In first instance, the objects selected are within 20 degrees of the galactic center. Three groups are distinguished. The first consists of nebulae for which the spectra were observed by us. The second are objects for which spectra have already been published elsewhere. The last group is formed by those nebulae for which the abundances have been available in the literature but for the sake of uniformity we have redone the analysis.
The sample of PNe studied by Gathier et al. (1983) forms the
bulk of this group. These authors demonstrated that at least 90% are
physically associated to the bulge region. The observations of this
sub-sample (
40 objects) were carried out during several different
sessions between 1982 and 1984 with the 3.6 m and 1.52 m telescopes at the
European Southern Observatory (ESO), La Silla, Chile. The remaining 10
nebulae were observed in 1988 with the 1.9 m telescopes at the South African
Astronomical Observatory (SAAO). This last set includes also some newly
discovered PNe discussed by Pottasch et al. (1988), Ratag et
al. (1990) and Kinman et al. (1988).
The ESO observations were made with the 3.60 m telescope (for the nebulae
1-4.2, 2-2.4, 2-3.3 (VLE), 2-3.5 (VLE), 2-4.1, 2-6.1, 3-4.3,
7+1.2, 353+6.2, 354+4.1, 357+7.1 and 359-1.2 (VLE) and with the
1.52 m telescope (for the rest of this group). Both telescopes were equipped
with similar instruments: a Boller and Chivens spectrograph equipped
with an IDS (Image Dissector Scanner) detector, and the same gratings
were used in both cases. Spectra were obtained in two parts: blue
spectra (range 
) were observed at 114 Å/mm, while red
spectra (
) were obtained at 171 Å/mm, with a
comfortable overlap between both spectra for consistency checks.
Exposures consisted of succession of pairs of scans observed through two
apertures (object and sky) separated by 60'', the object being flipped
from one aperture to the other between consecutive pairs of scans. The
only difference between the two telescopes was the size of the aperture
to cope with the different scales and yet achieve similar spectral
resolutions. At the 3.60 m, the apertures were 1
8 
4'' in
size, yielding spectral resolutions (projected slit widths) of 5.3 and
8.0 Å in the blue and in the red, respectively, while at the 1.52 m
they were 4
4''
with corresponding resolutions of 5.0 and 7.5 Å. As a consequence,
the H
fluxes obtained at the 3.60 m telescope are only lower
limits, but they are close to total fluxes for point-like objects
observed at the 1.52 m.
The data reduction was performed in the usual way for IDS data, the two
apertures being reduced independently to the end step of absolute fluxes
and only averaged afterwards. Sky is subtracted from object observed in
the same aperture, sequentially in time, but the integration times of a
few minutes between successive flips are short enough to ensure that sky
variations are negligible in normal observing conditions. Pixel to pixel
sensitivity variations were removed by division by a normalised flat
field, transformation to wavelength scale done with reference He-Ar
spectra obtained before and after the exposures on the sky and
conversion to absolute fluxes (with the shortcomings of small slits
mentioned above) made through spectra of standard stars observed in
exactly the same conditions. An initial correction to the raw data
(elevation to the power 0.96), specific to these IDS, has been applied
before any further reduction to take into account a probable
non-linearity discovered by Rosa (1985). While some hesitation about
the reality of the effect had persisted for some years, evidence is now
conclusive about it and its correction. A similar effect is found for
the Kitt Peak IITS (Intensified Image Dissector Scanner)
(see Peimbert & Torres-Peimbert 1987a) and has been
corrected for in the data of Aller & Keyes (1987) used below.
But more important, the growing CCD data base now shows (for objects
observed with both detectors) that the best agreement is obtained with
corrected IDS data (see Leisy & Dennefeld 1996, and references
therein). One remaining question is the exact value of the [OIII] 5007 to
4959 line ratio, observed with all data sets to be close to 3.00 (
0.08)
while the theoretical value is 2.88.
The remaining objects (2+1.1, 3-4.7, 3-4.3, 3+3.1, 4+6.2, 5+4.1,
357+4.1, 359-2.4, 359+3.4 and 359+4.1) were observed at SAAO with
the 1.9m telescope, equipped with the Cassegrain Grating Spectrograph and
the Intensified Reticon Photon Counting System. Object and Sky were
observed alternately through two entrance apertures of 18
6'', separated by 30'' on the sky. Two gratings were used:
the first one, giving 210 Å/mm covered the full range of
with 8 Å resolution (FWHM) while the second one,
yielding 100 Å/mm, covered either the blue or the red range at
4 Å resolution with the 18 entrance slit. The observing
strategy and the subsequent standard data reduction flow were similar to
the one
described above for the IDS data, therefore providing at the end an
homogeneous set of data.
