Annu. Rev. Astron. Astrophys. 2004. 42: 603-683 Copyright © 2004 by Annual Reviews. All rights reserved

4.3. Some "Bulges" Are As Flat As Disks

In Section 4.1, we repeatedly noted that pseudobulges examples were very flat, based on observed axial ratios or spiral structure. Secular formation out of disks does not require them all to be flat (Section 7.1), but it appears that we are fortunate and that many are flat.

This is seen in the distribution of observed bulge ellipticities derived by Kent (1985, 1987a, 1988). He decomposed major- and minor-axis profiles of disk galaxies into r^1/4-law bulges and exponential disks. The bulge and disk ellipticities were fit parameters that were allowed to be different. Figure 8 in Kormendy (1993) shows the following:

1. A majority of bulges appear rounder than their associated disks. These include the well known classical bulges in M 31, M 81, NGC 2841, NGC 3115, and NGC 4594 (the Sombrero galaxy).

2. Some bulges have apparent flattenings that are similar to those of their associated disks, as Kent noted.

3. Some bulges appear more flattened than their associated disks; these may be nuclear bars (Section 4.4).

4. The median ratio of bulge and disk ellipticities, _bulge / _disk, is smallest for Sas and increases toward later Hubble types. This agrees with other evidence that pseudobulges are more common in later-type galaxies.

5. However, the median _bulge / _disk for S0 galaxies is similar to that for Scs, not Sas. Kinematically disklike bulges also are more common in S0s than in Sas (Sections 4.6 and 4.7). Similar effects led van den Bergh (1976b) to develop his "parallel sequence" classification.

Bulge-disk decompositions should be interpreted with caution. The bulge and disk parameters are strongly coupled. Even when the bulge ellipticity is a fit parameter, it assumed to be constant with radius; this is necessary for computational stability. But Figures 10 - 12 and much other data show that this is a serious oversimplification. Also, most decompositions in the literature are not suitable. Some have too little leverage on the bulge. Non-parametric decompositions depend on the assumption that the bulge and disk have different flattenings; they force the bulge to be rounder than the disk. So we have few checks of the above results. Those that are available are consistent with points 1 - 5 but show a large dispersion in numbers. Here are two examples:

Fathi & Peletier (2003) carry out bulge-disk decompositions for 35 S0 - Sb and 35 Sbc - Sm galaxies based on HST NICMOS H-band images. The high spatial resolution provides good leverage on small "bulges". The results show that _bulge / _disk > 0.9 in 36 % of S0 - Sb galaxies and 51 % of Sbc - Sm galaxies. This is consistent with Kent's decompositions and confirms that flat "bulges" are more common in late-type galaxies.

In contrast, Möllenhoff & Heidt (2001) find that only 10 % of their decompositions imply _bulge / _disk > 0.9. These are K-band measurements of a sample of S0 - Sc galaxies weighted toward later Hubble types. The galaxies are relatively face-on; this reduces sensitivity to the flattening. However, the above results refer to the 39 galaxies that meet the selection criterion used for points 1 - 5, _disk < 0.14. So different authors get substantially different distributions of bulge flattening. On the other hand, Figures 10 - 12 clearly show that some pseudobulges are as flat as disks.