A Model of the Vela Supernova Remnant

Home , Vela Pulsar, Vela Supernova Remnant

A MODEL OF THE VELA SUPERNOVA REMNANT

VA S I L I I G VA RA M A D Z E ? Abastumani Astrophysical Observatory, Georgian Academy of Sciences, A.Kazbegi ave. 2-a, Tbilisi, 380060, Georgia; Sternberg State Astronomical Institute, Universitetskij Prospect 13, Moscow, 119899, Russia; Abdus Salam International Centre for Theoretical Physics, Strada Costiera 11, P.O. Box 586, 34100 Trieste, Italy

Abstract. A model of the Vela supernova remnant (SNR) based on a cavity explosion of a supernova (SN) star is proposed. It is suggested that the general structure of the remnant is determined by the interaction of the SN blast wave with a massive shell created by the SN progenitor (15−20 M )star. A possible origin of the nebula of hard X-ray emission detected around the Vela pulsar is discussed.

1. Introduction

The Vela SNR is one of the best studied SNRs, primarily because of its large angular extent and high brightness, and to no small degree due to its association with the well-known Vela pulsar. High-resolution observations made at the last ten years revealed many new details of the Vela SNR and revived interest to this remnant. However in despite of the considerable progress in observations there is still no a model proposed to explain the whole complex of the observational data, though many attempts were made to explain one or another peculiarity of the Vela SNR. In this paper‡ we try to make up this deﬁciency. In Section 2 we review the main observational data. Section 3 contains our model of the Vela SNR.

2. Observational Data

The angular size of the main body of the Vela SNR is about 7◦ in radio (Duncan et al., 1996, hereafter D96), optical (Seward 1990, hereafter S90) and X-ray (S90; Aschenbach et al., 1995, hereafter A95) ranges. In all these spectral ranges the Vela SNR shows a distinct asymmetry along the line perpendicular to the Galactic plane. The northeast half of the remnant (faced towards the Galactic plane) has a well

? Address for correspondence: Krasin str. 19, ap. 81, Moscow, 123056, Russia; E-mail: [email protected] ‡This paper summarizes the results published in Gvaramadze 1998a,b, 1999.

Astrophysics and Space Science 274: 195–203, 2000. © 2000 Kluwer Academic Publishers. Printed in the Netherlands. 196 VASILII GVARAMADZE determined circular boundary of radius of ' 3.5◦, whereas the opposite half is very disordered, with some (X-ray) structures (A95) stretched up to ' 6.5◦ from the center (associated with the Vela pulsar position). There also exist several features protruding far outside the northeast rim of the remnant (A95; D96; Gvaramadze, 1998a). At a distance to the Vela SNR of 500 pc (in what follows, we adopt just this value; the contradictory distance estimates for this remnant will be discussed elsewhere) 1◦ corresponds to about 9 pc. The only reliable age estimate of the Vela SNR comes from the estimate of the spin-down age of the Vela pulsar, which is ' (2 − 3) · 104 years (Lyne et al., 1996).

2.1. THE RADIO DATA

Early radio observations (e.g. Milne, 1968) showed that the radio emission of the Vela SNR comes from three radio components, which are known as Vela X, Y, and Z. The maximum of emission of the brightest radio component, Vela X, is displaced for ' 450 to the south of the pulsar. High-resolution observations of the Vela SNR showed that it is composed of many filamentary and looplike structures (e.g. D96; Bock et al., 1998a, hereafter B98), and that the northeast half of the remnant is enclosed by an arc of polarized emission which closely matchs its boundary (D96). The Vela X also consists of many strongly polarized filaments with a mean width of few arcminutes (Milne, 1995, hereafter M95; Frail et al., 1997, hereafter F97; B98). Some authors (e.g. Weiler and Panagia, 1980, hereafter W80; Dwarakanath, 1991; B98) believe that the spectrum of Vela X is much flatter than that of other parts of the Vela SNR, while M95 (see also Milne and Manchester, 1986) showed that the spectra of all radio components are nearly the same with a value of the spectral index between –0.4 and –0.8. Based on the high percentage of polarization and on the possible flat spectrum of the Vela X, W80 suggested that this radio source is a plerion, i.e. a pulsar-powered nebula. However Milne and Manchester (1986) pointed out that the large displacement of the Vela pulsar from the center of Vela X could not be explained by the proper motion of the pulsar, since the latter is too small and not in the proper direction (e.g. Nasuti et al., 1997), and concluded that the Vela X is a part of the Vela SNR’s shell.

