It would be very ungracious not to say a few words about this remarkable set of presentations, even though it would be cruel to the audience to try to review them at this late hour. I made sixteen pages of summary notes of the highlights. We have heard about enormous improvements in our knowledge of the behavior of complex molecules in space. May I remind you that the computations of chemical equilibrium presented by Professor Klemperer covered largely the diatomic molecules known many years ago. The radio astronomers, with incredible technical proficiency, have found many polyatomic molecules with molecular weights close to a hundred; in a solid particle that causes continuous absorption we must reach weights near 108.

This report will review space observations of stars and the interstellar medium and what we can expect to learn from instruments planned for the near future. Older observations are described in more detail by Underhill and Morton (1967), and Houziaux and Butler (1970). The interpretation of these measurements depends heavily on the results of theoretical and experimental atomic physics. Therefore the last section will describe some of the areas where new atomic data are essential for our understanding of space observations.

Rockets or satellites are necessary to detect all wavelengths between 0.1 and 3000 Å except for two bands from 2000 to 2200 Å and from 2700 to 3000 Å accessible at balloon altitudes. However, for most sources outside the solar system the bound-free absorption of atomic hydrogen will obliterate all wavelengths between 100 and 912 Å. For example, unit optical depth at 304 Å occurs at 12 parsecs even if the interstellar hydrogen density is as low as 0.1 atom cm-3.

Solar ultraviolet and X-radiation arriving at the Earth is emitted mainly by highly-charged ions in the outer layers of the Sun, where departures from thermodynamic equilibrium are extreme. Hence the diagnosis of physical conditions from spectral intensities entails the use of accurate cross sections for a variety of physical processes governing the excitation and ionization of emitting ions. Now that the full potential of satellites for solar data acquisition has begun to be realized, the need for atomic data of all kinds is more acute than ever before. In this review, I shall illustrate the importance of laboratory and theoretical investigations for the analysis of solar ultraviolet and X-ray data by referring to five problems to which space experimenters are currently devoting a great deal of attention:

• (1) the temperature minimum at the interface between the photosphere and the chromosphere,

• (2) the temperature and density profiles of the low chromosphere,

• (3) the transition zone at the chromosphere-corona interface,

• (4) the corona,

• (5) solar flares.

A review is given of some of the problems of classification and intensity analysis involving unidentified or recently identified lines in the ultraviolet solar spectrum. In particular, attention is drawn to the present position regarding two and three electron spectra in the soft X-ray region, and forbidden transitions within the ground configurations at longer wavelengths.

In the past three or four years a great deal of attention has been given to the interpretation of the solar UV continuum. Several sets of intensity and limb-darkening measurements made from satellite, rocket and balloon borne instruments are now available, which allow comparisons to be made with theoretical computations. The main purpose of this paper is to discuss these comparisons, we restrict this discussion to the wavelength region between 3000 Å and 1680 Å, in which a significant difference between the observations and the computations still exists. The lines in this wavelength region which are of relevance to the continuum opacity are listed as follows:

the Ly-α line of neutral hydrogen at 1216 Å which appears as a strong emission line;

the resonance doublet of Mgװ at 2795 Å and 2803 Å which causes a broad depression in the solar continuum and shows an emission core similar in shape to the Ly-α line;

the auto-ionization doublet of AlI at 1932 Å and 1936 Å, which appears as two broad absorption features;

several absorption lines such as Mgi 2852 Å Sii 2881 Å;

many FeI and FeII absorption lines.

There are also several discontinuities in the continuum emitted between the lines in the spectrum between 1216 Å and 3000 Å. The most important ones are located at 2500 Å, 2085 Å and 1680 Å, which correspond to the photoionization edges of MgI, AlI and SiI respectively.

In the present review I will discuss certain rather general aspects of the theory, and will leave it to other speakers to describe the results of detailed calculations. Further references to recent work are given in the Report of IAU Commission 14.

1. Atomic Wave Functions and Energy Levels

In order to calculate data of importance for ultra-violet and X-ray astronomy - wavelengths, transition probabilities, collision cross-sections, ionization and recombination rates, and line-broadening parameters - it is necessary to begin with the calculation of atomic wave functions. The interest is mainly in highly ionized systems.

