Discovery of an HI Counterpart to the Extended Tail of Mira
L. D. Matthews, Y. Libert, E. Gerard, T. Le Bertre, M. J. Reid
aa r X i v : . [ a s t r o - ph ] M a y Discovery of an H i Counterpart to the Extended Tail of Mira
L. D. Matthews , Y. Libert , E. G´erard , T. Le Bertre , M. J. Reid ABSTRACT
We report the detection of an H i counterpart to the extended, far-ultraviolet-emitting tail associated with the asymptotic giant branch star Mira ( o Ceti).Using the Nan¸cay Radio Telescope (NRT), we have detected emission as faras 88 ′ north of the star, confirming that the tail contains a significant atomiccomponent ( M HI ∼ × − M ⊙ ). The NRT spectra reveal a deceleration of thetail gas caused by interaction with the local interstellar medium. We estimatean age for the tail of ∼ . × years, suggesting that the mass-loss history ofMira has been more prolonged than previous observational estimates. Using theVery Large Array (VLA) we have also imaged the H i tail out to ∼ ′ (0.4 pc)from the star. The detected emission shows a “head-tail” morphology, but withcomplex substructure. Regions with detected H i emission correlate with far-ultraviolet-luminous regions on large scales, but the two tracers are not closelycorrelated on smaller scales ( < ∼ ′ ). We propose that detectable tails of H i arelikely to be a common feature of red giants undergoing mass-loss. Subject headings: stars: AGB and post-AGB – stars: Individual (Mira AB) —stars: winds, outflows – radio lines: stars
1. Introduction
Mira ( o Ceti) is a mass-losing star on the asymptotic giant branch (AGB). It is thearchetype of a class of pulsating, long-period variables, characterized by regular pulsations(with periods of order hundreds of days) and large-amplitude variations in optical brightness(by up to ∼ ∼ ′′ . ∼
54 AU; Matthews & Karovska2006) . Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA, USA 02138 LERMA, UMR 8112, Observatoire de Paris, 61 av. de l’Observatoire, F-75014 Paris, France GEPI, UMR8111, Observatoire de Paris, 5 Place J. Janssen, F-92195 Meudon Cedex, France All physical quantities quoted in this paper assume a distance of 107 pc (Knapp et al. 2003).
GALEX satellite, Martin et al. (2007; hereafter M07) discovered that Mira is surrounded by a bowshock structure and sports a spectacular cometary-like tail, stretching two degrees ( ∼ ∼
128 km s − with respect to the interstellarmedium (ISM); see § molecules that are collisionally excitedby turbulent mixing between the cool molecular gas and the electrons from a shock-heatedgas component.We recently observed Mira in the H i as part of a larger, ongoing H i survey program ofthe circumstellar envelopes of evolved stars (see G´erard & Le Bertre 2006; Matthews & Reid2007). H i is now known to be common in circumstellar environments and frequently showsevidence of extending to very large distances from the star ( > ∼ i by Bowers & Knapp (1988), and for more than adecade remained the only AGB star known to have associated H i emission. Unfortunately,the data of Bowers & Knapp had a signal-to-noise too low to permit a detailed investiga-tion of the morphology and kinematics of the circumstellar material, although these authorsdid report tentative evidence for an interaction between the circumstellar debris and thesurrounding ISM. More recently, NRT observations of Mira by G´erard & Le Bertre (2006)provided a significantly improved H i line profile showing a roughly triangular shape, similarto that previously seen in CO (e.g., Winters et al. 2003), as well as evidence for a northwardextension of the emission. However, the coarse spatial resolution of the Nan¸cay beam pro-vided little detail on the H i distribution close to the star. To better characterize the extentand morphology of the H i envelope of Mira, we therefore obtained new imaging observationswith the VLA. As we describe here, the fortuitous timing of our observations provides a pow-erful complement to the recent GALEX results for understanding the mass-loss history ofMira. To probe the most extended, lowest column density material in the recently discoveredFUV tail, we have also obtained new mapping observations of an extended region aroundMira using the NRT. The Very Large Array of the National Radio Astronomy Observatory is a facility of the National ScienceFoundation, operated under cooperative agreement by Associated Universities, Inc.
