|Annu. Rev. Astron. Astrophys. 1989. 27:
Copyright © 1989 by . All rights reserved
2.1. Sources of X Rays
Spiral galaxies are extended and complex X-ray sources with total luminosities in the Einstein band (~ 0.2 - 3.5 keV) of ~ 1038 erg s-1 to a few 1041 erg s-1 (Fabian 1981, Long & Van Speybroeck 1983 , Fabbiano 1984, 1986a). Observations of the Milky Way and of the Local Group galaxies (e.g. see Fabian 1981, Helfand 1984a, b) suggest that a good fraction of this X-ray emission is due to a collection of individual bright sources, such as close accreting binaries with a compact companion and supernova remnants, with luminosities ranging from ~ 1035 erg s-1 up to a few 1038 erg s-1. Stars also emit coronal X rays with luminosities of 1028 erg s-1 to 1033 erg s-1 (Vaiana et al. 1981). However, except perhaps at the lowest energies and in some starburst regions, stars do not contribute significantly to the total X-ray emission, since the X-ray to optical ratios measured in spiral galaxies (Long & Van Speybroeck 1983, Fabbiano & Trinchieri 1985) are larger than those expected from a normal stellar population (Topka et al. 1982, Helfand & Caillault 1982), and since the average X-ray spectrum of spiral galaxies appears harder than that of the stellar emission [kT > 2 keV in galaxies (Fabbiano & Trinchieri 1987); kT ~ 0.5 - 1 keV in stars (Helfand & Caillault 1982)]. Nuclear sources, either connected with star formation activity or with nonthermal Seyfert-like activity, can also be present and contribute various amounts to the X-ray emission. In this paper, however, we do not discuss "classical" Seyfert-type galaxies, where the emission is totally dominated by the nucleus (e.g. Elvis et al. 1978). Neither diffuse X-ray emission from inverse Compton scattering of the radio electrons off the optical-infrared photons nor synchrotron emission is likely to contribute significantly to the total X-ray emission (Fabbiano et al. 1982). A more likely source is diffuse thermal emission from a hot phase of the interstellar medium, heated by supernovae, and this is discussed later in this review.
2.2. X-Ray Observations of the Local Group
It is not surprising that the most detailed work on individual X-ray sources in galaxies and their identifications has been done as a result of the Einstein observations of Local Group galaxies (distance 1 Mpc). This work has been reviewed comprehensively by Helfand (1984a). Here I summarize the main results and review the more recent work not included in Helfand's paper. Table 1, adapted from Helfand (1984a), summarizes the results on the X-ray sources detected and identified with objects belonging or likely to belong to these galaxies. Although this summary offers a way to intercompare the different galaxies in the Local Group, the reader should take care in using it because of the different limiting sensitivities and completenesses of the different surveys. In particular, the Magellanic Clouds have been surveyed down to limiting luminosities for point-source detection of 1034 - 1035 erg s-1, whereas the other galaxies have limiting luminosities closer to 1037 erg s-1.