Based on the newly obtained spectra we also confirm the non-PN nature
of some of the objects. The spectra of PK 3-4.6 (
), PK 6+7.1
(
), and PK 7+1.2 (
) are typical of symbiotic stars as
noticed by Acker et al. (1987). No emission line object was
detected at the position of PK 0-02 (
).
Considering the relatively greater uncertainty in the measurement of
H
, only the Balmer line ratios H
/H and
H/H
were applied in the determination of
interstellar distinction. The theoretical values for the decrement were
taken from Brocklehurst (1971) for case B, 
=
104 K and 
= 104 cm-3, except that for PK
0-2.3 and PK 3-4.9 where the decrement for 
K was used. The adopted E(B-V) is obtained by averaging the
results from H/H ratio and from H/H, weighted
in the ratio 3 : 1, to allow for the greater observational accuracy of the
brighter Balmer lines (except for PK 355-2.4 and PK 353+6.2, where there
are strong indications of self-absorption in Balmer lines). For PK
355-2.4 and PK 353+6.2, the log
[F(H)/F(H)]-log[F(H)] diagram such as discussed
by Osterbrock (1974; see also Barker 1978a,
Fig. 1 (click here)) has been used to estimate the optical depth of H
due to Balmer line self-absorption as well as the "corrected'' E(B-V).
The 
(H
for both nebulae are between 10 and 15. The resulting
values of E(B-V) derived from the Balmer decrement are given in Col. 7 of
Table 2a. Another way of determining the extinction E(B-V) is by comparing
the radio continuum flux density with the H flux. Assuming that the
extinction is all external to the nebula, and by taking into account the
wavelength dependence of the extinction, we can derive the E(B-V),
which is listed in Col. 8 of Table 2a. The values computed using both
methods show, in general, a fair agreement, although in a small number
of cases a large discrepancy is noticed. This discrepancy is largely
caused by the uncertainty in the H flux measurement. Many objects
have been observed with a slit size smaller than the diameters of the
nebulae, and it is certainly very difficult to estimate the fraction of
the unobserved flux. This fraction depends on various factors such as
the detail morphology of the nebula, the exact position on the nebular
image where the slit has been centered on, and the surface distribution
of the H flux. In the cases of large discrepancy, we have
always preferred
the E(B-V) derived from the Balmer decrement method. In other
instances, the average values were generally adopted. These values, in
combination with the reddening curve of Savage & Mathis (1979)
are then used to calculate the reddening-corrected intensities. These are
given in Table 1 relative to 
.
Recently Webster (1988) and Acker et al. (1989a) published the spectra of a large number of PNe. The first survey is intended for the bulge region, while the other is for the whole sky. Those objects have been selected which are not included in the groups described in Sects.\ 2.1 and 2.3. Webster has derived the abundances of He, O and N in her sample by using the empirical ionization correction factors based on the similarities in ionization potential of some ions of different elements. In order to obtain a more homogeneous result we have reanalysed the spectra, using their published line intensities which are reddening-corrected.
Acker et al. published the uncorrected intensities but provide also the
logarithmic reddening corrections c at H. As in the former
sample, it is also possible to derive the extinction by comparing
radio frequency and H fluxes. The method used to derive the
adopted c values is similar to that described in Sect. 2.1. We have
used these c-values together with the reddening curve mentioned in
Sect. 2.1 to derive the unreddened intensities. We however note that,
because of the survey nature of their project, the relatively short
exposure times used by Acker et al. are generally not adequate to derive
high quality abundances from their data which furthermore, were not
corrected for the IDS linearity problem. In addition, the important
[OII]3727 line was not observed. For the discussion on the
uncertainties in
line intensity measurements, the reader is referred to the original
papers.
Objects in this category are taken from the work of Aller & Keyes (1987). They are those nebulae not included in the first group. As mentioned earlier, a similar algorithm to that employed in this work had been used to derive the abundances. It should also be noted that temperatures and densities derived from our data were used by Aller & Keyes (1987) for some objects in common, due to the uncertainties resulting from the low elevation observations at Lick Observatory. Therefore only objects not observed by us were retained from their work.
Further selection has subsequently lead to exclude those nebulae which
are likely to be foreground objects on the basis of luminosity (greater
than 
if placed at the GC) size (greater
than 20'' in diameter), and the available information on the
interstellar extinction along the line of sight. Objects rejected from
the sample are NGC 6369, 
, Hb5, 
, Tc 1, Hb4, Hb6,
NGC 6567, NGC 6578 and NGC 6629.
In total about 110 nebulae are included in the final sample. This is around 85% of the original sample. Although it is still possible that a very few foreground objects remain, the fraction is certainly much less than 10%. The number of PNe as a function of angular distance from the center is shown as a histogram in Fig. 1 (click here). They are mostly within 10 degrees of the Galactic Center. Selection effect, due to the very large extinction, probably explains the lack of objects within 4 degrees of the center.
Figure 1: Histogram of the angular distance from the Galactic center of
the bulge PNe studied in this program