Stellar atmospheres

THE observational test of ionization theory involves a considerable program of measurement, if the accuracy necessary for a quantitative test is to be attained. The present chapter contains a synopsis of new data obtained by the writer to supplement the material already published in Harvard Circulars.[403][404] The data here presented practically complete the available material for the strong lines of known series relations in the region of the spectrum usually examined.

LINE INTENSITY

The theory predicts the degree of absorption that will be produced by each atom at a given temperature, and the related quantity that is measured is the intensity of the corresponding Fraunhofer line in the spectrum of the star. Spectrum lines are differentiated by various qualities, such as width, darkness, and wings, and their conspicuousness is governed by the intensity of the neighboring continuous background. It is not easy to specify all these quantities on an intensity scale that is one-dimensional, and the various ways in which line intensities have been estimated represent different attempts to choose and express a suitable scale.

Many of the applications of so-called line-intensity, such as the estimation of spectroscopic parallaxes, have involved ratios between the strengths of various lines in the same spectrum. This method of comparison avoids most of the difficulties caused by differences of line character and continuous background, for the lines that are to be compared are chosen because of their proximity and comparability. Harper and Young[405] have standardized the method by comparing spectrum line ratios with line ratios on an artificial scale.

METHOD OF ESTIMATING INTENSITY

In a comparison of ionization theory with observation, some measure of line-intensity is required which can be compared from class to class. It seems probable that direct estimates of intensity, for spectra of the same dispersion, density, and definition, will be comparable within the limits of accuracy of the material.

Two series of spectra were measured by the writer in order to obtain material for the test of the theory of ionization. For the first group standard lines in the spectrum of Cygni were used for the formation of a direct intensity scale, and for the second group, comprising the cooler stars, a strip of the solar spectrum was similarly employed. An arbitrary scale was constructed by assigning a series of intensities to well placed lines in the spectrum, and using these as standards. A list of the lines used for the second group, the assigned intensity, and the intensity as given in Rowland’s table, are contained in the following table.

The estimates thus made might be defined as estimates of width-intensity-contrast between the line and the continuous background. On an ideal plate which was not burned out, such estimates would give a measure of the total energy of the line relative to the neighboring continuous spectrum. The accuracy attained by direct estimates of this kind appears to be as great as the material warrants.

ACCURACY OF THE ESTIMATES

It is not possible at present to evaluate the accuracy of these estimates with the same precision as for other physical quantities, but the consistency of the readings from comparable plates of the same star will at least give a measure of the value of the estimates. Table XVIII contains the measures on forty-three lines in the spectrum of Gruis, taken from six plates of the same dispersion, and comparable quality, density, and definition. Successive columns give the wave-length, the arithmetic mean intensity, and the standard deviation .

TABLEXVIII

These measures are strictly representative of the material as a whole, for the plates of Gruis were measured at wide intervals in the ordinary course of the work, and were selected for illustration because there was a greater number of suitable plates of this star than for any other.

HOMOGENEITY OF MATERIAL

The observational material on line-intensities follows in tabular form. The measures were made in two groups, comprising respectively the hotter stars and the stars cooler than Class , and different intensity scales were used for the two. The solar scale mentioned above was used for the second group of stars; the first group was referred to standard lines in the spectrum of Cygni. The distribution of the stars in the two groups among the spectral classes was as follows:

If it were possible to use a series of giants throughout, the task of determining the intensity maxima would be greatly simplified. Among the hotter stars the differences introduced by absolute magnitude are not great enough to make the maxima difficult to determine. With later classes, however, the changes with absolute magnitude are very marked. As will be pointed out in an ensuing chapter,[406] the actual strength of the lines differs considerably from giant to dwarf, owing to the difference in the effective optical depth of the photosphere. This difference in strength is in addition to the well-known “absolute magnitude effect” which is shown, for example, by the enhanced lines; it increases the difficulty of making estimates of line change from one class to the next, since, owing to selection, the available stars are far from homogeneous in absolute magnitude. In addition to this factor, there is the practical difficulty of making comparable estimates on the sharp narrow lines of a super-giant and those of a dwarf, since the lines of a dwarf tend to be hazy and lack contrast with the background.

It might be expected, from the distribution in luminosity of the stars used, that irregularities in the intensity sequence would probably occur in the classes and at . For the purpose of estimation of maxima, the classes are not of very great importance, as few of the maxima under present investigation occur there, but the irregularity at may well prove to be serious. There is indeed a general tendency for the intensity of, metallic lines to increase at . All the stars measured were of very high luminosity, and probably the rise of intensity is due to this feature, or rather to the increase of material above the photosphere that accompanies it. A maximum is only assumed to occur at when a line increases regularly through the types, as do the lines of neutral calcium. The iron and titanium maxima obviously occur earlier in the sequence, although the lines of both these elements are often noticeably strengthened at .

The following tabulation contains the data on line-intensity for all the lines of known series relations that have been measured up to the present. All the measures were made by the writer, excepting those for zinc, which are taken from Menzel’s paper.[407] Successive columns of the table contain the atom, the series relations, the wave-length, and the observed intensities in the various spectral classes. The column headed “Blends” is a direct transcription from Rowland’s tables, and contains details both of the line under consideration and of closely adjacent lines. The column headed “Remarks” contains the writer’s own conclusions, based on solar evidence, astrophysical behavior, and laboratory affinities, as to the source and maximum of the line that has been measured.

The recorded intensities, for classes cooler than , are derived from the selection of stars mentioned earlier in the present chapter. A list of the individual stars is contained in Appendix III. Four typical stars have been selected to represent that class. The figures in the final column refer to the notes to the table, which are listed under the respective atoms, and give the observed maximum, the intensities and origins of blended lines (in Rowland’s notation), and short remarks, which indicate whether or no the observed behavior is to be attributed to the line considered. Maxima that are obviously due to another line are placed in parentheses.

TABLE XIX