|Total LX||No. of||Binaries||Binaries||Unidentified|
|Galaxy||Type||(erg s-1)||sources||Interlopersa||SNR||(young Pop I)||(older population)||in galaxies||Referencesf|
|LMCb||Ir I||6.6 × 1038||102(52)||~ 46 (~ 19)||32(21)||7(6)||2(2)||~ 15 (~ 4)||1, 4, 5, 6|
|SMC||Ir I||6.1 × 1037||57||~ 14||~ 12||1||0||~ 30||1, 5, 7, 8, 9|
|M31||Sb||3.6 × 1039||117||~ 6||2||~ 26||~ 23GC+ ~ 60c||-||1, 5, 10, 11, 12|
|M32||E2||5.4 × 1037||1||-||-||-||1||-||2, 5, 12|
|M33||Sc||1.1 × 1039d||17||~ 3||1||~ 12||0||-||1, 5, 13, 14, 15|
|IC 1613||Ir I||-||0||-||-||-||-||-||1, 5|
|NGC 6822||Ir||1 × 1037||2||~ 1||~ 1||-||-||-||1|
|NGC 205||ES||< 9 × 1036||0||-||-||-||-||-||1|
|U Mi||E||3.2 × 1035||3||-||-||-||-||3||1|
|Maffei 1||E?||1 × 1039||3||-||-||-||-||extended||1|
|Milky Way||~ Sc||~ 3 × 1039||~ 125c||-||~ 10||~ 40||8GC+ ~ 67c||-||3, 5, 16|
|a These are either background or foreground sources not belonging to the galaxies.|
|b Numbers in parentheses are from the complete X-ray-flux-limited sample.|
|c GC are globular cluster sources.|
|d This includes the bright nuclear source.|
|e Approximate number of galactic X-ray sources with LX > 1035 erg s-1.|
|f References: (1) Markert & Donahue 1985, and references therein; (2) G. Trinchieri, private communication, 1988; (3) Fabian 1981; (4) Long et al. 1981a; (5) Helfand 1984a; (6) Cowley et al. 1984; (7) Seward & Mitchell 1981; (8) Inoue et al. 1983; (9) Bruhweiler et al. 1987; (10) Van Speybroeck et al. 1979; (11) Long & Van Speybroeck 1983; (12) Crampton et al. 1984b; (13) Long et al. 1981 (14) Markert & Rallis 1983; (15) Trinchieri et al. 1988; (16) Helfand 1985.|
In the Large Magellanic Cloud (LMC) most of the detected and identified sources are supernova remnants (see also Long & Helfand 1979, Helfand & Long 1980, Helfand 1982, 1984b, Tuohy et al. 1982, Mathewson et al. 1983, 1984, 1985, Cowley et al. 1984), of which three are Crab Nebula-like and one definitely contains a pulsar (Clark et al. 1982, Seward et al. 1984, Chanan et al. 1984). Studies of this sample of remnants has provided new insight on the supernova rate, the pulsar birthrate, and the evolution of the remnants (see Helfand 1984 a, and references therein). One fourth of the sources found in the Small Magellanic Cloud (SMC) are likely to be supernova remnants. The X-ray population of the Clouds also includes a number of massive young Population I close accreting binaries, to which belong the brightest X-ray binaries known in the pre-Einstein era (Clark et al. 1978), and a number of still unidentified fainter sources with LX ~ 1034 - 1036 erg s-1 (e.g. Long et al. 1981a, Bruhweiler et al. 1987). The latter authors speculate that these sources in the SMC may be analogous to the wide Be-neutron star binaries that are found in the Galaxy (White et al. 1982, Tuohy et al. 1988), or that they could be normal O and B stars whose X-ray luminosity might have been enhanced by a factor of ~ 104 because of the low metallicity of the SMC.
Supernova remnants instead constitute a less important component of the X-ray sources detected in the two spiral systems M31 (Figure 1) and M33, which appear to be dominated by binary X-ray source candidates. One should remember that the luminosity threshold for these galaxies is higher than in the Clouds (see above), and that at least in the edge-on M31 the soft X-ray emission of supernova remnants might be affected by interstellar extinction, so it is not possible to draw a strong conclusion from this result. However, seven remnants with LX > 1037 erg s-1 are found in the LMC, while only two identifications with supernova remnants have been reported for M31 in this luminosity range, and possibly one for M33 (Long & Van Speybroeck 1983, Long et al. 1981b). The pointlike X-ray sources detected in M33 appear to be associated with young Population I indicators, with the exception of a strong nuclear source that is discussed later (Long et al. 1981b, Markert & Rallis 1983, Trinchieri et al. 1988).
Figure 1. The circles show the positions of the X-ray sources of M31, superimposed onto an optical photograph (courtesy of L. Van Speybroeck). Notice the clustering of sources in the bulge.
