII 10 18.8 4072.16 OII 10 15.8 4074.05 OIII 23 2.2 4075.94 [SII] OII 20.8 4080.90 OIII 23 2.6 4081.90 1.0 4083.86 OiI 49 4.0 4085.24 0II 10 5.0 4087.19 0II 48 4.9 SiIV 1 4089.18 OII 48 10.9 4092.98 OII 10 3.1 4097.25 NIII 1 69.1 4101.67 H8 Too Bright 4103.41 OII 20 34.6 4104.90 OII 20 3.7 4107.01 OII 47 2.3 4109.82 1.2 4110.63 OII 20 1.9 4112.55 OII 21 1.3 4115.88 SiIV 1 1.0 4119.52 OII 20 7.7 4120.85 HeI 8.4 4123.55 [FeV] 158 0. 9 4125.30 0.9 4128.70 1.9 4132.66 OII 19 3.6 41353.1 1.1 4136.76 0.6 4139.94 1.5 4143.81 HeI 12.8 4146.00 1.4 4150.56 NeII 53 0 7 41535.56 OII 19 5.5 120

TABLE 7-2 (Continued) k Identification Multiplt Tentative I 4155.48 OII 19 357 4163,12 [KV] 15 1.1 4169.14 OII 19 2,8 4171.57 NII 431.1 4176.75 1.2 4181.15 NII 49 1.5 4185.51 OII 36 2.8 4186.94 CIII 4.5 4189*78 OII 36 3.9 4195.72 NIII 6 2.8 4199.90 HeII 3 16.7 4202.00 AII 8, 124 o.6 NIII 6 4215,80 I 363 01 33 4217.20 Nell 52 1.1 4219.79 NeII 52 2 8 4222,83 OI 33 4225.07 lo1 4227.44 [FeV] 4231.61 NeII 52 1.1 4233.83 OI 33 0.7 4237.04 NII 48 2.2 4241.24 NII 47,48 3.3 4245,28 0.8 4J248,84 0,8 4250*53 NeII 52 1 1 4253 88 OII 101 3.2 4267,11 CII 6 34.6 4275.57 OII 67 6,5 4276.73 OII 54,67 3,7 4277.74 OII 67 247 4272.88 1,1 4281.38 OII.40 0.9 4283.29 OII 54,67 2.4 4285*70 OII 78 2.5 4288.80 OII 54 1,2 4291.61 OII 78 2.0 4294*85 OII 54 3.2 4303.89 OII. 54 5.9 4307.33 OII 53 o.8 4309.21 AII 36 1.7 4311.50 CII 28 0 8 4313595 OII 78 1.2 121

TABLE 7-2 (Continued) X Identification Multiplet Tentative I 4315.36 OII 78 1.2 4317.24 OII 2 4.3 4319.81 OII 2 2.9 4325.83 OII.2 3.1 4327.63 OII 11 2.3 4331.39 OII 11 1.1 4333.0 O1II 65 1.6 4337.10 OII 2 1.9 4340.42 EH Too Bright 4344.51 OII 65 2.7 4345.72 OII 4.8 4347.80 OII 16 2.1 4349.46 OII 1 7.1 4351.75 OII 16 1.2 4355.47 0.7 41363.16 [OIII] 15 255.1 4366.97 OII 2 5.1 4369.20 OII 26 1.5 4371.73 OII 76 1.5 34576.67 OII 46 1.6 4379.20 NIII. 17 12.4 4387.90 HeI 21.7 4392. 00 NeII 57 3.3 4398.02 NeII 56 1.8 404oo. 91 4403.51 4408.31 1.8 4409.42 NeII 57 3.0 4412.95 NeII 55 1.5 4414.94 OII 5 4.0 4416.94 OII 5 34 NeII 61 41419 98 4425.43 0.6 4428.50 NeII 61,57 7.5 4431,09 NeII 56 2.4 4432.76 NeII 55 2.1 4434.87 1.1 4437.60 HeI 50 2,5 4439.95 Nell 61 0.9 4442.47 NeII 56 0.3 4446.68 NeII 56 0.9 4448.537 OII 35 1.7 122

