The ISM of the dwarf elliptical (dE) companions of M31 exhibits puzzling properties that are not yet understood. In NGC 205, the stellar component does not show rotation, while the HI does . In NGC185, stars and gas belong to the same kinematic system, while in NGC147 neither HI nor molecular gas were detected . In these three galaxies, the most recent measured star formation event took place 50 Myr, 100 Myr, and >1 Gyr ago, respectively (, , ), but these recent episodes cannot explain the differences in the ISM content. With 106 M, the amount of gas in NGC205 is a factor of 10 below what one would expect from normal mass loss through stellar evolution, and the kinematic differences between gas and stars make this an unlikely origin . In NGC185 stellar mass loss may provide an explanation for the HI. In both galaxies, the gas is less extended than the optical body, asymmetrically distributed, and clumpy on scales of less than 200 pc. NGC205 contains CO and dust, which are closely associated with HI concentrations on scales of 100 pc . While the molecular clouds in NGC205 closely resemble Galactic clouds, the HI envelopes of the CO clouds have much lower column densities of only ~ 1020 cm-2 due to the lower interstellar UV radiation field. This implies that in dwarf galaxies with low UV radiation molecular gas may form and survive even at column densities below 1021 cm-2 as less shielding is required (; see also ). The gas to dust ratio, which is similar to the one in the Milky Way, indicates that little dust gets destroyed in this environment. Even extremely cold dust with < 10 K was detected in NGC205 .
Owing to their prominent old stellar populations, some dIrrs resemble dwarf spheroidal (dSph) galaxies. But while dSph galaxies appear to be devoid of gas, these galaxies have been detected in HI. Thus they are classified as dIrr/dSph systems, galaxies that may be in transition from low-mass dIrrs to gas-less dSphs. LGS3, one of these transition types, is at a distance of 280 kpc from M31 and experienced low-rate star formation until 500 Myr ago . Its star formation rate was not significant enough to expel its gas, and the HI distribution is nicely centered on the optical galaxy. In contrast, the dIrr/dSph galaxy Phoenix (distance ~ 400 kpc from the Milky Way), which formed stars continuously until 100 Myr ago , does not contain HI within the main body of the optical galaxy. However, a nearby HI cloud with ~ 5 . 106 M has a velocity consistent with its having originated from Phe (, ). It may have been expelled through supernova explosions (though its regular shape seems to argue against this), or ram pressure stripping . Gas in two extended HI lobes appears to be within the tidal radius of Sculptor, a dSph without a young or significant intermediate-age population , matching its velocity . The amount of gas detected is consistent with expectations from stellar mass loss through normal stellar evolution. On the other hand, the surrounding field is filled with similar clouds, suggesting the possibility of mere coincidence .
The upper limits for HI in the other Local Group dSphs are at column densities of a few 1017 cm-2, well below even of expectations from mass loss through normal stellar evolution. At earlier stages in their evolution, these dSphs were evidently capable of forming stars and retaining gas over extended periods of time. Even those that are predominantly old show evidence for metallicity spreads . The Fornax dSph still formed stars < 200 Myr ago , while sensitive HI searches did not detect any gas  - perhaps a galaxy one step beyond Phe, just having completed its transformation into a dSph? Claims that gas exists at larger distances from dSphs  were not confirmed for several of them (e.g., LeoI ; AndV ). Searches for diffuse highly ionized gas only yielded upper limits (LeoI: ). Internal effects such as supernova explosions appear to be insufficient for removing the gas . Gas loss through tidal shocks during perigalactic passages close to a massive galaxy may rid dSphs of their gas  and provide an explanation for transition-type galaxies and the morphology-density relation, but environmental effects cannot explain isolated, gas-less dSphs like Tucana. Hence many interesting questions concerning the ISM in dwarf galaxies remain open.
It is a pleasure to thank the organizers for their kind invitation, Y.-H. Chu and J.S. Gallagher for a critical reading of the text, and Landessternwarte Heidelberg for a quiet office where this review was completed.