4 Results

4.1 Statistics

About 3/4 of the Sdm and Sm galaxies are not at all classifiable and the rest is highly uncertain, because the bulges of these galaxies -- if they have any -- are very small and faint and the influence of dust is very strong. Other structures like bars or star forming regions are more dominant. Therefore the selected sample is reduced to disk galaxies ranging from S0-Sd ( -3.5 < T < 7.5) resulting in 1224 galaxies which will be investigated in the following.

40.0% of these bulges are unclassifiable, almost 3/4 of them due to an inclination which is too far from edge-on. This is an expected effect since the selection criterion of our sample for the inclination is a lower limit (see above). Only a few galaxies with a classifiable bulge in the sample have an axis ratio which is near to the sample limit of $\log R_{25}\!=\!0.30$ (S0) and $\log R_{25}\!=\!0.35$ (Sa - Sd), respectively. This shows that the limits are well chosen. There are only two peculiar edge-on galaxies known (with $\log D_{25} > 2\hbox{$^\prime$ }$) which have classifiable bulges and are not in our sample due to their low axis ratio (IC 2560 and NGC 7123). Therefore our list of galaxies with b/p bulges and a diameter larger than 2 $\hbox {$^\prime $ }$ is almost complete and the number of non-included b/p bulges is small.

Stars in the foreground (type 5.2) are responsible for 13% of unclassifiable bulges. These stars can influence the projected light from the galaxy in a way that the shape of the bulge cannot be identified. Dust lanes (5.4) prevent classifications in the same way, but only for a few galaxies, in which the dust lane has a large extension in comparison with the bulge size (see also Sect. 4.3). Their fraction decreases from late types and is equal to zero for galaxies earlier than Sbc. Scd and Sd galaxies have too small bulges (5.5) for classification, if their diameter is around the limit of 2 $\hbox {$^\prime $ }$and their inclination is near to $90^{\circ}$. At this orientation the measured diameter (D₂₅) and therefore the ratio of disk to bulge length is maximal due to optical depth effects (Xilouris et al. 1999). The result is that galaxies of the same morphological type and diameter (D₂₅) have bulges of different sizes which depend on the inclination angle. However, the fraction of galaxies in the sample having such small bulges is negligible (Table 1). In a few cases galaxies are in the southern Galactic plane (5.6), where the DSS has a higher surface brightness limit due to the use of the SERC V plates (see above). Therefore the images of the galaxies in this region have frequently a very small signal-to-noise ratio and the bulges appear to be too faint for classification. Perturbations (5.3) by interactions are only in a very few cases the reason for bulges being not classifiable.

   

Table 1: Reasons for the unclassifiability of bulges

bulge type	5.1	5.2	5.3	5.4	5.5	5.6
frequency	74%	13%	2%	4%	4%	3%

   

Table 2: Classifiable bulges binned by morphological type

	1	21	3	4	$\Sigma$
S0/S0a	5	21	28	78	132
Sa/Sab	6	13	23	52	94
Sb/Sbc	14	44	65	136	259
Sc/Scd	5	28	53	100	186
Sd	0	9	16	38	63
$\Sigma$	30	115	185	404	734
%	4.1	15.7	25.2	55.0	100

¹: 10.4% of all type 2 are 2+.

734 galaxies of our sample are classifiable (Table 2). From these galaxies we get a frequency of $45.0(\pm4.5)$% b/p bulges (type 1 + 2 + 3) (Table 2). The distribution of all galaxies with b/p bulges binned by morphological type shows a weak maximum at Sb/Sbc galaxies (Fig. 2). The smallest fraction of b/p bulges is observed for early and late type disk galaxies. The fractions range from 40 to 48% in the maximum. From number statistics the errors are 4 - 7% for the frequencies of b/p bulges in each bin. Within these errors there is no dependence of the morphological type discernable. Regarding only galaxies with b/p types 1 and 2 the distribution is nearly the same. These results are in some aspects in contrast to the former investigations (see Sect. 5).

Figure 2: Frequency of b/p bulges binned by morphological type derived from Table 2. Dotted lines: type 1. Dashed lines: types 1 + 2. Solid lines: all b/p types ( ${\bf 1 - 3}$)

4.2 CCD observations

The optical CCD images were obtained in several observing runs between 1985 and 1998 at Lowell Observatory (1.06 m), Calar Alto (1.23 m), and ESO/ La Silla (0.9 m, 1.54 m, 2.2 m, and NTT). Standard reduction techniques for bias subtraction and flatfielding were applied. Individual short exposure frames of the galaxy were combined. The data are partly published in Barteldrees & Dettmar (1994), and Pohlen et al. (2000). Further data will be reported together with follow-up observations of b/p bulges in a separate paper (Lütticke et al. in prep.). In total we have observed 74 galaxies of our investigated RC3 sample up to now.

