THE ASTRONOMICAL JOURNAL, 117:410-428, 1999 January
© 1999. The American Astronomical Society. All rights reserved. Printed in U.S.A.

We present a millimeter-wave study of a cluster of bipolar CO outflows embedded in the western end of the Circinus molecular cloud complex, G317-4, that is traced by very high optical extinction. For an assumed distance of 700 pc, the entire Circinus cloud is estimated to have a mass of about 5 × 10⁴ M. The opaque western portion that was mapped in this study has a mass of about 10³ M, contains a number of embedded infrared sources and various compact 1.3 mm continuum sources, and has a remarkable filamentary structure with numerous cavities that appears to be the fossil remnants of past star formation activity. The most active star-forming region in this part of Circinus is centered around a compact cluster of millimeter continuum sources associated with IRAS 14564-6254 and IRAS 14563-6301, which lies about 7 to the south. This region contains two known Herbig-Haro objects, HH 76 and HH 77, and a profusion of overlapping high-velocity CO outflow lobes. Among these, we can clearly distinguish the two largest outflows in Circinus (flows A and B), which appear to originate from the two brightest IRAS sources. This region also contains at least two other prominent but overlapping bipolar CO outflows (flows C and C), one of which may be associated with IRAS 14564-6258. Two compact and relatively low-velocity CO outflows lie at the northern periphery of the Circinus core and are associated with IRAS 14563-6250 (flow E), a source also detected as a 1.3 mm continuum source, and with IRAS 14562-6248 (flow G). A small but prominent reflection nebula associated with the nebulous star vBH 65a and a coaxial Herbig-Haro jet, HH 139, is located at the southeastern edge of this cloud core and illuminates part of a cavity seen as a low-extinction region. A faint and low-mass CO molecular flow is associated with vBH 65a and HH 139 (flow F). The infrared source IRAS 14580-6303 drives a small CO flow (flow I). A second, active center of star formation is centered on the source IRAS 14592-6311, on the peculiar Herbig Ae/Be star vBH 65b, about 20 to the southeast of the main cloud core; four HH objects, HH 140 through HH 143, and a compact CO outflow are located here (flow D). About 5 farther south, IRAS 14596-6320 drives yet another outflow (flow H). Thus, the mapped portion of Circinus contains at least 10 CO-emitting molecular outflows. Assuming that star formation has continued at a steady rate for the last several hundred thousand years, the Circinus cloud is expected to have produced dozens of young stars. Their outflows have severely altered the structure and kinematics of this cloud, as evidenced by the multitude of prominent cavities and dust filaments that outline their boundaries. This level of star formation activity is consistent with the numerous postoutflow phase H emission-line stars that have been found in this region. The Circinus cloud complex is an archetypical case where star formation activity may have profoundly affected the structure of a molecular cloud, producing its strikingly filamentary and cavitated appearance and providing further evidence that star formation may be a self-regulated process.

Key words: stars: formationstars: premain-sequence

     ¹ Department of Astrophysical and Planetary Sciences, Center for Astrophysics and Space Astronomy, and Center for Astrobiology, University of Colorado, Campus Box 389, Boulder, CO 80309-0389.
     ² bally@casa.colorado.edu; http://casa.colorado.edu/~bally.
     ³ reipurth@casa.colorado.edu.
     ⁴ Harvard Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138.
     ⁵ clada@cfa.harvard.edu.
     ⁶ billawal@casa.colorado.edu.

     Star formation may be a self-regulated process in which jets, winds, and radiation produced by young stars may regulate the rate of gravitational collapse, possibly prevent collapse altogether, or even disrupt the star formation environment. Recent observations have shown the ubiquity of giant, parsec-scale outflows (cf. Reipurth, Bally, & Devine 1997; Devine et al. 1997; Eislöffel & Mundt 1997; Bence, Richer, & Padman 1996; Bally, Devine, & Reipurth 1996) that drive strong shocks into the surrounding medium where they may accelerate or even dissociate molecules, produce cloud turbulence, and in some cases blow out of the host cloud.

     The Circinus cloud complex is located toward l = 318, b = -4 along a relatively extinction-free line of sight in the southern Milky Way. In visual wavelength images, the cloud stands in stark contrast against a rich background of stars in the fourth quadrant of the Galaxy, and the cloud can be traced by its extinction over a region about 2° × 6° in extent. Our attention was drawn to this object by the filamentary appearance of the western portion of the cloud, which is very opaque on deep photographs (see Fig. 1). This core region contains dozens of cavities bounded by narrow dust filaments, about a half-dozen IRAS sources, and many signs of star formation including Herbig-Haro objects (Reipurth & Graham 1988; Ray & Eislöffel 1994) and H emission stars (Mikami & Ogura 1994). A fan of nearly north-southoriented filaments radiates from the southern periphery of the main core. A 25 long ridge of opaque material extends east-west from the southeast corner of the cloud core, and over a dozen narrow dust lanes cross this ridge at nearly right angles. The multiple cavities and filaments that fan away from the opaque core may indicate that this cloud has suffered extensive damage from the impact of dozens of outflows powered by young stars formed during the last few hundred thousand years.

FIG. 1.An R-band ESO Schmidt photograph of the high-obscuration western portion of the Circinus cloud

     In this paper, the first of a series of multiwavelength investigations of the Circinus complex is presented. We discuss new CO observations that reveal the presence of multiple high-velocity bipolar outflows powered by embedded young stars. In subsequent papers, we will present observations of new Herbig-Haro objects, shock-excited molecular hydrogen, and a census of the young stars produced by the Circinus cloud over the last few million years.

     We observed J = 10 ¹²CO, ¹³CO, and C¹⁸O transitions during 1988 April 28 to May 9 and during 1990 May with the 15 m diameter Swedish-ESO Submillimetre Telescope (SEST). We used a Schottky barrier heterodyne mixer providing a single-sideband receiver temperature of about T_ssb = 350500 K, which under typical observing conditions produced system temperatures ranging from about T_sys = 600 to over 900 K. We used a chopper to compare the sky emission to an ambient temperature absorber for calibrating the spectra and checked the resulting calibration by comparing SEST spectra obtained at several locations in M17 and in the Ophiuchus cloud with observations obtained with other telescopes in the northern hemisphere. These comparisons indicate that our calibration is accurate to about 15%. Observations of the compact high-velocity outflow in the Orion A core were used to estimate the main-lobe beam efficiency to be about = 0.7. However, our spectra, maps, and mass estimates have not been corrected for the main-lobe efficiency since most of the observed structures are large compared with the beam. The FWHM diameter of the SEST beam between 109 and 115 GHz is about 44. Spectra were recorded with a 2000 channel acousto-optical spectrometer that provided a channel spacing of 43 kHz. All observations were obtained in frequency-switching mode, in which the local oscillator was shifted by 20 MHz at a rate of about 5 Hz. In this mode, the SEST system provided flat and stable baselines, which for most spectra could be well fitted with a first-order baseline. Approximately 1200 ¹²CO, 1300 ¹³CO, and 350 C¹⁸O spectra were obtained during the two observing runs, with most spectra obtained on uniform 40 grids with integration times of 300 s per position, yielding spectra with a typical rms noise of about 0.25 K in 43 kHz channels. The data were reduced with the Bell Laboratories COMB data reduction package, which was used to fold frequency-switched spectra; fit baselines; and generate contour maps, images, data cubes, and mass estimates of the various cloud components.

