The question of the interstellar matter in dwarf galaxies was emphasized recently by the very interesting series of papers resulting from the ISOCAM survey described here by Tuffs. The intriguing result regarding extremely cold dust, with T < 10K, located in or around BCDs (Popescu et al. 2002) raises a number of questions, such as (a) what is the nature of the detected dust, i.e., is this really Galactic dust, in such large quantities as claimed and located in metal-poor galaxies, and (b) what are its origins, i.e., is this dust internally produced during a previous episode of SF or is it accreted from within the cluster. It is not clear whether any of this dust may be connected with extended HI distributions (e.g., Hoffman et al. 1996). Note also that the other ISO contribution on the subject of dust in galaxies (Boselli et al. 1998) showed the dearth of the very small grains or grain inclusions that are responsible for the mid-IR emission at 6.75 and 15 µm from low-mass dwarfs. There, this was explained as due to the destruction of this dust component by the UV photons, but it is possible that the carriers of these mid-IR features do not form at all in metal-poor DGs. In this context, note that Gondhalekar et al. (1998) did not detect CO emission from some members of a sample of compact and star-bursting galaxies. The CO emission was absent in the low-luminosity members of the sample, essentially those that would be classified as dwarfs. This also may point out a link with the low-metallicity property of these objects.
A final remark about dwarf galaxies concerns the ability of modern large-scale structure simulations to reproduce the dwarf galaxy population. Realizations of the Local Universe, e.g., Klypin et al. (2001), apparently reproduce well the Local Supercluster, Virgo, Coma, and Perseus-Pisces, both in terms of the mass distribution and in terms of the velocity fields (large scale flows). These simulations can now have very fine spatial resolution (~ 20/h kpc) and good mass resolution that can go as low as 40 km/sec. This is the dispersion velocity of small-mass halos, which presumably can simulate dwarf galaxies. The theoretical discipline could, therefore, simulate the local population of dwarf galaxies in and around the VC and throw some light on the question of Virgo infall and the formation of the dE and dI populations.
Regarding studies of galaxies in the Virgo cluster, and specifically the determination of the influence of the local, galactic ISM on X-ray measurements, I want to emphasize a caveat regarding the proper accounting for foreground material. We have shown (Brosch et al. 1999) that FIR observations in the direction of the center of the Virgo cluster indicate the presence of significant quantities of dust. Some of this dust does not seem to cause reddening of background galaxies, indicating probably the presence of large dust grains that produce gray extinction. If this is the case, then usual methods for detecting such dust and correcting for its influence (e.g., as done for the Cepheids in the HST Key Project) will underestimate the extinction and will produce wrong distance estimates. This is true also in the case of dwarf galaxies, and is difficult to properly account for this effect.