Annu. Rev. Astron. Astrophys. 1981. 19: 77-113 Copyright © 1981 by Annual Reviews. All rights reserved

5.2. Elliptical and SO Galaxies

5.2.1. METALLICITY AND ABSOLUTE MAGNITUDE     It is convenient to discuss elliptical and SO galaxies together, since they appear to have almost identical color and line-strength properties - somewhat surprisingly, perhaps, in view of the distinctive disc structure of SO's. We must distinguish two (possibly separate) effects: (a) the variation of color or line strength across a given system, and (b) the variation of some kind of mean color or line strength between systems, as a function of absolute magnitude or other property. The color-absolute magnitude effect is the more thoroughly investigated of the two. Since radial variations within galaxies exist, corrections are needed in order to compare colors of galaxies at some particular intrinsic photometric radius. These may be uncertain, particularly if elliptical galaxies are prolate objects for which we do not know the orientation, but nevertheless a strong correlation of (U - V), (V - K), and (U - R) colors with M_v is observed, in the sense that the brighter galaxies are redder (Baum 1959, de Vaucouleurs 1961, de Vaucouleurs & de Vaucouleurs 1972, Lasker 1970, Sandage 1972, Sandage & Visvanathan 1978a, b, Visvanathan & Sandage 1977, Frogel et al. 1978, Strom & Strom 1978). Sandage and Visvanathan find no evidence for any difference in the color-magnitude relations of E or SO galaxies, and no evidence for a difference between galaxies in the field and in clusters. Their relation implies (U - V) / M_v 0.10 (see Figure 4). Interpreted as a metallicity change this is equivalent to nearly an order-of-magnitude change in metals between M_v = - 23 and M_v = - 15, and a metallicity change is supported by Faber's (1973) observation of line-strength correlations with absolute magnitude. Tinsley (1978) suggests a metallicity-mass relation Z Z M^0.25. Larson, Tinsley & Caldwell (1980) suggest that there may be a significant difference in the scatter of the color-magnitude diagram for field and cluster galaxies, which could be due to residual star formation in field galaxies.

Figure 4. Color-magnitude relation for field and cluster elliptical and SO galaxies. Data are from Sandage & Visvanathan (1978b, Table 1). Colors are corrected to an apparent aperture of half the de Vaucouleurs diameter, uncertain values have been omitted, and where the E or SO classification is uncertain the point has been plotted as SO. The error bar is a rough estimate of photometric uncertainty in the colors. The color scale is transformed from (u - V) to (U - V) using Sandage & Visvanathan (1978a), and the metallicity scale from the calibration of Aaronson et al. (1978). The dashed line is Visvanathan & Sandage's (1977) mean relation, their assumed distance modulus to the Virgo cluster and H0 being 31.70 and 50 km s-1 Mpc-1 respectively.

There is some uncertainty whether the color-magnitude relation flattens off for very luminous galaxies. Baum's relation, which flattened at both luminosity extremes, was based on a small sample of (B - V) colors - which do not have nearly as strong a correlation with M_v as do (U - V) colors. de Vaucouleurs (1961) and de Vaucouleurs & de Vaucouleurs (1972) claim a flattening, although it is not immediately evident in their later data. Lasker (1970) and Frogel et al. (1978) both suspect a flattening, but it is not apparent in the extensive data of Sandage & Visvanathan (1978a, b) or Strom & Strom (1978).

In all the color-absolute magnitude relations there is considerable scatter about the mean relationship, and Terlevich et al. (1981) have shown a correlation between the line strengths at a given absolute magnitude and the velocity dispersion in the galaxy. Since the ellipticity also varies with velocity dispersion, the result implies that flatter ellipticals have lower metallicities than more spherical systems at a given luminosity. This effect had also been noted from Sandage & Visvanathan's colors by van den Bergh (1979). The implied differences in metal abundance depend on uncertain projection effects, but must be at least a factor of two.