     Figure 1 shows an R-band ESO Schmidt photograph of the Circinus molecular cloud covering about a 1 deg² field of view. The cloud is seen in silhouette against a rich background of stars and consists of a complex network of opaque filaments and translucent cavities. Figure 2a is a finder chart showing the locations of the various IRAS sources and outflows discussed in the text. The border of this chart is identical to the borders of the contour diagrams shown in Figures 2b2d, so that the locations of sources and flows can be easily identified on both contour maps and on the optical image by comparison with Figure 2a. Theaxes in these figures are labeled in arcminute offsets from a reference position (0, 0) that corresponds to (1950) = 14^h56^m249, (1950) = -6254^'59^''.

FIG. 2.(a) A finder chart showing the location of CO outflows in the western part of the Circinus complex. (b) A velocity-integrated 13CO J = 10 map of the Circinus cloud over the velocity range vlsr = -8 to -4 km s-1 superimposed on the image shown in Fig. 1. Contour levels are plotted from 1 to 14 K km s-1 at intervals of 0.5 K km s-1. Solid circles mark the locations of IRAS sources. The straight lines mark the location and orientation of the flows discussed in the text. Two lines (marked C and C) indicate two sets of overlapping outflow lobes that are hard to separate from each other. Therefore, for the flows labeled as C and C, the tables combine (sum) the emission from both the C and C flows. Squares mark the locations of seven Herbig-Haro objects listed in A General Catalogue of Herbig-Haro Objects (B. Reipurth 1994 [available via anonymous ftp to ftp://ftp.hq.eso.org/pub/Catalogs/Herbig-Haro]). The axes are arcminute offsets with respect to the position of a star located at (1950) = 14h56m249, (1950) = -62°5459 close to the millimeter-wavelength continuum source. (c) The 13CO map showing the anomalous velocity components southeast of vBH 65a superimposed on the optical image. The velocity range of the integration is vlsr = -10 to -8 km s-1. Contour levels are shown at intervals of 0.5 K km s-1 from 0.5 to 1.5 K km s-1. (d) A C18O map of the Circinus cloud superimposed on the optical image. Same as Fig. 2a but for C18O. Levels plotted from 0.25 to 2.5 K km s-1 at intervals of 0.25 K km s-1. Solid circles mark the locations of IRAS sources. The straight lines mark the location and orientation of the flows discussed in the text. Two lines (marked C and C) indicate two sets of overlapping outflow lobes that are hard to separate from each other. Squares mark the locations of seven Herbig-Haro objects listed in A General Catalogue of Herbig-Haro Objects.

     Neckel & Klare (1980) investigated the extinction toward this line of sight, finding an abrupt increase of A_V = 0.6 mag at a distance of about 170 pc and a larger jump with A_V greater than 2.0 mag toward stars with distances estimated to lie between 600 and 900 pc. The wall at 170 pc is unlikely to be sufficiently opaque to be associated with the Circinus complex. On the other hand, the low density of foreground stars toward the cores makes it likely that the cloud is closer than 900 pc. We will adopt a distance of 700 pc for use in estimating parameters for the Circinus cloud and its outflows while recognizing that this value is uncertain by nearly a factor of 1.5.

     CO emission from the Circinus cloud complex was first identified in the survey of the southern Milky Way by Dame et al. (1987). A 2° × 6° CO cloud centered at velocity v_lsr = -6 km s^-1 coincides with the region of optical obscuration. Using the integrated emission in the Dame et al. (1987) survey and a conversion factor of N(H₂) = 2.6 × 10²⁰I(CO) yields an estimated mass of about 4.7 × 10⁴d M for the mass of this entire complex, where d₇₀₀ is the distance in units of 700 pc.

     Figures 2b and 2c show velocity-integrated ¹³CO maps showing emission between v_lsr = -8 to -4 km s^-1 and v_lsr = -10 to -8 km s^-1, respectively, superimposed on the optical image of the Circinus cloud. Figure 2d shows a velocity-integrated C¹⁸O map showing emission between v_lsr = -10 to -4 km s^-1 superimposed on the optical image of the Circinus cloud. Figure 3 shows a closeup of the main core in ¹³CO integrated from v_lsr = -8 to -4 km s^-1. The rare CO isotopes should approximately trace the column density of molecular gas, and the contour maps shown in Figures 2b and 2d closely match the region of high optical obscuration.

FIG. 3.A 13CO contour map of the main core that contains the millimeter-wavelength continuum source with levels plotted from 2 to 14 K km s-1 at intervals of 1 K km s-1. Objects are marked as in Fig. 2a.

     The ¹²CO data contains additional faint and narrow-line cloud components at v_lsr = -24.0, -9.5, -6.5, and +3.5 km s^-1 toward Circinus. The main components associated with the region of obscuration shown in Figure 1 correspond to the bright line centered at v_lsr = -6.5 km s^-1. None of the other ¹²CO features exhibit spatial correlations with the region of high optical obscuration and are therefore likely to trace low-opacity foreground (3.5 km s^-1) and background (v_lsr = -24.0, -9.5 km s^-1) cloud components.

     Table 1 lists estimates of the dimensions, masses, and densities of the various major quiescent components of the Circinus cloud complex. The properties of the outflows will be discussed in the next section. We use the SEST ¹³CO measurements to estimate the total mass contained within the mapped region, as well as the masses of various subcomponents. Throughout this analysis, we assume a constant excitation temperature of T_ex = 10 K, which is consistent with the peak brightness of the ¹²CO line in this region and N(H₂)/N(¹³CO) = 7 × 10⁵. We use the column density for each region to estimate the average surface density of H₂ over the area of integration (the mass surface density and number density estimates include the correction for the cosmic abundance of helium: = 1.36). We present the formulae used in estimating the masses and column densities in the Appendix. On large scales, the various cloud substructures are elongated in our maps. We assume that the typical line-of-sight spatial extent of each region is similar to the projected width of each structure along its minimum dimension. We divide this length into the average surface density of each structure to derive an estimate for the average number density of H₂ in each subregion.