This report is a continuation of the preceding review talk. I shall describe general purpose computer programs that have been developed in Professor Seaton’s group at University College London. Other speakers will present results produced with them.

The programs require a minimum of input. General purpose means; not restricted to one ion or just one isoelectronic sequence. High speed is achieved by using approximations suitable for medium and highly ionized light atoms. The programs allow fully for configuration mixing, which plays an important role in ion calculations. Equations in M. J. Seaton’s review will be referred to as (S.number).

Using general purpose computer programmes developed by Drs. Eissner and Nussbaumer at University College London for the solution of atomic structure and atomic collision problems, collision strengths have been obtained for transitions between terms of the ground and first few excited configurations of ions of interest in studies of the solar corona. The atomic structure programme is written in a configuration interaction representation and uses a scaled Thomas-Fermi potential as described by Eissner and Nussbaumer (1969). The bound state wave functions generated by the structure programme are used by the collision programme (Eissner, 1970) to calculate the R-matrix solution of the collision problem by means of the distorted wave approximation (Saraph et al., 1969). This approximation should be valid for highly ionized ions such as exist in the corona and this prediction is fully confirmed where comparison can be made with calculations using the more rigorous but more time-consuming close coupling method (Burke et al., 1966; Petrini, 1969).

The collision strengths for transitions between levels of Fexiv have been calculated using the close coupling and the Coulomb-Born I methods. It is shown that the Coulomb-Born I approximation gives reliable results for highly ionized ions.

The electron excitation rate coefficient for the transition in the 3s2 3p configuration, giving rise to the green coronal line, has been investigated in detail. The quantum defect method, which takes into account the excitation via autoionization levels of FeXIII, has been used. This process greatly increases the excitation rate for low energies of the incident electron.

The statistical equilibrium populations of the FeXIV levels have been calculated under coronal conditions, and the effective excitation rate coefficient for the green coronal line is given for a wide range of temperatures and densities.

The permitted transitions 2s21S-2s2p1P and 2s2p3P-2p23P in the Bel-like ions CIII, NIV and OV have been observed for some years in the solar spectrum (Hall et al., 1963). Recently, intensity data have also been obtained for the intercombination line 2s21S-2s2p3P1 in these ions (Burton et al., 1970). A large number of excitation rate coefficients are needed before the intensity ratios of these transitions can be computed and compared with those observed. These excitation cross-sections are now becoming available (Osterbrock, 1970; Eissner, private communication), and the present paper gives the results of an analysis of the intensity data. Figure 1 shows a partial term scheme for the Bei-like ions and the observed transitions.

A summary is given of some recent measurements of the ionization and excitation of positive ions by electrons, and the detachment of electrons from H− by proton and electron impacts. The results are discussed and compared with theory. Conclusions are drawn which are relevant to atomic collision processes which occur in astrophyscial plasmas.

In the visible part of the spectrum, the methods to solve astrophysical problems by laboratory experiments are well known. Electric arcs or shockwaves allow the production of plasmas of known parameters, from which measurements of oscillator strengths, of line broadening effects, and of line shifts are possible. However, there was a difficulty to introduce elements into laboratory light sources which have no volatile compounds. This was particularly difficult for the metals. This difficulty was overcome only recently by burning the electric arc in argon gas containing a few percent of chlorine. When this gas mixture is led over heated metal, some metal-chloride is formed and fed into the arc. In this way, even a very small amount of metal can be investigated, so that the resonance lines of the metals are optically thin. As an example I would like to recall that during the past year Professor Richter at Kiel has succeeded in a redetermination of the absolute oscillator strengths of iron lines, which has allowed to correct the abundance of iron in the Sun’s atmosphere. The abundance turned out to be ten times larger than assumed hitherto, leading to a new conception of the conditions in the photosphere (Kock and Richter, 1968; Garz and Kock, 1969; Garz et al., 1969a, b; Baschek et al., 1970). With this abundance, there is no difference in composition between the solar photosphere, the corona and meteorites. The method used for iron was applied to a number of other elements. The abundance of nickel is of special interest. The ratio of iron to nickel is again the same in the photosphere, in the corona, and in meteorites, being 1:1.25±0.15 (see Figure 1).

