THE ARECIBO LEGACY FAST ALFA SURVEY. VII. A NEUTRAL HYDROGEN CLOUD COMPLEX IN THE VIRGO CLUSTER

The study of the cold gas content in extragalactic systems plays an important role in understanding their formation and evolution. Not only does the gas content reveal information on the star formation potential of a galaxy, it can also help trace the dynamic history and response to the surrounding environment, whether it inhabits a dense cluster or an unpopulated field. For example, the fingerprint of galactic collisions and ram pressure stripping effects can be seen in the form of detached gaseous fragments and clumps. The 21 cm spectral line of neutral hydrogen serves as an important tool for measuring the cool, neutral gas component of galaxies, as well as a redshift indicator and kinematical probe. The H i line can also be detected in low mass systems with little or no stellar emission. Blind H i surveys thus yield samples which include a population "underappreciated" in optical surveys.

A number of intergalactic H i clouds have been reported in the literature, some in pairs and groups or in the vicinity of large disk galaxies (e.g. Schneider et al. 1983; Sancisi et al. 1987; Kilborn et al. 2000; Ryder et al. 2001), others the result of disturbances from interactions within a cluster (Minchin et al. 2005, 2007; Oosterloo & van Gorkom 2005; Haynes et al. 2007). The serendipitous detection of HI1225+01 (Giovanelli & Haynes 1989; Chengalur et al. 1995) revealed a binary structure, with the northeastern component found to be associated with a dwarf galaxy, while the southwestern component has failed to yield any optical light down to a surface magnitude limit of 27 mag arcsec⁻² (Salzer et al. 1991). A recently reported H i feature, VirgoHI21, was detected north of the spiral galaxy NGC 4254 (Minchin et al. 2005) and interpreted as a fairly massive "dark galaxy." Arecibo Legacy Fast ALFA (ALFALFA) survey data, which provide full coverage of the region, indicate that VirgoHI21 is not an isolated dark galaxy, but rather a tidal tail of NGC 4254 resulting from the high-speed encounter of that galaxy with a now distant interloper (Haynes et al. 2007; Duc & Bournaud 2008).

A galaxy traveling through a cluster can be affected by a variety of mechanisms, the most commonly invoked being ram pressure stripping. The study of H i deficiency in Virgo spirals was pioneered by Davies & Lewis (1973), with the first reliably measured deficiencies reported by Chamaraux et al. (1980). Giovanelli & Haynes (1983) showed that H i deficient galaxies in Virgo had truncated H i disks. These results were extended to other clusters (Giovanelli & Haynes 1985; Haynes & Giovanelli 1986; Haynes et al. 1984) and mapped with higher resolution (Cayatte et al. 1990; Chung et al. 2007). Oosterloo & van Gorkhom (2005) recently reported the discovery of a H i cloud in the vicinity of the Virgo spiral NGC 4388 with a H i mass of 3.4 × 10⁸ M_☉, resolved into distinct clumps. They show a clear connection to NGC 4388. It has also been shown that galaxies can be "harassed" by the collective effect of the cluster potential and multiple high speed encounters with other cluster members (e.g. Moore et al. 1996; Mihos et al. 2005). The hot intracluster gas is a hostile environment for the cold material removed from cluster galaxies through these processes. The interaction remnants represented by the latter are thus transient features. A full census of their number and estimates of longevity can provide insights on the metal enrichment of the intracluster gas.

Virgo is the nearest rich cluster of galaxies, thus providing the best target for the detailed study of many environment driven effects on the evolution of galaxies. With over 1300 cataloged detections (Binggeli et al. 1985), the overdensity associated with the Virgo cluster extends over more than 300 deg² in the sky, distributed in several subclumps. The largest two clumps are centered around large elliptical galaxies, M87 for the main clump, and M49 for the southern subclump. The morphology–density relation is well established in the cluster, with late-type objects thought to be still falling into the cluster, as it dynamically evolves. The dynamically vulnerable, extended H i disks of late-type galaxies can be easy victims of the processes discussed above. Optical data bases, such as the Digital Sky Survey (Lasker et al. 1990) and the Sloan Digital Sky Survey (York et al. 2000), provide photometry and targeted spectroscopy of cluster objects, as does the compendium of Virgo galaxies data of GOLDMINE Web site (Gavazzi et al. 2003; http://goldmine.mib.infn.it/). Individual galaxy distances are growing in number thanks to the ACS Virgo Cluster survey, which focuses on early-type galaxies via the surface-brightness fluctuation method (Mei et al. 2007). Targeted H i surveys also have played an important role in the overall characterization of the cluster (see, e.g., GOLDMINE for references).

The H i Parkes All-Sky Survey (HI PASS; Barnes et al. 2001) has mapped 30,000 deg², part of which covers Virgo in the region 12^h< R.A. <13^h and +2°< decl. <+20°. HIPASS catalogs available from NED list 125 galaxies in this region. The ALFALFA Survey (Giovanelli et al. 2005) is more sensitive and has significantly higher angular and spectral resolution than HIPASS. It is thus providing a much more incisive look into the H i properties of the cluster. While the H i content of many individual spiral galaxies has been mapped in detail with interferometry (Cayatte et al. 1990, 1994; Chung et al. 2007), no survey has covered the Virgo cluster region with the sensitivity and completeness of ALFALFA. It is a primary goal of the ALFALFA survey to study and characterize the local H i universe, including the Virgo cluster. ALFALFA will provide a fair, well-sampled and homogeneous spectral dataset, covering 7000 deg² of the high galactic latitude sky out to a redshift of cz_☉ ∼ 18, 000 km s⁻¹. Approximately 250 deg² of the Virgo cluster in the range 11^h44^m< R.A.(J2000) <14^h00^m and +8°< decl.(J2000) <+16° have been completely surveyed as of late 2007, with catalogs of extracted source parameters already released for publication (Giovanelli et al. 2007; Kent et al. 2008). A key feature of the survey is its sensitivity to low H i mass objects—ALFALFA will provide a complete H i census down to 2 × 10⁷ M_☉ at the Virgo distance. Sampling the low-mass population will be key in determining the H i mass function in the cluster environment, as well as in the field. The ALFALFA dataset in Virgo has already yielded many new interesting results (Kent et al. 2007; Haynes et al. 2007; Koopmann et al. 2008). In this paper we give a detailed description of a multicomponent H i cloud complex in the Virgo cluster region initially reported by Kent et al. (2007). In Section 2 we describe the observations and data reduction method within the context of ALFALFA and the cloud detections made with that dataset. In Section 3 we detail follow-up observations conducted with the Very Large Array (VLA)¹¹. In Section 4 we discuss CO observations of the cloud. Section 5 discusses the environment of the cloud complex. Section 6 discusses possible formation mechanisms for the cloud complex. Section 7 summarizes the results of the paper. Throughout the paper we assume a distance to the Virgo Cluster of D_V = 16.7 Mpc, and refer to the heliocentric reference frame for all velocities.

