ON THE ORIGIN OF THE SUPERGIANT H i SHELL AND PUTATIVE COMPANION IN NGC 6822^*

Holes and shells in the neutral interstellar medium (ISM) of star-forming galaxies have been challenging to interpret for decades. While alternative creation mechanisms exist, and the details of the energetics and multi-phase interactions are very complex (see, e.g., discussion in Tenorio-Tagle & Bodenheimer 1988), recent results support a scenario where energetic processes associated with stellar evolution (feedback) are the primary mechanism by which such structures are created (a very short and incomplete list would include Weisz et al. 2009a, 2009b; Cannon et al. 2011a, 2011b; Warren et al. 2011; and the numerous references therein). Many of these results have been attained in low-mass dwarf galaxies, where the lack of large-scale dynamical processes (i.e., spiral density waves) allows the characteristic timescale of the gas structures to extend for long periods of time (longer than a few hundred Myr in the aforementioned studies). The stellar feedback cycle can be considered one of the primary internal catalysts of galaxy evolution.

A second driver of rapid galaxy evolution is gravitational interactions with companion systems. Interactions between massive galaxies can cause powerful starburst episodes (e.g., Heckman et al. 1990); similarly, interactions between dwarf and massive galaxies can be dramatic (e.g., the Magellanic Stream being pulled from the Small Magellanic Cloud (SMC)/LMC system by interactions with the Milky Way; Putman et al. 1998). However, interactions between low-mass systems, which were frequent in the early universe, are less common at low redshift. Part of the reason is that most present-day star-forming, gas-rich dwarf galaxies are found in regions of comparative isolation or in loose associations such as the Local Group. Thus, "isolated" dwarf galaxies (i.e., not satellites of massive systems) that are undergoing interactions with other low-mass objects are rare, with only a few well-studied examples (e.g., VCC 848; Gallagher & Hunter 1989). Interactions between companion low-mass systems are a relatively unexplored driver of dwarf galaxy evolution.

At a distance of 492 ± 20 kpc (1'' = 2.4 pc; Lee et al. 1993; A. E. Dolphin et al. 2012, in preparation), NGC 6822 is among the nearest star-forming dIrrs in the Local Group. This system is gas-rich, dark matter dominated (de Blok & Walter 2000), and resembles the SMC in size and luminosity (M_B ∼ –15.8; Hodge et al. 1991). It is not associated with the concentrations of galaxies surrounding M 31 or with the Milky Way (Mateo 1998). While the system is well studied (more than 600 journal references to date), and is in our cosmological backyard, the internal and external processes that drive its evolution remain poorly constrained.

NGC 6822 provides an opportunity to study a system that is currently being shaped by both internal (stellar feedback) and external (low-mass gravitational interaction) processes. The neutral gas disk is dominated by an intricate structure of holes and shells (see Figure 1 and de Blok & Walter 2000, 2006). The most impressive feature is a large hole spanning some 2 kpc × 1.4 kpc (labeled "H i Hole" in Figure 1); this is one of the largest known holes in the ISM of a nearby dwarf galaxy. Equally remarkable is the presence of a putative small companion galaxy ($M_{\rm H\,\mathsc{i}}$ ∼ 10⁷ M_☉) that appears to be interacting with the main disk (labeled "Putative Companion" in Figure 1). de Blok & Walter (2000) demonstrate a significant offset in velocity between this northwest (NW) H i cloud and NGC 6822, arguing for the object being a physically separate low-mass system (see further discussion below). NGC 6822 is thus exceptional in that it has both a giant H i hole and a putative low-mass companion in close proximity; it is one of the only known galaxies with both properties.

Zoom In Zoom Out Reset image size

The ground-based optical study by de Blok & Walter (2006) used deep Subaru data covering the entire H i disk to resolve the bright O and B stars in NGC 6822. That study revealed the surprising presence of numerous young stars (ages <10⁸ yr) throughout the entire H i disk, well beyond R₂₅. Young blue stars were found associated with the putative companion galaxy, indicating that star formation (SF) has been ongoing there for at least the last 10⁸ yr (see de Blok & Walter 2006). Further, there are also relatively young stars located in the outer H i disk.

