Envelopes of HI are often found around Galactic GMCs (1-10 pc scale; e.g. Andersson, Wannier, & Morris 1991; see Blitz 1993 for a summary) and have been ascribed to photodissociation of the GMC surface by the ISRF. Reach, Koo, & Heiles (1994) find evidence for H2 in some interstellar cirrus clouds with high column densities (N(HI) > 4 × 1020 cm-2) in quantities about equal to, or greater than the HI in spite of the fact that these clouds are likely to be fully exposed to the far-UV flux from the Galactic plane. In the very few instances where the viewing geometry is favorable, and when the line-of-sight superposition can be separated using radial velocity information, HI ``blankets'' can be seen between B Stars & GMCs (10-100 pc scale); a clear example of this is Maddalena's Cloud, which has been described as a large PDR of size ~ 50 × 200 pc dissociated by far-UV photons from one or two B5-O9 stars each located about 50 pc from the HI (Williams & Maddalena 1996).
2.2. ...and in other nearby galaxies
The first indication that photodissociation may be operating to affect the large-scale morphology (100-1000 pc) of the HI in galaxies was found in M83 by Allen, Atherton, & Tilanus (1986), who noticed a spatial separation between a particularly well-defined dust lane and the associated ridge of HI and HII in M83. Other studies have followed on M83 and on other galaxies (M51, M100; see Smith et al. (2000) for references) and have generally agreed that the initial interpretation in terms of photodissociation remains a viable option. The separation in the case of M83 is about 500 pc, and arises because of the difference between the spiral pattern speed and the rotation speed of the gas, coupled with the time for collapse of GMCs and the time that a massive young star lives on the main sequence.
The first study to successfully identify the characteristic PDR ``shell'' morphology of HI in close association with far-UV sources in a nearby galaxy was carried out on M81 by Allen et al. (1997). The problem is, of course, to obtain sufficient linear resolution (~ 100 pc) in the HI observations to permit one to identify the morphology of the PDR structures. An important point to note is that the best ``correlation'' is between the HI and the far-UV, not between the HI and the H.
A study similar to M81 but with more quantitative results has recently been carried out on M101 by Smith et al. (2000), who used VLA-HI and UIT far-UV data to identify and measure PDRs over the whole extent of the M101 disk. From these observations they derived the volume density of the H2 in the adjacent GMCs in the context of the PDR model. Figure 1 shows the best estimate of the H2 volume densities of GMCs near a sample of 35 young star clusters. The range in density (30 - 1000 cm-3) is typical for GMCs in our Galaxy, lending support to the use of the PDR picture, and also shows little trend with galactocentric distance.
There is also IR spectral evidence that PDRs are important for understanding the physics of the ISM in galaxy disks. KAO observations of the 158 µm CII line suggest that as much as 70%-80% of the HI in NGC 6946 could be produced by photodissociation (Madden et al. 1993), and ISO spectra in the mid-IR indicate that the bulk of the mid-IR emission from galaxy disks arises in PDRs (Laurent et al. 1999; Roussel, Sauvage, & Vigroux 2000; Vigroux et al. 1999).