One of these sources (M33 X-7) has a variable light curve that is consistent with a 1.8-day eclipsing binary period, similar to those of some massive galactic X-ray binaries (Peres et al. 1989). In M31 (Van Speybroeck et al. 1979, Van Speybroeck & Bechtold 1981, Long & Van Speybroeck 1983; see Helfand 1984a) most of the X-ray sources are instead likely to belong to an older stellar population. The luminosity of these sources is in the range of that of Galactic low-mass binaries, and some variability has been reported (Van Speybroeck & Bechtold 1981 , McKechnie et al. 1984), consistent with the hypothesis of their being powered by accretion onto a compact object. Of these X-ray sources in M31, 19 have been identified with globular clusters tabulated by Sargent et at (1977) and Battistini et al. (1980), and 4 more sources could also be associated with globular clusters (Crampton et al. 1984); ~ 30 sources are associated with the disk and the spiral arms; 19 pointlike sources plus a number of confused sources are detected in the inner bulge, within 2' (400 pc) of the nucleus; and 22 additional sources lie in the outer bulge. The average X-ray spectrum of the bulge sources is similar to those of low-mass X-ray binaries in the Galaxy (Fabbiano et al. 1987b, Makishima et al. 1989). Only a few sources have been detected in the less massive members of the Local Group, but this is not surprising if the X-ray source formation rate is somehow linked to the mass or to the stellar content of a galaxy (see later).
Table 1 also gives an approximate estimate of the X-ray source content of the Galaxy (from Helfand 1984a; see also Fabian 1981). These estimates undoubtedly suffer from the unavoidable biases of all Galactic observation - namely, the difficulties in estimating distances and membership in a given stellar population or galactic structure, and the obscuration of softer X-ray sources (among which are the supernova remnants) by the interstellar medium in the Galactic plane. Even if we keep in mind all the necessary cautions, however, a comparison of the different entries of Table 1 suggests differences in the X-ray source composition in galaxies of different morphology. In particular, it appears that there is a shift from X-ray sources belonging to the young Population I to X-ray sources belonging to an older stellar population, going from the later-type galaxies (LMC, SMC, and M33, which are dominated by the spiral arm and disk stellar component) to the earlier-type ones (the Milky Way and definitely M31, with prominent bulges, and a large number of globular clusters). Helfand (1985) shows that the percentage per unit mass of sources belonging to the spiral arm population increases with the morphological type in Local Group galaxies, going from bulge-dominated to disk-dominated galaxies; conversely, the percentage of bulge-type sources and globular cluster sources decreases with morphological type. Moreover, in M31 and the Galaxy the brightest X-ray sources belong to the older stellar population, while the opposite is observed in the Clouds and in M33. Another effect, first pointed out by Clark et al. (1978) (who ascribed it to the lower metallicity of the accreting gas), is the higher luminosity of the X-ray binary sources in the Clouds (see also Long & Van Speybroeck 1983, Crampton et al. 1984, Helfand 1984a). This effect could be responsible for the enhancement of the detection of young Population I sources in later-type galaxies. Alternatively, the reason could lie in an intrinsic higher mass for the accreting compact object: In particular, two of the four best black hole candidates are found in the LMC [LMC X-1 and LMC X-3 (Hutchings 1984; see Helfand 1985)].
