Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Mon. Not. R. Astron. Soc. 296, 225-230 (1998) 12-iim fine-structure emission line and continuum images of G333.6-0.2 Takuya Fujiyoshi, 1 Craig H. Smith, 1 Toby J. T. Moore, 1 ' 2 David K. Aitken, 1 ' 3 Patrick F. Roche 4 and Dale E.Quin& 'School of Physics, University College, The University of New South Wales, Australian Defence Force Academy, Canberra, ACT 2600, Australia Astrophysics Group, School of Electrical Engineering, Electronics & Physics, Liverpool John Moores University, Byrom Street, Liverpool L3 3AF 3 Division of Physical Sciences, University of Hertfordshire, College Lane, Hatfield, Herts AL10 9AB 4 Department of Astrophysics, Oxford University, Keble Road, Oxford OX1 3RH 2 Accepted 1997 October 1. Received 1997 June 23; in original form 1996 October 18 ABSTRACT We present high spatial resolution (-0.8 arcsec) diffraction-limited 12.8-Rm Ne ii finestructure emission line and 12.5-rim continuum images of the bright southern compact H ii region G333.6-0.2, taken with the mid-infrared imaging polarimeter NIMPOL. The two images show remarkably similar, compact, yet asymmetric, flux distributions. The [Ne ii] image shows a complex structure near the ionizing source(s) which we interpret in terms of the ionization structure of the H II region. It is found that G333.6-0.2 is more likely to be excited by a cluster of 0 and B stars than by a single star. Key words: Hit regions — ISM: individual: G333.6-0.2 — infrared: ISM: continuum — infrared: ISM: lines and bands. 1 INTRODUCTION G333.6-0.2 is a southern compact H II region which is bright at radio and infrared wavelengths but almost totally obscured in the visible (Goss & Shaver 1970; Becklin et al. 1973). Becklin et al. attributed the large mid- and far-infrared (MIR and FIR) excesses of the object to optically thin thermal emission from dust grains. The strong 12.8-Rm Ne II fine-structure emission line was first observed in the N-band (8-13 tim) spectroscopic study of the H II region by Aitken & Jones (1974). They interpreted the almost featureless spectrum (i.e. lacking the typical 10-Rm silicate feature) as re-absorption of thermal emission from optically thin dust containing silicates, by a cooler envelope of similar material surrounding the hotter emitting core. From these observations they proposed a core—halo model for G333.6-0.2. The H II region is also associated with a faint visible nebulosity (Churms et al. 1974) and, by examining extinction towards G333.6-0.2 at different infrared wavelengths, Hyland et al. (1980) developed the 'blister' model (an H II region on the edge of a dense molecular cloud — thought to be not uncommon in H ii regions — e.g. Habing & Israel 1979) viewed face-on. A number of attempts had successfully been made to map G333.6-0.2 at medium spatial resolutions in the [Ne ii] (at 3.4 arcsec: Aitken, Griffiths & Jones 1977; at 3.6 arcsec: Geballe et al. 1981). Both groups found that the [Ne II] map, and maps at other nearby wavelengths, showed remarkably similar distributions with a simple, compact and nearly symmetric source. The peaks in the maps were coincident, suggesting that the exciting source(s) must be concentrated within the central peak of line emission. The agreement of intensity distribution over a range of wavelengths is not surprising if G333.6-0.2 lies on the front surface of a dense © 1998 RAS cloud as concluded by Hyland et al. (1980), since the distribution observed is then due mainly to variation in the emission measure and not in the extinction. The agreement also applies to the FIR maps of Hyland et al. (1980, at 30, 50 and 100 lim) which showed radial symmetry of flux distribution at all wavelengths, again suggesting that the exciting star(s) are concentrated within the central region. To probe the nature of the exciting source more closely, we have made observations using the higher spatial resolution of an array camera. We present effectively diffraction-limited (-0.8 arcsec) 12.8-Rm Ne II fine-structure emission line and 12.5-Rm continuum images of G333.6-0.2. The [Ne n] image shows a previously unresolved complex structure within the inner 5 arcsec of the main peak. 