2. VLA Observations
Mira was observed in the H i ′ . The primary beam of the VLA at our observingfrequency of ∼ ∼ ′ .The VLA correlator was used in dual polarization (2AC) mode with a 0.78 MHz band-pass, yielding 256 spectral channels with 3.05 kHz ( ∼ − ) spacing. The band wascentered at a velocity of 25 km s − relative to the local standard of rest (LSR); the bandcenter was offset slightly from the systemic velocity of the star ( V sys , LSR =46.7 km s − ) toavoid placing a strong Galactic feature near the edge of the band.Observations of Mira were interspersed with observations of two phase calibrators (J0201-115 and J0220-019) approximately every 20 minutes. 3C48 was used as an absolute fluxcalibrator, and an additional strong point source (J2253+161) was observed as a bandpasscalibrator. To insure that the absolute flux scale and bandpass calibration were not corruptedby Galactic emission in the band, the flux and bandpass calibrators were each observed twice,first with the band shifted by +1 MHz and then by − ∼ ∼
13% of the observed visibilities. During our firstobserving session, roughly half of the baselines had to be flagged in all channels numberingintegral multiples of 12, 13, and 14, owing to a strong local RFI source that emitted aninterference “comb”. The source of this RFI was identified as the Small Radio Telescope atthe VLA Visitor Center, and was switched off during the subsequent two observing sessions.Our VLA data were calibrated and reduced using the Astronomical Image ProcessingSystem (AIPS). To avoid closure errors on VLA-EVLA baselines, we computed and applieda bandpass solution to the raw data before proceeding with any further calibration (G. vanMoorsel, private communication). A new frequency-averaged dataset was then computedand used to calibrate the frequency-independent complex gains (see Table 1). Followingthis, a second correction to the bandpass was computed and applied, and time-dependentfrequency shifts were applied to the data to compensate for changes caused by the Earth’smotion. Finally, prior to imaging, the u - v data were continuum-subtracted using a linear fit 4 –to the real and imaginary components of the visibilities. Channels 20-85 and 105-160 weredetermined to be line-free and were used for these fits. These channel ranges correspondto LSR velocities of 52.7 − − and 4.4 − − , respectively. The continuumsubtraction procedure was also effective at removing frequency-independent patterns in thechannel images caused by solar contamination.We imaged the VLA line data using the standard AIPS CLEAN deconvolution algo-rithm and produced data cubes using several different weighting schemes, two of which arepresented here (Table 2). We also produced an image of the 21-cm continuum emission inthe region using a vector average of the line-free portion of the band. 5 –Table 1. VLA Calibration Sources Source α (J2000.0) δ (J2000.0) Flux Density (Jy) Date3C48 a
01 37 41.2994 +33 09 35.132 15.88 ∗ All0201-115 b
02 01 57.1647 −
11 32 31.133 2.64 ± ± ± b
02 20 54.2800 −
01 56 51.800 3.33 ± ± ± c
22 53 57.7479 +16 08 53.560 14.43 ± † ± † ± † ∗ Adopted flux density at 1420.3 MHz, computed according to the VLA CalibrationManual (Perley & Taylor 2003). † Quoted flux density is the mean from the two observed frequencies; see text. a Primary flux calibrator. b Secondary gain calibrator. c Bandpass calibrator.
Table 2. Deconvolved Image Characteristics
Image R Taper θ FWHM
PA rmsDescriptor (k λ ,k λ ) (arcsec) (degrees) (mJy beam − )(1) (2) (3) (4) (5) (6)Robust +1 +1 ... 63 ′′ × ′′ − ′′ × ′′ +31 1.6-2.1Continuum +1 ... 64 ′′ × ′′ −
10 0.73Note. — Explanation of columns: (1) image or data cube designation usedin the text; (2) robust parameter used in image deconvolution (see Briggs1995); (3) Gaussian taper applied in u and v directions, expressed as distanceto 30% point of Gaussian in units of kilolambda; (4) dimensions of synthe-sized beam; (5) position angle of synthesized beam (measured east fromnorth); (6) rms noise per channel (1 σ ; line data) or in frequency-averageddata (continuum).