The ALFALFA observing strategy is described in detail by Giovanelli et al. (2005). The data are obtained in a meridian transit drift mode with the ALFA. Raw scans consist of 14 spectra (7 beams × 2 linear polarizations beam⁻¹), sampled at a rate of 1 Hz. The spectra span a 100 MHz bandwidth with 4096 channels per polarization, centered at 1385 MHz. The resulting spectral resolution is 24.4 kHz before smoothing, corresponding to δV = 5.1 km s⁻¹ at the rest frequency of the 21 cm H i line. Scans are calibrated, baselined, and flagged for radio frequency interference, and then combined into regularly sampled data cubes.

Details of the ALFALFA data processing can be found in Kent (2008). Briefly, frequency channels in each grid have been baselined with linear fits and then "flatfielded," where a median subtraction has been performed in both the right ascension and declination directions. A matched-filter algorithm was used for signal detection, with manual follow-up (Saintonge 2007). The careful attention to signal extraction, analysis, and data quality has proven to be invaluable in detecting faint objects and optimizing the output of the signal extraction process.

The cloud complex was discovered in the ALFALFA data obtained in the Spring 2005 campaign, which sampled the Virgo core region (12^h < α < 13^h, 8° < δ < 16°). The data presented here are taken from a 24 × 24 data cube centered at α = 12^h20^m, δ = 9°00' (J2000). The system temperatures of the ALFA receivers during the observations were in the range 26 K < T_sys < 30 K, yielding a root mean square (rms) noise of σ_m = 2.5  mJy beam⁻¹ in channels with δV = 5.1 km s⁻¹. The characteristics of the ALFALFA data cube from which the data are taken are summarized in Table 1.

Table 1. ALFALFA Observing and Data Cube Parameters

Parameter	Value
Sky center (J2000)	12 20 00, +09 00 00
Spectral range	25 MHz (−2000–3200 km s−1)
Effective integration time	48 s (beam solid angle)−1
Spectral resolution δV	24.4 kHz (5.1 km s−1)
Half-power beam size	33 × 38
RMS noise σm for δV = 5.1 km s−1	2.5 mJy beam−1

Download table as:  ASCIITypeset image

2.1. ALFALFA Observations

The cloud complex is located near α ∼ 12^h30^m, δ ∼ 9°30'—this places it 29 (845 kpc in projection) South of M87 and 15 (432 kpc in projection) north of M49. We discuss the optical environment of the cloud complex in Section 5.

The complex consists of five distinct emission features or "clouds," which we denote C1–C5. Together, they span approximately 35' (170 kpc in projection at the cluster distance) on the sky and 130 km s⁻¹ in velocity. Channel maps of the ALFALFA dataset in the vicinity of these detections are shown in Figure 1, and total intensity (zeroth moment) and intensity-weighted velocity maps of the region is in Figure 2. An integrated spectral profile for each cloud is presented in Figure 3.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The individual properties of the clouds derived from the ALFALFA data are given in Table 2; all parameters are computed in the manner described by Giovanelli et al. (2007). The spatial centroid of each cloud is in Column (2). Its accuracy depends on the source strength, and varies from an average of ∼15'' for the brightest features to ∼30'' for the faintest ones. The heliocentric velocity cz_☉, width at 50% of the peak W50 and total flux F_c of the integrated spectral profiles in Figure 3 are in Columns (3)–(5). The signal-to-noise ratio (S/N) of the detections is in Column (6), and is given by

$$\begin{equation} {\rm S/N}=\bigg ({1000 \, F_c \over W50}\bigg){w^{1/2}_{\rm smo}\over \sigma _{\rm rms}}, \end{equation} \tag{ 1 }$$

where F_c is in Jy km s⁻¹, W50 is in km s⁻¹, w_smo is a smoothing width equal to the number of 10  km s⁻¹  bins bridging half the signal, and σ_rms is the rms noise (in mJy) across the integrated spectrum at 10 km s⁻¹ resolution. The H i mass M_H i for each cloud is in Column (7), and is computed assuming that the clouds are optically thin and at the Virgo distance D_V= 16.7 Mpc:

$$\begin{equation} M_{{\rm H}\,\mathsc{i}}/M_\odot = 2.356 \times 10^5 \, D_V^{2} \, F_c, \end{equation} \tag{ 2 }$$

where D_V is in Mpc and F_c is in Jy km s⁻¹. The uncertainties on M_H i in Table 2 and elsewhere do not include that in the distance adopted, which is poorly constrained due to the large peculiar velocities of objects near or within the cluster.

Table 2. Single-Dish Cloud Properties from ALFALFA Survey

Cloud (1)	α, δ (J2000) (2)	cz☉ (km s−1) (3)	W50 (km s−1) (4)	Fc (Jy km s−1) (5)	S/N (6)	log(MHI/M☉) (7)
C1	12 30 25.8, +09 28 01	488 ± 5	62 ± 11	2.48 ± 0.07	21.2	8.21
C2	12 31 19.0, +09 27 49	607 ± 4	56 ± 7	0.72 ± 0.06	6.5	7.67
C3	12 29 42.8, +09 41 54	524 ± 7	116 ± 15	1.16 ± 0.07	8.6	7.87
C4	12 30 19.4, +09 35 18	603 ± 4	252 ± 7	2.56 ± 0.09	13.1	8.22
C5	12 31 26.7, +09 18 52	480 ± 10	53 ± 21	0.91 ± 0.06	7.6	7.77

Notes. Column (1): cloud name. Column (2): right ascension and declination of cloud centroid (J2000). Column (3): average heliocentric velocity of integrated spectral profile from Figure 3. Column (4): profile width, measured at 50 % of the integrated spectra profile peak and corrected for instrumental broadening as described in Giovanelli et al. (2007). Column (5): total flux of integrated spectral profile. Column (6): signal-to-noise ratio of the detection, computed using W50 and F_c via Equation (1). Column (7): base 10 logarithm of total H i mass, computed using F_c via Equation (2).