TABLE 1     PROPERTIES OF CLOUD CORES IN THE CIRCINUS CLOUD

     This method probably underestimates the density and mass since the C¹⁸O:¹³CO ratios in the cloud cores range from 1:3 to 1:5, indicating that along some lines of sight the ¹³CO lines may be optically thick. On average, the spectra obtained toward each local cloud maximum imply about a factor of 25 higher column density than the average column density derived in Table 1. Furthermore, the excitation temperature in the core regions could be substantially different from our assumption of 10 K. Local heating by embedded sources may raise the value of T_ex, which would have the effect of increasing the column (and volume) density derived from our data. On the other hand, in the interior of the cloud, well away from external UV or local heating sources, where the bulk of the ¹³CO emission is expected to be produced, T_ex may be lower than either in the ¹²CO photosphere [where (¹²CO) 1] or near embedded stars. Since most of the mass in the cloud is located in the diffuse regions relatively far from local heat sources, these effects are not likely to substantially influence our mass and average density estimates.

     The portion of the Circinus complex shown in Figures 1, 2, and 3 contains 10 molecular outflows that can be discerned in the ¹²CO data. The most prominent flows lie in the main cloud core centered near the (0, 0) position in Figures 2 and 3. Figures 4a and 4b show close-ups of the main core in the ¹²CO line wings that trace the lobes of high-velocity gas emerging from this region. The complex morphology of overlapping redshifted (dotted lines) and blueshifted (solid lines) emission can be resolved into three or four bipolar outflows, with opposing lobes located symmetrically about two IRAS infrared sources embedded within the main cloud core. We discuss these prominent flows in detail.

FIG. 4.12CO maps of the Circinus cloud core. All coordinates are arcminute offsets with respect to the position of a star located at (1950) = 14h56m249, (1950) = -62°5459. (a) The highest velocity 12CO-emitting gas. Solid contours correspond to the blueshifted gas in the velocity interval vlsr = -16.0 to -11.0 km s-1. Dashed lines correspond to the blueshifted gas in the velocity interval vlsr = -1.0 to 4.0 km s-1. Contour levels are shown at intervals of 1 K km s-1 from 1 to 10 K km s-1. (b) The low-velocity 12CO-emitting gas that can just be distinguished from the cloud core emission. Solid contours correspond to the blueshifted gas in the velocity interval vlsr = -12.0 to -10.0 km s-1. Dashed lines correspond to the blueshifted gas in the velocity interval vlsr = -3.0 to -1.0 km s-1. Contour levels are shown at intervals of 1 K km s-1 from 1 to 5 K km s-1. (c) Very low velocity 12CO-emiting gas. Solid contours correspond to the blueshifted gas in the velocity interval vlsr = -10.5 to -9.0 km s-1. Dashed lines correspond to the blueshifted gas in the velocity interval vlsr = -5.0 to -3.0 km s-1. Contour levels are shown at intervals of 1 K km s-1 from 1 to 10 K km s-1.

     Flow A.The brightest 100 m infrared source in the Circinus cloud, IRAS 14564-6254, at the northern end of the CO core, lies along the axis and in between the lobes of a 10 long CO flow with an axis at a position angle (P.A.) of 80° measured from the source to the centroid of the blueshifted lobe. This flow is highly collimated with a major-to-minor axis ratio larger than 5 to 1. The blueshifted lobe is larger and better defined than the redshifted lobe. IRAS 14564-6254 coincides with a compact cluster of 1.3 mm wavelength continuum sources with a peak flux density of 510 mJy and an area-integrated flux density of about 2.5 Jy (Reipurth, Nyman, & Chini 1996).

     Flow B.A major north-south molecular outflow is centered on a CO subcondensation located about 65 farther south near the position of IRAS 14563-6301. Flow B is highly collimated along a north-south axis with the blueshifted lobe oriented toward P.A. = 177°. This flow has a major-to-minor axis ratio greater than 8 to 1 and a total length of at least 12. The redshifted lobe, centered near (-1.2, -3), is considerably stronger than the blueshifted lobe. Toward its northern end, the redshifted lobe of flow B is confused with the redshifted lobe of flow A.

     Flows C and C.In addition to the above two rather prominent outflows, additional high-velocity features are centered close to and slightly north of the point of symmetry of flow B. These flow components are sufficiently confused with flow B and with each other that it is difficult to determine their sizes and the locations of their sources. The complexity of the emission indicates that at least two highly confused flows (C and C) exist in this region.

     The flow we designate as C originates from the same region as flow B, with a blueshifted lobe extending toward P.A. = 120°. Thus, the source IRAS 14563-6301 may be a binary, with one component driving flow B and the other driving flow C. The CO emission from flow C consists of a group of blueshifted clumps that extend toward the southeast. The redshifted counterflow is highly confused with the red lobe of flow B, but several knots of redshifted emission lying to the northwest of IRAS 14563-6301 and to the west of the well-defined B flow are also assumed to be associated with flow C. A low-velocity redshifted protrusion from the vicinity of IRAS 14563-6301 with a major axis close to P.A. = 120° is visible in Figure 4b and may also be associated with this flow. Mass estimates for the redshifted lobe of flow C given in Table 2 are broken into two distinct spatial regions to avoid the contamination of the much stronger B flow.

TABLE 2     PROPERTIES OF CO OUTFLOWS IN THE CIRCINUS CLOUD

     Figures 4a and 4b also show several redshifted and blueshifted knots to the north of flow C at P.A. = 135° that are blended with parts of the A, B, and C flows. The emission along this axis switches from redshifted to blueshifted near relative offsets of (-1, -3.5), which corresponds to (1950) = 14^h56^m161, (1950) = -62°5829. IRAS 14564-6258, which is detected at short IRAS wavelengths, lies near this location. Since this possible flow is highly confused with parts of flow C (and to a lesser extent with the redshifted lobe of the B flow), we label this candidate flow C in the figures. Flow C has a position angle of P.A. = 135°. Since the C and C flows are confused, the mass estimates in each velocity bin listed in Table 3 refer to the combined lobes of the C and C flows. However, in Tables 2 and 5, we have attempted to separate the total masses in the lobes of these flows, but the resulting estimates are highly uncertain because of source confusion.

TABLE 3     INCLINATION-CORRECTED PROPERTIES OF THE CIRCINUS CO OUTFLOW LOBES

TABLE 5     TOTAL MASS, MOMENTA, AND KINETIC ENERGIES CORRECTED FOR HIDDEN MASS

     The proposed four bipolar outflows (flows A through C) can account for most of the high-velocity emission from the main Circinus core. Each of these proposed outflows contains a redshifted and blueshifted lobe along a well-defined outflow axis. For the A, B, and C flows, the point of symmetry where the velocity shifts from the approaching to the receding lobe coincides with a bright and cool IRAS source with fluxes increasing toward longer wavelengths, which implies the presence of at least one Class I (or earlier) embedded young stellar object associated with each IRAS source. These sources are discussed in more detail in § 4. Additional outflow lobes may be hidden underneath these three or four dominant outflows.