The goal of the present investigation is to study the underlying stellar populations in the giant H i hole and in the putative companion, thus probing both the internal and external processes that are driving the recent evolution of NGC 6822. New Hubble Space Telescope (HST) Advanced Camera for Surveys (ACS; Ford et al. 1998) imaging is acquired in six positions along the H i major axis. Each position is labeled in Figure 1, from "grid 1" through "grid 6," moving from NW to southeast (SE); note that the grid 1 position is coincident with the putative companion, while the grid 5 position is coincident with the H i hole. The extracted resolved stellar photometry reaches well below the red clump, allowing an accurate measurement of the global star formation history (SFH) as well as the recent SFH (with high time resolution over the past ∼500 Myr). The SFHs we derive provide strong temporal constraints on the creation of the various structures observed in the optical and in H i. This is a unique opportunity to study multiple important evolutionary processes in dwarf galaxies.

Six positions spanning the H i major axis of NGC 6822 were observed for one orbit each in Cycle 18 for program GO-12180 (P.I. Cannon); these six positions, labeled as grid 1 through grid 6, are shown by red boxes in Figure 1. The HST/ACS was used to acquire images in the F475W and F814W filters in two visits (grids 1, 2, and 3 observed on 2010 October 22; grids 4, 5, and 6 observed on 2011 March 28–29). Two images were acquired with each filter at each position, with integration times differing slightly between grids (886 and 1346 s for F475W and F814 in grids 1 and 6, and 1119 and 1118 s for F475W and F814 in grids 2, 3, 4, and 5). No dithering strategy was applied; each filter's observation was CRSPLIT into two images to facilitate the removal of cosmic rays. Hot pixels are easily differentiated from stars using our photometric measurement tools (see below). Figure 2 shows color representations of each HST field of view; the presence of both young blue stars and a diffuse red stellar population is obvious in all grid positions.

Zoom In Zoom Out Reset image size

We performed stellar point-spread function photometry using the DOLPHOT software package, a modified version of HSTPHOT (Dolphin 2000) with an ACS specific module. To separate well-measured stellar sources from artifacts, we applied quality criteria to the raw photometric catalogs in each field. Specifically, a well-measured star satisfied the following criteria: S/R_F475W > 5.0 and S/R_F814W > 5.0, (sharp_F475W + sharp₈₁₄)² < 0.075, and (crowd_F475W + crowd_F814W) < 1.0. The 50% completeness levels and numbers of matched stars in each grid location are given in Table 1. To characterize the observational uncertainties and completeness functions, in each field we performed 500,000 artificial star tests. In order to compute the completeness function, the artificial star tests were culled using identical criteria to the photometry. The results of these tests are shown in Figure 3, which plots the photometric completeness as a function of magnitude for each of the six grid positions. The photometric depths and completeness functions are nearly identical for all six grid positions.

Zoom In Zoom Out Reset image size

Table 1. HST Observations of NGC 6822

Grid	R.A.	Decl.	F475W tint	F814W tint	No. of Matched	F475W 50%	F814W 50%
						Completeness	Completeness
No.	(J2000)	(J2000)	(s)	(s)	Stars	(mag)	(mag)
1	19h43m3350	−14°33m410	886	1346	4715	26.8	25.5
2	19h43m5402	−14°37m096	1119	1118	5824	26.8	25.5
3	19h44m1487	−14°40m397	1119	1118	15850	26.8	25.5
4	19h45m1025	−14°49m464	1119	1118	195588	26.2	24.9
5	19h45m3026	−14°53m143	1119	1118	52449	26.6	25.3
6	19h45m5113	−14°56m367	886	1346	13435	26.8	25.4

Download table as:  ASCIITypeset image

Figure 4 presents the color–magnitude diagram (CMD) of each grid position using a density contour approach (i.e., a "Hess diagram"). Our photometry is complete well below the red clump and fully samples the blue helium burning region of the CMD in each grid (see Weisz et al. 2008 for a complete discussion of these various CMD regions). Note the differences in stellar density from one grid to the next (e.g., grid 4 contains ∼40 times more matched stars than grid 1).

Zoom In Zoom Out Reset image size

An accurate CMD of a galaxy's stellar populations allows the derivation of the intensity of SF as a function of both time and of location, i.e., an SFH. To derive the SFH, we apply a maximum-likelihood technique that fits the observed features of the CMD with a set of synthetic CMDs. The SFH of the best-fitting synthetic CMD is the most probable SFH of the observed stellar population. We refer the reader to Dolphin (2002), Dolphin et al. (2003), and A. E. Dolphin (2012, in preparation) for detailed discussion and demonstrations of this technique and its ability to differentiate flat SFHs from those with temporal features.