5.2.2. RADIAL GRADIENTS     Radial variations in the metal content of stars between the central and outer parts of elliptical and SO galaxies are evident in both line-strength and color gradients. Colors redden (or are flat within experimental error) towards the center, although sometimes the nucleus may be somewhat bluer than its surrounding region - through the presence of emission lines or the ultraviolet contribution of a small population of young blue stars or from a nonthermal central source. Variations may be measured either by using apertures of different sizes, centered on the nucleus, or by using a small aperture at different positions. The color-aperture relation has been extensively discussed by de Vaucouleurs (1961), de Vaucouleurs & de Vaucouleurs (1972), de Vaucouleurs & Corwin (1977), Tifft (1963, 1969), and Frogel et al. (1978); and more data have recently been published by Persson, Frogel & Aaronson (1979). Although variable-aperture photometry gives a broad idea of the color trends, the small moveable aperture method (also effectively the method used in photographic photometry) probably gives a more accurate measurement of gradients. Miller & Prendergast (1962) used the rather insensitive (B - V), but later investigations (Burkhead & Kalinowski 1974, Strom et al. 1976, 1977, 1978, and particularly Strom & Strom 1978), use either (U - V) or (U - R). The number of galaxies studied individually in detail is still small, but Figure 5 shows the gradients observed by Strom et al. (1977) along both the major and minor axes of the SO galaxy NGC 3115. Not all ellipticals show large gradients (i.e. logZ / R 0.04 dex kpc^-1), although projection effects from the uncertain shapes (particularly as the colors observed centrally must be an integration along the line of sight through outer parts as well) make general statements difficult. In their extensive (U - R) photometry of Coma cluster galaxies, Strom & Strom (1978) suggest that only systems with M_v < - 21 (H₀ = 55 km s^-1 Mpc^-1) show particularly large color changes with radius, and then only in 20 percent of systems of this brightness. In systems with an obvious gradient such as NGC 4816 (Strom & Strom 1978) the continuous (U - R) change of about 1.2 magnitudes out to 70 kpc implies an order-of-magnitude variation in metal content. Color changes are undoubtedly present in systems of lower luminosity, although Strom & Strom suggest that these may often be small outside 2 kpc from the nucleus. The Stroms and their collaborators have also noted a tendency for the contours of constant color to be significantly flatter than the isophotes in systems with large color changes.

Figure 5. Radial gradients in the SO galaxy NGC 3115. The "mean" metallicity of the stellar population has been derived from the (U - V) photometry of Strom et al. (1977) using the calibration of Aaronson et al. (1978). Note that error bars represent photometric errors only. Radial distances are computed assuming H0 = 50 km s-1 Mpc-1.

Line-strength variations have been investigated by Spinrad et al. (1971), Spinrad, Smith & Taylor (1972), Spinrad & Stone (1975), Faber (1977), Cohen (1979a), and Burstein (1979). The general trend is for line strengths to decrease outwards, confirming the decreasing metallicity suggested by the colors, but both observation (limiting to central regions) and calibration are difficult. An exception to the general decrease may be the E0 galaxy NGC 1399 for which CN photometry (Spinrad & Stone 1975) shows little fall off, but the observations are uncertain. Cohen's investigation gives encouraging agreement of metallicity changes in NGC 3115 compared with Strom et al.'s (1977) color photometry, and she also demonstrates observational variation of lines sensitive to abundance while other line groups that are insensitive remain fairly constant. Burstein's observations on the Faber system suggest metallicity gradients 0.1 dex kpc^-1 in the six galaxies he examined.

It is not yet possible to decide the actual form of the variation of metallicity with radius in an early-type galaxy until the galaxian shapes can be disentangled, although as Figure 5 shows, a steady trend (steeper along the minor axis) is apparent. Strom et al. (1976) and Faber (1977) suggest that data are consistent with a linear log Z - logR plot although this relation must, of course, break down close to the nucleus. As mentioned in Section 4.3 above, the absolute scale of abundances is uncertain. But assuming a reasonably low metallicity for M71 and 47 Tuc, it seems likely that the central regions of the most luminous ellipticals exhibit colors characteristic of a population with probably twice or at most three times solar. For the center of the nearby elliptical M32 (M_v = - 16.3) recent discussions (Cohen 1978, 1979a, O'Connell 1980, Prichet & Campbell 1980, Heckman 1980) imply solar or somewhat less metallicity. At this absolute magnitude a lower metallicity might well be expected, although M32 does not conform well to color-magnitude relations, so that the tidal influence of M31 and evidence of comparatively young stars (O'Connell 1980) make such predictions uncertain.