TABLE 7-2 (Continued) x Identification Multiplet Tentative I 4453.29 1.0 4456.83 NeII 61 1.3 4459.04 1.3 4464.25 0.9 4465.65 OII 94 1.8 4471.50 HeI 184.4 4478.04 oII 68 1.5 4481.16 MgII 4 1.1 4487.85 OII 104 1.0 4489.70 OIl 86 1.3 4483.23 4491.89 OII 86 2.1 4495.58 4498.52 AII 136 1.0 4504.40 4510.98 NIII 3 6.9 4515.08 NIII 3 2.2 4517.52 NIII 3 2.0 4i521.07 0,3 4523.65. NIII 3 2.4 4527.82 NIII 3 0.2 4529.7 OIII 15 4530.32 NII 59 1.5 4535196 1.9 4534.57 3NII 3 2,7 4541.59 HeII 19.4 )444.80 NIII 12 2.6 4547.34 NIII 3 1.4 4552.54 NII 58 4557.72 0.9 4562.66 NeII 64 0.8 4564.43 AII 85 1.1 4571.10 MgI. 1 3,8 4574.58 SilII 2 4588.6 4590.97 OII 15 2.8 4596.17 OII 15 2.2 4602.1 OII 93 2.3 4609,9 OII 93,92 4.8 4613.30 4621.07 4631.10 4634.16 NIII 2 44,5 123

TABLE 7-2 (Concluded) Identification Multiplet Tentative I 4638.85 OII 1 1.5 4640.64 NIII 2 85.5 4641. 90go 2 25.7 OII 1 4647.40 CIII 1 9.1 4649.14 OII 1 24.7 4651.35 CIII 9.8 4658.42 4661.64 OII 1 8.3 4667.28 4671.09 4673.75 OII 1 2.7 4676.26 oil 1 6.7 46835 40 4.9 4685.68 Hell 1 Too intense 4698.27 4702.00 OIIl 25,40 63.3 4705.36 OIl 25 47o8.56 4711.34 AIV 142.5 4713.17 HeI 17.1 4714.80 1.7 4740.20 4782.539 14.0 4805.06 23.5 4861.530 H 4921.96 HeI 4958.95 124

4294.82 OII o',",'41.59 H6-.l -- 4303.82 oIl NIII 4544.80.5 —9 onl 4307.51 -'?597.9i....... 0el 4345.26. --'. 0 4552.547 NI7 - 4319.63 OII 4327.48 oil4 oIl 4331.47 - 4571.10 MgI 434o. 47 Hy i- 4349.49 o0II 4596.17 OII L.A. cIII 4376.78 - 4379.09 NIII HeI 4387.97 - 4391.94 NeI NI 4634.16463110 4397.94 NeII N6 460.85 oii 4409.30 NeIl OI l 4649.11- 4647.40 OIl 0ii 4414.91-".... iii0,1i... ~,',:'4416.98 OI 457.69. NeIl 4428.54 NeI 4432.26 - - 4430.90 NeI 4437.55 Hie oI 4673.75 4676.26 onl 24448.21 6II 4683.40 4685.68 HeII "- 4456.959 NeII OIl 4701.76 4698.27 4465.4 on0 4471..50 eI. Hel 4713.17 4 [v OIl 4477.88 - 4481.33 MgII 4491.25 OII 4498.55 AII 4740.20 [AIv] 4510.92 NIII NIII 4514.89 — i'-i 4518 18. NIII -- 4523.60 NIII 4531 96.45340 NIl - 4541.59 Hell 4782.69 Fig. 7-1. Spectru. of NGC 7009 -