     Flow D.The next flow to be recognized in the data is a small outflow associated with the source IRAS 14592-6311 embedded in the east-to-west ridge of molecular gas extending to the southeast of the main Circinus core. IRAS 14592-6311 and flow D are associated with the peculiar Herbig Ae/Be star vBH 65b. Four Herbig-Haro objects, HH 140 through HH 143, lie within a few arcminutes of this region (Ray & Eislöffel 1994).

     Flow E.Reipurth et al. (1993) reported 1.3 mm continuum emission from two other sources in the main Circinus core. The Class I IRAS 14563-6250, which is located near HH 76 at offset coordinates (-0.7, +4.9), has a 1.3 mm flux density of 153 mJy. A compact redshifted lobe of emission lies to the northeast, and a tongue of low-velocity blueshifted emission that extends to the southeast and blends with the outflow lobes of the A flow is associated with this region. This pair of lobes is located symmetrically about the IRAS source and form a compact CO outflow at P.A. = 220°.

     Flow F.The star vBH 65a is associated with a compact reflection nebula and a small optical jet, HH 139, that extends along the axis of symmetry of the reflection nebula at P.A. = 100° toward the east. This star is associated with a Class I source, IRAS 14568-6304, and a 1.3 mm continuum source with a flux density of 66 mJy. Although neither Figure 4a nor 4b shows a prominent outflow here, very low velocity CO lobes are visible between v_lsr = -11 and -8 km s^-1 on the blueshifted side of the cloud core and v_lsr = -5 and -3 km s^-1 on the redshifted side of the cloud core. These compact lobes show elongation along the axis of symmetry of the reflection nebula, with the blueshifted gas lying toward the east and redshifted gas lying toward the west.

     Flow G.The source IRAS 14562-6246 lies about 2 north of IRAS 14563-6250 and is associated with an outflow very similar in size, appearance, and orientation to flow E. This outflow has a position angle of P.A. = 210°. The E and G flows are hard to separate from the relatively broad core of the ¹²CO spectra in the data cube, and as a result, their spatial extents are very difficult to determine. Their highest velocity emission protrudes only a few km s^-1 beyond the edge of the general ¹²CO emission of the main Circinus core. Thus, the mass estimates given in Tables 2 and 5 are highly uncertain and may be in error by several factors of 2. Furthermore, the corrections for hidden mass, total momentum, and kinetic energy listed in Table 5 are even more uncertain and should be treated as order-of-magnitude estimates only. To determine a best estimate for the amount of mass hidden behind the cloud core, we assume that the missing mass correction scales like that for the prominent lobes of the A, B, and C flows (see below).

     Flow H.At the southeastern corner of the mapped field, we found subtle high-velocity wings associated with IRAS 14596-6320.

     Flow I.Finally, between vBH 65b and the main Circinus core lies a pair of infrared sources, IRAS 14580-6303 and IRAS 14582-6305. Subtle high-velocity lobes are associated with the first and brighter IR source. A small but prominent cavity can be seen in Figure 1 that extends due west of IRAS 14580-6303. However, the orientation of the CO flow is hard to determine because of the small extent of our ¹²CO map near this source and the presence of high-velocity gas throughout the mapped field. Using the orientation of the maximum velocity gradient in the line wings, we estimate that the orientation of the blueshifted lobe is about P.A. = 300°, but this figure is uncertain by at least 30°.

     Figure 5 shows spatial-velocity cuts at several position angles through the complex of outflows. Figure 5a is centered on IRAS 14564-6254, and it illustrates the kinematics of flow A. This figure displays that the full width at 0.25 K is about 20 km s^-1. Figure 5b shows a north-south cut through flow B, which has a full width at 0.25 K of about 24 km s^-1. The A and B flows are confused on the right-hand side of this diagram. Figure 5c shows a cut along the axis of flow C and Figure 5d a cut along the proposed axis of flow C. Finally, Figure 5e shows a cut through the ¹³CO data cube that illustrates the typical line width of the Circinus cloud in this isotope.

FIG. 5.12CO spatial velocity maps of the Circinus cloud. (a) Flow A at P.A. = 80° centered on IRAS 14564-6254. Contour levels are shown at intervals of 0.25 K from -0.5 to 5.75 K. (b) Flow B at P.A. = 180°. Contour levels are shown at intervals of 0.25 K from -0.5 to 5.5 K. (c) Flow C at P.A. = 120°. Contour levels are shown at intervals of 0.25 K from -0.25 to 5.0 K. (d) Flow C at P.A. = 135°. Contour levels are shown at intervals of 0.25 K from -0.25 to 5.0 K. (e) A 13CO spatial velocity map from (25, -16) to (-10, -8) that illustrates the velocity structure of the core emission and shows the anomalous velocity component southeast of vBH 65a. Contour levels are shown at intervals of 0.25 K from -0.25 to 5.25 K.

     Figure 6 shows the spectra of all observed CO isotopes at the positions of the sources of the outflows discussed above.

FIG. 6.Spectra of 12CO (thin line), 13CO (medium line), and C18O (thick line) emission toward the sources of the A, B + C, C, D, E, F, G, H, and I flows obtained by forming an average over a 1 diameter region centered on each associated IRAS source.

     Our data reveal ¹²CO emission-line components at several radial velocities. These features limit our ability to analyze the structure of bipolar CO outflows emerging from the Circinus cloud to velocity ranges free from contamination by the background or foreground features. Table 2 lists the properties derived from the analysis of the CO line wings from the four outflows over the velocity ranges where contamination from gas unrelated to the Circinus cloud is minimal. For the most prominent outflow lobes, Table 3 breaks this analysis into v = 1 km s^-1 wide radial velocity channels.

     We use two procedures to estimate the outflow masses, momenta, and energies, one well known, and the other new. The first procedure simply assumes that I(¹²CO)/I(¹³CO) is independent of the radial velocity of the gas and has a mean value determined by integrating the ¹³CO emission over the spatial and velocity extent of the ¹²CO emission from each lobe of an outflow in a velocity range that excludes the emission from the cloud core. In contrast, the second procedure fits a polynomial function to the velocity dependence of the I(¹²CO)/I(¹³CO) ratio in the average spectrum of each outflow lobe in the portion of the line profile that traces the high-velocity emission. This function is used to extrapolate the I(¹²CO)/I(¹³CO) ratio as a function of velocity into the far line wings, where the observed intensity of ¹³CO is too faint to be directly measured. Although these procedures produce similar total mass estimates, the resulting velocity dependences of the mass, momentum, and flow kinetic energy are different.

