As mentioned in Section 12.1, the optical morphology of a galaxy certainly bears some relevance to the HI content and distribution, but it is important to remember that the optical light and the HI emission do not necessarily arise from the same locations. Most of the detected HI in our own galaxy represents a galaxian population clearly associated with the disk but not coextensive with other conspicuous large-scale components of the interstellar medium, such as molecular clouds or HII regions. The neutral, atomic hydrogen is typically found farther out in the disk than any of its other directly measurable components.
The HI distribution in galaxies with a well-established disk in the stellar component is also disk-shaped. The outlines of spiral arms are often conspicuous in the HI gas, as beautifully illustrated by the two-armed spiral pattern of the map of M81 of Rots and Shane (1975). In nearby galaxies for which the linear resolution of observations is particularly high, details that bear on the structure and dynamics of the interstellar medium are discernible. For example, in the case of M31, bubblelike features are observed, with sizes ranging from the map resolution limit (80 pc) up to 1 kpc (Brinks and Bajaja 1986).
A notable aspect of galaxian HI disks is the occurrence of strong central surface density depressions. Such depressions tend to appear more conspicuous in high-luminosity early-type spirals which possess large bulges. In lenticular galaxies the central depressions are frequently larger than the optical disks themselves, in which case the term "HI disk" tends to be modified to "HI ring." Although these outer HI rings are usually concentric with the optical disks, they can be strongly inclined to the latter. The azimuthal distribution of HI within the rings can be rather irregular, to the point that segments of the ring may not even be detected. In some S0's, diffuse polar rings are seen, orthogonal to the plane of the normal stellar disk, and are occasionally rich in HI. The trend of central depressions growing with earlier optical morphology is not entirely systematic, as summarized by van Woerden et al. (1983). The presence of the central depression in M31, an intermediate spiral, can be appreciated in Figure 12.2; notice how in M31 the region dominated by the stellar bulge is nearly devoid of HI. In our galaxy, the HI surface density starts to decrease at approximately the radius at which that of molecular hydrogen reaches a peak. In the case of early-type disks, not enough molecular gas is found to compensate for the locally anemic HI distributions. At the other extreme of the Hubble sequence, irregulars and gas-rich dwarfs exhibit an HI distribution similar in character to that of their light: clumpy and disorganized.
A frequently used azimuthally symmetric model for the HI surface density H(r) which gives a useful rough representation for most spirals is the sum of two Gaussians, the inner one with amplitude and width respectively -0.6 and 0.5 of the corresponding values for the outer one. This simplified model glosses over individual details and asymmetries and some large-scale features that may be common to most spiral disks. Sancisi (1983) has shown that a common feature seen in the outer regions of spirals is an HI "shoulder": a steep drop in H(r) occurs usually near the optical radius (normally the "Holmberg radius" at 26.5 mag arcsec-2), followed by a gentle decrease at larger radii. Such shoulders are especially noticeable among galaxies that have very extended HI distributions. The onset of warps (see Section 12.2.3) tends to coincide with the "shoulder" radius. Gas in the shoulder may differ dynamically from the inner HI, being characterized by offset eccentric orbits instead of circular ones, and may form an envelope to the whole galaxy. Sancisi thinks that very large HI rings, as seen in lenticulars, may represent extreme examples of "shoulders." Interpretations of those features in evolutionary terms are not widely accepted. In special cases, episodes of accretion of gas-rich systems by a dominating galaxy are invoked, as in the case of polar rings. However, the relatively frequent occurrence of some features-such as the "shoulders" described by Sancisi, which have in several instances been observed to begin at a truncation edge of the stellar disk-suggests that internal causes should perhaps be sought within the framework of a slowly infalling disk model or of one where at some critical radius the onset of dynamical instabilities precipitates noncircular motions.