The synthetic CMDs are a linear combination of simple stellar populations that have been convolved with observational biases as determined by artificial star tests. We generate each synthetic population by assuming a stellar initial mass function and a binary star fraction, and designating a number and resolution of bins in both age and metallicity. Further, the observed and synthetic CMDs are binned to facilitate statistical comparisons between the two.

In the case of NGC 6822, we have adopted a Kroupa IMF (Kroupa 2001) within a mass range of 0.15–120 M_☉, a binary fraction of 0.35, and the Padova stellar evolution models (Girardi et al. 2002, 2010). To balance computational efficiency with the desire for a detailed SFH, we selected 36 time bins with uniform logarithmic spacing between log (t) = 6.6 and 10.15 and an isochronal metallicity range from 0.0001 Z_☉ to 0.0019 Z_☉ with a resolution of 0.1 dex. The data do not contain significant main-sequence turnoff features for populations older than ∼5 Gyr, and we thus have only poor constraints on the lifetime chemical evolution. Therefore, we required that the metallicities monotonically increase to minimize potential non-physical chemical models resulting from the red giant branch (RGB) age–metallicity degeneracies (e.g., Gallart et al. 2005). We have designated the 50% completeness limits as the faintest photometric depths considered.

Intervening Milky Way foreground stars can contaminate the observed CMD. To mitigate this effect, we have included a model foreground CMD in the SFH measurements. This CMD was constructed with the Dartmouth isochrones (Dotter et al. 2008), which extend to slightly lower masses of 0.1 M_☉, and follow the color and magnitude distributions detailed in de Jong et al. (2010). Finally, given the relative sparseness of the tip of the red giant branch (TRGB) populations in each field, we fixed the distance to previous TRGB determinations (see Section 1). However, we allowed the code to solve for the optimal extinction value to each field, as NGC 6822 appears to have significant variations in extinction from field to field.

In this study, we are interested in comparing the SFHs derived for each grid position; each of the six CMDs contains data of comparable photometric depths and similar error functions (see Figure 3). The similarity of data quality at each position means that the uncertainties on the SFHs reflect the observational errors and the random uncertainties due to the number of stars. The random uncertainties are calculated by performing 50 Monte Carlo tests that use a Poisson random noise generator to create a random sampling of the best-fit CMD. We do not consider the systematic uncertainties introduced by choice of stellar evolution models that are typically calculated in SFH derivations; in the limit of comparable quality data these choices do not affect the differential measurements from one position to another (e.g., Weisz et al. 2011).

The cumulative SFHs of all six grid positions are presented in Figure 5. This shows, for look-back times beginning at the big bang and decreasing until the present day, the fraction of the total stellar mass in each field that formed at or before a given time. All grid positions are normalized to a cumulative value of unity at the present day. When a given region reaches a cumulative SF of 0.5, then 50% of the total stellar mass in that region has formed; we hereafter refer to this timescale, measured back from the present day, as τ₅₀. The solid lines represent the best-fit cumulative SFH for each grid; the dotted lines represent the 1σ error bounds (technically the bounds at the 16th and 84th percentiles) derived from Monte Carlo simulations.

Zoom In Zoom Out Reset image size

While some variations exist, the cumulative SFHs of the six grid positions are qualitatively consistent with one another from the Hubble time up until ∼6 Gyr ago. This is strong evidence for the entire NGC 6822 system evolving as a single entity during this period. If the putative companion in grid 1 were in fact undergoing its first episode of SF at the present time, then such a signature would be very obvious in the plot as a long period of quiescence followed by a rapid increase in the cumulative SF in recent epochs. No such pattern is seen for the stars detected in grid 1. We note that our age resolution is most coarse for these ancient bins; however, this caveat strengthens our interpretation. Even if the details of the ancient SFH are less certain than those at recent times, Figure 5 shows that the field containing the putative companion has formed the highest fraction of its stars at old ages compared to the other fields. In fact, this is exactly what is expected if the outer field contains just the extended disk and not a companion. Numerous studies of the resolved stars of nearby dwarf galaxies have shown that the outer regions have higher fractions of older stars (e.g., Dohm-Palmer et al. 1997, 1998; Gallagher et al. 1998; Tolstoy et al. 1998). This nearly universal characteristic of dwarf galaxies has been interpreted as the result of the active star-forming region of the galaxy shrinking with time (Hidalgo et al. 2003) and has been supported by very detailed spatially resolved SFHs (e.g., Hidalgo et al. 2009).