2.2. THE OPTICAL DATA

The [OIII] λ5010 filter photographs of the Vela SNR by Parker et al. (1979, hereafter P79) combined in a mosaic and presented in Figure 1 show a very intricate design composed of a multitude of arclike and looplike filaments. It is generally accepted that the origin of optical filaments is connected with the interaction between adiabatic (Sedov-Taylor) SN shocks and interstellar clouds (e.g. Bychkov and Pikel’ner, 1975), or with projection effects in radiative shocks, whose fronts are rippled due to the refraction and reflection by density inhomogeneities in the ambient medium (e.g. Pikel’ner, 1954). The mean expansion velocity of the Vela SNR, inferred from ultraviolet absorption observations, is about 100 km s−1 (e.g. VELA SUPERNOVA REMNANT 197

Jenkins et al., 1984, hereafter J84), whereas some portions of the remnant’s shell expand with much higher speeds (e.g. J84; Danks and Sembach, 1995). The same observations revealed that the line of sight component of the shell velocity does not gradually decrease towards the edges of the remnant (as it should be according to the standard Sedov-Taylor model), but rather shows a chaotic behaviour. J84 suggested that the ‘unusual’ velocity ﬁeld inherent to the Vela remnant’s shell indicates that some process induces transverse motions in the shell.

2.3. THE X-RAY DATA

The soft X-ray emission of the Vela SNR is dominated in the northwest and southeast quadrants of the remnant, and shows some limb-brightening on the northeast edge (Kahn et al., 1985; hereafter K85). K85 found also that the X-ray emission is distributed in a patchy way, with significant spectral and intensity variations on angular scales ranging from several arcminutes to a few degrees. Another interesting finding of K85 is that the soft X-ray structures generally correspond with optical filaments. The characteristic X-ray temperature was estimated as ' 0.2keV.Itis generally believed that the soft X-ray emission of the Vela SNR is produced by the hot plasma behind the adiabatic (Sedov-Taylor) SN shock wave (e.g. K85). In this case, the X-ray temperature of 0.2 keV corresponds to the shock velocity of ' 400 km s−1. Subsequent observations of the Vela SNR reveal (S90; A95) that the X-ray extension of this remnant is much larger than that which was found by K85. Particularly, S90 found a long, faint arc in the western part of the SNR, which was interpreted as a rim of the SNR’s shell. A95 discovered a number of radial X-ray structures protruding far beyond the main body of the Vela SNR and interpreted them as bow shocks produced by fragments of the exploded SN star. Willmore et al. (1992; hereafter W92) found an elongated structure of hard X-ray emission (in the 2.5–25 keV band) in the center of the remnant. This structure stretches for about 1◦ on either side of the pulsar in the northeast-southwest direction. W92 suggested that this structure is a synchrotron nebula. Then Markwardt and Ögelman (1995, hereafter M95) discovered a filamentary structure of irregular shape extended from the Vela pulsar position to the center of the Vela X. This feature appears at energies above 0.7 keV (M95) and is visible to energies of at least 7 keV (Markwardt and Ögelman, 1997, hereafter M97). M95 (see also M97; F97) interpreted the X-ray filament as a one-sided jet emanating from the Vela pulsar and supplying the energy from the pulsar into the Vela X. Therethrough they supported the proposal of W80 that the VelaX is a plerion. An important result of X-ray studies of the Vela SNR is the discovery of a net of arcs and loops of intense soft X-rays which covers the whole remnant (Aschenbach, 1997). And finally, an analysis of X-ray images of the Vela SNR presented by Aschenbach (1998) shows that some features visible at radio, optical and/or soft X-ray wavelengths (e.g the X-ray protrusion labelled by A95 as D/D’ and the U-type optical filament on the south edge of the remnant (see Figure 1)) have obvious hard (≥ 1.3 keV) X-ray counterparts. 198 VASILII GVARAMADZE

