The Molecular Outflow from R Mon^*

R Mon was mapped on 2018 December 8 in good weather conditions (precipitable water vapor, pwv = 0.71 mm) in ¹³CO(3–2) and ¹²CO(3–2) using the 7 pixel LAsMA array on APEX in Chile (Güsten et al. 2006). LAsMA is a 7 pixel single polarization heterodyne array tunable from 270–370 GHz. The array is arranged in a hexagonal configuration around a central pixel with a spacing of about two beam widths between the pixels. It uses a K mirror as de-rotator. The backends are the new Fast Fourier Transform fourth Generation Spectrometers (Klein et al. 2012) with a bandwidth of 2 × 4 GHz. We observed ¹²CO(3–2) and ¹³CO(3–2) by placing ¹²CO in the upper sideband and ¹³CO in the lower sideband. The mapping was done in total power on-the-fly mode using a clean reference position, +120'', −2050'' relative to R Mon. We mapped a region of 65 × 72. The map was scanned in both R.A. and decl. with a spacing of 6'', resulting in a uniformly sampled map with good fidelity. The data were reduced using CLASS.⁵ All spectra are calibrated in T_mb, see Table 1.

Table 1.  Observation Details: Frequencies, Receivers, Half Power Beam Widths (HPBWs), Beam Efficiencies, and rms Sensitivities

Receiver	Transition	Frequency	θHPBW	ηmb	rmsa
		(GHz)	( '' )		(K)
PI230	CO(2–1)	230538.000	27.3	0.70	0.017
	13CO(2–1)	220398.684	28.5	0.70	0.014
	C18O(2–1)	219560.354	28.6	0.70	0.013
LAsMA	13CO(3–2)	330.587965	19.0	0.70b	0.15c
	CO(3–2)	345.795990	18.2	0.70b	0.13c
FLASH+	13CO(3–2)	330.587965	18.5	0.69	0.035
	CO(3–2)	345.795990	17.7	0.69	0.025
FLASH+	CO(4–3)	461.040768	13.3	0.58	0.14

Notes.

^aThe velocity resolution is 0.5 km s⁻¹ for all spectra. ^bAverage for all pixels. ^cRms per resolution element, 91 × 91.

Download table as:  ASCIITypeset image

Long integration spectra (16 minutes, on+off) of CO(2–1), ¹³CO(2–1), and C¹⁸O(2–1) in good weather conditions (pwv = 2.2 mm) were obtained on 2018 December 10 with the PI230 receiver on APEX. We observed two positions: one centered on R Mon and a second one on the bright far-infrared (FIR) peak at offset −40'', +58'' on the western side of NGC 2261. PI230 is a dual polarization, dual sideband receiver with good image band rejections. Each band has a bandwidth of 8 GHz and therefore each polarization covers 16 GHz. Each band is connected to two FFTS4G spectrometer (Klein et al. 2012) with 8 GHz bandwidth and 131,072 channels, providing a frequency resolution of 61 kHz, i.e., a velocity resolution of ∼0.079 km s⁻¹ at 230 GHz. The PI230 receiver can be set up to simultaneously observe CO(2–1), ¹³CO(2–1), and C¹⁸O(2–1). Observations details are summarized in Table 1.

Long integration spectra (20 minutes) of CO(3–2), ¹³CO, and CO(4–3) were obtained with FLASH⁺ on APEX on 2018 December 12 of the same positions observed with PI230. The weather conditions for the CO(4–3) observations were marginal. Therefore these observations were repeated on 2019 April 18 in dry weather conditions (pwv = 0.8 mm). The integration time for the April observations was 20 minutes. FLASH⁺ is a dual channel heterodyne SIS receiver operating simultaneously—on orthogonal polarizations—in the 345 GHz and in the 460 GHz atmospheric windows. Both bands employ state-of-the-art sideband separating SIS mixers. The mixers provide large tuning ranges enabling us to simultaneously observe ¹³CO and ¹²CO(3–2) in the 345 GHz window and ¹²CO(4–3) in the 460 GHz window. The backends are FFTS4G spectrometers, proving a velocity resolution of 0.053 km s⁻¹ at 345 GHz and 0.040 km s⁻¹ at 461 GHz. The CO(3–2) spectrum is an average of the December and April observations. Observation details are summarized in Table 1.