Download table as:  ASCIITypeset image

2.2. Cloud Morphologies and Kinematics from ALFALFA Data

Aperture synthesis observations of the central region of the cloud complex detected by ALFALFA were obtained with the VLA in two campaigns. On 2005 July 11, 5 hr on-source were obtained via rapid response observations in C configuration. On 2006 January 18–23, dynamically scheduled observations during the D–A configuration change yielded ∼1.5 hr on-source with antennas that had not yet been moved. All observations had common pointing and spectral centers of α = 12^h30^m45^s,  δ = +9°26' 0'' (J2000) and ν = 1417 MHz, and online Hanning smoothing was applied to yield 48.8 kHz channels over a total bandpass of 3.125 MHz. The observing setup was chosen to maximize the number of ALFALFA cloud detections accessible to the VLA in a single pointing: all of the clouds in Table 2 except C3 fall within the resulting field-of-view and bandpass.

The data from the runs was reduced using the Astronomical Image Processing System (AIPS; Greisen 2003). Standard flux, phase and bandpass calibration routines were applied. Continuum emission was removed from the data via linear fits to the average visibilities in line-free channels spanning 245 kHz at either end of the calibrated bandpass, yielding a net bandwidth of 2.2 MHz sensitive to H i emission.

After calibration, the data from each run were combined into a single UV data cube and imaged using a variety of spatial and spectral weighting schemes. The dirty beam pattern was deconvolved from the data using the Multi-Scale Clean algorithm implemented in AIPS (Cornwell 2008), in which components are extracted from a series of tapered images of the visibilities. We find that the resulting maps are not sensitive to the details of the deconvolution process, and therefore analyze the highest sensitivity, naturally-weighted cube with a synthesized beam width of 22'' (1.8 kpc at the Virgo distance). All maps and derived parameters are corrected for the attenuation of the primary beam, and averaged over two or three spectral channels to yield resolutions of δV' = 20.7 km s⁻¹ or δV' = 31.0 km s⁻¹, respectively. A summary of the aperture synthesis observing and map parameters is given in Table 3. For clarity, all variables denoting parameters derived from the VLA observations are primed.

Table 3. Aperture Synthesis Observing and Data Cube Parameters

Parameter	Value
Pointing center (J2000)	12 30 45, +09 26 00
Total time on-source	265 minutes
Net bandpass	2.2 MHz (378–822 km s−1)
Maximum spectral resolution δV'	48.8 kHz (10.3 km s−1)
Natural Weighting:
Synthesized beam	222 × 207 @  −20°
σ' m at pointing center, δV' = 20.7  km s−1	0.33 mJy beam−1
σ'm at pointing center, δV' = 31.0  km s−1	0.29 mJy beam−1

Download table as:  ASCIITypeset image

3.1. H i Aperture Synthesis Detections

We make two detections in the VLA data, which correspond to the clouds C1 and C2 identified in the ALFALFA survey data. Channel maps of these detections are shown in Figures 4 and 5. Contours are at multiples of the median rms map noise σ'_m in the primary beam-corrected maps at that location, and negative contours are indicated by dashed lines. We find that the emission associated with C1 and C2 spans multiple synthesized beams over contiguous (but largely independent¹²) channels irrespective of the visibility weighting or image deconvolution scheme adopted: they are credible detections in the VLA data.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Total intensity and intensity-weighted velocity maps for the clouds are shown in Figures 6 and 7. The data cubes are blanked before these moments are computed: a mask is generated for each frequency channel by smoothing the data to half the angular resolution in Table 3 and blanking regions with fluxes less than 2σ'_m of that image. In addition, the intensity-weighted velocity maps in Figures 6(b) and 7(b) are computed only at locations with column density $N_{H{\scriptscriptstyle I}}^{\prime }$ ⩾10²⁰ cm⁻². Integrated spectral profiles for C1 and C2 from the VLA observations are in Figure 8. The 1σ error bars on each point in Figure 8 reflect σ'_m integrated over the emission region in that channel and a 5% calibration uncertainty.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The properties of C1 and C2 derived from the VLA data are given in Table 4. Unless otherwise indicated, the parameters are computed in the same manner as their ALFALFA counterparts (see Section 2.1). The location of the peak N'_HI in the total intensity maps of Figures 6(a) and 7(a) is given in Column (2). The centroid cz'_☉ of the integrated profiles of Figure 8 is in Column (3), and W50' of the profiles is in Column (4). The values of W50' are corrected for instrumental effects by assuming that the unbroadened profile is gaussian. The integrated flux density F'_c and H i mass M'_H i are in Columns (5) and (8), respectively. The maximum angular extent a'_H i of each cloud is in Column (6). We adopt the outermost locations where N'_HI = 10²⁰ cm⁻² in Figures 6(a) and 7(a) as the cloud edges, and correct the measured values for beam smearing. The position angle PA'_HI at which a'_HI is measured is in Column (7). An estimate of the dynamical mass M'_dyn of each cloud is in Column (9), and is computed via

$$\begin{equation} M_{\rm dyn}^{\prime } = (3.39 \times 10^4)\,a_{{\rm H}\,\mathsc{i}}^{\prime } D_V \left(\frac{W50^{\prime }}{2} \right)^2 \;\;, \end{equation} \tag{ 3 }$$

where a'_H i is the object diameter in arcminutes, W50' is in km s⁻¹ and the Virgo distance D_V is in Mpc. We note that M'_dyn has physical meaning only if the clouds are self-gravitating and in dynamical equilibrium; these two assumptions may not be valid for C1 and C2 (see Section 6).

Table 4. Aperture Synthesis Cloud Properties from the VLA Observations

Feature (1)	(α, δ)' (J2000) (2)	cz'☉ (km s−1) (3)	W50' (km s−1) (4)	F'c (Jy km s-1) (5)	a'HI (') (6)	PA'HI (°) (7)	log(M'HI/M☉) (8)	log(M'dyn/M☉) (9)
C1	12 30 24 +09 28 20	488 ± 6	78 ± 11	2.14 ± 0.07	2.7 ± 0.3	36	8.15	9.36
C2	12 31 18 +09 29 25	597 ± 3	38 ± 5	0.57 ± 0.06	1.5 ± 0.3	146	7.57	8.48

Notes. Column (1): cloud name. Column (2): right ascension and declination of the peak N'_HI (J2000). Column (3): average heliocentric velocity of integrated spectral profile from Figure 8. Column (4): profile width, measured at 50 % of the integrated spectral profile peak and corrected for instrumental effects assuming that unbroadened profile is gaussian. Column (5): total flux of integrated spectral profile. Column (6): maximum linear extent of region with N'_HI ⩾ 10²⁰ cm⁻² in the total intensity maps (Figures 6(a) and 7(a)). Column (7): position angle at which a'_HI was measured. Column (8): base 10 logarithm of total H i mass, computed using F'_c via Equation (2). Column (9): base 10 logarithm of the dynamical mass, computed using W50' and a'_HI via Equation (3).