     The signal-to-noise ratio in the high-velocity outflow lobes is much higher in the ¹²CO data than in the ¹³CO data. Therefore, the ¹²CO spectra are used to estimate the mass in each outflow lobe and in each velocity bin. The area-averaged and velocity-integrated ¹³CO and ¹²CO intensities are measured over the spatial extents of each outflow lobe as determined from their extents in the ¹²CO maps. In the line wings (excluding the emission from the cloud core) the area- and velocity-averaged ratio, I(¹²CO)/I(¹³CO), is found to vary from about 12 to over 33 with a median value of 20, indicating that the outflow lobes are optically thick in ¹²CO with a mean optical depth of about 4. We use this constraint to estimate the mean opacity of the ¹²CO line in each lobe and to correct the ¹²CO-based mass estimates for optical depth. In practice, this is achieved by dividing the ¹²CO integrated fluxes (integrated over the area of each outflow lobe and the velocity interval of interest) by 20, the mean I(¹²CO)/I(¹³CO) ratio, and treating the resulting spectrum as if it were a ¹³CO spectrum (equivalent to estimating the ¹³CO spectrum from the ¹²CO data). The mass in the high-velocity lobe is computed from the scaled (by the mean ¹³CO/¹²CO ratio) ¹²CO flux integrated over the solid angle and velocity range of the outflow lobe using the same assumptions as for the analysis of the ¹³CO data for the quiescent cloud. Estimates of the mass, momentum, and kinetic energy of each major outflow lobe using this method are presented in Tables 2 and 3.

     For the most prominent outflow lobes, the analysis is done over a set of v = 1 km s^-1 wide radial velocity intervals from the highest velocity (with respect to the cloud core), where the line wings can be seen to the lowest velocity at which the line wings are free from contamination from the cloud core or from unrelated cloud components. The area- and velocity-integrated ¹²CO emission is scaled in the COMB data reduction package to produce the mass in each velocity and area interval. These are listed in Table 3.

     We assume that the flows are each inclined by an angle of i = 60 with respect to the line of sight (the statistically most probable angle for random flow directions), so that the observed radial velocity is exactly one-half of the true velocity. Thus, as an example, the mass observed in the radial velocity interval v_lsr = -16 to -15 km s^-1 is displaced from the cloud core velocity (v_lsr = -6.5 km s^-1) by an observed radial velocity difference of v_rad = 9 km s^-1. When corrected for the assumed inclination angle, this mass element has a physical velocity of v = v_rad/ cos i = 18 km s^-1. The momentum and kinetic energy associated with the motion of the mass in a given radial velocity range with a mean velocity displacement v is given by P = M(v)v and E = 0.5M(v)v². These inclination-corrected velocity differences are used to calculate P and E in Tables 2 and 3 for each calculated M(v), the mass observed in the specified velocity interval integrated over the solid angle indicated in Table 3. The total observed lobe momenta and kinetic energies listed in Table 2 are obtained by summing over the observed velocity ranges and are not corrected for the contribution of the velocity ranges over which the outflow lobes are hidden by the foreground and background emission from the cloud.

     The mass, momentum, and kinetic energy in each v = 2 km s^-1 wide velocity interval (corrected for inclination) can be represented as a power-law dependence on the flow velocity with the form M(v) = M₀v^-, P(v) = P₀v^-, and E(v) = E₀v^- over the velocity range of the measurements. The redshifted and blueshifted lobes have slightly different power-law indices. The mean power-law indices (averaged over the five best defined lobes) are = 2.7 ± 0.2, = 1.7 ± 0.2, and = 0.7 ± 0.2.

     We use our power-law relations derived above to estimate the amount of mass, momentum, and kinetic energy hidden by the bright emission associated with the quiescent cloud. The masses, momenta, and kinetic energies in Tables 2 and 3 were estimated over a range of inclination-corrected radial velocities extending from close to the redshifted and blueshifted edges of the ¹³CO and C¹⁸O lines, where the particular outflow lobes become difficult to distinguish from the wings of the ¹²CO line (v₂; this and the other v parameters refer to the difference between the inclination-corrected velocities of a fluid element and the line centroid velocity of the cloud core), and the velocity where the ¹²CO emission is lost in the noise (v₃). In the A, B, and C + C^' flow lobes, v₂ = 6 km s^-1 on the redshifted side and v₂ = 8 km s^-1 on the blueshifted side of the core and v₃ = 18 km s^-1. We seek to estimate the amount of mass, momentum, and kinetic energy that lies hidden between v₂ and the lowest velocities where a parcel of gas is likely to be still considered as part of the outflow, (v₁) set by the turbulent velocity of the core. We estimate v₁ from the FWHM width of the ¹³CO and the width at the base of the C¹⁸O lines, finding that in the A, B, and C + C^' lobes, v₁ = 2.0 km s^-1 (see Fig. 6).

     The ratio of total mass in the interval v₁ to v₃ to the observed mass in the interval v₂ to v₃ is given by



For the above values of v₁, v₂, and v₃ and a mass power-law index = 2.7, the best estimate of the true mass of gas in each outflow lobe is about 10 times greater than the observed mass (M_obs) in the velocity range between v₂ and v₃ listed in Table 2. The correction factor for the momentum is about 5 and for kinetic energy is about 2.2. Applying these corrections for hidden mass to the lobe parameters in the top of Table 2 leads to the total mass, momentum, and kinetic energy estimates listed in Table 5.

     The second, new method of analysis to estimate the masses in each outflow lobe is based on the realization that the ¹³CO observations also can trace the low-velocity inner wings of the outflows. Averaged over the ¹²CO-determined spatial extents of the redshifted and blueshifted lobes of the A, B, and C + C^' flows, the ¹³CO emission can be used to estimate lobe masses wherever the lobe-averaged ¹³CO emission-line exceeds 1 above the rms noise for the averaged spectrum of the flow in question (0.012 K for flow A). Between the velocity where this criterion is met and the velocity where the emission from the turbulent cloud core dominates the line profile, we estimate the lobe column density and mass by assuming that ¹³CO is optically thin.