We retrieved Herschel Space Observatory PACS and SPIRE imaging of R Mon from the Herschel data archive. The SPIRE imaging was obtained in Small Map mode with a scan speed of 30''/s on 2013 March 24 (OD 1411; AOR-ID 1342268324). The PACS imaging was done in cross scan mode with medium scan speed, 20''/s, for both 70/160 μm (AOR-IDs 1342269841 & 1342269842), and 100/160 μm (AOR-IDs 1342269843 & 1342269844). These observations were carried out on 2013 April 11 (OD 1428). Both the SPIRE and PACS observations were part of an OT1 program by G. Meeus (OT1_gmeeus_1). The half power beam widths (HPBWs) for PACS are 56, 68, and 113 for 70, 100, and 160 μm, respectively. For SPIRE the HPBWs are 184, 252, and 367 for the 250, 350, and 500 μ bands, respectively.

Cantó et al. (1981) mapped the R Mon region in CO(1–0) and found that R Mon is located in a small elongated molecular cloud, ∼0.8 × 0.4 pc. Based on their observations they argued that the stellar wind of R Mon has created a bipolar cavity, with blueshifted emission to the north and redshifted emission to the south. In their interpretation the conical nebula NGC 2261 is shining in light reflected from the cavity walls of the northern outflow, a view which is commonly accepted. They also found another small cloud ∼7' to the north of R Mon, where the outflow becomes visible as the Herbig–Haro complex HH 39. This cloud has a velocity of ∼7.5 km s⁻¹, while R Mon has a V_lsr of 9.5 km s⁻¹, i.e., they are separate clouds.

Our high spatial resolution CO(3–2) and ¹³CO(3–2) maps do not go far enough to the north to show the northern cloud. The SPIRE images, however, capture most of the northern cloud and HH 39, see Figure 1. The SPIRE images confirm that there is a region free of dust emission between the R Mon cloud and the HH 39 cloud.

Our high spatial resolution SPIRE and CO (3–2) images show that the cloud in which R Mon is embedded has a filamentary structure with one narrow filament running from NE to SW through R Mon and NGC 2261. A second filament extends at least 35 to the south from R Mon (Figures 1 and 2).There is another faint filament to the east of the southern filament, which is outside the area mapped in CO, see Figure 1. ¹³CO(3–2) is less extended than the ¹²CO emission. The southern filament is not seen at all in ¹³CO, suggesting that this part of the cloud has lower column density. The Hubble nebula stands out as a cavity in both dust and CO emission, with the western side being rather prominent both in dust and ¹³CO, while CO(3–2) is stronger on the eastern side of the nebula. It is likely that the CO(3–2) emission may be enhanced by the outflow at velocities close to the systemic velocity of the R Mon cloud.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

One can see from the three-color image of the R Mon cloud (Figure 3) that the Hubble variable nebula is hotter than the surrounding cloud, which appears to have a relatively uniform temperature. To further characterize the cloud and the reflection nebula, we have done simple graybody fits (see e.g., Sandell 2000 and Section 4.2) using flux densities deduced from the PACS and SPIRE images. The flux densities for the cloud surrounding R Mon were integrated over the same area we mapped in CO(3–2) and ¹³CO(3–2) after subtracting emission from the reflection nebula. We also did a graybody fit of the warmer dust emission associated with the reflection nebula after subtracting out the emission from R Mon (Table 2). The graybody fit to the flux densities of the reflection nebula gives a dust temperature, T_d = 31 K, and a dust emissivity. β = 1.38, resulting in a mass (gas + dust) of 2.6 ${M}_{\odot }$, assuming standard Hildebrand dust opacity. For the surrounding cloud we derive T_d = 18.7 K, β = 1.49, and a mass of 66 ${M}_{\odot }$, or ∼70 ${M}_{\odot }$ in total.

