5. The extent of the HI distribution

It was a great surprise to find that the nearly featureless galaxy NGC 2841 which, judged from optical photographs, should be the spiral galaxy in axial symmetry, is not only more than twice as large in the HI, but also possibly warped in the outer parts (cf. chapter 4.6). Also for a few other galaxies the HI extends much further out than does the optical image on photographs such as the prints of the Palomar Sky Survey. In this section we investigate whether we can derive quantitative information on HI sizes. Such information is not only interesting by itself but is also relevant for the interpretation of quasar absorption line systems (intervening galaxy hypothesis).

A convenient way to define the HI size is to specify the size at an isophote level, similar to the way optical sizes of galaxies are defined. We have estimated the HI size at two isophote levels: 5 × 10²⁰ atoms cm^-2 (274.3 K km s^-1) and 1.82 × 10²⁰ atoms cm^-2 (100 K km s^-1): These levels have been chosen because they occur in the outer parts of galaxies where the surface density of neutral hydrogen decreases with radius. The results for a number of galaxies are given in Table 2. The estimates are rather uncertain because the contours of the HI maps have an irregular shape. Comparison of the radius at the 1.82 × 10²⁰ atoms cm^-1 level (hereafter denoted as R_1.82) with the Holmberg radius, R_H, reveals that for a number of galaxies the ratio R_1.82 / R_H is larger than 1.5, and for others it is smaller than 1.5. The occurrence of what can be called a large HI envelope, i.e. R_l.82 / R_H > 1.5 is not related with morphological type, as can be seen in Table 2.

A better comparison can be made if we correct the observed diameters to face-on diameters; the results are given in columns 7 and 8 of Table 2. We have corrected the observed isophote levels with cos i, where i is the inclination of the galaxy taken from De Vaucouleurs et al. (1976). For the optical diameter we took D(0) as given by De Vaucouleurs et al. We have estimated the diameter,of the isophote level corresponding to a face-on value of 1.82 × 10²⁰ atoms cm^-2 from the maps of the HI column density distribution. We present the estimates rather than values available from the curves of _HI(r), because information on the latter is available for a smaller sample of galaxies. There is, however, a good agreement (within 10%) between the latter values and our estimates. In column 8 of Table 2 we present the ratio D_1.82 / D(0).

The average value of the ratio D_1.82 / D(0) is 2.2 ± 1.1. There is no significant type dependence of D_1.82 / D(0) (and also D_1.82 / D_H), contrary to earlier statements that later type spirals have larger HI envelopes than earlier type spirals. Bottinelli (1971) found an increase of a factor 2 in the ratio HI diameter / Holmberg diameter from Sb to Irr galaxies. She defined the HI diameter by a gaussian half-width, D_GH, which is not the same as an isophotal diameter. Her data were obtained with the Nancay telescope which has a beam elongated in the north-south direction. If we omit the galaxies oriented on the sky such that the position angle of the major axis lies between -35° and +35° then the correlation between D_GH and D_H is not significant for the spirals in Bottinelli's sample. The magellanic type galaxies might still have a larger ratio D_GH / D_H than do the spirals.

Kellman and Black (1973) argued, on the basis of Bottinelli's data, that log < _HI >, which is taken equal to log M_HI - 3 log D_HI, where < _HI > , M_HI and D_HI are the mean volume density, mass, and size of the HI, respectively, is not a function of type. This conclusion does not agree with the one of Sandage, Freeman and Stokes (1970) who calculated an increase of < _HI > with type. For 12 galaxies in our sample, for which we could derive fairly reliable HI masses and sizes (D_1.82),we have calculated log < _HI > = log M_HI - 3 log D_1.82 and log < _HI > = log M_HI - 2 log D_1.82. We find that these < _HI > and < _HI > values are constant within a factor of 3. Although our sample. is small we conclude that neither the extent of the HI distribution nor the mean density of neutral hydrogen is clearly correlated with Hubble type.

We briefly comment on some results obtained for the HI absorption lines in quasars. As an example we use NGC 3067, apparently associated with 3C 232 located 1.9' north of it. Haschick and Burke (1975) reported the detection of a narrow HI absorption dip at a radial velocity of 1411 km s^-1, which is close to the expected radial velocity at the position where the line of sight to the quasar intersects the disk of NGC 3067. The estimated HI column density at this position, at a radius of about 50 kpc (H_o = 75 km s^-1 Mpc^-1) which is three times the Holmberg radius, is a few times 10¹⁹ atoms cm^-2. This is not unusual: if we extrapolate the curve of log _HI(r) for NGC 3198, a galaxy of similar size as NGC 3067, out to that radius we find a similar column density.

This raises the question of the extent of the HI distribution once more: does in the outer parts the scalelength of the radial HI distribution remain constant or does it fall off abruptly? The existence of clumps of HI gas (ails, outlying clouds, cf. chapter 6) suggests that the _HI(R) does not gradually fall off to infinity Moreover, in the galaxies with large envelopes there are clumps within the envelope (e.g. in M81 and NGC 4449), and the deviations from circular motion are large as well (warp or large scale asymmetry). At present it is puzzling why these envelopes occur only around about 30% of the galaxies and not around all of them. This must be related in a general sense to the formation of galaxies.