     At velocities where the ¹³CO emission, integrated over the spatial extent of each outflow lobe as determined from the ¹²CO emission, falls below the level where it can be reliably measured (below 0.012 K for flow A), the ¹²CO data must be used for column density and mass estimation. However, the ¹²CO emission must be corrected for optical depth. Instead of assuming that the isotope ratio is constant and independent of velocity as in the first mass estimation method, we extrapolate the velocity dependence of the isotope ratio from the inner line wings where its velocity dependence can be reliably determined. A second-order polynomial is fitted to the velocity dependence of the isotope ratio over the velocity regime where the ¹³CO line is reliably detected in each outflow lobe (see Fig. 7). Only those data that trace gas outside the line core and with W(¹³CO) = I(¹³CO) and W(¹²CO) = I(¹²CO) greater than their respective rms noises are used in the polynomial fit. The resulting fit, R_12/13(v), is used to extrapolate the isotope ratio into the velocity regime where only the ¹²CO line is reliably detected. However, for the purpose of extrapolation, this function is truncated when it reaches a value of 89, the assumed intrinsic ¹²CO to ¹³CO isotopomer ratio. We thus replace the observed ¹³CO line profile with the estimated ¹³CO profile obtained by calculating the quantity T(¹²CO)/R_12/13(v) for those velocities where ¹²CO is reliably detected but the ¹³CO line falls in the rms noise criterion.

FIG. 7.Power-law fits to the 12CO/13CO intensity ratios averaged over the respective areas of each of the five most prominent outflow lobes. Diamonds and crosses are used to denote the line wings in each region (blueshifted or redshifted). The solid curves show the polynomial fits to the observed isotope ratio averaged over the spatial extent of the dominant outflow lobe in each region. The lower right-hand panel shows the isotope ratio for the entire outflow complex in the core region, combining both lobes of the A, B, and C + C' outflows. This polynomial fit to the isotope ratio is shown as the dashed curve in the other panels. The pair of vertical dashed lines in each panel show the velocity extent of the emission from the Circinus cloud core.

     Results of this second procedure are also tabulated in Tables 2, 3, and 4. Power-law fits to the mass as a function of velocity in each outflow lobe are shown in Figure 8. Only those points that do not lie within the cloud core velocities (-8 to -5 km s^-1) are used in the fit. For each lobe, the mass functions for both redshifted and blueshifted gas are plotted, but the more prominent wing is used to label each outflow lobe as red- or blueshifted.

TABLE 4     IRAS SOURCES IN THE CIRCINUS CLOUD

FIG. 8.Plots of the mass (M) in 2 km s-1 wide velocity intervals as a function of the deprojected flow velocity () derived using the assumption that the 12CO/13CO varies as a function of velocity as determined in Fig. 7. Symbols are as in Fig. 7. The solid lines are fits to the data point shown. See text for detailed discussion.

     Several dozen IRAS sources located toward the Circinus complex are listed in Table 4. The brightest 100 m sources are associated with CO outflows. The driving source of flow A is IRAS 14564-6254, which was observed at 1300 m by Reipurth, Nyman, & Chini (1996). They clearly resolved the object into four distinct sources, suggesting that a small stellar aggregate is being formed there. An H emission star is located close to the IRAS position (star 5 in Mikami & Ogura 1994), but this object is almost certainly unrelated to the IRAS or 1300 m sources. The B and C flows have axes that intersect almost precisely at their centers of symmetry, and at that location we find the Class I source IRAS 14563-6301. This suggests that the source is multiple, with at least two components simultaneously driving outflows. A chain of redshifted and blueshifted lobes that is confused with the C flow and with the redshifted lobe of the B flow may delineate a flow (C) centered at (-1, -3.5). At least three more much fainter flows (E, F, and G) are associated with IRAS sources and are nearly lost behind the overlapping lobes of the stronger outflows and the broad emission from the relatively turbulent main Circinus core. IRAS 14563-6250 (near the Herbig-Haro object HH 76) drives flow E, and IRAS 14568-6304 (near vBH 65a and HH 139) drives a very faint flow F. Both IRAS sources are also associated with compact 1.3 mm continuum sources. Flow G lies several arcminutes north of flow E and is very similar to it in size, appearance, and orientation. Three additional flows lie east and south of the main core. The source of flow D, IRAS 14592-6311, coincides with a visible star with bright reflection nebulosity, vBH 65b (van den Bergh & Herbst 1975), which is associated with four HH objects (Ray & Eislöffel 1994). Flows I and H are driven by IRAS sources embedded in a prominent dust condensation easily seen in Figure 1.

     It is interesting to note that a prominent and visually very opaque core near (14, -12) [(1950) = 14^h58^m287, (1950) = -63°0659] neither harbors a known IR source nor exhibits ¹²CO line wings indicative of an outflow. Its high opacity, compact size, and strong C¹⁸O and ¹³CO emission indicate that it is very dense. The integrated line intensity is 4.6 K km s^-1 in ¹³CO and 1.0 K km s^-1 in C¹⁸O, a ratio characteristic of the densest cores in Circinus and indicating that ¹³CO may be optically thick. It is possible that this condensation is still in a precollapse phase and may therefore be an ideal candidate for the investigation of a cloud core in a prestar-forming state.

     We can estimate the timescale for star formation in Circinus, the number of stars formed in the Circinus complex, and the cloud star formation efficiency from the observed outflow and cloud parameters. The field included in the ¹³CO SEST map is estimated to have a mass of about 900 M (see Table 1) and a turbulent line width of v = 2.5 km s^-1 estimated by averaging all our ¹³CO spectra. If the product of these two numbers, P_cloud = 2.2 × 10³ M km s^-1, is the result of acceleration of the cloud by the radiative shocks powered by N outflows, the number of outflows required to generate the observed motions is N = P_cloud/P_flow. The four most massive flows discussed here in detail (A, B, C, and C) dominate the mass, momentum, and energy injection into the cloud (see Tables 2 and 5). They have an average observed P_flow = 30 M km s^-1 in each flow. This value sets a lower bound since it is uncorrected for hidden mass and excludes the impact of the smaller flows. Table 5 lists estimates for this quantity corrected for hidden mass, and these values imply P_flow = 200 M km s^-1. Thus, this range of values for P_flow implies that between 11 and 73 stars similar to the driving sources of the three or four most massive flows are required to have formed in the portion of Circinus mapped with the SEST to produce the observed internal motions. When we include lower mass stars that power some of the weaker flows, the estimated number of stars required to have formed may range from 25 to over 100 stars. If the median stellar mass associated with the more powerful flows is 1 M, the implied star formation efficiency in this portion of the Circinus cloud, _SFE = M_cloud/M_*, ranges from 1.2% to 8% when only the most massive stars (such as those that power the A, B, C, and C flows) are considered, or _SFE = 5%20% when the sources of all 10 outflows are considered. If the three or four most massive flows discussed represent a typical steady state, and outflows have a lifetime of 10⁵ yr (Bally & Lada 1983), then the duration of this star formation activity is about _SF = N_*flow/N_flow (0.52) × 10⁶ yr for N_* = 25100 stars, and N_flow = 5 active flows present at any one time. Since some of the outflows may have blown clear of the cloud and deposited their flow energy in the surrounding lower density intercloud medium, this estimate is a lower limit. However, if the outflows do blow out of the parent cloud, the cavity walls left behind still absorb a substantial fraction of the energy of the flow and, at most, blowout will increase the required number of stars and the star formation efficiency by about a factor of 2.