Zoom In Zoom Out Reset image size

The long integration spectra toward R Mon were fit with two-component Gaussians, while only one velocity component was needed for the FIR peak. The results of these Gaussian fits are given in Table 3. Using the equations given in Nishimura et al. (2015) we can now use these results to derive gas temperatures and column densities of CO and its isotopologues. Nishimura et al. only discuss CO(2–1) and its isotopologues, but we have expanded them to higher J-transitions of CO using the generalized derivations for how to calculate column densities presented in Mangum & Shirley (2015). CO(2–1) gives a gas temperature of 20 K for the cloud emission both toward R Mon and the FIR peak. This is in good agreement with the dust temperature, 18.7 K, derived from the graybody fit. We find that ¹³CO(2–1) is significantly optically thick at both positions. The ¹³CO(2–1) optical depth of the molecular cloud is 0.74 toward the FIR peak and 0.76 toward R Mon. C¹⁸O(2–1) is essentially optically thin with optical depths of 0.05 and 0.07 toward R Mon and the FIR peak, respectively. ¹³CO(3–2) is optically thick in the dense compressed gas surrounding the outflow. Toward R Mon the optical depth is 1.55, although it is possible that the ¹³CO peak emission could include some of the outflow. Toward the FIR peak the optical depth is 0.66, i.e., more similar to the ¹³CO(2–1) optical depth.

Table 3.  Gaussian Fits to Long Integration Spectra

Line	$\int {T}_{\mathrm{mb}}^{* }{dV}$	${T}_{\mathrm{mb}}^{* }$	VLSR	ΔV
	(K km s−1)	(K)	(km s−1)	(km s−1)
(0'', 0'')
CO(4–3)	28.59 ± 1.26	9.46	9.27 ± 0.02	2.84 ± 0.06
	15.62 ± 1.20	2.03	9.68 ± 0.10	7.23 ± 0.44
CO(3–2)	28.14 ± 0.11	10.97	9.36 ± 0.00	2.41 ± 0.01
	9.54 ± 0.14	1.43	9.48 ± 0.02	6.29 ± 0.00
CO(2–1)	32.77 ± 0.05	14.98	9.35 ± 0.00	2.06 ± 0.01
	6.28 ± 0.08	1.20	9.35 ± 0.02	4.85 ± 0.01
13CO(3–2)	10.18 ± 0.02	6.58	9.35 ± 0.00	2.41 ± 0.01
	3.11 ± 0.05	0.77	9.34 ± 0.04	3.77 ± 0.11
13CO(2–1)	7.99 ± 0.02	6.41	9.47 ± 0.00	1.17 ± 0.00
	3.44 ± 0.01	1.43	9.50 ± 0.00	2.26 ± 0.02
C18O(2–1)	0.69 ± 0.00	0.72	9.55 ± 0.01	0.89 ± 0.01
	0.26 ± 0.02	0.11	9.51 ± 0.09	2.26 ± 0.19
(−40'', +58'')
CO(4–3)	15.73 ± 0.11	8.94	9.16 ± 0.01	1.65 ± 0.01
CO(3–2)	19.78 ± 0.04	11.16	9.19 ± 0.00	1.67 ± 0.00
CO(2–1)	25.96 ± 0.02	14.75	9.09 ± 0.00	1.65 ± 0.00
13CO(3–2)	6.77 ± 0.03	6.55	9.04 ± 0.00	0.97 ± 0.01
13CO(2–1)	7.79 ± 0.01	7.15	9.11 ± 0.00	1.02 ± 0.00
C18O(2–1	0.86 ± 0.01	1.12	9.11 ± 0.00	0.73 ± 0.01

Download table as:  ASCIITypeset image

Since we have mapped the cloud in ¹³CO(3–2), we can estimate the mass of the cloud seen in ¹³CO. Integrating the ¹³CO(3–2) map over the central 2 km s⁻¹ (i.e., the map shown in the top panel of Figure 2) we get 65,320 K km s⁻¹× arcsec². Of this emission ∼35% is optically thick, essentially from the fourth contour up in Figure 2, with an average optical depth of ∼0.5. We estimate the mass for optically thin emission with an excitation temperature of 20 K. By assuming a standard CO to H₂ abundance ratio of 10⁻⁴ and a ¹²C/¹³C isotope ratio of 50 (Pilleri et al. 2012), we derive a cloud mass of 67 ${M}_{\odot }$, after correcting the optically thick fraction of ¹³CO, i.e., the optical depth correction τ/(1 − e^−τ) (Goldsmith & Langer 1999), which gives a correction factor of 1.25. The total mass of the cloud estimated from ¹³CO(3–2) agrees remarkably well with what we estimated from dust emission.