2 OBSERVATIONS AND DATA REDUCTION The observations were made on 1994 May 27 and 1995 May 14 using the f/36 focus of the 3.9-m Anglo-Australian Telescope (AAT) at Siding Spring Observatory, Australia, with our MIR imaging polarimeter NIMPOL. The instrument utilizes an Amber Engineering 128 x 128 Si:Ga AE159 Focal Plane Array, which is liquid-He cooled but usually heated to about 18 K for optimum efficiency. Under normal conditions the camera provides adequately oversampled diffraction-limited images on the AAT with a pixel size of 0.25 x 0.25 arcsec 2 and a field of view of 32 x 32 arcsec 2 . Wavelength selection is provided by a cold CVF and discrete filters (see Smith, Aitken & Moore 1994 for further detail). We observed G333.6-0.2 using a narrow-band 12.8-p.m Ne ii filter with AX 2 per cent — 0.3 xm (the full width at the half 226 T. Fujiyoshi et al. power point) and a 12.5-Rm filter with AX 0.8 rim. Both the skychop and beam-switch were 30 arcsec north-south off the detector. The coordinates were RA = 16 h 22m 095 and Dec. = -50°06'00" (J2000) at the 12.5-Rm continuum peak. Total on-source integration time was -15 min for the 12.8-Rm image and -30 min for the 12.5-Rm image. This produced a signalto-noise ratio (in a 4.3-arcsec software aperture centred on the main peak) of more than 200 in the 12.8-rim image and over 300 in the 12.5-tim image. Standard point sources on both nights of observation showed full widths at half maximum (FWHMs) of about 1 arcsec while the diffraction limit of the AAT at these wavelengths is -0.8 arcsec (the Rayleigh criterion). In order to restore obvious seeing and/or tracking errors, where applicable, we have re-registered images in blocks of a few seconds, involving shifts of up to 3 pixels. In order to subtract the continuum contribution from the [Ne II] image, we have used the 8-13 spectrum of G333.6-0.2, taken with a 4.3-arcsec beam centred on the continuum peak (Fujiyoshi et al., in preparation), to estimate the line-to-continuum flux ratio in the 12.8-Rm filter. This ratio was then used to scale the 12.5-iim image to produce an approximation of the 12.8-Rm continuum distribution, after measuring the 12.8-Rm line-plus-continuum flux in a 4.3-arcsec software aperture. The 12.5-Rm filter also covers the wavelength of [Ne II], but the line contributes less than 15 per cent to the total flux in this band and the resulting error is acceptably small. The maximum entropy method (MEM, see Sehgal & Morris 1992 and references therein) was employed to smooth and enhance the final images. An image of the standard star 13 Grus, similarly treated, showed that MEM was capable of achieving a spatial resolution of 0.6 arcsec (FWHM), further restoring the blurring caused by seeing and/or tracking errors. However, the images were eventually smoothed by a 0.8-arcsec FWHM Gaussian back to the diffraction limit. Since the images were oversampled this could be done without loss of information. The absolute calibration uncertainty of the spectrum used for the calibration of the images is estimated to be approximately 10 per cent. This increases only slightly in the [Ne II] intensity due to the continuum subtraction process, since the relative uncertainty between the spectral points is better than 1 per cent. 3 RESULTS The 12.8-Rm Ne II fine-structure emission line and 12.5-tim continuum images of G333.6-0.2 are presented in Figs 1 and 2, respectively. As Aitken et al. (1977) found at medium spatial resolution, the overall level of similarity between the two is high, especially in the fainter regions. However, the higher spatial resolution achieved here shows up quite large differences in the central 5 arcsec. Before discussing each feature, it is worth recalling that the continuum in the N-band originates from the thermal emission of dust grains such as silicates and graphite, whereas the fine-structure line emission traces the hot, ionized gas. The face-on blister geometry proposed by Hyland et al. (1980) offers a possible explanation for the large-scale similarity between the two images. Assuming little dust exists within the H ii region itself, the dust producing the observed continuum emission must lie in a narrow