3. Model of the Vela SNR

The main point of our model of the Vela SNR is the statement that the general shape of the Vela SNR might be explained as a result of the interaction between the SN ejecta/shock and the pre-existing wind-driven shell created by the SN progenitor star in an interstellar medium with a density gradient perpendicular to the Galactic plane; the existense of a large-scale region of enhanced density to the northeast of the Vela SNR (i.e. towards the Galactic plane) follows from observations (e.g. Dame et al., 1987). Though the large-scale density enhancenment strongly affects the appearence of the Vela SNR (see below), we can neglect its existense in the qualitative analysis of the wind-driven bubble evolution. Preliminary analysis suggests that the Vela SNR is a result of type II supernova explosion, and that a progenitor star was a 15 − 20 M star with mechanical 34 −1 luminosity in the range Lw = (0.3 − 3) · 10 ergs s . The ionizing radiation of the progenitor star creates an H II region, the inner, homogenized part of which gradually expands due to the continuous photoevaporation of small-scale density inhomogeneities in stellar environs (McKee et al., 1984). If Lw is much smaller ∗ ' 34 2 1/3 −1 than some characteristic wind luminosity, Lw 10 (S46/n) egrs s ,where 46 −1 S46 is the stellar ionizing ﬂux in units of 10 photons s and n is the mean density the ambient medium would have if were homogenized, the stellar wind ﬂows through a homogeneous medium with the number density ' n and temperature T = 8000 K. E.g. for a B0.5 V star with S46 ' 7 (Osterbrock, 1989) and 34 −1 −3 Lw ' 10 egrs s ,andforn=0.1cm (e.g. Wallerstein and Silk, 1971) one ∗ has that Lw Lw, therefore we can use the self-similar solutions by Weaver et al. (1977) to describe the early evolution of a wind-driven bubble. Initially the bubble is surrounded by a thin, dense shell of swept-up interstellar = 1/5 −1/5 3/5 ' 3/5 = 34 gas of radius Rb(t) 11 L34 n t6 pc 17 t6 pc, where L34 Lw/(10 −1 6 ergs s ), t6 = t/(10 years), but eventually the gas pressure in the bubble becomes comparable to that of the ambient medium, and the bubble stalls, while = 1/2 −1/2 ' the shell disappears. This happens at the moment ts,6 0.3 L34 n 1, i.e. ts is more than ten times smaller than the time spent by the SN progenitor star as a core hydrogen burning star, which is ' 1.5 · 107 years (see e.g. Vanbeveren et al., 1998, hereafter V98). The radius of the stalled bubble is Rb(ts) = Rs = 1/2 −1/2 ' 5.5 L34 n pc 17 pc. Since the star continues to supply the energy in the 1/3 bubble, the radius of the bubble continues to grow, Rb(t > ts) = Rs(t/ts) , until the radiative losses in the bubble interior becomes comparable to Lw. It can be 7 shown that this happens for t>trad,wheretrad ' 10 years. Then the bubble recedes to some stable radius Rr, at which radiative losses exactly balance Lw = 6/13 −7/13 ' (D’Ercole, 1992, hereafter D92): Rr 2.2L34 n 8 pc. The great bulk of radiative losses is connected with a thin spherical layer close to Rr, where nearly the whole mass of the bubble gas is concentrated. The main body of the bubble is occupied by a hot, tenuous gas with the number density ' 10−2 n (D92) and temperature ' 100 T . VELA SUPERNOVA REMNANT 199

Figure 1. Mosaic of [O III] λ5010 ﬁlter photographs of the Vela SNR (P79). Position of the Vela pulsar is indicated by a cross. North is up, east at left. The characteristic size of circular ﬁlaments ◦ (‘blisters’) in the central part of the remnant is about 1 (' 9pc).