As Figure 5 shows, NGC 6822 has formed roughly half of its stars within the most recent 5 Gyr (i.e., τ₅₀ ≃ 5 Gyr). This is broadly consistent with the value τ₅₀ = 6.9 Gyr reported by Orban et al. (2008) in a reanalysis of archival HST images near the center of NGC 6822, who found that 57% of the lifetime SF took place in the past 5 Gyr. Our derived τ₅₀ value can be compared with measurements of the same timescale for other nearby galaxies. Using a sample of seven Local Group dIrr galaxies, Weisz et al. (2011) calculate a characteristic timescale of ∼6 Gyr for the formation of 50% of the stellar mass; NGC 6822 thus is very similar to other Local Group dIrrs. Interestingly, the τ₅₀ level for 25 other dIrr galaxies from the Weisz et al. (2011) sample (mostly outside the Local Group) shows an average τ₅₀ that is about a factor of two longer (i.e., a greater fraction of early SF).

The more recent cumulative SFHs (from ∼6 Gyr to the present) do show some appreciable differences between grids 1, 2, and 5 and the other three. Specifically, grids 1, 2, and 5 show a period of relative quiescence (i.e., little to no increase in the cumulative SF) between ∼6 and ∼3 Gyr, while grids 3, 4, and 6 show an increasing cumulative SF during the same interval. Around ∼2–3 Gyr ago, grids 1, 2, and 5 show a significant increase in cumulative SF. This could be interpreted as weak evidence for an interaction involving the material now located in grids 1 and 2 with the material currently located in grid 5. More specifically, this could be interpreted as weak evidence for an interaction scenario leading to the creation of the large H i hole at the position of grid 5. We discuss this hypothesis in more detail below.

The overall similarity of the cumulative SFHs of the six NGC 6822 grids, especially in the epochs more ancient than ∼6 Gyr, can be interpreted as the lack of evidence for inside-out growth of its disk. Within the errors, the outermost field is as ancient as the inner fields. This contrasts with the results of Williams et al. (2009), who detect the signature of increasing disk scale length with cosmic time in M 33. The cumulative SFH of NGC 6822 can be considered to be intermediate between the extreme Local Group dwarfs LGS-3 (which formed ∼90% of its stars in an intense SF event more than 11 Gyr ago; Hidalgo et al. 2011) and Leo A (which formed 90% of its stars more recently than 8 Gyr ago, with a peak in the most recent ∼1.5–4 Gyr period; Cole et al. 2007).

The present study focuses on the recent patterns of SF throughout NGC 6822. The recent SFHs (in M_☉ yr⁻¹) in each grid location as a function of time, over the last 500 Myr and 100 Myr, are presented in Figures 6 and 7, respectively. Note that since each panel presents the SFH derived from all stars in the entire ACS field of view at each grid location, the SFHs can also be considered in units of star formation rate (SFR) per unit time per unit area (the ∼202'' × 202'' field of view of the ACS translates to ∼482 × ∼482 pc or ∼0.23 kpc² at the distance of 492 pc). From the longer-duration SFHs (Figure 6) it is clear that the SFR in each grid (with the exception of grid 4) has been essentially flat over the last 500 Myr; no major SF events are apparent in any grid location (except perhaps grid 4) until the most recent ∼50 Myr interval.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

As Figure 7 shows, all six fields show evidence of an increased rate of SF in the last 50 Myr (although with considerable differences in the amplitude of increase from one grid location to the next). Such a global increase in SF, which is happening on a time scale which is short compared to the dynamical time scale (∼140 Myr; McQuinn et al. 2010) is often associated with an interaction. Thus, we next investigate the possibility that the NW H i concentration is a companion galaxy.

The original interpretation of the NW region as a companion galaxy by de Blok & Walter (2000, 2003, 2006) was based on the observed neutral gas velocity offset (compared to the rest of the H i disk) in this region and on the presence of a significant number of spatially coincident young stars. de Blok & Walter (2006) derive a total luminosity of M_B = −7.8 for the stars in the NW cloud and estimate a total stellar mass of ∼10⁵ M_☉. This putative companion thus is very gas-rich ($M_{\rm H\,\mathsc{i}}$/L_B > 10 depending on model assumptions).

The similarity of the ancient SFHs in all six grid positions discussed above appears to contradict the interpretation of this system as a companion galaxy. However, the more recent cumulative SF rates in each grid position show some weak characteristics that might be consistent with an interaction scenario. We now use the information on the stellar populations in the six grid positions to address the question: is the NW region a bona fide companion galaxy that is interacting with NGC 6822 at the present time?