Hell 3833.80 - -. 3835.39 H 40062.90 - Nell HeI 383:09 - OIl 4072.1610 il }4075.87- OHel3580 3i- 856.02 Sill oIl 4083.91 - 4085.12 OIl HeII -858.075 S OII 4+087.16 -.4089.30 OII 3862.59 S480 4092.94 OII. 3868.76 [NelIIl] I 4097.31 NIII - 871.62 CII H6 4101.74 - -'-^^p^^^- *'v h-inh. 7h _^H —B^^ — 4103.02 Oil 3882.20 Oil 4107.07 oII [Fev] 3891.28 889.5 H 46. siv 3895.52 [FeV] l 119.22 -- i 4120.81 HeI 3907.15 OIl- 4132.81.l 4140.5 HeI 41+43.776 5- 3920.68 CII 4146.05 HeiI 3923.18 --::::: - 926.537'e 0 1+3.30 -'- 1 — 4156.54 Oel - 4163.30. [KV] OIl 391+5.05 - 169.2. OIl 3948.15 CII Nl1 6 39541.37 1Oil 4181.17 NII Oil 4285:;i::.15 — - * 4185.46 Oil eI.3964.75' —-'CIII 118705 1856 OIl HeI.3964.73 -1 4189.79 OIl H- 3970.07. - 3967.7 [NeI] NIII 195.70. [SI] 46.39730-27'oII -f- 46.4199.83 HeII - 3982.72 OIl NeIl 1+217 15 - 1+::~ 4219.76 NeIl 3995.00 4- 4227.49 [FeV]: —4003.64 NIII 4241.79 NIl Fi2.19 He1 oNeII 4250.68- —:2:'' " 402619 HeI e: I --- 1035.09 OII 401+3ii:.00' — 401+.31 OII::''l:::''- 14267.15 CII 4043-00" — A. - 1+273.17 OII OIl 4275.52 - - 4277.90 OII 4- 1281.40 OIl 4:: ~285.70 O — -'4062.90 NeII OII 4288.83 4291.25 0II [SII] 4068.60 +069.64 oil oil 4294.83 OII 4072.16 --—:O 1+ - -- 075.87.ol - 1303.82 oII Fig. 7-1. Continued. 126

[NeIII] 3... 3.40.74 OIII 5355.05!Ne'-l?3587.40 HeI 3355.05 NeIII -- 3367.00 oII 3613.64 HeI 3634.24 HeI 0 34 1-3 —'3405.74 0111 OIII 3408-13 r: oIV 3411.76- L.A....i —- 3415.29 0III L.A. 3417.9, - q IOIII 3428.67 — H 3666.loi 667.6 H 3430.60 oiii H 3669.47 3671.48 H H 36736.47 -H;:~ 36~ 73676.36 H 3:;- 3440.9 0 H 36793I 36. H NIII 3444. -- 3-82.81 H 3447.59 HeI H 5i686.83 3691.56 H NeIl 3694.22 01! 3702.75' —-_ 3697.14 H )73.85 H Hel 3705.001 - 0 NeII 3709.64 3707 NeII 3713.09 3711.97 H Hi. 34789 Hel 3721.94 H 3487.72 Hel oii32.5 788 Ol -- 3512.51 He 0111o 3754.673 3757.21 OIII 0III 3759.87- 3762.41 SiIV'NeIl 3766.29 353o —-49 HeI 8 II o3770.63 H -504 3777.16 NeIl SIll 3778.90 - 37.6784.89 -eI?~:~ ~"~~~~~...........3791.26 0III / —3554.52 HeI HeII 3796.33 3797.90 H:~:~Ji~~~~~~~~~ ~~3805.77 Heil 3568.53 NeII SIII:: 377 — 3813.50 Hell f iil — ~3819.61 HeIr Fig 7-1. Continued 1;27

3121.71 OIII 3132.87 0111 - 3187.74 Hel - 326o.98 oiii OIII 5265.46 -- - 3299.36 OIII - 3312.30 OIII Nell 53335.87 - 3350.7 OIII Fig. 7- 1 Concluded. 128

6aT 4650.16 CII 40 Ir. 4649.1 4647.40 C m _ 4640.64 N]' 4641.90~ < 4638.85 0 I — j^, - - ~4634.16 Nm ] >4921.93,HeI. 4 6 4610.14 0 U 4602.1 Oil r i 4596.17 0?I 4861.33 HI _. g 4571.1a4571.0 Mg I o. (D 4544.80 N fm 0 P ^4541.60 He p > 4534.57 N mII P. 4530.40 N 11 4782.69 --. 4529.7 0 om c1+ 1 4523.60 N IIm CD co 34518.18 0 t 4514.89 N mI c-F 4510.92 Nm d _I, 4740.19 Cs <| ~ 4724 8 0 -^ ^4498.55 A II ~~~~~~~~~~*, ^^~~~~~~~4711.34 [A T].4491.25 00 > 4489.48 0 2__ —- -- _-4698.27,r448772 OII | ~ -- --------- 4685.68 ^Q 54481.33 Mg nIIl —>4477. 88 0 l= 467626 0 D S^ E ~~~, 4673.75 01 4471.50 He'-' - 4661.64 0 On ---------- ( ~~~~~~~~465769