Detailed comparisons have been made between the different X-ray source populations of the two most massive members of the Local Group: the Galaxy and M31 (Van Speybroeck & Bechtold 1981, Long & Van Speybroeck 1983, Battistini et al. 1982, Crampton et al. 1984). In particular, Long & Van Speybroeck (1983) remarked that M31 appears to have both a larger population of bulge sources and a larger and more luminous population of globular cluster sources than does the Galaxy. These results should be explainable in terms of the differences between these two galaxies: M31 has a larger population of globular clusters [~ 600 (Crampton et al. 1985) versus 180 in the Galaxy (Harris & Racine 1979)]. Crampton et al. (1984) remark that the fractions of X-ray-emitting globular clusters in the two galaxies are similar, and that the higher X-ray luminosities seen in the M31 globular clusters may simply reflect the fact that M31 contains more globular clusters and that the X-ray-bright ones, which are also the optically brightest and more condensed, lie on the high-luminosity tail of an overall distribution that is similar to the one in the Galaxy. Other authors (Long & Van Speybroeck 1983, Huchra et al. 1982, Battistini et al. 1982) have debated whether these M31 globular clusters are peculiar, and whether a metallicity-X-ray luminosity effect is possible. The bulge of M31 has a central concentration of ~ 20 sources within ~ 400 pc that appear more luminous on average than a similar number of sources in the outer bulge (Van Speybroeck et al. 1979). If these sources are distributed spherically, they would constitute a structure unlike any seen in the Galaxy, where only a few sources are seen in a similar volume (Long & Van Speybroeck 1983). These authors, however, suggest the alternative that the bulge X-ray sources in both galaxies might form similar, flattened barlike structures (see also Van Speybroeck & Bechtold 1981). The coincidence of this enhanced source distribution in M31 with a reported hole in the distribution of optical novae led Vader et al. (1982) to suggest that these sources could be the result of the evolution of a dead cataclysmic variable population that had long since undergone its nova phase. More recent observations, however, have dispelled the notion of a nova hole in M31 (Ciardullo et al. 1987). These authors suggest that the enhanced X-ray source content and the high specific nova rate that they find in the bulge of M31 could both be connected with the disruption of globular clusters in the bulge, which would then release the binaries that had formed in their cores. Although controversial (van Paradijs & Lewin 1985, Vader et al. 1982), a similar mechanism has been suggested by Grindlay (1984, 1985) for the formation of low-mass binaries in the Galaxy.
The results discussed so far are concerned with the detectable individual source content of the Local Group galaxies. However, below the single source threshold, it is still possible to detect the integrated emission of the fainter, individually undetectable sources and of any truly diffuse component, such as a hot phase of the interstellar medium. This type of emission has been reported in the Galaxy, where it accounts for < 10% of the integrated emission of the resolved sources in the 2-10 keV range [the "Galactic ridge" (e.g. Worrall et al. 1982, Warwick et al. 1985, Koyama et al. 1986)]. A soft diffuse X-ray background (E < 0.284 keV), which is likely to be of Galactic origin, has also been observed (e.g. McCammon et al. 1983, Marshall & Clark 1984). A soft (0.25 keV) X-ray survey has revealed the presence of diffuse emission in the LMC (Singh et al. 1987). This emission has also been seen in the Einstein survey and has a total X-ray luminosity of 3 × 1038 erg s-1 (D. Helfand, private communication, 1988). The only other galaxy of the Local Group for which diffuse emission has been reported to date is M33 (Trinchieri et al. 1988), where the diffuse component accounts for ~ 1/3 of the total nonnuclear emission (which is not surprising given the much higher source detection threshold of these observations) and can be separated into a spectrally hard (kT > 2 keV) and a soft (kT < 1 keV) component. The hard component could be due to the integrated contribution of several lower luminosity compact accreting systems and young supernova remnants; the soft component is most likely due to the integrated emission of stellar coronae, with a possible contribution from a hot phase of the interstellar medium.
2.3. Detailed Observations of Bright Spiral Galaxies
Only a few very bright X-ray sources can be detected in the Einstein images of more distant galaxies, which typically appear as extended X-ray emission regions (e.g. Fabbiano & Trinchieri 1987). However, there is reason to believe that most of the X-ray emission of these galaxies is due to sources akin to those detected in the Local Group. A comparison of the fraction of the X-ray emission resolved in individual bright sources versus that which appears diffuse is consistent with the bulk of the emission originating from individual sources below threshold: More distant galaxies, with higher point-source thresholds, have a relatively larger "diffuse" emission than do less distant galaxies (Fabbiano 1988a). Moreover, the X-ray spectra of these galaxies, although ill defined, are consistent with the hard spectra expected from binary X-ray sources (Fabbiano & Trinchieri 1987, Trinchieri et al. 1988, Fabbiano 1988b).