     Some of the weaker flows may be much older than the most active flows. For example, flow F appears to be associated with a fairly massive star, vBH 65a, which may be similar in mass to the sources of the more prominent flows. However, this star, and its outflow, may be much older than the A, B, C, and C flows since the star is visible, has associated visible HH objects, and may have formed a large cavity devoid of molecular gas. Therefore, the dynamic ages of some of the weaker flows may underestimate the true age of the source star or its outflow.

     Mikami & Ogura (1994) found 14 H emission-line stars toward the portion of the Circinus clouds covered by our survey. B. Reipurth (1998, private communication) has identified additional but fainter emission-line stars. These stars provide evidence for previous star formation activity, and when these sources were embedded young stars, they probably produced flows similar to those now observed. As the outflow from each source subsides, the ejected and entrained gas decelerates in momentum-conserving interactions with the ambient medium to form a dense shell of swept-up gas surrounding cavities of low-density material. The prominent dust filaments may correspond to the fossilized walls of previous generations of outflows, some of which may have been powered by the visible young stars in this field.

     Figures 2 and 3 show that the dust filaments correspond to prominent features visible in the ¹³CO and C¹⁸O maps. Thus, these structures have relatively high column densities, with N(H₂) ranging from 10²¹ to 10²² cm^-2. Inspection of the radial velocity field in ¹³CO shows that most of these filaments have velocities of less than 1 km s^-1. Many of the filaments are about 1 pc long and about 0.10.2 pc wide and enclose cavities with dimensions of about a parsec. A good example of a prominent cavity lies along the axis of the HH 139 jet associated with vBH 65a. This cavity may be young compared with the other cavities evident in Figure 1, but old compared with the other active outflows in Circinus, since it is associated with an active stellar source, a jet, but only a very weak CO outflow.

     Our ¹³CO data shows a 2 × 5 clump of emission at v_lsr = -9 km s^-1 centered about 4 southeast of vBH 65a at (6, -12) and elongated along an axis pointing back toward this star. This feature coincides with a nearly north-south (P.A. = 165°) and opaque filament of dust at the southern periphery of the cavity that may by associated with vBH 65a. A second but fainter clump with a similar size and elongation is located at (11, -16). These features trace an anomalous velocity component that is blueshifted by about 3 km s^-1 with respect to the emission produced by the bulk of the ¹³CO emission from the Circinus complex. Nowhere else is a component at this velocity observed in the mapped field. One possible interpretation is that this anomalous ¹³CO emission component traces gas that was expelled a long time ago by an ancient outflow. If the source was vBH 65a, then the outflow orientation from this source must have changed by about 25°, the dynamical age of the feature centered on (6, -12) is about 2 × 10⁵ yr, and the age of the feature at (11, 16) is at least 5 × 10⁵ yr.

     For cavity dimensions and wall velocities of 1 pc and 12 km s^-1, the cavities are expected to survive for about 0.31.0 Myr before the random motions in the cloud distort their shapes or fill in their voids. Thus, the cavities have lifetimes about an order of magnitude longer than the dynamic ages of the visible CO outflows. Thus, if star formation has continued at a steady state for a significant fraction of a million years, there ought to be dozens of cavities and filaments that delineate cavity walls. This is consistent with the presence of dozens of prominent dust filaments in the Circinus cloud that may therefore trace the walls of cavities produced by old and long extinct fossilized outflows. Their numbers are consistent with the above predictions for the number of young stars produced by the Circinus cloud and with the number of observed H emission-line stars.

4.4. Emission-LineWing Power-Law Indices

     Several investigators have found that the high-velocity ¹²CO emission in the lobes of bipolar outflows can be characterized by a broken power law (cf. Lada & Fich 1996, 1998; Fich & Lada 1997; Shepherd et al. 1998; Bachiller 1996; Stahler 1994) with a mass index = 1.8 at low velocities and > 4 above the break. Zhang & Zheng (1997) have modeled the structure of the CO line wings in the optically thin limit for models of the molecular outflow in which the emitting gas is entrained from the surrounding medium by a bow shock. The resulting models produce broken power laws with an index of = 1.8 at low velocities and = 5.6 above the break, as observed in a variety of bipolar molecular outflows. One interpretation of this result is that the radial velocity where the break occurs is related to the location of the emitting gas with respect to the bow shock that entrains the surrounding medium. Gas below the velocity where the break occurs is associated with the low-velocity material entrained by the far bow-shock wings, while gas at higher velocities is associated with the mostly forward-moving head of the bow shock. However, in considering the application of the Zhang & Zheng models, it is important to remember that optically thin high-velocity emission is assumed. The line wings produced by the Circinus outflow lobes are demonstrably optically thick in the lower velocity portions of their lobes.

     The Circinus power-law indices derived by the first method of mass estimation are higher (by about 1.0 in the slope) than the power-law indices derived for the inner wings of many other bipolar outflows (Stahler 1994; Bachiller 1996) but considerably lower than the indices found for the highest velocity gas at velocities beyond the velocity where the power law breaks. The indices in Circinus imply that most of the mass of each outflow lobe lies at the lowest velocities that are likely to be hidden by the cloud core.

     The second method of mass estimation produces power-law indices that are much steeper than the first method, with indices that range from nearly 3.0 to slightly over 5.0, about 13 times steeper than the indices found by using the first mass estimation method. This difference is a direct result of the assumptions made about the variation of optical depth with velocity. In the second method, the polynomial fit results in rapidly decreasing line optical depth with increasing velocity, while in the first method, the optical depth is assumed to be independent of velocity.

     The second method produces power-law indices that approach the power-law indices found for outflows at velocities above the velocity where the power-law index breaks. This suggests that the broken power-law indices found for other outflows may be an artifact of the assumptions used in the analysis of the mass functions. The low power-law indices in the inner portions of the line profiles may trace those parts of the flow where the ¹²CO emission is optically thick, and the steeper, outer power-law indices may trace the parts of the outflow where the ¹²CO emission is optically thin. The momentum and kinetic energy power-law indices can be found by adding 1.0 or 2.0 to the indices shown in Figure 8. Future measurements of the velocity structure of the lobes of outflows with arcsecond angular resolution provided by interferometry are needed to clarify the origin of the observed power-law indices and of the broken power laws seen in some CO outflows.