OCT I04336.87 O 4465.40 0 I 4332.76 0 E >4331.47,31.13 0 u 4459.1 4327.48 0 U\ 4456.95 Ne EL 4325.77 OII ~~<>~~, \ ^~~~4453.2 4319.63 0 n 4317.14 0 II,4448.21 0 11 4315.35 0 En 4446.46 Ne U 4313.50 CE 4313.43 n0 4312.43, 4442.67 Ne l " 4309.11 AR1 4307.31 0 H 4 443755 He I 4303.82 0 U 4432.26 Ne E 4430.90 Ne n 4428.54 Ne I 4294.82 0 n1 4291.25 0 H 4288.83 0 1 428570 0 u4419.98 >4285.70 On 11 441.98 0 I, 441677 Ne nL 4283.13 0 1 4414.91 0 I l ~-".'-4413.20 Ne It F,;1..~~~ P^1^g~~~ 54412.54 Ne I ~Q'~^ %^7~'"onP-4409.30 Ne II * i 4276.71 O E 4409.30 Ne Z 4275.52 O 1 4408.3 73 Fs4273.17 0 Q * 4267.15 C H 4397.94 Ne I1 43911.94 Ne II (D > {S, ^^_____________ _4387.97 He I "4253.98 0 lO 4253.75 O11 4250.68 Ne II g4248.84<- t _ ___ 4379.09 N m 4245.28 > 19^^~~~~~~~~~~~4376.78 C U 4241.79 N 1 ^^^r~~~ 1\~~~~~4371.65 on 4237.05 N 14369.28 0I 423693 N Er 4233.32 0 1 4436 6.90 0 n 4231.60 Ne II 4363.21 m0 422749 [Fe ] Los Anaeles J I4225.07 1% > 4355.47.~ 4219.76 Ne l7,421715 Ne 11 l 4349.49 0 n'4215.69 Nm 4347.43 E0 4209~ 4345.56 0 AI 4344.42 0 [ <^f > 1 ~~~~~~~~~~~~4340 Hy 4201.99 AD:[

T^T 4201.99 A 11 406896 ] n ^ ---— 64 0111 4199.83 Hell 4195.70 N m 4062.90 Ne II 4060.98 OIl 4060.58 0 1 4189.79 0 U 4187.05 CML 4185.46 0 11 4181.17 Nil 4180.87 4176.75 4043.00 404131 On U417161 NI 4171.6!1 N 11 4169%23 OnI 4035.09 011 4163.30 [K V] 402619 He I.- {-~^^4|56.54 OS 4156.54 O IL 4153.30 O1 4150.67 Ne I 4146.05 P.h 4143.76 He l -^ r —---- ~4009.27 He I >4140.5 Ro )> 4136.8 4135.5 4003.64 N mD 0 j4001 -- 4132.81 On 0 + >3999.1' 127 H A 399736 Fe[S] >4128.74 g ^~~3995.*OO~~~ S4125.35 g,$ ~~~ p ^~3993.~~1 l~>4123.9 [Fe V] 4120.81 Hel 4119.22 011 3986.2 4116.10 Si If 398272 0 1 4112.03 0 n 3979.71 ->4110.80 OI 4109.9 3976.5 < >4107.07 0 II 4104.74 011 4103.02 O[S ____ 4101.74 HJ 4097.31 BJL 4092.94 0 n 4089.30 OHl 5408716 O1I 4085.12 0 n 4083.91 On 401.10 oin C^""______~4075.87 OnI < 407390 z^-~ 07. —- ----- * 4069.64 4^0' j _________ — S~40-68.60[Sn]]