Only a very few of the sources detected in spiral galaxies can be chance superpositions of background or foreground objects, given the statistics of the serendipitous Einstein source detections (Gioia et al. 1984; see Fabbiano & Trinchieri 1987). Most of these sources are therefore in the galaxies, often in the spiral arms, and their X-ray luminosities can be very high indeed. They are typically well above the Eddington limit for accretion onto a 1 - M compact object, which is ~ 1.3 × 1038 erg s-1, and can be as bright as a few × 1039 erg s-1. These sources have been detected in a number of spirals, including M83, M51, NGC 253, NGC 4631, NGC 6946, M101, and M81 (Long & Van Speybroeck 1983, Fabbiano & Trinchieri 1984, 1987, Trinchieri et al. 1985, Palumbo et al. 1985, Fabbiano 1988a), and eight bright sources have also been reported in the central starburst region of M82 (Watson et al. 1984). We exclude bright nuclear regions from the present discussion and concentrate instead on the more puzzling galactic sources. One of the most extreme cases is that of a source reported in M100 with an X-ray luminosity of ~ 1040 erg s-1 (Palumbo et al. 1981). However, a very recent analysis of the same data by this reviewer, and of a subsequent longer exposure with the same high resolution, suggests that this source might be spurious: It is not detected in the longer exposure, and it is only marginally detected as an extended feature in the first observation. A similarly luminous source, however, is detected in NGC 4631 (Fabbiano & Trinchieri 1987). About 36 sources more luminous than ~ 2 × 1038 erg s-1 have been reported to date, 16 of which are more luminous than 1039 erg s-1.
What are these sources? In one case the answer is simple: One of these sources is SN 1980K, which was detected in NGC 6946 ~ 35 days after maximum light (Canizares et al. 1982b). The variability reported for some bright sources in M101 suggests pointlike objects [possibly bright accretion binaries (Long & Van Speybroeck 1983)]. If these sources are mostly complex emission regions, we would be faced with several bright sources (e.g. 1037 erg s-1) in volumes with typical dimensions of a few hundred parsecs to a kiloparsec (Fabbiano & Trinchieri 1987). These sources are not typically in bulges, where such crowding could be expected [e.g. M31 (see earlier discussion)]. If these sources are truly single objects, they could indicate the presence of massive black holes in these galaxies. It is possible, however, that the distances of some galaxies might have been overestimated, which would make these sources appear more luminous than they are in reality. For instance, estimates of the distance of NGC 4631 range from 12 Mpc (Sandage & Tammann 1981) to 3 Mpc (Duric et al. 1982); using the lower estimate, the luminosity of the source reported by Fabbiano & Trinchieri (1987) as ~ 1.4 × 1040 erg s-1 would become ~ 9 × 1038 erg s-1. However, this still exceeds the Eddington luminosity of an accreting neutron star.
The observations of the Local Group suggest differences in the X-ray sources in galaxies of different morphological type. However, differences also seem to occur in galaxies of similar morphology. An example is given by a comparison of the X-ray properties of the two Sb galaxies M31 and M81 (Fabbiano 1988a). The latter shows a number of individual X-ray sources, all more luminous than the most luminous sources of M31, and it is highly unlikely that this is due to an overestimate of the distance of M81. With the present data it is impossible to discriminate between an intrinsically more luminous X-ray source population in M81 or a more numerous population with the same luminosity function as that of M31. An overall comparison between these two galaxies also shows that M81 is overluminous in both X-ray and radio continuum emission, and to a lesser extent in far-infrared emission, relative to their optical luminosities. This suggests differences in the star formation history of these two galaxies, which resulted in a more efficient production of X-ray, cosmic ray, and far-infrared sources in M81. All these results are very tantalizing and show the importance of future sensitive X-ray observations in furthering our understanding of the global properties of spiral galaxies and of the detailed physical properties of their stellar remnants.