Download table as:  ASCIITypeset image

Position–velocity slices through the C1 and C2 datacubes are shown in Figures 9 and 10, respectively. For C1 in Figure 9, one 3'' wide slice is oriented along PA'_HI, and the other is perpendicular to this axis. We note that the emission is unresolved both spatially and spectrally along the axis perpendicular to PA'_HI for C2, and we therefore show only the slice oriented along PA'_HI in Figure 10.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

3.2. H i Morphologies and Kinematics of C1 and C2

3.3. Non-Detections in the VLA Follow-Up Data

An exploratory observation of the CO(3–2) emission line consisting of a single pointing (R.A.(J2000) = 12^h30^m27^s, decl. (J2000) = +9°28'345) was carried out with the APEX Telescope (Güsten et al. 2006) located in the Chajnantor Plateau in northern Chile. This pointing was chosen because corresponds to the peak of the H i emission in the VLA map. The observations took place during the night of 2008 January 1, using the the 345 GHz DSB heterodyne receiver APEX-2A. The beamsize (FWHM) of the APEX 12 m telescope is 18'' at 345 GHz. The receiver was tuned to CO(3–2) at V_helio = 490 km s⁻¹, with a total bandwidth of 1550 km s⁻¹ and spectral resolution of 0.8 km s⁻¹. Observations were performed with an atmospheric opacity at 220 GHz of 0.34, with a mean PWV ∼ 2.1 mm yielding to a T_sys(DSB) = 370 K. Pointing using Saturn as reference was regularly performed and was better than 2''. The total on-source integration time was 36 minutes, reaching an rms of 48 mK with a spectral resolution of 0.9 km s⁻¹. The antenna efficiency was estimated at 70%.

Data reduction and analysis were performed using the CLASS package from the GILDAS software (Pety 2005). Reduction consisted of baseline subtraction and the integration of the individual scans. The adjacent channels were re-binned to a velocity resolution of 22.9 km s⁻¹, leading to a rms of 12.9 mK in T_mb. No clear detection was found.

These exploratory CO(3–2) observations allow us to place an upper limit to the mass in molecular form in C1. In order to estimate a 3σ upper limit, we assume an expected line width of 50 km s⁻¹, yielding to a I_CO(3−2) = 1.93 K km s⁻¹. We assume a conversion "X" factor between the CO(1–0) line and the molecular hydrogen column density $N_{\rm H_{2}}$ of 2.8×10²⁰ cm⁻² (km s⁻¹)⁻¹ (Bloemen et al. 1986). We also adopted a CO(3–2)/CO(1–0) integrated line intensity ratio of 0.4 which is the typical value of the Galactic disk (Sanders et al. 1993). These values yield a H₂ column density limit for the observation of $N_{\rm H_{2}} <$ 1.3×10²¹ cm⁻², averaged over a beam of 18″. At a distance of 16.7 Mpc, this column density over a 1′  diameter is equivalent to a molecular hydrogen upper mass limit of $M_{\rm H_{2}} < 4.0 \times 10^{8}\;M _{\odot }$, which is comparable to the H i mass (see Table 2.). Allowing for uncertainties in the value of the X factor, the metallicity of the cloud gas, and the CO(3–2)/CO(1–0) integrated line ratio, we can conclude that the gas mass in the cloud C1 is unlikely to be dominated by the molecular component.

The optical feature of unknown redshift appearing in the field of C1 and discussed in Section 3.2 appears to be just above the SDSS g-band detection limit. If the feature is approximately ∼10'' across, then its g-band luminosity at the Virgo distance would be L_g ∼ 10⁶ L_☉. The stellar M/L ratios models of Bell et al. (2003) give M*/L*∼ 1.6. If this feature is the optical counterpart of C1, then the stellar to H i mass ratio is less than 0.01 and M_H i/L_g > 170. An analogous exercise for C2 and the putative counterpart VCC 1357 yields a stellar to H i mass ratio of less than 0.05.

As for the possible optical counterparts of clouds C1 and C2, attempts to obtain optical redshifts were made in two occasions with the Palomar 5 m Hale telescope. The first time was unsuccessful due to limited sky transparency, the second due to lack of any discernible H_α emission or of any other measurable spectral signature. The possibility of an association between H i and optical features remains open.

Figure 11 shows the Virgo Cluster X-ray emission in the vicinity of the H i cloud complex (Snowden et al. 1995). The large X-ray emission regions are centered on M49 and M87, with the H i cloud complex indicated with the crossed circles. Near the projected location of the cloud complex, Vollmer et al. (2001) estimate that the hot intracluster gas (ICM) density is n_icm ∼ 10⁻⁴ cm⁻³, while Shibata et al. (2001) have measured an ICM temperature of T₈ = 0.235 in units of 10⁸ K.

Zoom In Zoom Out Reset image size

Several galaxies are projected in the vicinity of the cloud complex. Figure 12 shows all cataloged objects within 1° of C1 and with heliocentric velocities between 100 and 900 km s⁻¹: galaxies detected in the 21 cm line are plotted with a circle of area proportional to the H i mass; galaxies not detected in ALFALFA with redshift information from other sources are plotted with crosses, most of which are early-type systems. It is assumed that all galaxies are located at the distance of the Virgo cluster. Note that the velocities of the H i clouds range between 480 and 607 km s⁻¹, i.e. if they are part of the Virgo cluster their line-of-sight velocity with respect to the cluster reference frame is directed toward us, as the heliocentric velocity of the cluster is 1150 km s⁻¹(Huchra 1988). Under the assumption that the clouds originated in the disk of a galaxy moving at high speed through the cluster, the velocity of the parent galaxy should also be incoming in the cluster reference frame, and most likely doing so at a larger velocity than the clouds themselves, as the latter would be decelerated by ram pressure after stripping. In the heliocentric reference frame, the parent galaxy should then have a lower velocity than those in the cloud complex. Two objects satisfy that condition: NGC 4424 and NGC 4445. Their locations are indicated by a plus sign (NGC 4424) and X symbol (NGC 4445) in Figure 11. In addition, as we discuss below, they are also extremely gas deficient for their size, which is evidence that they have been stripped within the last passage through the cluster.