     For the fastest visible gas (v = 18 km s^-1; corrected for inclination), the dynamical ages of these flows ( = R/v) range from 1.4 × 10⁴ yr (flow D) to 6.7 × 10⁴ yr (flow B). On the other hand, the lowest observed velocities have dynamical ages about 23 times longer and gas that is just moving above the turbulent velocity of the core and remains hidden has a dynamical age about 410 times longer. As with other bipolar CO flows, these estimates are highly uncertain (to a factor of 23) and depend in detail on the nature of the flow dynamics that is assumed.

     The observed momentum and kinetic energy can be divided by the dynamical ages of each lobe to estimate the momentum injection rate and mechanical luminosity of each outflow. Using the inclination- and hidden masscorrected values in Table 5, and an assumed flow lifetime of about 10⁵ yr, implies that each of the three major Circinus outflows entrains mass at a rate 1 × 10^-3 M yr^-1, they inject momentum into the cloud at a rate 3.5 × 10^-3 M km s^-1 yr^-1, and the resulting mass motions have a mechanical luminosity of 1.5 L. The corrected outflow masses are comparable to the total masses estimated for the cloud cores in which the sources are embedded. This implies that at the presently observed stage, a large fraction of the core masses may be influenced by the mass flows driven by the young stellar objects.

     The Circinus cloud appears to have been severely modified by extensive star formation over the past few million years. The formation of dozens of stars has shredded and churned the cloud, producing dozens of fossil outflow cavities surrounded by dense filaments of compressed gas that may be responsible for the "Swiss cheese" appearance of the cloud. Even with low efficiency, star formation is capable of producing the observed chaotic motion and structure in the cloud.

     We have come to recognize that star formation is highly correlated and that even in low-mass clouds, stars form in very dense groups that frequently produce multiple overlapping outflow lobes. When our first Circinus data were obtained nearly 10 years ago, we did not have the fortitude or confidence to attempt to resolve the confusion of overlapping outflow lobes and line wings into individual outflow lobes. In part, we were hindered by the lack of efficient three-dimensional visualization tools needed to analyze the phase space of our observed data cubes. But more significantly, we lacked the experience that has recently been gained from the detailed investigation of environments such as OMC 2/3, NGC 1333, and Cepheus A, where dozens to hundreds of young stars produce a confusion of overlapping outflows. The arcsecond angular resolution of visual and near-infrared images was needed to resolve the confusion in these regions and has greatly aided in the deconvolution of overlapping outflow lobes in the millimeter-wavelength maps of these regions. These data have taught us to first identify the dominant flows, then to proceed to search for ever weaker and more subtle signs of additional but fainter flows, especially in the vicinity of very cool IRAS sources. It is this procedure that has finally enabled us to understand the structure of the Circinus outflow complex.

     We thank David Theil for assistance in the preparation of the figures.

     Here, we provide the formulae that describe the procedures used in estimating column densities and masses from the data. The column density is given by (cf. Margulis & Lada 1985)



where the excitation temperature T_ex is given by



with T₀ (=h/k) = 5.3 K for the ¹³CO line. The mass is given by



where = 1.36 is the mean molecular weight of hydrogen corrected for helium and other trace constituents, m_H is the mass of hydrogen, N(H₂) is the H₂ column density, A is the spatial area of the flow (where A = 4.68 × 10⁴²dA_sr pc², d₁₀₀ is the distance to Circinus in units of 100 pc, and A_sr is the area of the flow in steradians), and N(H₂) = 7 × 10⁵N(¹³CO) cm^-2.

     When only the ¹²CO data are available, we estimate masses from this line by estimating the equivalent ¹³CO line using the assumed ¹²CO/¹³CO ratio (using either of the methods described in the text; this is equivalent to correcting the ¹²CO-based column density or mass estimate for optical depth). The estimated ¹³CO profile is obtained by calculating the quantity T(¹²CO)/R_12/13(v) for those velocities where ¹²CO is reliably detected but the ¹³CO line falls in the rms noise criterion. Thus, ¹³CO column density estimated from the ¹²CO profiles is given by

• Bachiller, R. 1996, ARA&A, 34, 111 First citation in article | Crossref | ADS
• Bally, J., Devine, D., & Reipurth, B. 1996, ApJ, 473, L49 First citation in article | IOPscience | ADS
• Bally, J., & Lada, C. J. 1983, ApJ, 265, 824 First citation in article | Crossref | ADS
• Bence, S. J., Richer, J. S., & Padman, R. 1996, MNRAS, 279, 866 First citation in article | Crossref | ADS
• Dame, T. M., et al. 1987, ApJ, 322, 706 First citation in article | Crossref | ADS
• Devine, D., Bally, J., Reipurth, B., & Heathcote, S. 1997, AJ, 114, 2095 First citation in article | Crossref | ADS
• Eislöffel, J., & Mundt, R. 1997, AJ, 114, 280 First citation in article | Crossref | ADS
• Fich, M. & Lada, C. J. 1997, ApJ, 484, L63 First citation in article | IOPscience | ADS
• Lada, C. J., & Fich, M. 1996, ApJ, 459, 638 First citation in article | Crossref | ADS
• . 1998, ApJS, 117, 147 First citation in article | IOPscience | ADS
• Margulis, M., & Lada, C. J. 1985, ApJ, 299, 925 First citation in article | Crossref | ADS
• Mikami, T., & Ogura, K. 1994, MNRAS, 270, 199 First citation in article | Crossref | ADS
• Neckel, T., & Klare, G. 1980, A&AS, 42, 251 First citation in article | ADS
• Ray, T.P., & Eislöffel, J. 1994, A&A, 290, 605 First citation in article | ADS
• Reipurth, B., Bally, J., & Devine, D. 1997, AJ, 114, 2708 First citation in article | Crossref | ADS
• Reipurth, B., Chini, R., Krügel, E., Kreysa, E., & Sievers, A. 1993, A&A, 273, 221 First citation in article | ADS
• Reipurth, B., & Graham, J. H. 1988, A&A, 202, 219 First citation in article | ADS
• Reipurth, B., Nyman, L. Aa., & Chini, R. 1996, A&A 314, 258 First citation in article
• Shepherd, D. S., Watson, A. M., Sargent, A. I., & Churchwell, E. 1998, ApJ, 507, 861 First citation in article | IOPscience | ADS
• Stahler, S. W. 1994, ApJ, 422, 616 First citation in article | Crossref | ADS
• van den Bergh, S., & Herbst, W. 1975, AJ, 80, 208 First citation in article | Crossref | ADS
• Zhang, Q., & Zheng, X. 1997, ApJ, 474, 719 First citation in article | IOPscience | ADS