The presence of bulge (and globular cluster) X-ray sources on the one hand, and of often very bright X-ray sources associated with the spiral arms and H II regions on the other, is immediately demonstrated by the X-ray images of the Local Group and of other relatively nearby galaxies. The close resemblance between the radial profile of the X-ray surface brightness of a few face-on spirals and that of the optical light of their exponential disk suggests the presence of a third component of the X-ray emission, one associated with the stellar population of the disks (see Fabbiano 1986a). This effect was first seen in M83 (Figure 2; Trinchieri et al. 1985) and then in M51 (Palumbo et al. 1985) and possibly NGC 6946 (Fabbiano & Trinchieri 1987) and M81 [Fabbiano 1988a; see also M33 (Trinchieri et al. 1988)]. In particular, the X-ray profile in M51 is significantly different from the H profile (where the arms are very prominent) and follows the exponential disk distribution, as do the radio continuum and the CO profiles (Figure 3). These observations open interesting possibilities for our understanding of the origin of low-mass X-ray binaries. The nature of these sources, which constitute ~ 60% of the Galactic sources and which are also called "galactic bulge sources" and "Population II" sources in the X-ray literature (e.g. van den Heuvel 1980, Helfand 1984a), is one of the open problems of "classical" X-ray astronomy. Suggestions on their origin have included capture of neutron stars in the Galactic bulge (van den Heuvel 1980), remnants of disrupted globular clusters (Grindlay 1984, 1985), or the evolved remnants of low-mass binary systems in the Galactic disk (e.g. Gursky 1976, Rappaport et al. 1982, Nomoto 1984, van den Heuvel 1984, and references therein). The observations of external galaxies morphologically similar to the Milky Way suggest that at least a good fraction of these sources may originate from the evolution of binary systems belonging to the disk stellar population, rather than from dynamical evolution (Fabbiano 1985a, 1986a, Trinchieri et al. 1985). However (Fabbiano 1985a, 1986a), there is a relative excess of X-ray emission over the disk emission in the innermost disk region seen in both M83 and M51 (Trinchieri et al. 1985, Palumbo et al. 1985), which, if we scale by distance and dimensions, roughly coincides with what has been called the X-ray Galactic bulge. This excess emission could either (a) indicate an intrinsically brighter population of X-ray sources, analogous to the bright "bulge" Galactic sources, (b) point to an enhanced past episode of star formation in the inner disk, in contrast perhaps with steady star formation in the disk as a whole (e.g. Vader et al. 1982), or (c) suggest the presence of an additional component of the X-ray binary source population, which could be related to the disruption of globular clusters (Grindlay 1984, 1985).
However, there is one case in which the X-ray surface brightness profile clearly does not follow the optical light but instead follows quite well the radio continuum profile: NGC 253 (Fabbiano 1988b). Enhanced radio and X-ray emission is observed in the inner disk of NGC 253, and I discuss in the next section the implications of this and other observations for a connection between X-ray sources and radio continuum emission.
Figure 3. Radial profile of the X-ray surface brightness of M51 [the points (Palumbo et al. 1985)], together with radial profiles at other wavelengths from Scoville & Young (1983) and Klein et al. (1984).
Another source of X-ray emission, besides the contributions of individual sources, has been predicted in spiral galaxies. This is the thermal emission of the interstellar medium, heated by supernovae, which release ~ 1042 erg s-1 in the galaxy. It has been suggested that hot gaseous coronae, or galactic fountains, could be produced and should be visible in soft X-rays in the Einstein range (e.g. Spitzer 1956, Cox & Smith 1974, Bregman 1980a, b, Corbelli & Salpeter 1988). There is evidence of soft thermal diffuse emission both in the Galactic plane and in the LMC (e.g. McCammon et al. 1983, Marshall & Clark 1984, Singh et al. 1987), and perhaps in M33 (Trinchieri et al. 1988). A search for this type of emission in more distant galaxies has been made by Bregman & Glassgold (1982) and McCammon & Sanders (1984). The former searched two edge-on galaxies, NGC 3628 and NGC 4244, for coronal emission and set limits of 1039 and 2 × 1038 erg s-1, respectively, on the X-ray luminosity of a gaseous halo; the latter instead analyzed a large face-on galaxy, M101, and concluded that the limits on the diffuse X-ray emission require that the temperature of any hot gas that is radiating 10% of the average supernova power be less than 1057 K, and that hot bubbles occupy at most 25% of the region between 10 and 20 kpc from the galactic center. Subsequently, Cox & McCammon (1986) have used these measurements to constrain the density of the interstellar medium of M101 and the characteristics of the population of supernova remnants evolving in the disk. The lack of intense diffuse soft X-ray emission could imply that most of the supernova energy is radiated in the unobservable far-ultraviolet (Cox 1983). The only reported instance of this type of soft X-ray emission in a spiral galaxy is in the edge-on NGC 4631, where this component could have an X-ray luminosity of 5 × 1039 erg s-1, which represents ~ 13% of the total emission in the Einstein band (Fabbiano & Trinchieri 1987).