Zoom In Zoom Out Reset image size

The SBa galaxy NGC 4424 (also known as UGC 7561 and VCC 9079) is located at $\alpha =12^{\rm {h}}27^{\rm {m}}11\mbox{$.\!\!^{\mathrm s}$}6, \delta =+09^{\circ }25^{\prime }14^{\prime \prime }$ (J2000). This places NGC 4424 at a projected distance of 236 kpc from the cloud center, and 900 kpc from M87. This peculiar galaxy is at a velocity (cz_☉ = 437 km s⁻¹), near that of the main cloud. The angular size of the stellar component, 36 × 18, translates to a linear size of 17 × 9 kpc at the Virgo distance. A detailed study by Cortes et al. (2006) characterizes it as having disturbed morphology; they also suggest that CO observations are indicative of noncircular gas motions and discuss convincingly the vulnerability of the galaxy to ram pressure stripping (see also Kenney et al. 1996). NGC 4424 has a H i mass of 2.5×10⁸ M_☉ and estimations of H i deficiency values have been reported between of 0.75 (Chung et al. 2007) and 1.09 (Helou et al. 1984). Those figures indicate that NGC 4424 has lost between 80% and 90% of its H i gas. Chung et al. (2007) noted that NGC 4424 exhibits a one-sided tail extending to the southeast, which is also detected by ALFALFA, and that the H i disk is truncated at a radius smaller than that of the stellar component of the galaxy. The direction of the tail shows the southeast to northwest direction of motion for NGC 4424; this motion is not compatible with the position of the cloud complex.

NGC 4445 (also known as UGC 7587 and VCC 1086) is an edge on spiral, possibly of type Sab, located at $\alpha =12^{\rm {h}}28^{\rm {m}}15\mbox{$.\!\!^{\mathrm s}$}9, \delta =+09^{\circ }26^{\prime }11^{\prime \prime }$ (J2000). Its heliocentric radial velocity is c z_☉ = 354 km s⁻¹and the H i mass 5.5 × 10⁷ M_☉ (Kent et al. 2008). Solanes et al. (2002) estimate its H i deficiency between 0.98 and 1.11, i.e. NGC 4445 has lost about between 90 % and 93 % of its H i gas. With a major axis angular diameter of 26, its linear size is about 13 kpc. We are not aware of H i synthesis data of this object; none are found in the VLA data archive.

One possibility that merits consideration is that the clouds are gravitationally bound structures, embedded in their own dark matter halos. If an H i cloud is gravitationally bound, its total dynamical mass within the H i radius can be computed from Equation (3), assuming spherical symmetry. Such mass is 2.3×10⁹ M_☉  for C1 and 0.3×10⁹ M_☉ for C2 as calculated from the aperture synthesis measurements. The density of such objects would be 1.4 × 10⁻²⁵ g cm^-3 for C1 and 1.0 × 10⁻²⁵ g cm^-3 for C2. These densities can be compared to the critical density, ρ_crit = (1.9 × 10⁻²⁹ g cm^-3)h², where h = 0.7. Were dark matter to provide the gravitational binding and to extend well beyond the visible, baryonic component, the total halo mass of, e.g., C1 would exceed 10¹⁰ M_☉. In that case, it would be surprising that the VLA data do not exhibit any trace of ordered motions in the gas, nor does the visible matter show any degree of central concentration, thus violating the mass–concentration relation for halos. A similar issue can be raised for the whole cloud complex. If bound by self-gravity, its total dynamical mass would be of order of 10¹¹ M_☉  and its mean density 2.1 × 10⁻²⁷ g cm^-3. Again, the cloud complex shows no ordered motions. The possibility that the clouds in the complex are a group of optically dark galaxies embedded in their own dark matter halo appears relatively implausible.

For a purely gravitational event, we may expect stream-like topology in the removed gas, which is not observed. High speed galaxy–galaxy encounters tend to produce relatively small damage in terms of gas mass loss by the victim, albeit spectacular in spatial extent as in the case of NGC 4254/VirgoHI 21. For NGC 4424 and NGC 4445, the damage would have been quite substantial (more H i is seen in the cloud complex than in either of the galaxies themselves) and the topology of the remnant is not particularly streamlike. These are not however very strong arguments and the possible origin through a high speed close encounter cannot be excluded.

Ram pressure stripping is very likely if the galaxy's orbit takes it deep in the cluster potential well, and can result in severe gas depletion from its outer disk (Cortes et al. 2006). With a radial velocity differing from the systemic velocity of the cluster by 700 km s⁻¹and a projected location half way between M87 and M49, the likelihood that ram pressure is currently very effective is marginal; it could however have been far more effective a few 10⁸ years ago, if the orbit dipped closer to the cluster center, independently on whether or not an encounter or merger took place. As we will see, that likelihood is even higher for NGC 4445.

6.1. Orbital Motion Through the ICM: a Simple Simulation

We consider the motion of a galaxy as it dips into the ICM on a parabolic orbit of pericluster distance r_p. Parametrized by the azimuthal angle ϕ, as illustrated in Figure 13, the distance of the galaxy from the focal point (M87) is

$$\begin{equation} r=\frac{2r_p}{1+{\rm cos}\, {\phi }} \end{equation} \tag{ 4 }$$

while the orbital escape velocity is

$$\begin{equation} v_g=\sqrt{\frac{2GM_{\rm clu}}{r}}, \end{equation} \tag{ 5 }$$

where M_clu is the cluster mass. For simplicity, we assume the cluster mass is concentrated in M87 and equal to 10¹⁴ M_☉, rather than integrating the motion for a changing cluster mass within the variable r, which is sufficient for our purposes. For the ICM density n_icm we assume the β–model (Schindler et al. 1999; Vollmer et al. 2001), given by

$$\begin{equation} n_{\rm icm}=n_{0} \big(1+ r^2\big/r^2_{\rm clu}\big)^{-3\beta /2} \end{equation} \tag{ 6 }$$

with a central density of n₀ = 0.04 cm⁻³ and a core radius of r_clu = 21 kpc, appropriate for the SW sector of the cluster and r_clu = 13.4 kpc for the cluster as a whole.

Zoom In Zoom Out Reset image size

As the galaxy travels through the ICM, we assume the galactic gas is stripped at pericluster; from there on, we estimate the ram acceleration and integrate the growing separation Δr and Δv between the galaxy and clouds. Figure 14 shows the orbital parameters for pericluster distances of r_p = 400, 500, and 600 kpc. The best orbit in this simple exercise is fit with r_p∼ 500 kpc. In that scenario, a galaxy reaches a distance r = 1100 kpc, 1.0 Gyr after pericluster passage. The velocity of the galaxy at r = 1100 kpc is 1100 km s⁻¹. Velocity (Δv) and spatial (Δr) separation values of 200 kpc and 300 km s⁻¹respectively are obtained from Figure 14. Corrections for the inclination along the line of sight for θ∼ 45° give values of Δr∼ 140 kpc and Δv∼ 200 km s⁻¹. These numbers are in general agreement with the observed values.