Only small differences (< 10%) between the classification of bulges comparing DSS (RC3 sample) and CCD images (Fig. 3 or Fig. 1 [top] and Fig. 4 [top]) are detected. Three bulges turned out to be type 4 rather than type 3 and two bulges are type 3 rather than type 4. This could be explained by the higher resolution of the CCD images and the unsharp nature of the border between these two classes. However, the total frequency of b/p bulges is not changed. Bulges of type 2+ in the DSS images can be classified more accurately by the high resolution of the CCD images. If there is any depression along the minor axis (e.g. NGC 1886, Fig. 3), the bulge type is changed to 1. If not, the bulge can be classified as (type 2). Bulges which have only a depression along the minor axis on one side are still classified as type 2+.

Figure 3: Inspecting the DSS image (top) it is hard to decide, if there are depressions along the minor axis of NGC 1886 (first classified as bulge type 2+) on both sides of the galaxy plane. The contour plots of the CCD images (NGC 1886: ESO/2.2 m, 30 min in r) reveal that NGC 1886 has a peanut bulge (bulge type 1)

The comparison of the CCD with the DSS images shows that neither the low resolution, nor the saturated bright parts of some galaxies, nor the low signal-to-noise ratio, nor unresolved structures (small stars in the foreground) strongly influence the classification. Therefore the results of the statistics, which are derived from the RC3 sample and inspected with the DSS, are strongly supported by the CCD observations. Furthermore, the 33 galaxies observed in different optical filters reveal the same shape of the bulge. Therefore the classification of bulges at optical wavelengths is independent of the filter.

47 bulges of galaxies not included in the investigated RC3 sample (45 have $D_{25} < 2\hbox{$^\prime$ }$, additionally, one galaxy has $\log R_{25} < 0.35$, and one S0 galaxy has $\log R_{25} < 0.30$) are classified on CCD images of our observing runs (Table 7).

4.3 NIR observations

The essential advantage of investigations in the NIR is the small extinction by dust at these wavelengths (Knapen et al. 1991). Therefore the possible influence of the dust lane near the galactic plane in edge-on galaxies on the bulge shape is largely reduced by NIR observations. Our NIR sample of galaxies with classifiable bulges consists of 60 galaxies (Lütticke et al. 2000a, hereafter Paper II). It reveals that 75% of the bulges have the same bulge type in the optical as well as in the NIR (Table 3). 21% of the bulges are classified in the NIR to the next lower class, these bulges are less box-shaped. The change of the bulge type is for two bulges in the opposite way, one of them from a boxy to a peanut bulge likely due to the low resolution and saturation of the DSS image of this galaxy (NGC 5166).

   

Table 3: Differences in the classification of bulges in the optical and NIR

opt. bulge type	4	4	3	3	3	2	2	2	1	1
NIR bulge type	4	3	4	3	2	3	2	1	2	1
no. of galaxies	12	1	7	9	0	5	12	1	0	9

Figure 4: The peanut bulge of ESO 443-42 is prominent in optical CCD images as well as in the NIR. Top: 0.9 m/ESO, 10 min in R. Bottom: 2.2 m/ESO, 23 min in K'

All peanut bulges prominent in the optical have the same shape in the NIR (e.g. ESO 443-42, Fig. 4). A detailed inspection of the galaxies having less boxy bulges in the NIR shows that dust is a possible explanation only for up to one third of these bulges. Differences in classification can be explained mainly by general uncertainties in the classification, the low resolution in the DSS -- there is no difference between classifications derived from optical CCD and NIR images --, and low quality of some NIR images (bad seeing). The bulges whose classes are changed can frequently be marked as intermediate types (types ${\bf 2 - 3}$ and ${\bf 3 - 4}$), respectively. Additionally, seeing conditions around 3 $\hbox{$^{\prime\prime}$ }$ weaken the boxy structure and over 50% of the bulges whose class changes were observed under such conditions.

Images of galaxies observed in all three NIR filters J, H, and Kreveal no difference in the bulge shape. Therefore the bulge classification seems to be same in a large region of wavelengths from the optical (350 nm) to the NIR (2.2 $\mu$m).