Fundamental to our interpretation is the assumption that if this is a bona fide companion galaxy, then it should harbor an ancient stellar population. This assumption is based on the detection of such a population (i.e., red clump and RGB stars) in every galaxy for which adequate data exist, even extending to the most metal-poor (e.g., I Zw 18; see Östlin 2000; Fiorentino et al. 2010, and the various references therein) or the most extreme SFH cases (e.g., Leo A, with the bulk of its SF delayed until the most recent few Gyr; see discussion above and in Cole et al. 2007).

If the putative companion of NGC 6822 is a bona fide dwarf galaxy, then our ACS images of grid 1 should reveal an overdensity of old stars at this location. Further, the ground-based Subaru imaging of the outer regions of NGC 6822 by de Blok & Walter (2006) should verify this signature. We thus explore both data sets in an attempt to identify this ancient stellar overdensity.

Figure 8 shows the spatial distribution of all RGB stars detected in the Subaru images; the locations of the six ACS grids are overlaid. While we are cognizant that crowding will be more severe in ground-based images compared to our new ACS images, the quality of the CMDs obtained by de Blok & Walter (2006) allow an unambiguous separation of the RGB component from the blue plume. This figure shows no evidence for a spatial concentration of RGB stars at the location of the putative companion.

Zoom In Zoom Out Reset image size

A direct comparison of the ACS and Subaru images is challenging not only due to the differences in crowding but also due to the difference in photometric depth; HST will more easily detect a faint overdensity of red stars. With these caveats in mind, Figure 9 shows the smoothed distributions of RGB stars detected by both HST and Subaru in the position of grid 1. We stress that a one-to-one agreement between these panels is not expected due to differences in crowding and depth. Again, the grid 1 position shows no obvious overdensity of RGB stars indicative of a stellar companion.

Zoom In Zoom Out Reset image size

It is conceivable that the ancient stellar population of the putative companion has a sufficiently low surface brightness that it is below the sensitivity of the Subaru data to detect. Further, it is conceivable that the ancient stars of this putative companion are distributed over an area larger than the projected size of the ACS field of view, making it difficult to identify by eye. So, as a final test, we examine the total number of red clump stars in each of the ACS grids. Figure 10(a) shows the grid 4 Hess CMD with the red clump selection region overlaid; the same selection region is applied to the CMD and photometry of each grid. Recall that our photometry is complete to more than 2 mag below the red clump; such a population is easily detectable in our data.

Zoom In Zoom Out Reset image size

In Figure 10(b), we plot the number of red clump stars in each grid position as a function of distance from the dynamical center of NGC 6822 (19^h44^m5804, −14°49^m189 (J2000) from Weldrake et al. 2003). The number of red clump stars per unit area falls exponentially as a function of radius, with a scale length of 545 pc; this compares favorably with the scale length of 430 pc found by Letarte et al. (2002) using the carbon star population. The number density of red clump stars at the grid 1 position is well within the 1σ error bar of the exponential fit. This plot is exactly what is expected for an exponentially declining surface brightness distribution as a function of radius. We therefore fail to detect evidence of an ancient stellar population that can be attributed to a bona fide low-mass companion object at this location. This is in agreement with the observation by Weldrake et al. (2003) that the velocity field in NGC 6822 is very highly symmetric in that the rotation curves derived from the approaching and receding sides are nearly identical. A true interaction would be expected to distort the velocity field in an observable manner. Given the high degree of symmetry, the most likely explanation for the H i morphology is a warp in which the outer H i is more highly inclined than the inner H i (in agreement with the tilted ring model results found by Weldrake et al. 2003).

Taken as a whole, the evidence above challenges the hypothesis that the putative companion is a bona fide dwarf galaxy that harbors an ancient stellar population. The origin of the global SF event during the most recent ∼50 Myr (which varies in intensity from one grid position to another) may be caused by a gravitational interaction, but we do not find evidence that the object in the NW is the interacting partner. This event may simply be stochastic—a result of the small number of young blue stars required to increase the SFR by the observed amounts in the grids with weaker SFRs.

A powerful diagnostic of the nature of this putative companion would be an observed metallicity difference between it and the main stellar body. The coarse metallicities of each field derived by our SFH show no evidence for such an offset. This further weakens the companion hypothesis. Of course, direct abundance estimators would be preferable; in this regard, deep long-slit spectroscopy of the faint H ii regions in the putative companion and/or of the RGB stars therein would be most useful in identifying any abundance difference between this object and the main stellar body.