zone between the foreground ionized gas and the background molecular cloud. This dust is heated by a combination of direct stellar illumination and Lyman alpha photons otherwise trapped in the ionized gas, and the emission arising from it should therefore appear spatially correlated with the density and ionization structure of the H ii region. Alternatively, the similarity may simply imply that the gas and dust are well mixed in the region. G333.6-0.2 has a compact central peak from which most of the flux originates in both images. It is elongated north-south and has a steeper gradient to the east, which may be due to the presence of dense absorbing materials in that direction (Hyland et al. 1980). In fact, in our spectropolarimetric study of the nebula (Fujiyoshi et al., in preparation) we find that there is an increase in the absorption optical depth in the region to the east of the main peak. In the 12.5-iim continuum image the central emission peak measures -1.7 arcsec (FWHM) east-west and twice as large in the north-south direction. Assuming a distance to the object of 3.0 kpc (Colgan et al. 1993), 1.7 arcsec translates approximately to 0.025 pc or 5.1 x 10 3 au. These values are smaller than that found by Hyland et al. (1980) with a slightly lower spatial resolution of 1.1 arcsec. The peak flux density is (1.5 ± 0.2) x 10 -14 W M -2 IIM -1 in the continuum, and the peak flux is (1.6 ± 0.2) x 10 -12 W M -2 in the line image, both in a 4.3-arcsec diameter beam centred at the coordinates of the main continuum peak. There is also a much fainter secondary peak, also elongated north-south, about 4 arcsec east of the main peak. Although the secondary peak in the [Ne image is not as prominent as that in the continuum, it is still clearly separated from the main peak by a small gap. It is on the pedestal of the main peak and may only have the appearance of a separate emission feature since the gap may be produced by a band of high extinction, as mentioned above. Another common feature is the extension to the west and southwest, which generally has circular-shaped contours centred on the continuum peak in both images. It is worth noting here that the positions of the north extension to the secondary peak, the main peak and the west and south-west extension are approximately those of the three star-like objects seen in the near-infrared and Ha photographic images of Churms et al. (1974), in the 1-Rm continuum CCD image of Geballe et al. (1981), and also in our data in preparation. The similarity between the distributions of [Ne ii] and 12.5-iim continuum fluxes also extends to the faint halo around the more intense central region. This is clearly seen in the tail-like structure arising just below the south-west extension and curling up to the east. In the medium spatial resolution line maps of Geballe et al. (1981) this structure was less well resolved and appeared as a plateau to the south-south-east of the central peak. Aitken et al. (1977) found that the Ne II equivalent width increases with aperture size, indicating that the halo is more intense in line than in continuum radiation. Our images show a similar trend even at the finer scale. Fig. 3(a) shows the cross-sectional views of both the line and continuum images in the central region. The [Ne II] image has a rather gentle slope while the continuum peaks sharply. This phenomenon can be seen more quantitatively in Fig. 3(b), in which the continuum flux density flattens much faster as the aperture size is increased than does the line flux. This observation is consistent with some suggestions that emission lines are usually more diffuse than dust emission in H II regions (e.g. Natta & Panagia 1976). The line emission is less affected by the proximity of the exciting source because the electron temperature (T e ) only changes slowly across the H ii region and the collisionally excited line emission is only weakly dependent on Te anyway. Only variations in the local density and line-of-sight column density of material are important to the [Ne II] emission, whereas the dust thermal continuum emission, while also determined by the column density, is strongly dependent on the radiative equilibrium © 1998 RAS, MNRAS 296, 225-230 12-pm images of G333.6-0.2 227 10.0 5.0 - 10 0 . - 15.0 15.0 10.0 5.0 0.0 -5.0 - 10.0 RA (aresee) Figure 1. Grey-scale image of G333.6-0.2 in the 12.8-ilm Ne II fine-structure emission line. The contours are drawn at 2.8, 3.8, 5.0, 6.2, 7.7, 9.3, 11, 13 and 15 x 10 -14 W m -2 arcsec -2 . temperature of the dust grains, and so on distance from the heating source. 