3. VLA Results3.1. The Morphology of Mira’s H i Envelope and Tail
Figure 1 presents H i total intensity contours for Mira derived from our VLA imaging,overlaid on the GALEX
FUV image from M07. H i data with velocities from V LSR = 40 . V LSR = 50 . − were included in these images. To improve signal-to-noise in derivingthe H i maps, we rejected pixels in the original data cubes whose absolute values fell below1.5 σ after smoothing spatially with a Gaussian kernel of width 3 pixels (30 ′′ ) and spectrallywith a Hanning function.Our lower resolution H i map (left) reveals a distinct “head-tail” structure, stretchingroughly 12 ′ ( ∼ i is significantly greaterthan seen here (see § i emission is concentrated near the position of Mira itself. A trail ofemission then extends to the northeast, following the same position angle as the FUV tail.In our higher resolution H i map (Figure 1, right), some fraction of the total emissionis lost (as it falls below our rejection threshold), but we see that on smaller scales the H i morphology of Mira becomes clumpy and complex. The location of the peak intensity ofthe H i emission shows a small but statistically significant offset to the southwest of thestar’s FUV position: (∆ α, ∆ δ )=( − ′′ . ± ′′ . − ′′ . ± ′′ . i emission is not symmetrically distributed aboutMira, but exhibits an enhancement to the northwest. An enhancement in K i emission wasalso seen along this direction by Josselin et al. This type of asymmetry might arise in partfrom anisotropies in the outflowing wind and/or density gradients in the surrounding ISM(see Vigelius et al. 2007). As the H i emission branches off to the north, it roughly follows aridge of bright FUV knots (part of what M07 term the “North Stream”), before bifurcatinginto two lobes. A few additional isolated clumps of H i are also visible to the north.All of the H i emission detected from Mira with the VLA overlaps with the FUV lightseen by GALEX , although the detailed relationship between the two tracers is unclear. H i is seen concentrated along the western side of the tail where the FUV emission is also thebrightest. However, a significant fraction of the FUV tail shows no H i counterpart, includingthe bow shock, the southeastern edge of the tail, and the FUV-bright region lying betweenMira and the bow shock (termed the “South Stream” by M07). Moreover, on smaller scalesthere is no obvious correlation between the observed column density of the H i emission and 7 –the surface brightness of the FUV emission. Detection of H α emission from the UV-brightknots by M07 suggests that most of the gas at these locations is likely to be partially ionized.In the case of the South Stream, given that this region has a different FUV − NUV color thanthe rest of the tail, the medium here may be very highly ionized, and that the FUV emissionfrom this location may have a different origin (e.g., hot plasma emission). i Emission Surrounding Mira
Individual H i channel maps from our VLA imaging observations are shown in Figure 2.We find that near the position of the star, the emission detected in the central velocity maps(44.3-47.5 km s − ) has a larger spatial extent than in the outer velocity channels, as wouldbe expected for an expanding envelope. At the same time, several of the channels showadditional emission extending toward the North that arises from the near-tail. The velocityfield of the latter component appears complex, suggesting that the small-scale motions ofthe tail gas may be affected by turbulence. This is consistent with the interpretation of thetail as a turbulent wake (e.g., Wareing et al. 2007b).Figure 3 shows the global H i spectrum of Mira derived from the VLA observations. Thespectrum shown as a thick black line was derived from the “Tapered” data cube (Table 2) bysumming all emission within a 8 ′ (E-W) × ′ . α J . = 2 h m . s , δ J . = − ◦ ′ ′′ . Uncertainties on the total flux densities in each channel are ∼ ± i profile agrees well with the NRT line profile derived toward Mira and is dis-cussed further in § i Absorption in the Tail
The 21-cm continuum emission within a 30 ′ region surrounding Mira comprises a numberof weak point sources with a total observed flux density of ∼ α by M07. The brightestcontinuum source in the region lies at α J2000 . = 02 h m . s , δ J2000 . = − ◦ ′ ′′ . ± GALEX , but lies outside the regionwhere H i was detected in emission with the VLA. We have examined a spectrum towardthis source and detect a weak ( ∼ σ ) absorption feature (Figure 4). Based on a Gaussian fit,this feature has a peak flux density S = − . ± . v = 7 . ± − , 8 –and a central velocity V LSR = 44 . ± − . Both the central velocity and the width ofthe line feature are consistent with the H i gas observed in the tail of Mira in emission (seeFigures 3 and 5).Detection of H i in absorption in the tail of Mira allows us to obtain a constraint on thespin temperature of the gas. For the “Robust +1” data, the limiting H i column density for adetection of H i in emission , integrated over a Gaussian line profile with FWHM 7.1 km s − ,is N HI < ∼ . × cm − (3 σ ). Under the assumption that the absorbing gas at the positionof the continuum source has an equal or lower column density than gas detected in emission,one may then write: T s ≤ N HI (1 . × ) R τ ( v ) dv K (1)where T s is the spin temperature of the atomic hydrogen and τ is its optical depth (e.g.,Dickey et al. 1978). The assumption of a Gaussian line shape yields a line-integrated opticaldepth for the H i absorption profile of ≈ T s < ∼
37 K. Clumping of theabsorbing material would further reduce this limit, implying that a component of the tailgas is rather cool.