Long integration spectra toward R Mon, (Figure 4, left panel) shows high-velocity wings even in C¹⁸O(2–1). The outflow velocities are modest, ∼3.5–15.5 km s⁻¹ in CO(2–1) and CO(3–2), and somewhat larger in CO(4–3), i.e., 0–18 km s⁻¹. This is because the molecular outflow is almost in the plane of the sky. Close et al. (1997) estimate an inclination of 20° ± 10°. The outflow wings in CO become brighter for higher J-transitions, which to some extent is due to better beam filling at higher J-transitions (smaller beam size), but mostly because the outflow is hot. Jiménez-Donaire et al. (2017) detected 26 CO transitions in PACS and SPIRE spectra of R Mon, the highest CO transition being CO(34–33), a clear indication of very hot gas. From a CO rotational diagram analysis they deduced the temperature components in R Mon: 949 ± 90 K, 358 ± 20 K, and 77 ± 12 K. Since the SPIRE spectrometer saw extended CO emission up to CO (8–7), they argue that the bulk of the emission must originate from the outflow and not the disk. The 77 K component is the one we see in our low J CO data. Alonso-Albi et al. (2018), who analyzed the same PACS and SPIRE data as Jiménez-Donaire et al., did not consider the outflow at all. Instead they incorporated the data into their disk model, which then required a large inner cavity.

Zoom In Zoom Out Reset image size

We also obtained long integration spectra of the FIR peak on the northwestern side the NGC 2261, i.e., at offset −40'', +58'' from R Mon. At this position the CO lines are very narrow without any hint of high-velocity wings (Figure 4). The only exception is CO(2–1), which shows a faint blueshifted wing, probably because the beam is broad enough to catch some emission from the R Mon outflow.

3.2.1. Outflow Morphology

Figure 5 shows the R Mon outflow over three different 1 km s⁻¹ wide velocity intervals: ±1.9 km s⁻¹, ±2.75 km s⁻¹, and ±3.75 km s⁻¹ from the cloud velocity overlaid on a SPIRE 250 μm image. At low velocities the blue outflow fills the reflection nebula, which is shaped by the outflow. At higher velocities the outflow becomes more collimated with a position angle close to 0°. The blueshifted outflow extends ∼160'' from R Mon. Position–velocity plots of CO(3–2) and ¹³CO(3–2) (Figure 6) show that essentially all the gas inside NGC 2261 is part of the blueshifted outflow from R Mon. There is some faint CO(3–2) emission from the cloud although it is rather diffuse, since it is not seen in ¹³CO, except close to R Mon. Although there is strong outflow emission on R Mon, there is very little high-velocity gas until ∼15''–20'' from the star, where the outflow becomes visible. Closer to the star, the outflow emission is dominated by the optical jet, which Movsessian et al. (2002) show extends up to 18'' north of R Mon. The outflow shows three peaks of enhanced emission at ∼50'', 80'', and 120'' from R Mon (Figure 5), suggesting that there may have been (possibly periodic) episodes of enhanced outflow activity.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

To the south of R Mon the outflow is far less pronounced and redshifted, almost certainly because the molecular cloud is rather diffuse. Even though we see the cloud in CO(3–2) and dust emission (Figure 2), it is not detected in ¹³CO(3–2). This shows that the molecular cloud south of R Mon is mostly low column density. Hence there is very little gas for the outflow to interact with. The redshifted outflow stops ∼120'' south of R Mon. It appears more wide-angled, and more eastward than the blueshifted flow. The p.a. for the red outflow is ∼168°, while the blueshifted outflow is close to 0°.

3.2.2. Outflow Properties

Previous estimates of the bolometric luminosity of R Mon (Harvey et al. 1979; Cohen et al. 1985; Natta et al. 1993) using ground-based near/mid-IR observations and FIR photometry from the Kuiper Airborne Observatory ranged from 700 to 860 ${L}_{\odot }$. Now there are much more accurate near-, mid-, and far-IR photometry available both from ground-based observations (Close et al. 1997), as well as from satellite mission like the Midcourse Space Experiment (MSX), Spitzer (Audard et al. 2007), the Wide-field Infrared Survey Explorer (WISE), and the Herschel Space Observatory. We have therefore assembled reliable photometry from the literature and the PACS and SPIRE images analyzed in this paper, see Table 2 and Alonso-Albi et al. (2009) for additional flux densities. Integrating the SED from 0.5 μm band to 1100 μm gives a total bolometric luminosity of 860 ${L}_{\odot }$, almost completely dominated by the emission in the near- and mid-IR. The FIR (30 μm–1 mm) portion of the luminosity, 150 ${L}_{\odot }$, is less than 20% of the total bolometric luminosity. Even if we assume that some of the far-UV radiation escapes the accretion disk/envelope surrounding R Mon, and use the flux densities within an 80'' radius around R Mon, we get a total bolometric luminosity of 960 ${L}_{\odot }$, an increase of only 100 ${L}_{\odot }$. At least some of this "extra luminosity" must arise from dust being heated externally by the general interstellar radiation field. Since more than 30% of the bolometric luminosity could come from the accretion disk (Natta et al. 1993), we conclude that R Mon cannot be an early B star. It simply does not produce enough luminosity to be one. It must be a late B star or even an early A star.

