6 Discussion

The new classification of bulges is very similar to the types used by Jarvis (1986) and Shaw (1987). The frequency of 45% b/p bulges is consistent with Dettmar & Barteldrees (1988), Dettmar (1989) and the frequency of 20% prominent b/p bulges with Shaw (1987). However, the derived frequency is now based on a much larger sample of 734 galaxies. Furthermore, for the first time a large fraction of b/p bulges in galaxies as late as Sd is found, and some previously unknown peanut bulges (type 1) are listed (Tables 6 and 7).

The large fraction of b/p bulges (45%) shows that such bulges are not that peculiar but rather quite normal. Therefore very common processes are required to explain the origin of b/p bulges.

In order to check possible formation processes relating b/p bulges with bars, we have compared the frequency distributions of galaxies with b/p bulges and barred galaxies. The frequency distribution of barred galaxies is derived from RC3 (only face-on galaxies) and contains 8587 galaxies (Fig. 5). Both distributions binned by morphological type show the same general dependence. The maximum for barred galaxies is also at Sb/Sbc and the minimum at S0/S0a galaxies. This minimum is more distinct for barred galaxies as for galaxies with b/p bulge. However, for all galaxies the fraction of barred galaxies is 55% (nearly 2/3 of them are strongly barred [SB] and the rest weakly barred [SAB = SX]) in relation to 45% of galaxies with b/p bulges. This high percentage of barred galaxies is also recently confirmed by Knapen et al. (2000). Furthermore, it is remarkable that the percentage of barred Sd galaxies is relatively high compared to the minimal frequency of galaxies with b/p bulges for Sd galaxies. However, it should be mentioned that our error of this bin is the largest due to low number statistics (Table 2). Other statistics of barred galaxies taken from older catalogues (Sellwood & Wilkinson 1993) show the same general dependences, although the percentages vary mainly due to different fractions of weakly barred galaxies in the catalogues. However, former samples contain a much smaller number of galaxies (Sellwood & Wilkinson 1993). Unfortunately, the galaxies are not all classified as SB, SX, or SA (unbarred galaxy), but there is also a group of galaxies without any classification (S. = 28%). These galaxies can be likely interpreted as unbarred galaxies (as done in Fig. 5), because in former times unbarred galaxies were simply called "S'', but they add uncertainties to the quantitative conclusion.

Figure 5: Frequency of barred galaxies binned by morphological type. All face-on galaxies of the RC3 are included, i.e., all galaxies with a diameter (D25) larger than 1 $\hbox {$^\prime $ }$ and an axis ratio smaller than 1/3 ( $\log R_{25} < 0.477$). Solid lines: strongly barred (SB) plus weakly barred (SX) galaxies. Dashed lines: only SB galaxies

   

Table 6: List of galaxies in the RC3-sample with bulge classification

(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Object	RA	DEC	bulge	NIR	earlier	T	log	log
	(2000)	(2000)	type	bulge	detections		D25	R25
				type
IC 5376	00 01	+34 31	4			2.0	1.30	0.76
NGC 7814	00 03	+16 08	4f			2.0	1.74	0.38
NGC 7817	00 03	+20 45	5.1			4.0	1.55	0.58
ESO 293- 34	00 06	-41 29	5.4			6.1	1.50	0.51
.......	.. ..	.. ..	.	.	...	...	...	...
.......	.. ..	.. ..	.	.	...	...	...	...
IC 2531	09 59	-29 36	1c		J P, SA I, S	5.3	1.84	1.09
NGC 3079	10 01	+55 40	1		S	5.03	1.90	0.74
NGC 3098	10 02	+24 42	3f	3	SA I	-2.0	1.36	0.57
.......	.. ..	.. ..	.	.	...	...	...	...
.......	.. ..	.. ..	.	.	...	...	...	...

The complete Table 6 with its notes and Table 7 are published in electronic form at CDS.

Bars as origin for b/p bulges (Combes et al. 1990; Raha et al. 1991; Pfenniger & Friedli 1991) are supported by the similarity in frequency distributions of galaxies with b/p bulges and barred galaxies. The fraction of b/p bulges is large enough to explain the b/p bulges by bars. The higher frequency of barred galaxies can easily be explained by the aspect angle of bars. End-on bars result for edge-on galaxies in elliptical shaped bulges, bars with intermediate aspect angles in boxy bulges, and edge-on bars in peanut bulges (Combes et al. 1990; Pfenniger & Friedli 1991, Paper II). Therefore our statistical results are consistent with recent studies stating a strong correlation of bars and b/p bulges. Bureau & Freeman (1999) and Merrifield & Kuijken (1999) find in observations of gas kinematics the characteristic "figure-of-eight'' rotation curve, which is a strong signature for the presence of a bar, in many galaxies with b/p bulges. Direct kinematic evidence for streaming motions of a bar in two galaxies with a b/p bulge are reported by Veilleux et al. (1999). Kinematical bar diagnostics in edge-on spiral galaxies using simulations of families of periodic orbits and hydrodynamical simulations confirm the connection between bars and b/p bulges (Bureau & Athanassoula 1999; Athanassoula & Bureau 1999). Additionally, we find in our NIR study (Paper II) a strong correlation of bar signatures with b/p bulges. However, by finding a few bulges with a very complex morphology, we do not exclude that some b/p bulges result from a recent merger event (Dettmar & Lütticke 1999, Paper III).

The optical CCD images verify our classification of bulges and thereby the statistics derived by the DSS images. For the analysis of faint structures and more objective parameters of b/p bulges (e.g. parameter for the depression at the minor axis) CCD images are necessary. The best parameter for an objective classification of bulges would be the minimum of the a₄ parameter determined in the NIR (minimizing the dust influence) and after subtraction of a modelled disk and bar. However, models of high quality need images with a high signal-to-noise ratio. This would result in unreasonably observing time for the whole RC3 sample. Therefore the classification by visual inspection seems to be the best method until CCD data of the whole sky are available (e.g. for the northern sky: Sloan Digital Sky Survey).

Furthermore, it is shown by NIR observations that dust is not an important factor for the classification of b/p bulges. Therefore they are indeed present in disks of late type spirals as pointed out previously by Dettmar & Ferrara (1996). This result is not in contrast to Baggett & MacKenty (1996) who verify b/p structures in the NIR only in four out of six galaxies because their sample contains several misclassifications from older literature lists (Sects. 2.1 and 5). The comparison of optical and NIR images of b/p bulges reveals that there is no difference between the shape of bulges in different wavelengths bands or the differences are smaller than the uncertainties of the classification.

Additionally, as a byproduct a new catalogue of edge-on galaxies with $D_{25} > 2\hbox{$^\prime$ }$confirmed by visual inspection is formed by our classification of bulges. This method gives a homogeneous sample of highly inclined galaxies compared to samples selected only by axis ratio and morphological type. This catalogue includes all galaxies with a bulge of types 1 - 4 and 5.2 - 5.5 (Table 6).