5.1. A Stellar Feedback Origin

5.2. Alternative Creation Mechanisms

We have presented new HST/ACS imaging of six fields that span the H i major axis of the Local Group dwarf galaxy NGC 6822. These positions are selected in order to probe the stellar populations within the putative companion in the NW, the giant H i shell in the SE, and the extended tidal material in the outer disk of the galaxy. The photometry reaches two magnitudes below the red clump and is extremely sensitive to SF within the most recent 500 Myr in each of the grid positions.

Remarkably, the SFHs of the six grid locations are essentially identical to one another. While the relative SFRs change from one grid location to another, each of the six observed positions underwent a period of heightened SF activity in the most recent ∼50 Myr interval. This interval is shorter than the dynamical timescale of the system (∼140 Myr) and argues for a galaxy-wide event at this time.

The cumulative ancient SFHs of the six grid positions are qualitatively similar and reveal that NGC 6822 has formed half of its total stellar mass within the most recent ∼5 Gyr. This argues against the inside-out growth of the disk in this system, as seen in some other nearby galaxies. The similarity of the ancient cumulative SFH of the putative companion and the other observed regions suggests that this region is indistinguishable from the outer halo of NGC 6822.

This interpretation is strengthened by the lack of a coherent ancient stellar population at the location of grid 1. We find no overdensity of RGB stars at the location of the putative companion, using either our new HST data or the previously published Subaru data (de Blok & Walter 2006). We capitalize on the photometric depth of our HST images to derive the number of red clump stars in each of the six grid positions; the number of red clump stars per unit area falls off as a perfect exponential function as one moves into the outer disk, and the number of stars in grid 1 is well within the 1σ error bound on this fit. We conclude that the putative companion is not a bona fide galaxy (i.e., one containing an ancient stellar population).

The stellar populations within the giant H i hole show a strong and well-populated RGB and red clump. While the blue plume is weak, there is nonetheless evidence for low-level SF activity over the last 500 Myr. Feedback from SF in the most recent 100 Myr is not energetic enough to produce the H i hole. Only when we integrate over a much longer time interval (>500 Myr, i.e., significantly longer than the dynamical timescale) and assume a high feedback efficiency do we recover sufficient feedback energy to meet the ∼10⁵³ erg requirement derived by de Blok & Walter (2000).

The giant H i hole in NGC 6822 is an especially important test case for the hypothesis that stellar feedback can create such structures. We find evidence for extended SF that gives rise to this structure. The timescale for this feedback exceeds the dynamical time of the galaxy. This adds further evidence for long-lived H i structures in solid-body rotating galaxies (see, e.g., Weisz et al. 2009b; Cannon et al. 2011b; Warren et al. 2011). We propose that long-lived holes and shells in the neutral ISM of dwarf galaxies are the norm; once formed they remain coherent for long periods of time. The lack of dynamical processes to remove these structures allows them to persist for hundreds of Myr in the outer disks of low-mass systems.

We consider various other creation mechanisms for the H i hole in an attempt to explain the enigmatic properties of the NGC 6822 system; none is as successful as feedback from extended SF. The overall neutral gas morphology is indicative of a tidal interaction; the kinematic discontinuities identified near the position of the putative companion by de Blok & Walter (2000) support this interpretation. From our HST imaging we rule out the putative companion as being a bona fide galaxy that took part in this interaction. However, an interaction with a gas cloud without an ancient stellar component (e.g., a high velocity cloud analog) remains a viable solution; the confusion of NGC 6822 with Galactic H i may allow such a companion to remain hidden in the current datasets.

We are observing NGC 6822 at a unique time in its evolution. Numerous open questions about the evolution of low-mass galaxies can be addressed by further studies of this enigmatic object. Of particular value would be higher spatial and spectral resolution H i imaging of the entire gas disk, facilitating detailed kinematic modeling of the tidal arms, the putative companion, and the giant H i hole. Such data would help to differentiate between the companion hypothesis for the NE H i cloud and a more simplistic, non-uniform H i distribution with changing inclination angle as a function of galactocentric radius.

We thank the referee, Antonio Aparicio, for insightful comments that improved this manuscript. Partial support for this research was provided by NASA through grant GO-12180 from the Space Telescope Science Institute, which is operated by AURA, INC., under NASA contract NAS5-26555. J.M.C. and E.M.O. thank Macalester College for research support. This investigation has made use of the NASA/IPAC Extragalactic Database (NED) which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration, and NASA's Astrophysics Data System.