The mass of the accretion disk/envelope surrounding R Mon has been estimated to be in the range 0.1–0.24 ${M}_{\odot }$ (Natta et al. 1993; Mannings 1994; Sandell et al. 2011). Since there is now good photometry all the way into the FIR, we can do a better graybody fit than Sandell et al. (2011). We have accurate photometry from millimeter wavelengths to the FIR allowing us to simultaneously fit the cold outer envelope and warm inner core using an isothermal two-component model (see e.g., Sandell 2000; Howard et al. 2013). The least-squares fitting program allows us to put constraints on the dust temperature and size of the dust emitting regions, and fits simultaneously the dust temperature, source size, and the dust emissivity index, β. For the warm component we assume β = 1.5. In our fitting we use published millimeter and submillimeter photometry, FIR photometry from this paper (Table 2), and mid-IR photometry. We restrict the short wavelength end to 15 μm to avoid picking up hot dust from the inner accretion disk. We also omit mm-aperture synthesis data, because they filter out most of the emission from the extended envelope. The flux densities at millimeter wavelengths are corrected for free–free emission, as in Sandell et al. (2011). We find that the cold envelope has a dust emissity, β = 1.5, a temperature, T_d = 36 K, and a radius of 25 ± 1.2. The temperature of the inner warm envelope is 150 K within a radius of 05. The fit is shown in Figure 7. Assuming standard Hildebrand dust opacity (κ_{250 μm} = 0.1 cm² g⁻¹) and a gas-to-dust ratio of 100 (Hildebrand 1983), this corresponds to a total mass (gas + dust) of 0.34 ${M}_{\odot }$.

Zoom In Zoom Out Reset image size

Our observations show that the blue outflow from R Mon has a p.a. of ∼0° at high velocities, suggesting that the R Mon accretion disk is essentially e–w, i.e., at a p.a of 90°. This is also supported by the high-angular resolution polarization imaging by Close et al. (1997), who found a "polarization disk" of low polarization at p.a. ∼90° (see their Figure 3). Furthermore, the imaging of the [S ii] jet with the Multi-Pupil Fiber Spectrograph on the 6 m telescope of the Special Astrophysical Observatory (Movsessian et al. 2002) shows that the jet has a p.a. of ∼0°, i.e., similar to what we see in CO.

The commonly assumed orientation for the R Mon outflow, p.a. = 350° comes from the assumption that HH 39 is on the symmetry axis of the outflow, which does not seem to be the case. However, based on this assumption, Fuente et al. (2006) assumed that the R Mon accretion would have a p.a. = 80°. It is therefore not surprising that they see blueshifted emission east of R Mon and redshifted emission to the west, giving the appearance of Keplerian rotation. However, as we have shown in Section 3.2.2, the outflow emission is optically thick in CO and even somewhat optically thick in ¹³CO in the R Mon envelope, therefore completely hiding any emission from the underlying accretion disk.

Jiménez-Donaire et al. (2017), who analyzed Herschel PACS and SPIRE spectra, detected 26 CO lines, 6 OH lines, 4 H₂O lines, [C i], [C ii], and strong [O i] 63 and 145 μm emission toward R Mon. The ratio [O i] 63 μm/145 μm is 9.9. Such a low ratio excludes a photon-dominated region (PDR) origin for [O i] because the ratio is typically 20 or higher (Keenan et al. 1994; Kaufman et al. 1999). The line ratio could be higher if the 63 μm line is self-absorbed, which is not very likely in a low-luminosity system like R Mon. There is overwhelming evidence that the line emission in R Mon is shock excited: strong, broad, and variable H i lines, and in particular the strong broad forbidden [S ii] and [O i] 6300 Å lines have to be shock excited. The observed [O i] line ratio suggests a gas density of ∼10⁴ cm⁻³, if the main collision partner is atomic H (Nisini et al. 2015).