4. Observations with the Nan¸cay Radio Telescope
To achieve greater sensitivity to extended, low surface brightness H i emission in thevicinity of Mira, we made observations at several positions along its tail with the NRT (seeTable 3). Our pointings were selected using the GALEX map from M07 as a guide. Theseobservations were obtained between 2007 September and 2007 December as part of a Targetof Opportunity program.The NRT is a meridian-transit-type telescope with an effective collecting area of roughly4000 m . At 1420 MHz, its half-power beam width is 4 ′ in right ascension and 22 ′ indeclination for a source at the declination of Mira ( − ◦ ). Typical system temperatures are ∼
35 K. Further properties of the NRT are described in van Driel et al. (1997). The largecollecting area of the NRT and the good match between the N-S extension of the beam andthe direction of Mira’s wake make the NRT well-suited to searching for extended, low columndensity material.Our observational strategy for mapping the tail consisted of position-switched measure-ments at each pointing, with beam-throws of ± ′ or ± ′ in the E-W direction. One-third ofthe time was devoted to the on-position and two-thirds of the time to the off-source compar-ison spectra. A full NRT spectrum has a bandwidth of 165 km s − and a spectral resolutionof 0.08 km s − . A total of 44 hours of data were obtained along the tail. Fortunately, there 9 –is minimal Galactic H i emission near the LSR velocity of Mira; this provides flat baselinesthat permit us to detect weak signals efficiently. Data processing was performed using theCLASS software and consisted of subtracting a linear baseline from each spectrum beforeaveraging.The results of our NRT mapping are summarized in Table 3, and we show a sampling ofour spectra in Figure 5. We have clearly detected H i emission from Mira’s tail as far as 88 ′ north of the star. Moreover, we see the peak velocity of the emission becomes progressivelyblueshifted with increasing distance from the star, indicating an overall deceleration (seealso § i in any of the NRT pointings that have no overlap withMira’s FUV tail, consistent with the material giving rise to the FUV light and the H i beingspatially coupled along the full length of the tail. 10 –Table 3. NRT Mapping of Mira’s Tail Position offset: a Integration time rms noise Velocity Line width F peak (arcmin E, arcmin N) (hours) (mJy) (km s − ) (km s − ) (mJy)(0,0) 49 4.56 45.4 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a The adopted coordinates of Mira were α J . =02 h m s , δ J . = − ◦ ′ ′′ .
11 –
5. Results and Interpretation5.1. The Global H i Line Profile and Total H i Mass of Mira’s CircumstellarMaterial
The region near the position of Mira has been extensively observed with the NRT since2000 (see also G´erard & Le Bertre 2006). The H i line spectrum we derive by integrating theemission throughout a 12(E-W) ′ × ′ (N-S) region agrees well with the integrated line profileobtained with the VLA (Figure 3). Based on Gaussian fits to the global line profiles from thetwo telescopes, we find line centroids of 45.69 ± − and 45.41 ± − for theNRT and VLA, respectively. These centroids are slightly blueshifted compared with the valuederived from CO(2-1) line observations by Winters et al. (2003; V CO = 46 . ± . − ).We note however that the CO line is somewhat asymmetric and appears to be comprisedof multiple components. Based on a two-component fit to the CO(3-2) spectrum, Knapp etal. (1998) find the broader component to be slightly blueshifted ( V CO = 46 . ± − ),making it consistent with the H i centroid to within uncertainties. The FWHM of the H i profiles are 6.64 ± − (NRT) and 6.13 ± − (VLA), comparable to linewidthsmeasured from CO data (Knapp et al. 1998; Winters et al. 2003). However, whereas the COlinewidths directly gauge the expansion velocity of the stellar wind, the H i profile width maybe affected by turbulent motions in the tail gas ( § ± − (VLA) and 0.51 ± .