Before a 15 M star exploded as a supernova it becomes for a short time, 5 tRSG ' 7.5 · 10 years, a red supergiant (e.g. V98). The H II region outside the bubble cools off and recombines, while the radiative losses in the bubble interior are negligible on time-scales of tRSG. As a result, the bubble supersonically reex- pands in the external cold medium and creates a new dense shell (see D92; cf. Shull et al., 1985). Let us assume that before the supernova exploded, the shell expands up to the radius of Rsh ' 30 pc, i.e. nearly to the current radius of the Vela SNR. = 3 − 3 ' Then the mass of the shell is Msh 4π Rsh Rr mHn/3 250 M . It is known that the mass of a shell is a decisive factor, which determines the evolution of a SN shock (e.g. Franco et al., 1991). If the mass of a shell is smaller than ' 50Mej,whereMej is the mass of a SN ejecta, the SN shock overruns the shell and continues to evolve adiabatically (as a Sedov-Taylor blast wave). For more massive ones, the SN shock merges with the shell, and the reaccelerated shell evolves into a momentum-conserving stage. We believe that just this situation takes place in the Vela SNR. Indeed, a 15 M progenitor star ends its evolution as a ' 5 M red supergiant (e.g. V98), therefore the mass of the supernova ejecta is ≤ 4 M ,i.e.Msh > 50 Mej. The impact of the SN ejecta/shock with the shell causes the development of Rayleigh-Taylor instability. One can show (Gvaramadze, 1999, hereafter G99) that the characteristic scale of deformations of the SNR’s shell is ' 2pc.Note,however, that this scale depends on the mass (as well as the (regular) magnetic field) distribution over the SNR’s shell (which is inhomogeneous due to the large-scale density enhancement to the northeast of the remnant), and should undergo considerable changes from place to place. We see from Figure 1 that the instability initiates small-scale deformations in the northeast (more massive) half of the shell, 200 VASILII GVARAMADZE thereby does not significantly disturb its initially circular shape. The situation is more dramatic in the southwest (less massive) half of the SNR. The dynamical force of the SN ejecta strongly deforms the shell and even disrupts it in places (see Figure 1). A hot gas escapes through the gaps in the SNR’s shell and forms radial outflows partially bounded by filamentary structures (see Gvaramadze, 1998a), where an overlay of contours of soft X-ray emission (A95) with a mosaic of optical photographs (P79) is presented). The large-scale domelike deformations of the remnant’s shell, if looked at sideways, appear as U-type or arclike filaments; numerous examples of such structures exist along the southeast, south and southwest edges of the Vela SNR (we suggest that some of X-ray protrusions discovered by A95 (namely A,B,C, and D/D’) could be interpreted as the shell deformations, while the rest two protrusions (E and F) are connected with outflows of a hot gas escaping from the interior of the Vela SNR through the breaks in the shell). The same deformations but viewed head-on appear as circular filaments; two prominent circular filaments are visible in the central part of the Vela SNR (see Figure 1). We interpret these partially intersecting filaments as outlines of two ‘blisters’ on the surface of the SNR, and suggest that these ‘blisters’ interact with each other due to the lateral expansion. The ‘blisters’ meet each other in the area shifted for about 450 to the south-southwest from the pulsar (note that the pulsar is situated just on the (northern) edge of the eastern ‘blister’). It is of interest that this area, outlined by distinctly visible optical filaments, coincides both with the ‘head’ of the X-ray ‘jet’ discovered by M95 and with the center of the radio source Vela X. Moreover, the optical filament on the east edge of the area traces about 60 percent of the X-ray ‘jet’ (see Figure 2 and Figure 3 of G99). Another interesting fact is that most of radio filaments of VelaX are lying within the region bounded by two circular filaments (the outlines of the ‘blisters’) and showing a fairly good correlation with optical ones (Gvaramadze, 1998b). These facts suggest that the filamentary structures visible throughout the Vela SNR in radio, optical and X-ray ranges have a common nature, and that their origin is connected with projection effects in the Rayleigh-Taylor unstable shell. In our model, the optical emission is expected to come from the outer layers of the shell, where the transmitted SN shock slows to become radiative, while the soft X-ray emission represents the inner layers of the shell heated by the SN shock up to X-ray temperatures (cf. Shull et al., 1985). This explains the general correlation of optical filaments and soft X- ray structures. However, it should not be one-to-one correspondence of optical and X-ray filaments (cf. B98). For example, it is quite possible that the less massive portions of the SNR’s shell were completely overtaken by the transmitted SN shock, and therefore are too hot to emit in the optical range. Apparently this is the reason why we do not see an optical counterpart to the long, faint arc of soft X-rays found by S90 in the west (less massive) half of the Vela SNR. Our model naturally explains the spectral and intensity variations of the soft X- ray emission (such variations arise when our line of sight crosses the deformations of the shell) as well as the fact that the soft X-ray emission of the Vela SNR does not VELA SUPERNOVA REMNANT 201 show limb-brightening except on the northeast edge. Our explanation is as follows. The shell deformations result in the increase of the effective ‘thickness’ of the shell. The ‘thickening’ is however not uniform around the shell. The shell deformations are less pronounced on the northeast edge of the remnant, and we see the limb- brightening in this direction. The farther from the Galactic plane, the larger the scale of deformations, and the larger the ‘thickness’ of the shell. This manifests in two wide regions of soft X-ray emission in the northwest and southeast directions. It follows also from our model that the main volume of the hot interior of the Vela SNR does not significantly contribute to the overall X-ray emission of the Vela SNR (see however the last paragraph of this section). Therefore we conclude that the hard X-ray emission (correlated with soft X-ray and optical filaments) might be connected with the innermost layers of the deformed SNR’s shell. The growth of the deformations of the shell is accompanied by the mass redistribution over their surfaces. The matter streams down from the tops of deformations towards the periphery and accumulates there. The origin of dense sheets of matter (the so-called spikes) is the result of this process. The most dense spikes arise in the regions of collective interaction of several deformations. We suggest that the hard X-ray emission is thermal and originate in the hot medium evaporated from the shell, and, in particular, from the dense spikes. We also suggest that hard X-ray emission should appear as bright spots in places where spikes penetrate deeply enough in the hot interior of the remnant to be effectively evaporated and heated to high temperatures, and where the line of sight extent of the evaporated gas exceeds that of the gas evaporated from the adjacent parts of the SNR’s shell. Some regular structures of relatively hard X-ray emission could arise in regions of collective interaction of deformations. The length and radial extent of these structures should be of the same order of magnitude as the scale of deformations, whereas the width is about the same as the thickness of the shell. It should be mentioned however that hard X-ray structures should be less regular than soft X-ray structures since the regions occupied by the hot evaporated gas are wider than spikes from which the gas was evaporated. Proceeding from the aforementioned, we suggest that the Vela X-ray ‘jet’ is a dense filament originated along the interface of domelike deformations of the Vela SNR’s shell and projected by chance near the line of sight to the Vela pulsar. There are two main arguments in support of this suggestion. The first one is that the X- ray ‘jet’ correlates (G99) with the optical and radio filaments, which should be the parts of the shell. The second one is that the spectra of the ‘jet’ and its surroundings (the SNR’s shell) are ‘virtually identical’ (M97). Our model offers a natural explanation for the ‘unusual’ velocity field inherent to the Vela SNR’s shell. As we already mentioned in Section 2.2, J84 suggested that some process should induce transverse motions in the shell. The existence of