Zoom In Zoom Out Reset image size

6.2. Possible Formation Mechanisms

We can now consider two possible scenarios of cloud formation as the galaxy moves through the cluster: (1) removal by ram pressure and (2) a high speed tidal interaction of the type responsible for VirgoHI21 (Duc & Bournaud 2008; Haynes et al. 2007).

Ram pressure stripping will occur when the ram pressure arising from the motion of the galaxy at velocity v_gal,3d with respect to the intracluster medium of number density n_icm, n_icmm_pv²_gal,3d (where m_p is the proton mass), exceeds the restoring gravitational force that binds the gas to the galaxy: 2πGΣ_ismΣ_d, where Σ_ism is the gas surface density of the galactic disk and Σ_d is the total surface mass density of the disk. This condition can be rewritten as

$$\begin{eqnarray} \bigg ({n_{\rm icm}\over 10^{-3}\, \rm {cm}^{-3}}\bigg) \bigg ({v_{{\rm gal},3d}\over 10^3\, \rm {km s}^{-1}}\bigg)^{2} &\ge& 11 \bigg ({M_{d}\over 10^{11} M_\odot }\bigg)^2 \bigg ({R_{\rm gal}\over 10\, \rm {kpc}}\bigg)^{-4}\nonumber\\ &&\times \bigg ({M_{\rm ism}\over 0.1M_{d}}\bigg), \end{eqnarray} \tag{ 7 }$$

where M_d and M_ism are now the total disk and gas masses within the radius R_gal.

At the location of the cloud complex, some 845 kpc from M87, the density of the intracluster gas (ICM) is significantly lower than near the center of the cluster. For a β–model (Equation (6)) at the present projected distance of the clouds from M87, n_icm ∼ 10⁻⁴ cm⁻³. In the case of NGC 4424, v_gal,3d ≃ 700/cos θ km s⁻¹, where θ is the angle between the line of sight and the galaxy velocity vector. We estimate a disk mass enclosed within R_gal ≃ 10 kpc of about 10¹⁰ M_☉ (a generous estimate, since the total velocity width of the H i emission of the galaxy is 61 km s⁻¹(Kent et al. 2008) and the inclination of the disk, albeit unknown, is unlikely to be smaller than 45°) and assume M_ism/0.1M_d ∼ 1. With those numbers, the ram pressure term roughly matches the restoring force at the larger galactocentric radii. Chung et al. (2007) also compute ram pressure effects in the range between 125 <n_icmv²_gal< 175 cm⁻³ (km s⁻¹)² and a restoring force range between 40 <Σ_ismV²_rot/R_gal< 475 cm⁻³ (km s⁻¹)² for NGC 4424.

In the case of NGC 4445, v_gal,3d ≃ 800/cos θ km s⁻¹. The radial extent of the H i disk is currently unknown, but we can estimate the ram pressure and the restoring force at R_gal ≃ 6 kpc, as well as its disk mass from a rotational velocity of 107 km s⁻¹, from the measured velocity full width of the H i line of 213 km s⁻¹. Again, the ram pressure term is comparable with the restoring force at the galactocentric radii at which most of the H i would be expected to be found in an unperturbed disk.

It is interesting to point out that NGC 4522, a very clear case of ram pressure stripping carefully studied by Kenney et al. (2004), is projected on the same region of the cluster (R.A. = 12^h33^m397, decl. = 09°10'30'') albeit in a different velocity regime (cz = 2307 km s⁻¹). At the present location, Kenney et al. argue that the ICM density as described by the smooth β-model of Equation (6) is likely to be inadequate to explain the stripping. They invoke the possibility that the ICM may be locally denser than implied by the model.

6.3. Timing Considerations

Estimates of the tidal effects can be made by considering two nearby clouds in the vicinity of a cluster (Figure 15). The two clouds will disperse at a rate Δv = gΔt under the gravitational tidal acceleration g from the cluster potential, where g is

$$\begin{equation} g=\frac{2GM_{\rm cluster}x}{R_{\rm cluster}^{3}}, \end{equation} \tag{ 8 }$$

where M_cluster is the mass of the cluster within radius R_cluster, and x is the initial separation of the two clouds. The timescale of the encounter can be taken as Δt ∼ R_cluster/v_c,3d, where v_c,3d = v_c/cos θ is the velocity of the cloud complex with respect to the cluster. Taking Δv = gΔt, we can write a simple relation for the separation rate of the two clouds

$$\begin{equation} \Delta v \approx \frac{2GM_{\rm cluster}x}{R_{\rm cluster}^2 v_{c,3d}}. \end{equation} \tag{ 9 }$$

Two clouds initially separated by 10 kpc, located 845 kpc from the center of the cluster of mass 10¹⁴ M_☉ and moving at a velocity of 600/cos θ km s⁻¹  with respect to the cluster would roughly double their separation in a time comparable with the cluster crossing time. Tidal forces related to the cluster potential alone are unlikely to account for the spatial dispersion of the clouds in the complex, if they were stripped from a single galaxy.

Zoom In Zoom Out Reset image size

The cloud complex is separated by Δr ∼ 235/sin θ kpc and by Δv ∼ 100/cos θ km s⁻¹ in velocity from NGC 4424; the corresponding numbers for NGC 4445 are respectively Δr ∼ 150/sin θ kpc Δv ∼ 200/cos θ km s⁻¹. This separation would increase from zero to the observed value over a time of order

$$\begin{equation} \Delta t_{\rm sep,4424}\sim 2.3(\tan {\theta })^{-1} \,\rm {Gyr} \end{equation} \tag{ 10 }$$

for NGC 4424 and

$$\begin{equation} \Delta t_{\rm sep,4445}\sim 0.7(\tan {\theta })^{-1} \,\rm {Gyr}. \end{equation} \tag{ 11 }$$

for NGC 4445. For relatively large values of θ, they can be comparable with the time it takes for the galaxy to cross the inner regions of the cluster R_cluster/v_gal,3d. Under constant acceleration, the time necessary to reach the observed values of Δv and Δr would be about twice Δt_sep as computed above. Either of those assumptions are unrealistic if ram pressure acts upon the clouds, as they travel through an ICM of variable density. However, two preliminary conclusions can be made already: (1) if the clouds originated from NGC 4445 or NGC 4424, the stripping took place at least a few 10⁸ years ago; (2) simple kinematic timescales make NGC 4445 as a more palatable candidate as a parent galaxy than NGC 4424. In the following, we concentrate our discussion on the assumption that the clouds were stripped from NGC 4445, although we maintain the inferred relations parametrized form for easy application to NGC 4424. Next, we derive the acceleration due to ram pressure.