Shostak et al. (1982) interpret an arc-shaped structure aligned with a ridge of radio continuum emission in NGC 1961 as evidence of a shock-heated interstellar medium that is being stripped by a hot intergalactic medium. However, without clear spectral confirmation, this interpretation is not unique: Enhanced star formation resulting from the interaction could also be responsible for the excess emission (e.g. Fabbiano et al. 1982; see next section). Similarly, a plume of extended emission emanating from NGC 4438, a spiral galaxy in the Virgo cluster, is reported by Kotanyi et al. (1983) as evidence of ram pressure sweeping of the interstellar gas from the disk of the galaxy. This plume, however, seems to emanate from the bright nucleus of this galaxy and could perhaps be similar to the gaseous plumes detected near the starburst nuclei of NGC 253 (Fabbiano & Trinchieri 1984) and M82 (Watson et al. 1984). I discuss these later in this review.
2.4. Average Sample Properties and Correlations With Other Wavebands
Although the Einstein images of some galaxies allow us to study them in detail and so reach some understanding of their X-ray emission components, most of the data are not of this high quality. Some 50 spiral galaxies were surveyed as part of the original Einstein observing program, and the results of these observations have been used to study the average properties of the sample and to explore correlations with the emission at other wavebands. This sample is not statistically complete; however, it can be regarded as representative of "normal" spiral galaxies of different morphologies and absolute magnitudes (Fabbiano & Trinchieri 1985). These galaxies have X-ray luminosities ranging between 1038 erg s-1 and 1041 erg s-1, which are linearly correlated with their emission in the optical B band. This correlation is similarly tight for early-type bulge-dominated spirals and for late-type disk/arm-dominated galaxies: For all of them the ratios of monochromatic (2 keV) X-ray to optical (B) flux densities cluster around 10-7. This result suggests that the X-ray emission is mostly due to sources constituting a constant fraction of the stellar population, in agreement with the conclusion of the detailed X-ray observations discussed above, which show that the X-ray-emitting population is likely to be dominated by binary X-ray sources (Long & Van Speybroeck 1983, Fabbiano et al. 1984b, Fabbiano 1984, Fabbiano & Trinchieri 1985). Even Sa galaxies follow this correlation, which suggests that their X-ray emission is due to the same type of sources responsible for the general emission of spiral galaxies and does not require an additional large gaseous emission component, such as is seen in bright elliptical galaxies (see later in this review).
Correlations have also been found between the X-ray emission and other variables, including the radio continuum, the near-infrared H band; and the far-infrared IRAS emission (Fabbiano & Trinchieri 1985, Fabbiano et al. 1988). These correlations are all very tight in late-type galaxies. In the subsample of bulge-dominated galaxies, however, the correlations between radio continuum and/or far-infrared luminosities with any of the other emission bands show a considerable amount of scatter and sometimes also a shift in zero point. In particular, for a given X-ray luminosity, there is a clear deficiency of radio continuum emission when bulge-dominated spirals are compared with disk/arm-dominated galaxies, whereas no differences are seen in the X-ray and optical (B) correlations. Since most of the optical and near-infrared emission of early-type spirals is dominated by the emission of the bulge (Kent 1985), these differences suggest that the radio continuum and the far-infrared are mainly related to the stellar population of the disk, whereas the X rays originate in both the disk and the bulge components. The latter conclusion is in agreement with the results of the X-ray observations of M31, which were discussed earlier.
