The faintness of the northern filaments is almost certainly due to obscuration by the inclined disk of the galaxy (cf. [Hennessy 1996]). The inclination angle of M82 is estimated to be 81.o5 ([Lynds & Sandage 1963]), such that the nearer edge of the galaxy is projected on the northwest side of the nucleus. We observe the nuclear regions of the galaxy through the southern side of the disk, and therefore expect the inner filaments to be much brighter there. If we assume a line-of-sight dimension for the optical disk equal to its linear dimension on the sky (11.'2; [de Vaucouleurs et al. 1991]), we expect the northern outflow to be at least partially obscured to projected radii of . This is consistent with the similar morphologies of the filaments beyond this radius in the north and south in Figure 8.
Although complicated by obscuration in the north, the inner filament structure also appears to differ between the two lobes in terms of collimation. The bright inner kiloparsec of the southern lobe is relatively well confined to the minor axis, whereas the northern outflow filaments cover a much wider range of opening angle. This is seen particularly well in the H flux map from the Fabry-Perot data (Fig. 1), and is probably not an obscuration effect, but rather a difference in the physical morphology of these two lobes. One possible interpretation is that the nuclear starburst is located slightly above the galactic mid-plane. The smaller mass of covering material to the north would make collimation of an expanding wind more difficult, resulting in an almost immediate ``breakout'' of the wind from the disk.
Finally, we have noted that the [NII]/H ratio map and [OIII] flux maps indicate that the southern outflow involves two distinct components, each originating from one of the central bright emission regions. This may be a ``limb brightening'' effect as has been suggested for the pair of central emission regions themselves. This implies that the emitting filaments are distributed along the outer surface of the outflow, rather than throughout the volume. On the other hand, it is likely that the two bright central regions from which the outflow streams appear to originate are stellar ``superclusters'' ([O'Connell et al. 1995]) which merely happen to be physically located on either side of the kinematic center of the galaxy, from our point of view. Regardless, we do not observe emission enhancement along the outflow axis, as would be the case for a volume brightened distribution.
If we do not include emission arising within approximately 8'' (125 pc) of the disk, our line fits encompass an H flux of ergs s-1 cm-2. After a rough subtraction of the halo model given in the previous section, this implies a total filament luminosity ergs s-1 (cf. Tab. 1). Assuming the filaments are completely ionized, using a case B recombination coefficient for K of cm3 s-1 ([Osterbrock 1989]), and employing the outflow geometry to be discussed in Section 4.3.4 below, we calculate an rms filament density cm-3. A very rough estimate for the filament filling factor of would suggest a mean electron density in the filaments cm-3, easily consistent with [SII] doublet ratios in the low-density limit throughout much of the large volume of outer filaments. This mean density implies a filament mass M distributed in a volume pc3, and a kinetic energy in the ionized filaments ergs. These values are consistent with those first computed by [Lynds & Sandage 1963], Mand ergs, respectively. The entire mass of outflowing gas is estimated to be a couple orders of magnitude larger ([Heckman, Armus, & Miley 1990]). Note that the kinetic energy in the filaments is only about one percent of the estimated input supernovae energy ( ergs; [Watson, Stanger, & Griffiths 1984]).