∗ Note that some signs of such spots were detected by W92. 202 VASILII GVARAMADZE laterally expanding deformations of the shell provides such a process (cf. Meaburn et al., 1988). Another issue which should be discussed is the nature of the Vela X. In our paper (Gvaramadze, 1998b) we came to the conclusion that the VelaX is a part of the SNR’s shell and suggested that its high brightness is a result of the interaction of domelike deformations of the shell, which gives rise to bright radio filaments. We also suggested that discrepancies in the determination of a spectral index for Vela X are connected with smoothing of radio filaments on low-frequency maps, that leads to low values of the index. This instrumental effect is less pronounced at high frequencies. So M95 showed that the spectral index of Vela X has a value close to the spectral indices of Vela Y and Vela Z. However recent comparison of MOST and VLA images of Vela X by Bock et al. (1998b) suggests that the filaments of Vela X have flat spectral indices. But one could admit that this result (obtained with different instruments and with a small frequency baseline) is affected by large calibration errors and therefore the spectra could be steeper. In conclusion let us discuss the origin of the hard X-ray nebula found by W92. As mentioned in Section 2.3, W92 suggested that this nebula is a plerion. Their suggestion was based on the partial overlapping of the nebula with the radio source Vela X, and on the spectrum of the nebula, which was fitted by a power-law model (though it was stressed that the data do not allow to discern the thermal and nonther- mal forms of the spectrum). What we propose is that the nebula is a dense material lost by the SN progenitor star during the red supergiant stage, and reheated to the observed temperatures after the SN exploded (G99).

Acknowledgements

I am grateful to D.C.-J. Bock and A. D’Ercole for useful discussions.