3.1 The Ne ii central dip The most obvious difference between the two images is the presence of a circular dip in the central region in the [Ne II] image. The Ne II central dip measures approximately 1.2 arcsec FWHM, or 1.7 x 10 -2 pc — 3.5 x 10 3 au at a distance of 3.0 kpc, in RA. Its circular shape is very suggestive of its association with the central exciting star or compact cluster of stars. If the [Ne II] image is smoothed further most of the central structure disappears and the smoothed image resembles remarkably the medium-resolution line maps of Aitken et al. (1977) and Geballe et al. (1981), supporting the consistency of our finding. The central dip in the [Ne II] image is not likely to be a product of the continuum subtraction as two extreme cases were examined by constructing two further [Ne II] images. One has 10 per cent more continuum subtracted from a 12.8-µm image with 10 per cent less flux level, and vice versa. The results showed that the features are consistent and relatively insensitive to the level of the continuum subtraction. Among the most plausible explanations for the existence of a central minimum in the [Ne II] image are: locally reduced neon abundance; high optical depth in foreground material or in the line; a cavity due to stellar wind; and ionization structure in the H II region. We consider the last option most likely, as there has been a © 1998 RAS, MNRAS 296, 225-230 detection of the Ne iii emission line at 36.0 iim in G333.6-0.2 (Colgan et al. 1993). The position, circular shape and size of the central Ne II minimum strongly suggest that it may be the ionization boundary of Ne in. Although there is no information available on the precise spatial origin of the detected [Ne In], its relatively high ionization potential (IP = 41.0 eV) requires that it should arise in the vicinity of the main ionizing source(s), that is near the centre. The other possibilities lack supporting evidence at this stage. 4 DISCUSSION 4.1 Ionization structure - the Ne in ionization boundary radius As a first approximation we take a simple case of a pure neon nebula without dust. For some heavy elements, the geometrical dilution of the ionizing radiation field is a more important factor in determining their ionization structure in H II regions than the StrOmgren-like self-absorption of the ionizing photons above a certain IP. Assuming constant density and spherical geometry in the innermost core of G333.6-0.2, and further that the above is the case for neon, we estimate the ionization transition boundary radius, where the numbers of Ne iii and Ne II ion species are equal, by solving the relevant ionization equilibrium equation (Rubin 1983). - 1/2 R! a l (Ne II, 017 ,6- f Rib (Ne iii) = with n e cy(Ne IL Te) 228 T Fujiyoshi et al. 10.0 5.0 0.0 - 10.0 -15.0 15.0 10.0 5.0 0.0 -5.0 -10 0 . RA (aresec) Figure 2. Same as Fig. 1 but for the 12.5-tim continuum. The contours are drawn at 1.2, 1.9, 2.8, 4.2, 5.9, 8.1, 11, 14 and 18 x 10 -13 W m -2 p,m -1 arcsec -2 . The coordinates are given relative to the position of the 12.5-Rm continuum peak. where Rib is the ionization boundary radius, 1?,, the stellar radius, p th the threshold frequency of ionizing photons, a l the ground state photoionization cross-section, Fp the stellar photon flux, n e the electron density in the Ne iii zone, and a the total recombination coefficient to Ne ii. Following Rubin et al. (1994), who developed a semi-empirical method with which to analyse nebular properties from FIR spectroscopy and subsequently applied it to G333.6-0.2, we use the effective temperature Teff = 36 000 K, log surface gravity log g = 4.0, and Te = 8000 K. Rubin, Kunze & Yamamoto (1995) calculated the total number of ionizing photons from several different model stellar atmospheres