Fabbiano & Trinchieri (1985; see also Fabbiano et al. 1984b) find that the correlation between X-ray and B-band emission is stronger than those between X-ray and either H-band or the B - H color, which are both indicators of older stellar content and/or galaxy mass (e.g. Whitmore 1984). This result suggests that the X-ray sources belong predominantly to the blue-emitting stellar Population I. A link with the youngest Population I, to which the massive binary X-ray sources belong, is suggested in particular by a comparison between spiral galaxies with "normal" average colors and galaxies with blue peculiar colors, indicative of extensive and recent star formation activity (see Larson & Tinsley 1978). For a given optical luminosity, the X-ray luminosity is enhanced in relatively bluer galaxies (Fabbiano et al. 1982, 1984b, Fabbiano & Panagia 1983). These results also have implications for the nature and evolution of the low-mass X-ray binaries, which represent a very large component of the X-ray emission of the Milky Way: In particular, they suggest that most of these sources are likely to belong to the old Population I and possibly originate from the evolution of native binary systems, rather than having a dynamical origin. These conclusions are supported by the presence of an exponential disk in the X-ray emission of face-on spiral galaxies, as discussed earlier (see Fabbiano 1985a). Of all the correlations studied in late-type spirals, only two - the X-ray/B and the radio/far-infrared - imply strict proportionality between the emission at the two different wavelengths; the others follow power laws with exponents significantly different from unity. Trying to understand which correlations are intrinsically stronger, and therefore more directly connected to the underlying phenomena we wish to discover, and why not all the correlations scale simply with luminosity can give us new insight on the stellar components and evolution of spiral galaxies. Implications for the initial mass function (IMF), the average dust content of the disks, and the presence of compact star-forming regions are reviewed and discussed by Fabbiano et al. (1988). In particular, these correlations suggest the preferential occurrence of obscured starburst components in the more luminous galaxies.
An interesting possibility, raised both by single-galaxy studies and by statistical comparisons, is that of a connection between X-ray and radio continuum emission in spiral and irregular galaxies (Fabbiano et al. 1984b, Fabbiano & Trinchieri 1985, 1987, Palumbo et al. 1985, Fabbiano 1988b). This connection could be through recent star formation, but it is possible that there could be a more direct link between X-ray sources and cosmic-ray production (Fabbiano & Trinchieri 1985, Fabbiano et al. 1988). In particular, there is strong evidence of particle acceleration in X-ray binaries: Relativistic jets have been detected in the massive X-ray binary SS 433 and have been suggested to explain the radio morphology of the low-mass binary Sco X-1 (Geldzahler et al. 1981, Hjellming & Johnston 1981, Watson et al. 1983); and gamma-ray emission has been reported from the X-ray binaries Cyg X-3, Her X-1, and Vela X-1, suggesting intense cosmic-ray production (Samorski & Stamm 1983, Dowthwaite et al. 1984, Protheroe et al. 1984). Although strong, the X-ray/radio correlation has a power law exponent different from unity. Fabbiano & Trinchieri (1985), assuming proportionality between the sources of cosmic rays and the X-ray-emitting population, suggested that this result could imply a luminosity dependence of the intensity of the magnetic field of spiral galaxies. However, the proportionality between radio and far-infrared emission suggests that the sources of cosmic-ray electrons are a constant fraction of the stars responsible for heating the dust to far-infrared temperatures, and therefore that the nonlinear X-ray/radio correlation could be the result of these obscured regions contributing relatively less to the X-ray emission than to the radio and far-infrared emission (Fabbiano et al. 1988).
Although limited to relatively small samples, these comparisons have shown the potential of a multifrequency approach to the study of global galaxy properties. Their extension to larger and better defined samples, through systematic searches of the Einstein data bank (e.g. D. Burstein et al., in preparation, 1989) and through future X-ray observations, will be needed to confirm some of the present results and to gain a more general understanding of the structure and evolution of spiral galaxies.