If we compare the observed PACS line intensities with the shock model predictions by Flower & Pineau des Forêts (2015), we find that the CO intensities agree well with a non-dissociative C-shock with shock velocities of 20–25 km s⁻¹ and pre-shock gas densities of 10⁴ cm⁻³. The observed OH intensities can still be explained with a C-shock although it requires higher shock velocities, 35 km s⁻¹, and higher pre-shock densities, 10⁴ cm⁻³. The observed OH lines can also be explained by a slow dissociative J-shock, 15 km s⁻¹, and high pre-shock densities, 10⁵. The observed [O i] intensities are far too high to be produced in a non-dissociative C-shock. They require a J-shock with a shock velocity 25–30 km s⁻¹ and a pre-shock density of 10⁵ cm⁻³. It therefore seems that in order to explain the observed line intensities the jet must power both J- and C-shocks.

We obtain a mass-loss rate from the CO outflow of ∼(1–3) × 10⁻⁵ ${M}_{\odot }$ yr⁻¹. This estimate is rather uncertain, because both the blue- and the redshifted outflow extends past the molecular cloud of R Mon, where it becomes too tenuous to be detected in CO. Furthermore, the dynamical timescale is very hard to estimate, see Section 3.2.2. We can get another estimate of the mass loss using the [O i] 63 μm PACS observations of R Mon (Jiménez-Donaire et al. 2017; Alonso-Albi et al. 2018). The [O i] luminosity is directly proportional to the mass-loss rate, if the jet velocity is high enough to produce a dissociative J-shock (Hollenbach 1985). From the previous paragraph we know that this is the case. We can therefore use the relationship

$$\begin{eqnarray*}&&L([{\rm{O}}\,{\rm{I}}]\,63\,\mu {\rm{m}})=0.1\times {\dot{M}}_{-5}\,{L}_{\odot },\end{eqnarray*}$$

which is valid over a wide range of physical conditions. In the above expression ${\dot{M}}_{-5}$ is the mass-loss rate in units of 10⁻⁵ ${M}_{\odot }$ yr⁻¹. Using the observed [O i] luminosity, 53.2 × 10⁻³ ${L}_{\odot }$ (Jiménez-Donaire et al. 2017), we get a mass-loss rate of 0.5 10⁻⁵ ${M}_{\odot }$ yr⁻¹, i.e., similar to what we derived from CO.

The accretion rate is even more uncertain. Spectra of R Mon show no Balmer jump and there are no spectral features, which would allow us to determine an accretion rate. If we assume that the outflow is jet driven, which appears likely based on the observed properties of the outflow (Section 3.2.1), and further assume that the accretion rate is 0.1–0.3 times the outflow rate (Shu et al. 1994; Ferreira et al. 2006), we get (1–9) × 10⁻⁶ ${M}_{\odot }$ yr⁻¹.

Mendigutía et al. (2011) were not able to derive an accurate accretion rate for R Mon, but estimate it to be in the range 3 × 10⁻³–7 × 10⁻⁶ ${M}_{\odot }$ yr⁻¹. Our value is close to their low range and about ten times less than what Natta et al. (1993) estimated from theoretical modeling, 4 × 10⁻⁵ ${M}_{\odot }$ yr⁻¹.

Mendigutía et al. (2011) show that there is a clear correlation between accretion luminosity and Br γ luminosity for HAEBE stars, which appears to be somewhat flatter than for classical T Tauri stars. Although the empirical relationship derived by Mendigutíia et al. shows considerable scatter, it can at least provide us some estimates of the accretion luminosity. Br γ is much less affected by extinction and veiling than lines like Hα and [O i] 6300 Å, which all correlate well with accretion. There are several published observations of Br γ in the literature. Evans et al. (1987) only got an upper limit in 1985, <3 × 10⁻¹⁶ W m⁻², while Carr (1990) detected it at approximately the same level in 1986, 3.6 × 10⁻¹⁶ W m⁻². Nisini et al. (1995), who observed the line in 1991 December got 10.2  × 10⁻¹⁶ W m⁻². This is about four times brighter than when it was observed by Carr (1990) indicating that there can be substantial variations in the accretion rate. If we extinction correct the observations for an A_V = 13.1 mag and use the relationship between accretion luminosity and Br γ luminosity in Mendigutía et al. (2011) we get an accretion luminosity of 140–350 ${L}_{\odot }$. This agrees quite well with what Natta et al. (1993) estimated from their theoretical modeling.