03 Jy km s − (NRT). Assuming the H i is opticallythin, the total H i mass contained within the portion of the circumstellar material imagedby the VLA is M HI ≈ . ± . × − M ⊙ . Using the NRT measurements summarized inTable 3, we can also estimate the additional amount of atomic material in the extended tailto be M HI ∼ . × − M ⊙ . (Here we have multiplied the observed emission by a geometriccorrection factor of 2 to account for the fact that we have not fully sampled the tail). Thecombined H i mass for the circumstellar envelope and tail of Mira is then M HI ∼ × − M ⊙ . A key result of our NRT mapping is that the spectra reveal a clear slowing-down of thematerial in the tail with increasing distance from Mira (Table 3 and Figure 5). At ∼ ◦ . i signal is ∼
14 mJy, with V LSR = 27 . ± . − , whereas the centroidof the H i signal at the center position is 45.4 ± − . This finding is consistent withthe model of Wareing et al. (2007b), who predicted an increasing velocity lag with respect 12 –to the velocity of Mira itself with increasing distance along the tail. This result also impliesthat the tail is older than the age of 3 × years derived by M07 under the assumption thatthis material is stationary with respect to the ISM.From Table 3, we can extrapolate to estimate the radial velocity of the H i material at2 ◦ from Mira (i.e., at the most extreme position where GALEX detected emission) to be V LSR ∼ ± − . Adopting the stellar radial velocity determined from CO observations( V LSR =46.7 km s − ; Winters et al. 2003), the proper motion from Perryman et al. (1997),and the solar motion parameters from Dehnen & Binney (1998), we estimate Mira’s velocityin the plane of the sky, corrected for solar motion, to be V t ≈
120 km s − . The velocitylag for the outermost tail material is therefore 23 ± − in the radial direction and ∼ . ± − in the plane of the sky. Finally, assuming a uniform deceleration ofthe stellar gas, we derive an age of t ∼ (1 . +0 . − . ) × years for the material detected by GALEX at 2 ◦ from Mira. This calculation does not take into account a possible variationof the mass-loss rate or the turbulence of the interstellar medium that the stellar gas mayencounter.Our age estimate for Mira’s tail exceeds previous observational estimates for Mira’s totalmass-loss duration by roughly an order of magnitude or more (see Young et al. 1993; Bowers& Knapp 1988; G´erard & Le Bertre 2006; M07). Furthermore, the tail age approachesthe expected interval between two thermal pulses; the relatively modest change in surfacebrightness over the length of Mira’s tail then suggests that the predicted growth in themass-loss rate for AGB stars between thermal pulses (e.g., Vassiliadis & Wood 1993) maybe smaller than previously assumed. Adopting the mass-loss rate for Mira derived from CO observations ( ˙ M ∼ . × − M ⊙ yr − ; Ryde & Sch¨oier 2001) and assuming this mass-loss rate has remained constant in time,the age derived in § ∼ . × − M ⊙ . After adjusting our current H i measurements for the mass of He, we thenestimate that neutral atomic material accounts for ∼
25% of Mira’s circumstellar envelopeand tail. We now briefly comment on some possible implications of this finding.Previous observations have shown that Mira’s wind is likely to be predominantly molec-ular as it leaves the star (e.g., Bowers & Knapp 1988; Josselin et al. 2000; Wood et al.2002). However, as discussed by Josselin et al., the bulk of Mira’s circumstellar material isexpected to be dissociated by the interstellar radiation field at radii of r > ∼ × cm from 13 –the star. Therefore, unless the wind is very clumpy (thereby increasing the survival time ofthe molecules), it is expected that atomic matter will comprise a significant fraction of thematerial that is ultimately swept by ram pressure into the tail.Under the assumption that the FUV light from Mira’s tail arises entirely from collisionalexcitation of H by hot electrons (M07), an expected by-product will be rapid dissociation ofmolecules (see also Raymond et al. 1997), thus providing an additional atomic contributionto Mira’s tail. Indeed, the dissociation rate of ∼ . × s − assumed by M07 shouldhave produced roughly a factor of four more H atoms during the past 1 . × years thanwe observe. Assuming some fraction of the wind is atomic before being swept into thetail, this raises some difficulty in how to maintain a sufficient supply of H to power Mira’sFUV luminosity over its inferred lifetime. A significantly lower molecular dissociation rate( ∼ T ∼ K) “surface” of the tail. Future multi-wavelength observations and modelling should help to clarify these issues.