t 3839.52 [Fe 2] 1 3838.09 He I 3835.393 HH 3970.07 H 3833.80 He 1_ 3 3831.9_ __ 3967.47 [Ne I) 3829.36 Mg r 382.83 A 1 3964.73 He I o,x, i3960.7 C<~_ _ 73819.61 He I _ 3954.37 0 1 3813.50 He L -===:: >381350He^^ ~~~~3948.15 C II 3810.96 o0 m 3757.21> 0 Li3945.05 IL -f._ - ___S 6 3935.02 3797.90 H ) 3932.72 [Fe 1] 7596.33 He II3 3728.80^_ (03928.87 _ 3791.26 0 m 3926.53 He i3' ^.= 3923.48 He I IJ S 3784.89 He I S 3920.68 C II 04I. > 3918.98 C I > 3778.90 sm *i t37 377 16 Ne 113915.1 3539152 3774.00 om1. 773.13 Si IE QT _ < _ 3907.45 0 1 0 3770.63 H 3906. 3906.1 c ~-3768.81 He I P3 70 3766.29 Ne I P - 3762.41 Si T13 -3759.87 m-. - 3895.52 [Fe Y] | - __ -a- 3757.21 0 mr ^_ 3754^~.67 0am I > 3891.28 [Fe -] 3889.05 H s>_ _______________3750.14 H_ 6 II ---- 388220 0 IsL 3746.46 A II, 3739.92 011 3871.62 C E 3734.37 H \ \ 3870.8 S 373285 Hel __ 3868.76 [Ne mE]J >3730.853728.80 [0 Il_^[ p>3866.8 He_"-| ^~ / =s86 3862.59 Si 3I 37260 5 [0 O] __ 6 ~ "t_::::::::::::::__ 3858.07 He: 3721.94 H _ --- 3856.02 Si 1:3 1 < 3853.0 _~ ~ 3715.27 / 3714.03 0 im ~,^3713.09 Ne I1 3711.97 H >3709.64 Ne I * ^ —- ~ ~__3707.24 0 m

3705.00 He I 3703.85 H 3568.53 Ne - -.75 oI 3697.14 H 3694.22 Ne 11 369156 H 3554.52 He I 3530.49 He 1 3664.68 —- -- -— _~. t~36 86.83 H 3^ ------------ 5'>~~ 12.51|; - — ~3682.81 H o^ -----— _______ 63424~~~~~~~ ~ He-3679.35 H c-I —---— 3496.64 He_____ — ~ _3676.36 H 3494.66*~ 0- C -_. __.___._U_.._ 3673.76 H 3,r ^ _ _ __ - 3671.48 H 3 478. 973669 47 H 3667. 68 H 34664- 3666.10 H 3530.49 He I \3664.68 H L.A. * >\ *^-366 H 3661.22 H 3450.94 ~;> 03660.28 H 344759;7=>~' He 3659.42 H 3512.51 VHe I -3 3504.235 0',_ - - " ^4.24 Hel r ( ^Cy 3498.64 He I 3494.66 0 I 3487.72 He 3613.64 Hel 3478.97 He I 3471.80 He I > 3468.44 3466.1 4 o0 3464.27.. 4 L = 3587640 Hel; =.3583.03 3450.94 0 fl] m? s344459 He I ~ 3578.08 3444.10 0 l'r'[ 3442.0,~3574.29 Ne n > 3440.39 0 ff, 3438.35 P

343835' —-3299.36 m3430.60 3428.67 0 m 342553 0 IV 34177 34179 3415.29 o0 3411.76 0 IO (?^~fiQ~~ ~o8. o3408.13 0 r i3405.74 0 ml 3399.80 3267.3 339790 Ne I:3365.46 0 mrl 3263.43 Ne U 3392.78 Ne II 3260.98 0 I ~~~~a~~~~~~~~~~~ oo'o':5 t >336700 On ('2 H'\^.3355.05 Ne El >3342.5 [Ne i],~ _ -- "^ 3340.74 0 m1,- 3334.87 Ne El._____ - --------- 3203.10 HeI t= — _31 -3187.74 He I' 3312.30 0 m i:

Hv HelI IIi %4686 0434 OMI )4363 Hel 7U4471 H6 I A IV NeIII 4 101 h3969 Ho SUI An A fl 1 I11 ~~~~~~~~~~~Hel h-1 OBJECT:. NGC 7009 DATE: 8/31/56 TELESCOPE: 100-inch SLOT WIDTH: 12A HS |l NeIII SCAN SPEED: 90A/min \3889?.3967 He 73969 0II H?X.3727 SII Hel H OIII FSs55 Ng7 k34255 BAAi R (Ha). CONTINUUM Fig. 7-5. Spectral scan of NGC 7009.