Historically, the first observation of a relativistic effect in atomic spectra was probably the discovery by Fraunhofer of the splitting of the Nai doublet in the solar spectrum. Thus theoretical understanding of these effects has long been important for interpreting astrophysical spectra, and it is especially important for the ultraviolet and X-ray spectra which form our subject today, for these often arise in highly ionized atoms. Because the relative importance of relativistic to non-relativistic terms is proportional to Z2, where Z is the nuclear charge, relativistic effects often play a major role in such atoms.

The relativistic effects on bound levels of atomic systems with particular application to systems of astrophysical importance are investigated by including the most important contributions to the low Z Pauli Approximation for the Hamiltonian. This work is an extension of a general purpose automatic computer program for calculating non-relativistic energy levels of complex atoms, described by Eissner and Nussbaumer (1969). The program should give fairly good intermediate coupling wave functions and it may thus be used as a basis for the calculation of intermediate coupling collision strenghts and oscillator strenghts. Numerical results are displayed for the Sodiumi sequence.

A review is given of the occurrence and transition probabilities of spectrum lines arising from magnetic quadrupole radiation. Magnetic dipole radiation in relativistic quantum mechanics is also discussed.

Most of the lines in the UV spectra are lines of ions which are formed in high temperature regions where the pressure broadening is caused by electrons and protons. This is the case in O and B type stars for which the theoretical calculation of the width of all the strong UV lines is important in determining both the blanketing effect and the abundances of the elements.

The cores of these strong lines are formed in non-LTE layers near the surface where the electron density is very low. The wings of some of the lines are more easy to interpret, being formed in deeper layers of the star, where one can assume LTE and where the electron density - or in the Sun, the neutral hydrogen density - is such that the pressure broadening is much more important than the natural width.

Two opposite approximations have been applied to the line broadening problem; the impact approximation is generally valid for electrons when the perturbations are so rapid that the collision time τc is very small compared to the typical time, Δω-1, of importance in computing the profile at the frequency Δω = ω – ω0 measured from the line centre. On the contrary, when τc≫Δω−1 the quasi static approximation may be assumed. Both of these approximations have been considerably improved and efforts have been made recently to develop a unified theory valid from the impact regime to the static regime.

The electron damping constants of some UV lines of astrophysical interest are calculated within the impact and semiclassical approximations and are compared with experimental results where these are available (NII, UV(1) and CII, UV (1)).

From the spectral intensity distribution of solar X-radiation one can conclude that there is a non-thermal contribution, which is effective during the initial phase of a flare. It is reasonable to suppose that the electrons producing the non-thermal radiation have a non-isotropic velocity distribution. Indeed, one should assume that the electrons, during their acceleration, obtain a preferred direction and that only afterwards does their velocity distribution become Maxwellian by means of collisions. Hence the short-wavelength continuous X-radiation in the initial phase of a flare, consisting mainly of bremsstrahlung, should be polarized and should have a non-isotropic angular distribution.

In papers presented to the IAU-Symposia in Budapest 1967 (Elwert, 1968) and in Leningrad last May, the polarization of X-radiation from electrons of a few keV accelerated in solar flares has been calculated using energy distributions based on the observations of Pounds et al. obtained with the satellites Ariel I and OSO-4.

This Joint Discussion is in two parts. The first part is a description of recent observations which illustrate the need for atomic data and the second part is a description of what atomic data are available or could readily be produced by the latest theoretical and experimental methods. The purpose of this summary is to highlight the immediate requirements for atomic data of current observations which are not met by our present knowledge and thereby indicate where further work is necessary in providing atomic data.

Although this is a discussion of atomic data the problems of assuming local thermodynamic equilibrium have inevitably been raised. The importance of considering whether departures from LTE are significant or not in the interpretation of observations is is clearly illustrated by the work of the Harvard group on the Lyman continuum emitted by the Sun which has shown a departure coefficient as large as 200 for the ground level of hydrogen.