emerging over various spectral ranges, including the Ne ii ionizing (i.e. Ne iii producing) photons for the above stellar parameters. The photon flux is also a function of R. Rubin et al. (1994) estimated that, in order to reproduce the observed FIR flux, the central star (if single) in G333.6-0.2 would have to be as large as 12,, = 30.8 R o . The photoionization cross-section data were derived from analytical fits of Verner et al. (1996). For the electron density, we have adopted the value of 4570 cm -3 obtained from observations of another doubly ionized species (0 ill, Colgan et al. 1993). This measurement was taken on the Kuiper Airborne Observatory using a large beam (-45 arcsec) and therefore more sensitive to the lower density outer regions. Further, since the second IP of oxygen (35.1 eV) is slightly lower than that of neon, this value of n e is likely to be smaller than in the Ne iii zone. We therefore take it as a lower limit to the electron density. In the situation being considered, the relevant recombination processes are direct radiative recombination (rate coefficient a R ) and dielectronic recombination (rate coefficient a D ). Values for these rate coefficients, along with the other parameters, are provided in Table 1. Inserting the values into the expression above gives R ib (Ne 8.0 x 10 -2 pc. Now we consider the effects of dust and other abundant heavy elements which will reduce the size of the Rib calculated for a pure neon nebula. If dust grains coexist with gas in the ionized region, absorption of ionizing photons by dust is thought to be important, especially in compact, bright H II regions (e.g. Mathis 1986). In optically thin cases at MIR wavelengths, the optical depth due to dust grains can be determined by comparing the intrinsic dust emission flux density with the dust temperature blackbody radiation intensity (Hildebrand 1983). To de-redden the 12.5- continuum flux density (to allow for the observed cold overlaying silicate absorption) we take the silicate optical depth at 12.5 T 12.5 12.5 = 0.58, and the dust temperature, Td = 250 K, both determined from the fit to the N-band spectrum (Fujiyoshi et al., in preparation). Comparison of de-reddened flux density and the blackbody intensity at 250 K gives r i2.5 through the nebula which is then divided by two to give that along the line of sight to the central exciting source(s). Roche & Aitken (1984) showed that the peak optical depth in the 9.7-µm feature is closely correlated with visual extinction in the form of A v = 18.5 x T9 7, where r9.7 = 2.3 x T12 5, assuming the Trapezium silicate emissivity function of Forrest, Gillett & Stein (1975). Once the visual extinction is known it can be extrapolated to the extreme-ultraviolet (EUV) wavelength range of Ne ii ionizing . . © 1998 RAS, MNRAS 296, 225-230 12-pm images of G333.6-0.2 229 (a) 60 65 Column number (pixel) 70 75 4.2 The ionizing source in G333.6 0.2 (b) - 10 8 C a) x 6 Jr_ 0 4 LL 2 0 5 15 10 Beam diameter (arcsec) 20 Figure 3. (a) Normalized flux distributions of the central region of G333.60.2 for the 12.8-Rm Ne II fine-structure emission line and 12.5-Rm continuum. It is a cross-sectional view across RA at Dec. offset = 0.0. The error bars shown for the [Ne II] data are ± 1 o (too small on the scale for the continuum data). (b) Flux or flux density as a function of beam diameter. The 12.8-11m data are in x10 -12 W M -2 , and the 12.5-p.m data are in x10 -11 W m -2 Rm -1 . - photons using EUV extinction curves (e.g. Voshchinnikov & Il'in 1993) which indicate for the standard MRN graphite—silicate mixture (Mathis, Rumpl & Nordsieck 1977) AEUV/Av 2. The extinction due to dust in the EUV in this way was estimated to be about 0.8 mag in the central 4.3 arcsec. In other words, more than 50 per cent of the initial Ne ill producing photon flux is absorbed by dust which will be re-emitted at longer wavelengths. The expression for Rib also neglects the effect of the presence of any other heavy elements, so Rib will be smaller since other species will compete for the ionizing photons to varying degrees depending on their abundance and photoionization cross-section. When this effect is taken into account