As a spherical cloud of radius r_c travels through the ICM at speed v_c,3d = v_c/cos θ, the force acting on its surface can be written as the ram pressure times the area, given by

$$\begin{equation} F = n_{\rm icm} m_{p} v_{c,3d}^{2} \pi r_c^{2}. \end{equation} \tag{ 12 }$$

The resulting acceleration due to ram pressure is

$$\begin{equation} a_{\rm rp} = \Delta v / \Delta t = \frac{\rho _{\rm icm}v_{c,3d}^2 \pi r_c^2 \eta }{ M_{{\rm H}\,\mathsc{i}}}, \end{equation} \tag{ 13 }$$

where the factor η accounts for the fraction of gas in form other than H i and M_H i is the H i mass of the cloud. The acceleration can then be written as

$$\begin{eqnarray} a_{\rm rp} &=& 2.5 \times 10^{-9}\eta \bigg(\frac{n_{\rm i\, cm}}{10^{-3} \rm {\, cm^{-3}}}\bigg) \bigg(\frac{r_c}{\rm {kpc}}\bigg)^{2} \nonumber\\ &&\times \bigg(\frac{v_{c,3d}}{10^3 \,\rm {\rm {km\,s^{-1}}}}\bigg)^2 \bigg(\frac{M_{{\rm H}\,\mathsc{i}}}{10^8 M_{\odot }}\bigg)^{-1} \rm {[cm\,s^{-2}]}. \end{eqnarray} \tag{ 14 }$$

Once the gas is stripped, while the motion of the galaxy is relatively unimpeded by the ram pressure, the clouds fall behind, affected by the acceleration a. In order to accumulate a velocity difference from the galaxy Δv, at constant acceleration a time on order of Δt_v = Δv/a will be required, while in order to accumulate a separation Δr, a time on order of Δt_r = (2Δr/a)^1/2 will be required:

$$\begin{equation} \Delta t_v = 1.3\times 10^8\; \rm {yr}\,\bigg(\frac{\Delta v}{10^2 \, {\rm {km\,s^{-1}}}}\bigg) A\, {\rm cos}\, \theta \end{equation} \tag{ 15 }$$

$$\begin{equation} \Delta t_r = 5.0\times 10^8\; \rm {yr} \bigg(\frac{\Delta r}{10^2 \,\rm {kpc}}\bigg)^{1/2} A^{1/2}{\rm cos}\, \theta \sin ^{-1/2} \theta \end{equation} \tag{ 16 }$$

where

$$\begin{eqnarray} A&=&\eta ^{-1} \bigg(\frac{n_{\rm icm}}{10^{-3} \,\rm {cm^{-3}}}\bigg)^{-1} \bigg(\frac{v_c}{10^3 \,\rm {\rm {km\,s^{-1}}}}\bigg)^{-2}\nonumber\\ &&\times \bigg(\frac{r_c}{\rm {kpc}}\bigg)^{-2} \bigg(\frac{M_{{\rm H}\,\mathsc{i}}}{10^8 M_{\odot }}\bigg). \end{eqnarray} \tag{ 17 }$$

For cloud C1, M_H i = 1.6 × 10⁸ M_☉. The maximum radius out to which the VLA detects H i emission in C1 is about 13, which translates to 6.5 kpc at the Virgo cluster distance. However, the effective radius of the cloud is very likely smaller than 6.5 kpc. A cursory inspection of Figure 6 shows that only about 1/3 of the area subtended by the emission out to a 1' radius exceeds a column density of gas at half the peak level, which would yield an effective radius of order of 3 kpc. In addition, from the topology of the H i gas in our own ISM, it is found that while the majority of the H i emission arises in denser regions, those clouds' volume filling factor is quite small. Such structure would be unresolved by the 22'' (1.8 kpc) beam of the VLA observations. Values of order r_c = 2–4 kpc appear appropriate. The parameter η, which accounts for the mass fraction of He and other than H i gas, should be in the range 0.4–0.7. Near the current location of the clouds, the ICM density is unlikely to exceed n_icm ≃ 10⁻⁴ cm⁻³, unless the ICM is very clumpy, as suggested by Kenney et al. (2004). As for the parameters specific for the galaxies, for NGC 4424 v_c = 700 km s⁻¹, Δr = 235 kpc and Δv = 100 km s⁻¹, while for NGC 4445 v_c = 800 km s⁻¹, Δr = 150 kpc and Δv = 200 km s⁻¹. Note that by imposing Δt_v ≃ Δt_r ≃ 2Δt_sep we can obtain a constraint on the combination of physical parameters

$$\begin{equation} 0.26 A^{1/2} \left(\frac{\Delta v}{10^2\; \rm {\rm {km\,s^{-1}}}}\right) \left(\frac{\Delta r}{10^2\; \rm {kpc}}\right)^{1/2} \sin ^{1/2}\theta =1. \end{equation} \tag{ 18 }$$

As we discussed earlier, accumulation of distance and velocity offsets between clouds and parent galaxy, Δr and Δv, at either constant rate or constant acceleration are used under outlined above, and do not take into account projection effects, which would account for the orientation of NGC 4445 and the clouds with respect to M87 at the cluster center. It is important to note that a connection between NGC 4445 and the cloud complex is somewhat ambiguous, especially when compared to other disrupted gas features like NGC 4388 which show a clear connection in the form of a tail.

6.4. Evaporation Timescale

According to our findings, if the clouds were stripped from a parent galaxy, most of the stripping took place a fair fraction of a Gyr ago. In that case, it is reasonable to investigate the survival of the clouds in the ICM. Differential ram pressure forces will lead to ablation and dilution of the gas clouds. In addition, upon removal from the galaxy, the cold gas can be heated by conduction by the intracluster gas and the gas mass of the cloud would thus be progressively reduced by evaporation, if (presumably absent) stellar mass loss does not replenish the gas. The evaporation rate in M_☉ yr⁻¹ can be written as (Cowie & Songaila 1977)

$$\begin{eqnarray} \dot{M_{\rm ev}}&\simeq& 16\pi \mu {\rm m}_p \,\kappa \,r_{\rm kpc}/25 k\nonumber\\ &\simeq& 35\, T_8^{5/2}\, r_{\rm kpc} \,(40/\ln \Lambda)\, M_\odot\,\rm { yr}^{-1}, \end{eqnarray} \tag{ 19 }$$

