The presence of significant amounts of molecular gas in the INNER parts of a galaxy may not be so remarkable in the context of the conventional wisdom about the state of the ISM in galaxies; however, I think most of the HI observers would expect that the HI present in the far OUTER parts of a galaxy is much more ``primordial'', and that star formation away out there is going to be rare. Recent results by Cuillandre et al. (2000) contradict both of these bits of conventional wisdom. These authors used a large-format CCD on the CFHT to observe stars and background galaxies in a field covering the outer parts of M31 from 23 to 33 kpc, beyond R25, a field in which the HI was mapped more than 25 years ago (Emerson 1974; Newton & Emerson 1977) with one of the pioneering HI imaging synthesis radio telescopes, the ``1/2-mile telescope'' at Cambridge, England. In particular, Cuillandre et al. compared V-I color-magnitude diagrams for HI-rich regions with the diagrams for HI-poor regions, and discovered evidence for dust mixed in with the HI in amounts corresponding to 0.3 - 0.4 of the amounts in the Solar neighborhood. This gas is therefore not primordial, at least not in the usual sense of not having been processed in stars. Furthermore, they discovered young B stars correlated with the HI, confirming that massive star formation is going on at the present time. The close association between HI and the young stars in the outer disk of M31 is shown in Figure 2. The presence of current, ongoing star formation implies that H2 must therefore also be present, mixed in with the HI, in order to form these massive stars.
Figure 2. The distribution of blue stars over the extreme outer disk of M31 from Cuillandre et al. (2000). This is a 28' × 28' field, and covers a distance range from 23 to 33 kpc along the SW major axis of M31. The points represent the locations of stars in the range 20.5 < V < 24.5 with -0.5 < V - I < 0.2; larger dots are brighter stars. The contours are HI column density from Newton & Emerson (1977), with a resolution of 3.6' EW × 5.8' NS FWHM, and have here been corrected for their primary beam attenuation. The HI contours are drawn at levels of 2 through 18 in steps of 2, plus contours at 13 and 19, in units of 7.7 × 1019 atoms cm-2. There is some contamination by quasars and Galactic white dwarfs, which are visible e.g. in the halo field at the extreme south-west of the image. A small area in the NW corner of this image shows no stars because this part of the CCD mosaic was not used.
Could the photodissociation picture developed by Smith et al. (2000) for M101 also work in the outer disk of M31? A back-of-the-envelope calculation using the formalism in the Smith et al. paper is encouraging; compact clouds of H2 with densities of order 100 cm-3 are needed, similar to M101. However, in M31 the field is clean and unconfused, and a complete census of all far-UV-producing stars can be done. Here we have perhaps the best laboratory yet for testing the photodissociation picture; a quantitative analysis is in progress.