Mon. Not. R. Astron. Soc. 000, 000–000 (0000) Printed 15 December 2018 (MN LATEX style file v2.2) A Dark Galaxy in the Virgo Cluster Imaged at 21-cm. R. F. Minchin1, J. I. Davies1, M. J. Disney1, A. R. Marble2, C. D. Impey2, P. J. Boyce3, D. A. Garcia1, M. Grossi1, C. A. Jordan4, R. H. Lang1, S. Roberts1, S. Sabatini5, W. van Driel6 1 School of Physics and Astronomy, Cardiff University, Cardiff, CF24 3YB, UK 2 Steward Observatory, University of Arizona, 933 N. Cherry Ave., Tucson, AZ 85721-0065, US 3 Planning Division, Cardiff University, Park Place, Cardiff, CF10 3UA, UK 4 Jodrell Bank Observatory, University of Manchester, Macclesfield, Cheshire, SK11 9DL, UK 5 Osservatorio Astronomico di Roma, via Frascati 33, I-00040, Monte Porzio, Italy 6 Observatoire de Paris, GEPI, CNRS UMR 8111 and Universit´eParis 7, 5 place Jules Janssen, F-92195 Meudon Cedex, France 15 December 2018 ABSTRACT Dark Matter supposedly dominates the extragalactic, yet no totally dark structure of galactic proportions has ever been convincingly identified. Earlier (Minchin et al. 2005) we suggested that VIRGOHI 21, a 21-cm source we found in the Virgo Cluster at Jodrell Bank using single-dish observations (Davies et al. 2004), was probably such a dark galaxy because of its broad line-width (∼ 200 kms−1) unaccompanied by any visible gravitational source to account for it. Now we have managed to image VIR- GOHI 21 in the neutral-hydrogen line, and indeed we find what appears to be a dark, edge-on, spinning disc with the mass and diameter of a typical spiral galaxy. Moreover the disc has unquestionably interacted with NGC 4254, a luminous spiral with an odd one-armed morphology, but lacking the massive interactor invariably responsible for such a feature. Published numerical models (Vollmer, Huchtmeier & van Driel 2005) of NGC 4254 call for a close interaction ∼ 108 years ago with a perturber of ∼ 1011 solar masses. This we take as completely independent evidence for the massive nature of VIRGOHI 21. Key words: dark matter – galaxies: individual (VIRGOHI 21) – radio lines: galaxies arXiv:astro-ph/0508153v1 5 Aug 2005 1 INTRODUCTION terparts have been known about for many years; these in- clude high velocity clouds (Wakker & van Woerden 1997), Cold Dark Matter (CDM) models of galaxy formation the Leo Ring (Schneider et al. 1983), and various gas clouds predict many more dark matter halos than are ob- close to bright galaxies (Kilborn et al. 2000; Boyce et al. served as galaxies (Kauffmann, White & Guiderdoni 1993; 2001; Ryder et al. 2001). However, none of these objects Moore et al. 1999; D’Onghia & Lake 2004). Whether stars have the characteristics of a galaxy, i.e. detectable emission can form in dark matter halos depends critically on the frac- over galaxy sized spatial scales and a velocity structure con- tion of baryons (md) that can be trapped and retained by sistent with a rotating and gravitationally bound disc. each halo (for previous discussions about the existence and formation of dark galaxies see (Jimenez et al. 1997; Hawkins 1997; Verde et al. 2002)). Some of those that retain a small fraction of their original baryons (md < 0.05) are able to form stable gaseus disks, but because of the low gas den- In a previous paper (Minchin et al. 2005) we described sities they are not able to form stars. The physical state Jodrell Bank and Arecibo observations of VIRGOHI 21 an of the gas in the disc is very dependent on its density and HI source discovered during a survey of the Virgo cluster temperature with the latter influenced by the intensity of (Davies et al. 2004). VIRGOHI 21 has all of the HI charac- any ionising background. Models predict that galaxies can teristics of a typical rotating disc galaxy, but no detectable form with gas column densities that prohibit star formation optical emission. Our conclusion was that VIRGOHI 21 was yet provide some self shielding from the ionizing background an extremely promising candidate for the first dark galaxy. (Davies et al. in prep.). Such dark galaxies are potentially In this paper, we present high resolution HI observations detectable by blind 21cm surveys of the sky. of this source, which add to the conclusions drawn in the Objects detected at 21cm but with no optical coun- earlier paper that this is, indeed, a massive dark galaxy. c 0000 RAS 2 R. F. Minchin et al. 2 OBSERVATIONS AND ANALYSIS North and in particular to the South could either be part of the interaction or be clumps further out in the disc. Note The new data