where μ is the mean molecular weight, m_p the proton mass, κ the thermal conductivity, T₈ the intracluster gas temperature in units of 10⁸ K, r_kpc the effective cloud radius in kpc (proportional to the square root of the area of the cloud exposed to the ICM), k the Boltzmann constant and ln Λ the Coulomb logarithm. For T₈ = 0.235 (Shibata et al. 2001) and 40/ln Λ ∼ 1, the evaporation timescale is

$$\begin{equation} t_{\rm ev} = M_{\rm gas}/\dot{M_{\rm ev}} \ge 1.7\times 10^8 r_{\rm kpc}^{-1} \eta ^{-1}\psi ^{-1} \rm {yr,} \end{equation} \tag{ 20 }$$

where ψ ⩽ 1 is a correction factor roughly proportional to the fraction of the time the cloud's orbit—between the time the gas is removed from the disk and the time of observations—places it in a part of the ICM within which the conductivity coefficient yields the evaporation rate in Equation (19). This estimate is very uncertain due to the largely unknown geometry of the cloud, an exact knowledge of its orbit through the ICM, the unknown fraction of the gas in ionized form and the possible presence of magnetic fields, which would inhibit conduction across field lines. It is clear from Equation (20) that if the H i gas is stripped from the galaxy in a region deep enough in the cluster ICM, conduction can obliterate evidence of the cold gas on timescales of <10⁸ yr. However, the rate in Equation (20) is approximately valid only in vicinity of the cluster core. Away from the central and denser parts of the cluster conduction becomes decreasingly effective, due to the rapid increase of the mean free path of electrons and conduction becoming saturated: this is thought to take place a couple of cluster ICM core radii (∼ 50 kpc) out from the center of the ICM distribution (Sarazin 1986). This is why in the simulation described in the preceding section, we have maintained the closest approach of the galaxy to M87, r_p, at values greater than a few hundred kpc; the evaporation timescale remains higher than the kinematical ones and the clouds can be preserved.

We have utilized datasets from the ALFALFA survey and follow-up observations with the VLA to obtain both single dish and aperture synthesis observations of a new H i cloud complex in the Virgo Cluster. The results of these observations are summarized as follows:

• 1.  
  Five separate H i clouds in the ALFALFA survey have been discovered with radial velocities between cz_☉∼ 500 and 600 km s⁻¹. The complex subtends an angle of 35' (170 kpc), the individual cloud H i masses range from 0.48 to 1.7 ×10⁸ M_☉  and the overall H i mass of the complex is 5.1 × 10⁸ M_☉, at the distance of the Virgo cluster. The clouds' velocity widths vary between 50 and 250 km s⁻¹. In the latter case, the wide spectrum is likely to be due to a blend of several poorly resolved clumps.
• 2.  
  Several cluster galaxies are found in the vicinity of the cloud complex; the most likely candidates for association with the clouds are NGC 4424, a disturbed SBa with c z_☉ = 437 km s⁻¹at a projected distance of 235 kpc from the center of the cloud complex, and NGC 4445, an edge on spiral at a projected distance of 150 kpc, with c z_☉ = 354 km s⁻¹. Both galaxies are extremely H i deficient, apparently having lost between 80% and 93% of their H i gas. The total H i mass of the cloud complex is equivalent to between 1/2 and 1/3 of the H i lost by NGC 4424 and about 4/5 of the H i lost by NGC 4445.
• 3.  
  Two of the clouds in the complex (C1 ad C2) have been detected with the VLA. They have angular sizes of 27 and 15 respectively. The synthesis maps do not exhibit any degree of symmetry in the gas distribution or in the velocity field. Faint optical features are found in the vicinity of the two clouds. Lack of optical redshifts prevent us from establishing a physical association. If the optical features were associated with the two clouds, their H i mass to g-band luminosity ratios would be respectively M_H i/L_g ∼ 170 for C1 and M_H i/L_g ∼ 30 for C2, in solar units. A handle on the H i size allows estimates of the dynamical masses of the clouds, contained within the H i radius and under the hypothesis that they are self-gravitating. Those values are 2.3 × 10⁹ M_☉ for C1 and 0.3 × 10⁹ M_☉ for C2.
• 4.  
  The possibility that the clouds constitute a group of primordial structures, embedded in their own dark matter halos, appears unlikely. A more plausible scenario is that the complex is material removed from a galaxy traveling through the cluster at high speed, such as NGC 4424 or NGC 4445. A ram pressure stripping event is preferred to a purely gravitational one.
• 5.  
  A simulation of plausible orbital parameters for the putative parent galaxy was carried out showing that a nearly radial orbit that would take it to within 400 kpc from M87 would produce effects comparable with those observed. Nearest approach to M87 would have resulted ∼ 1 Gyr ago, at which stripping of most of the gas is assumed to have taken place. The nearest approach to the cluster center would be approximately 500 kpc. At that distance from M87, thermal conduction would be ineffective at evaporating the clouds after removal from the parent galaxy. At pericluster passage, the velocity (not corrected for line-of-sight inclination) of the galaxy would be 1100 km s⁻¹. The inclination of the orbital plane to the line of sight of the best-fit simulation would be approximately 45°.
• 6.  
  Dynamical timescales, based on relative velocities and spatial displacements of the clouds from the putative parent galaxy, suggest that the most likely such parent is NGC 4445.

This research 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. Skyview was developed and maintained under NASA ADP Grant NAS5-32068 under the auspices of the High Energy Astrophysics Science Archive Research Center at the Goddard Space Flight Center Laboratory of NASA. K.S. acknowledges support from a Jansky Fellowship during the completion of this work. This work has been supported by NSF grants AST–0307661, AST–0435697, AST–0607007 and the Brinson Foundation.

This research has made use of Sloan Digital Sky Survey (SDSS) data. Funding for the SDSS has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Aeronautics and Space Administration, the National Science Foundation, the U.S. Department of Energy, the Japanese Monbukagakusho, and the Max Planck Society. The SDSS Web site is http://www.sdss.org/. The SDSS is managed by the Astrophysical Research Consortium (ARC) for the Participating Institutions. The Participating Institutions are The University of Chicago, Fermilab, the Institute for Advanced Study, the Japan Participation Group, The Johns Hopkins University, the Korean Scientist Group, Los Alamos National Laboratory, the Max Planck Institute for Astronomy (MPIA), the Max Planck Institute for Astrophysics (MPA), New Mexico State University, University of Pittsburgh, University of Portsmouth, Princeton University, the United States Naval Observatory, and the University of Washington.

The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc.

The National Astronomy and Ionosphere Center is operated by Cornell University under a cooperative agreement with the National Science Foundation.