6. Discussion: Are H i Tails Ubiquitous Features of Evolved Stars UndergoingMass-Loss?
We have reported the detection of an extended tail of neutral, atomic hydrogen associ-ated with the AGB star Mira. This H i i tail may represent an extreme example of a rather common phenomenon for evolved starsundergoing mass-loss.G´erard & Le Bertre (2006) already reported evidence that H i emission associated withcircumstellar envelopes may be offset from the position of the central star. In addition,Matthews & Reid (2007) previously reported the detection of an H i “plume” stretching ∼ ∼
18 km s − and ∼
100 km s − ,respectively), indicating that unusually high space motion is not a prerequisite for tail for-mation; indeed, it may require only that the stellar space velocity exceeds the expansionvelocity of the wind.While the sample of stars imaged in H i is presently small, evidence of interaction be-tween the circumstellar envelope and the ISM has also been seen in the global H i spectra ofa number of H i -detected stars (e.g., G´erard & Le Bertre 2006 and references therein). Ob-served H i line profile shapes are frequently inconsistent with a classic spherically symmetricmodel of mass-loss at a constant outflow speed, and may show velocity centroids offset fromthose observed in CO. As shown by Gardan et al. (2006) and Libert et al. (2007), theseprofiles can be well reproduced once the effects of ISM interaction are accounted for. Theimportance of ISM interactions in the evolution of circumstellar envelopes has also been un-derscored by the numerical simulations of Villaver et al. (2002) and Wareing et al. (2007a,c),and by the discovery of a far-infrared bow shock associated with the AGB star R Hya (Uetaet al. 2006). We therefore propose that extended gaseous tails may be ubiquitous featuresof evolved stars undergoing mass-loss.
For stars with low space velocities, hot companions ,and/or largely atomic winds, these tails may lack associated bow shock structures and/ora detectable FUV counterpart, but should in many instances be readily detectable via H i GALEX
FUV image and to J. Raymond for valuable discussions. The Nan¸cay Radio Observatory isthe Unit´e Scientifique Nan¸cay of the Observatoire de Paris and is associated with the FrenchCentre National de Recherche Scientifique (CNRS) as the Unit´e de Service et de Recherche(USR), No. B704. The Observatory also gratefully acknowledges the financial support ofthe R´egion Centre in France. The VLA observations presented here were part of programAM887. Mira’s hot companion, Mira B, is unlikely to significantly affect the composition of Mira’s wind and tailowing to the small extent of the ionized zone surrounding it (see Matthews & Karovska 2006).
15 –
REFERENCES
This preprint was prepared with the AAS L A TEX macros v5.2.
17 –Fig. 1.— H i total intensity contours overlaid on false color GALEX
FUV images of Mirafrom M07. The
GALEX image has been smoothed with a 3 × ′′ . × ′′ .
5) boxcarfunction. The full extent of the FUV emission is not shown. The left panel shows theH i contours derived from the “Tapered” image while the right panel shows those from the“Robust +1” image (see Table 2). Contour levels are ( − , . , − , , . , , ... . × . − m s − (left); ( − , − . , . , , ... . × . − m s − (right). A black starsymbol designates the position of Mira ( α J . =02 h m s , δ J . = − ◦ ′ ′′ . -02 455055-03 000510 51.4 KM/S 50.8 KM/S 50.1 KM/S 49.5 KM/S 48.8 KM/S-02 455055-03 000510 48.2 KM/S 47.5 KM/S 46.9 KM/S 46.3 KM/S 45.6 KM/S D E C L I NA T I O N ( J2000 ) -02 455055-03 000510 45.0 KM/S 44.3 KM/S 43.7 KM/S 43.0 KM/S 42.4 KM/S02 20 00 00-02 455055-03 000510 41.7 KM/S 41.1 KM/SRIGHT ASCENSION (J2000)02 20 00 0040.5 KM/S 39.8 KM/S02 20 00 0039.2 KM/S Fig. 2.— H i channel maps near the systemic velocity of Mira, taken from the VLA “Tapered”data cube (Table 2). Contour levels are ( − − . , . , × . − . A starsymbol indicates the position of Mira. 19 –Fig. 3.— H i spectra toward Mira. The thin red line shows the NRT spectrum obtained bysumming the measurements over a 12(E-W) ′ × ′ (N-S) region; the thick black line showsthe VLA spectrum obtained by summing within a 8 ′ . × ′ . i absorption spectrum toward the continuum source at α J2000 . = 02 h m . s , δ J2000 . = − ◦ ′ ′′ .
8. The flux from the continuum source itself has been subtracted. Thestronger, blueshifted absorption feature near − − is due to Galactic interstellar ma-terial along the line-of-sight, but the weaker, redshifted feature has a velocity and linewidthconsistent with the circumstellar material surrounding Mira. The thick line shows a Gaus-sian fit to the latter feature (see text for details). An arrow indicates the stellar systemicvelocity of Mira determined from CO observations. 21 –Fig. 5.— NRT H i spectra along Mira’s tail. The spectra shown have been smoothed to avelocity resolution of 0.32 km s − . Gaussian fits to the emission from Mira are overplotted.Note the bottom panel has a different vertical scale. Features blueward of V LSR ≈
10 km s −1