were taken in March 2005 at the Westerbork the lack of connection between galaxy ‘C’ and VIRGOHI 21. Synthesis Radio Telescope (WSRT) in two full 12-hour syn- theses and reduced using the Miriad package. The data were flagged for shadowing and on two of the fourteen 25- metre antennae one polarisation was flagged due to problems 3 DISCUSSION with the gain. A spectral bandwidth of 10 MHz covered the 1 Combining the new data with the old, we argue: velocity range 930 - 3070 km s− . Removal of the noisier end (a) If attributed to gravitation, changes in velocity of channels left 230 useful channels of width 8.2 km s−1 each, 1 galactic size over galactic scales, as seen here, require masses giving a velocity resolution of 10 km s− over the range 980 −1 of galactic proportions. On dimensional grounds the veloc- to 2890 km s . Continuum removal was carried out in the −1 ity wrench ∆V (∼ 200 km s ) seen at VIRGOHI 21 over a UV plane using Uvlin. The standard source 3C147 was used 22 conservative length-scale ∆x (∼ 14 kpc ≃ 5×10 cm ≃ 3 ar- for calibration. Cleaning used a robust setting of 1, close to cmin at the Virgo Cluster distance of 16 Mpc (Minchin et al. normal weighting (Briggs 1995). The cleaned cube was gaus- 2 10−11 2005)) implies a mass M ≥ (∆V ) ∆x/G ≃ 10 so- sian smoothed spatially and Hanning smoothed in velocity, lar masses. More specifically the gravitational (free-fall) and then used for a second-pass deeper cleaning which gave 3 2 1 2 timescale (∆x) / /(GM) / needs to be ≤ ∆x/V , the time the final cube used in the analysis. The synthesised beam it takes for gas travelling at V (relative to M) to change was 99′′ × 30′′ in size (extended North-South) and the noise −1 −1 velocity by ∆V . And where ∆V ∼ V , as it appears to was 0.3 mJy beam channel , giving a 5σ column-density 2 1 19 2 be here (see Fig. 2b), M ≥ (∆V ) ∆x/G again. For non- to sources 25 km s− wide of 2×10 Hydrogen atoms cm− . gravitational alternatives see below. i Fig. 1a shows a neutral Hydrogen (H ) contour map of (b) The bridge between NGC 4254 and VIRGOHI 21 re- the field superimposed on a negative optical image. VIR- veals that they have interacted. According to published nu- GOHI 21 is the elongated structure in the centre (which is merical simulations (Vollmer et al. 2005) (below) the mor- at about 2000 km s−1). A faint bridge can be seen stretch- 1 phological peculiarities of NGC 4254 require a perturbing ing down to the prominent spiral NGC 4254 (2400 km s− ) 11 1 mass of ∼ 10 M⊙ – providing independent evidence that while the other two sources, NGC 4262 (1500 km s− , upper 11 −1 VIRGOHI 21 weighs ∼ 10 M⊙. The simulation implies an left) and the faint galaxy ‘C’ (1750 km s , immediately to interaction 3 × 108 years ago. As the projected length of the the left of VIRGOHI 21) appear unconnected. Fig. 1b shows bridge is 120 kpc this would imply it has been drawn out at the velocity-declination projection of the full 3 dimensional a projected speed of 390 km s−1, which is comparable to the data cube. Now NGC 4254 is at the bottom right while VIR- radial velocity difference between NGC 4254 and VIRGOHI GOHI 21 is the angular structure in the centre. C is to its 21 of 400 km s−1. left. Far more detail can be seen on an animation which is (c) NGC 4262, the spiral to the North-East, is not in- available at http://www.astro.cf.ac.uk/groups/galaxies, for volved. There is no bridge to it (see animation) and the instance the bridge is clear and obvious, as is the lack of any radial velocity difference between NGC 4254 and NGC 4262 connection between VIRGOHI 21 and either NGC 4262 or is too large (900 km s−1) to generate one. C. The apparent alignment of NGC 4262 with VIRGOHI 21 (d) If object C, the H i galaxy just to the east (left) of and NGC 4254 in Fig. 1b is a consequence of the particular VIRGOHI 21 (Fig. 2) were involved its mass, by (a), must be projection shown 11 ∼ 10 M⊙. However its measured H i velocity distribution 9 The ∼ 25 arcmin filamentary bridge stretches from the and size suggest a probable mass ≤ 10 M⊙, while it has a 8 6 low-velocity (western) edge of NGC 4254, falling gently in luminosity of only 10 L⊙ and a MHI of 4×10 M⊙. It is two −1 ◦ ′ radial velocity from 2250 km s at declination +14 20 to- orders of magnitude too undermassive and underluminous to −1 ◦ ′ wards 1900 km s at +14 41 where it is suddenly arrested. explain the dynamics of VIRGOHI 21, while there is no sign ◦ ′ Then at +14 46 it is abruptly wrenched upwards again to- of its interacting.