for other abundant heavy elements (C, N, 0), assuming the Orion abundance (Rubin et al. 1991) and considering the cross-sectional dependence of photoionization derived from analytical fits of Verner et al. (1996), we estimate a further reduction in the photon flux of more than 80 per cent. That is, when we include the effects of both dust and other abundant heavy elements competing for the Ne II ionizing photons, the initial flux is reduced by more than 90 per cent. The reduced number of photons would still give an Rib larger than the dip by a factor of —2; however, as we mentioned earlier the estimate of the electron © 1998 RAS, MNRAS 296, 225-230 density in the Ne iii zone is a lower limit. Clearly the greater the n e value, which one would expect towards the more highly ionized core of the nebula, the smaller will the Rib be. Furthermore, if three star-like structures seen in the Ha and optical images of Churms et al. (1974), the 1-Rm continuum CCD image of Geballe et al. (1981), and our data in preparation represented the three separate stars or star clusters, then using the single stellar radius of 30.8 R o would be inappropriate. This radius was chosen in order to reproduce the FIR flux from the nebula, but when used for the central peak only would give an overestimated number of photons, which would in turn result in an overestimate of the Ne HI ionization boundary radius. Given these considerations it is most likely that the Ne ii dip does represent the Ne ni ionization boundary and that the parameters in Table 1 are roughly representative of the situation in G333.6-0.2. An interesting combination of compact symmetric appearance, high luminosity, and low excitation in G333.6-0.2 has been noted by many observers. The 12.5-Rm continuum image (Fig. 2) shows a compact peak which measures 1.7 arcsec or 0.025 pc across at 3.0 kpc. However, our images clearly show the asymmetric nature of the object. The total luminosity was estimated to be 1.9 x 10 6 L o (Colgan et al. 1993), while detailed modelling (Rubin et al. 1994) suggested a relatively low excitation temperature, Teff, of about 36 000 K, which is typical of an 08 star. There are a number of possible explanations for this discrepancy, i.e. that the exciting source consists of a compact cluster of a large number of mainsequence stars, a few super-luminous pre-main-sequence objects, or a single star still contracting to the main sequence. The H II region is associated with H2 and CO molecular emissions (Storey 1983; Storey et al. 1989), and OH and H 2 O masers (Batchelor et al. 1980), all of which are common in massive star-forming regions. If a single star alone is responsible for ionizing the nebula, the combination of Teff 36 000 K and R,, = 30.8 R o implies that the exciting source is more likely to be a supergiant than a newly formed star. There are some suggestions that high-mass stars form preferentially in clusters rather than in isolation (e.g. Larson 1990). If that is indeed the case in G333.6-0.2, and assuming that the stars have reached the main-sequence stage, the ionizing photon flux used in the previous section to estimate the Ne iii ionization boundary radius requires approximately 13 08 stars. This figure is consistent with the total luminosity which suggests about 11 08 stars. Perhaps it is improbable to expect a cluster of identical stars, and a combination of 0 and B stars is more appropriate. Such a combi- nation, the OB cluster, is thought to be at the centre of a number of compact H II regions (e.g. G29.96-0.02: Lumsden & Hoare 1996). If the exciting source is pre-main-sequence, say at the zero-age main-sequence (ZAMS) stage, then the total luminosity requires about 40 ZAMS stars. Rubin, Hollenbach & Erickson (1983) did not favour a large number of ZAMS stars as the ionizing source in G333.6-0.2 since not only does it involve a large mass but also it requires an unusual initial mass function. Therefore, we consider the OB cluster the best candidate for the exciting source in G333.6-0.2. 