I

REFERENCES 1. Bowen, I., Minkowski, R., and Aller, L, H. Ap. J., 1955. 2. Bowen, I. S., and Wyse, A. Bo Lick Obs. Bull. 19, 1, 1939. 3. Wyse, A. S. Ap. J. 95, 356, 1942. 4. Aller, L, H. Gaseous Nebulae, Wiley, New York, 1956, p. 30. 5. Aller, L. H. Abundances of the Elements Interscience, New York, 1961, po 71 et seq. 137

SECTION VIII REPORT ON RESEARCH CARRIED OUT AT THE UNIVERSITY OF MICHIGAN AND SUPPORTED BY THE U.S.A.F. UNDER CONTRACT NO. AF 49(638)-807 by D. Mugglestone

I

An extensive and detailed study of the most recent literature in the field of stellar and solar abundances was made-particularly of the literature brought to light and forming the bases of controversial discussion at the recent International Astronomical Union meeting at Berkeley, California. Particular attention yas given to (1) considerations of the effects of saturation on the derived abundances of elements from the observed absorption lines, and (2) the very controversial question of the possible deviations from local thermodynamic equilibrium (L.T.E.) in the solar atmosphere. (1) A paper by Neven, brought to the author s attention at the I.A.U. meeting, severely criticised published material by the author on solar atmospheric abundances. The theoretical background of this criticism was studied and, except in minor matters, was found to be erroneous. A program was initiated to recalculate Neven's results in the case of oxygen and nitrogen (this is nearing completion at the University of Queensland) and to establish numerically the source of Neven's error. (2) A major controversial topic at the IoA.U. meeting was the question of the applicability of Local Thermodynamic Equilibrium (L.T.E.) to the outer atmospheric layers of the sun. Astrophysicists were strongly divided on the question, as some groups of research workers (notably the Paris group) require strong non-LT.E. effects to reconcile the solar abundances they derive from weak and strong absorption lines, whereas others (notably the Michigan group) find close agreement between weak and strong lines while still maintaining that L.T.E, applies. It is felt that the situation can be clarified by a study of one element (oxygen has been selected) by theoretically predicting the detailed profile of the absorption lines, not only at the centre of the solar disk but also at various disk positions toward the limb of the sun, and then establishing whether or not non-L.T.E. effects must be invoked to reconcile these predictions with observations. The basic theory necessary to accomplish this objective has been closely studied and completed in detail, and a program has been initiated to utilize. high-speed computor techniques to deal with the lengthy and complex calculations necessary for this analysis. Very accurate observational line profiles are necessary and will.be obtained by Dr. Edith Miller. The high-dispersion spectrograph of the McMath-Hulbert Solar Observatory (The University of Michigan) will be used and corrections will be made macro-turbulence in the solar atmosphere ("wiggly-line" analysis). The programing will be carried out on the new G.E. 225 computor of the University,of Queensland. It is fully expected that the analysis will provide valuable information on the applicability, or non-applicability, of the concept of Local Thermodynamic Equilibrium in the outer layers of the sun. I would like to express my appreciation for valuable discussions with the Department of Astronomy of The University of Michigan, particularly Professor Aller and Dr. MIller, who gave so freely of their time. I would also like to thank the United States Air Force Office of Scientific Research, whose financial assistance made so profitable a visit possible. D. Mugglestone Senior Lecturer in Theoretical Physics University of Queensland Brisbane, Australia 141

UNIVERSITY OF MICHIGAN 3 901 1111111111111111111111115 02499 5543 3 9015 02499 5543