References

Aschenbach, B.: 1997, in: F. Makino and K. Mitsuda (eds.), X-Ray Imaging and Spectroscopy of Cosmic Hot Plasmas, Univ. Acad. Press, Tokyo, 333. Aschenbach, B.: 1998, Nature 396, 141. Aschenbach, B., Egger, R. and Trümper, J.: 1995, Nature 373, 587 (A95). Bock, D.C.-J., Turtle, A.J. and Green, A.J.: 1998a, Astron. J. 116, 1886 (B98). Bock, D.C.-J., Frail, D.A., Sault, R.J., Green, A.J. and Milne, D.K.: 1998b, Mem. Soc. Astron. Ital. 69, 919. Bychkov, K.V. and Pikel’ner, S.B.: 1975, Sov. Astron. Lett. 1, 14. Dame, T.M., Ungerechts, H., Cohen, R.S., et al.: 1987, Astrophys. J. 322, 706. Danks, A.C. and Sembach, K.R.: 1995, Astron. J. 109, 2627. D’Ercole, A.: 1992, Mon. Not. R. Astron. Soc. 255, 572 (D92). Dwarakanath, K.S.: 1991, J. Astron. Astrophys. 12, 199. VELA SUPERNOVA REMNANT 203

Duncan, A.R., Stewart, R.T., Haynes, R.F. and Jones, K.L.: 1996, Mon. Not. R. Astron. Soc. 280, 252 (D96). Frail, D.A., Bietenholz, M.F., Markwardt, C.B. and Ögelman, H.: 1997, Astrophys. J. 475, 224 (F97). Franco, J., Tenorio-Tagle, G., Bodenheimer, P. and Ró˙zyczka, M.: 1991, Publ. Astron. Soc. Paciﬁc 103, 803. Gvaramadze, V.V.: 1998a, in: D. Breitschwerdt, M. Freyberg and J. Trümper (eds.), The Local Bubble and Beyond, Springer-Verlag, Heidelberg, 141. Gvaramadze, V.V.: 1998b, Astron. Lett. 24, 178. Gvaramadze, V.V.: 1999, Astron. Astrophys. 352, 712 (G99). Jenkins, E.B., Wallerstein, G. and Silk, J.: 1984, Astrophys. J. 278, 649 (J84). Kahn, S.M., Gorenstein, P., Harnden, F.R. and Seward, F.D.: 1985, Astrophys. J. 299, 821 (K85). Lyne, A.G., Pritchard, R.S., Graham-Smith, F. and Camilo, F.: 1996, Nature 381, 497. Markwardt, C.B. and Ögelman, H.: 1995, Nature 375, 40 (M95). Markwardt, C.B. and Ögelman, H.: 1997, Astrophys. J. 480, L13 (M97). McKee, C.F., Van Buren, D. and Lazareff, R.: 1984, Astrophys. J. 278, L115. Meaburn, J., Hartquist, T.W. and Dyson, J.E.: 1988, Mon. Not. R. Astron. Soc. 230, 243. Milne, D.K.: 1968, Aust. J. Phys. 21, 201. Milne, D.K.: 1995, Mon. Not. R. Astron. Soc. 277, 1435. Milne, D.K. and Manchester, R.N.: 1986, Astron. Astrophys. 167, 117. Nasuti, F.P., Mignani, R., Caraveo, P.A. and Bignami, G.F.: 1997, Astron. Astrophys. 323, 839. Osterbrock, D.E.: 1989, Astrophysics of Gaseous Nebulae and Active Galactic Nuclei, Univ. Science Books, Mill Valley. Parker, R.A.R., Gull, T.R. and Kirschner, R.P.: 1979, An Emission-Line Survey of the Milky Way, NASA SP-434, Washington (P79). Pikel’ner, S.B.: 1954, Izv. Krym. Astroﬁz. Obs. 12, 93. Seward, F.D.: 1990, in: M. Elvis (ed.), Imaging X-Ray Astronomy, Cambridge Univ. Press, 241 (S90). Shull, P., Dyson, J.E., Kahn, F.D. and West, K.A.: 1985, Mon. Not. R. Astron. Soc. 212, 799. Vanbeveren, D., De Loore, C. and Van Rensbergen, W.: 1998, Astron. Astrophys. Rev. 9, 63 (V98). Wallerstein, G. and Silk, J.: 1971, Astrophys. J. 170, 289. Weaver, R., McCray, R., Castor, J., Shapiro, P. and Moore, R.: 1977, Astrophys. J. 218, 377. Weiler, K.W. and Panagia, N.: 1980, Astron. Astrophys. 90, 269 (W80). Willmore, A.P., Eyles, C.J., Skinner, G.K. and Watt, M.P.: 1992, Mon. Not. R. Astron. Soc. 254, 139 (W92).