5 CONCLUSIONS High spatial resolution (-0.8 arcsec) diffraction-limited 12.8-tim [Ne ii] and 12.5-gm continuum images of the bright southern 230 T Fujiyoshi et al. Table 1. Parameters for the Ne iii ionization boundary radius (see text for details). Parameter Teff Te R4Td1) al ne D a R a Value Reference 36,000 K (log g = 4) 8,000 K 7.506 x 10 43 s -1 (R s, = 30.8 R o ) 8.1 x 10 -18 cm 2 4,570 cm -3 4.314 x 10 -13 cm 3 s -1 1.772 x 10 -12 cm 3 s -1 Rubin et al. (1994) Rubin et al. (1994) Rubin et al. (1995) Verner et al. (1996) Colgan et al. (1993) Nussbaumer & Storey (1987) Pequignot, Petitjean & Boisson (1991) compact H II region G333.6-0.2 are presented. The images show remarkably similar, compact, yet asymmetrical emission distributions. The central peak, from which the most flux originates, measures -1.7 arcsec (FWHM) in RA in the 12.5-Rm continuum, which is approximately 2.5 x 10 -2 pc or 5.1 x 10 3 au. The [Ne image shows for the first time the complex structure within the inner 5 arcsec of the line peak. We interpret this in terms of the ionization structure of the H II region. The size of the central dip compares well with the estimated Ne iii ionization boundary when effects of dust and other abundant heavy elements are taken into account. From the model parameters used we conclude that the ionizing source in G333.6-0.2 is more likely to be a cluster of 0 and B stars than a single star. ACKNOWLEDGMENTS We would like to thank Professor Walt Duley, Dr Chris Wright and Dr Rob Smith for valuable discussions, and the staff at the AAT for their efficient assistance in observation. We would also like to thank the anonymous referee for useful comments. REFERENCES Aitken D. K., Jones B., 1974, MNRAS, 167, 11P Aitken D. K., Griffiths J., Jones B., 1977, MNRAS, 179, 179 Batchelor R. A., Caswell J. L., Goss W. M., Haynes R. F., Knowles S. H., Wellington K. J., 1980, Aust. J. Phys., 33, 139 Becklin E. E., Frogel J. A., Neugebauer G., Persson S. E., Wynn-Williams C. G., 1973, ApJ, 182, L125 Churms J., Feast M. W., Glass I. S., Harding G. A., Lloyd Evans T., Martin W. L., 1974, MNRAS, 169, 39P Colgan S. W. J., Haas M. R., Erickson E. F., Rubin R. H., Simpson J. P., Russell R. W., 1993, ApJ, 413, 237 Forrest W. J., Gillett F. C., Stein W. A., 1975, ApJ, 195, 423 Geballe T. R., Wamsteker W., Danks A. C., Lacy J. H., Beck S. C., 1981, ApJ, 247, 130 Goss W. M., Shaver P. A., 1970, Aust. J. Phys., Astrophys. Suppl., No. 14, 1 Habing H. J., Israel F. P., 1979, ARA&A, 17, 345 Hildebrand R. H., 1983, QJRAS, 24, 267 Hyland A. R., McGregor P. J., Robinson G., Thomas J. A., Becklin E. E., Gatley I., Werner M. W., 1980, ApJ, 241, 709 Larson R. B., 1990, in Capuzzo-Dolcetta R., Chiosi C., Di Fazio A., eds, Physical Processes in Fragmentation and Star Formation. Kluwer, Dordrecht p. 389 Lumsden S. L., Hoare M. G., 1996, ApJ, 464, 272 Mathis J. S., 1986, PASP, 98, 995 Mathis J. S., Rumpl W., Nordsieck K. H., 1977, ApJ, 217, 425 Natta A., Panagia N., 1976, A&A, 50, 191 Nussbaumer H., Storey P. J., 1987, A&A, 69, 123 Pequignot D., Petitjean P., Boisson C., 1991, A&A, 251, 680 Roche P. F., Aitken D. K., 1984, MNRAS, 208, 481 Rubin R. H., 1983, ApJ, 274, 671 Rubin R. H., Hollenbach D. J., Erickson E. F., 1983, ApJ, 265, 239 Rubin R. H., Simpson J. P., Haas M. R., Erickson E. F., 1991, ApJ, 374, 564 Rubin R. H., Simpson J. P., Lord S. D., Colgan S. W. J., Erickson E. F., Haas M. R., 1994, ApJ, 420, 772 Rubin R. H., Kunze D., Yamamoto T., 1995, in Adelman S. J., Wiese W. L., eds, ASP Conf. Ser. 78, Astrophysical Applications of Powerful New Databases. Astron. Soc. Pac., San Francisco, p. 479 Sehgal A., Morris L. R., 1992, in Worrall D. M., Biemesderfer C., Barnes J., eds, Astronomical data analysis software and systems i. Astron. Soc. Pac. San Francisco, p. 232 Smith C. H., Aitken D. K., Moore T. J. T., 1994, in Crawford D. L., Craine E. R., eds, Proc. SPIE 2198, Instrumentation in Astronomy VIII. SPIE, Bellingham, WA, p. 736 Storey J. W. V., 1983, MNRAS, 202, 105 Storey J. W. V., Harnett J. I., Lugten J. B., Crawford M. K., Stacey G. J., Watson D. M., Genzel R., 1989, MNRAS, 237, 1001 Verner D. A., Ferland G. J., Korista K. T., Yakovlev D. G., 1996, ApJ, 465, 487 Voshchinnikov N. V., Il'in V. B., 1993, AZh, 70, 38 This paper has been typeset from a TEX/L A TEX file prepared by the author. © 1998 RAS, MNRAS 296, 225-230