6 Discussion

For bright-line determinations of oxygen abundances, the reported values for each galaxy are means of values derived from the McGaugh and the Pilyugin calibrations. New oxygen abundances are reported for eight galaxies from the southern sample: in the Cen A group, A1334-277, DDO 161, and NGC 5264; in the Scl group, AM0106-382, ESO347-G017 (but see Appendix A), and ESO348-G009; and finally, ESO358-G060, IC 2032, and NGC 2915.

   
6.1 Comments on individual galaxies

All of the galaxies in the southern sample are presented below in alphabetical order.

Figure 5: Difference in oxygen abundance from various methods versus log O32 (top panel), and versus log R23 (bottom panel). Each point represents an H II region from dwarf galaxies in the southern sample. "Direct'' denotes oxygen abundances derived from [O III] $\lambda 4363$ measurements, "McG'' denotes oxygen abundances derived using the bright-line method by McGaugh (1991), and "Pil'' denotes oxygen abundances derived using the bright-line method by Pilyugin (2000,2001a).

6.1.1 A0355-465 (ESO249-G032)

Webster et al. (1983) obtained spectra of two H II regions (#A and #B). Unfortunately, their [O III] $\lambda 4363$ measurement in H II region #B is very uncertain with their quoted error in excess of 40%; they derived 12+log(O/H) = 8.41 and $\rm log(N/O) = -1.70$. From our spectrum of H II region #B, we obtain 12+log(O/H) = 8.12 and log(N/O) = -1.61 using the bright-line method. While our I([O III] $\lambda\lambda 4959,5007$)/I(H$\beta$) ratio is about 80% of that reported by Webster et al. (1983), our I([O II] $\lambda 3727$)/I(H$\beta$) ratio is about a factor of two lower.

There is currently no velocity listed for this galaxy in the NED database. From the emission lines in the spectrum, the heliocentric velocity was estimated to be 1168 km s^-1. Assuming that A0355-465 is isolated and that the Hubble constant has a value of 75 km s^-1 Mpc^-1, the estimated distance is 15.6 Mpc and the distance modulus is 31.0. This distance is comparable to that of the Virgo Cluster and the Fornax Cluster. The estimated absolute magnitude in B is -14.7, which is similar to that of the Local Group dI NGC 6822. The angular dimensions from the NED database are $2\hbox{$.\mkern-4mu^\prime$ }3$ by $1\hbox{$.\mkern-4mu^\prime$ }4$, corresponding to linear dimensions 10.4 kpc by 6.3 kpc. This is comparable to the dimensions of the Magellanic dwarf NGC 4532 (VCC 1554) in the Virgo Cluster (Hoffman et al. 1999).

An upper limit to the total 21-cm flux was obtained by Longmore et al. (1982), though another measurement should be obtained for confirmation. Nevertheless, A0355-465 exhibits one of the lowest $M_{\rm HI}/L_B$ in the present sample. This dwarf lies $9\hbox{$.\mkern-4mu^\prime$ }6$ away from the face-on SB(rs)cd galaxy NGC 1493, which has a measured heliocentric velocity of 1054 km s^-1, roughly similar to our estimated velocity for A0355-465. In fact, the dwarf may be a member of a very small group of galaxies with at least four catalogued members (LGG 106; Garcia 1993).

6.1.2 A1243-335 (ESO381-G020)

Webster et al. (1983) identified three H II regions in this Cen A group dwarf irregular galaxy. The brightest (labelled #A) was found in the southeast corner of the galaxy; H II regions #B and #C were found in the northwest and southwest part of the galaxy, respectively. They obtained spectroscopy for all three H II regions and detected [O III] $\lambda 4363$ only in H II #A. While their quoted uncertainty for the [O III] $\lambda 4363$ detection is rather large (>40%), their derived oxygen abundance was 12+log(O/H) = 8.04. H II regions #A and #B appear to be spatially coincident with local maxima in H I (Côté et al. 2000). The H I extent is 4.5 times larger than the optical galaxy.

Our spectrum of H II region #A revealed the bright oxygen lines, but no [O III] $\lambda 4363$. The derived bright-line oxygen abundance is 12+log(O/H) = 7.78, which is almost 0.3 dex lower than the Webster et al. (1983) value. The adopted nitrogen-to-oxygen abundance ratio is log(N/O) = -1.62.

6.1.3 A1324-412 (ESO324-G024)

For this Cen A group dwarf irregular, Côté et al. (1997) measured the total H I flux, and Karachentsev et al. (2002b) derived a TRGB distance. However, our spectrum only revealed H$\beta$ and H$\alpha$ emission; the subsequent derived reddening is E(B-V) = +0.14 mag.

6.1.4 A1334-277 (ESO444-G084)

The oxygen abundance reported here (12+log(O/H) = 7.40) is the first ever published for this Cen A group member and is the third most metal-poor galaxy in the present study. While the upper limit to the [N II]/[O II] intensity ratio may indicate an upper branch abundance for this Cen A group dwarf, a low metallicity is assumed, because the I([N II] $\lambda 6583$)/I(H$\alpha$) and I([S II] $\lambda 6716$)/I(H$\alpha$) ratios are small, i.e., less than 3% and 2%, respectively. Indeed, bright-line calibrations using I([N II] $\lambda 6583$)/I(H$\alpha$) (van Zee et al. 1998b; Denicoló et al. 2002) yield lower branch abundances.

While this dwarf galaxy is relatively metal-poor, it is not particularly isolated. The TRGB distance of this dI is 4.61 Mpc, which puts this galaxy 380 kpc from the spiral galaxy M 83 (Karachentsev et al. 2002b). While the H I extent is almost six times larger than the optical extent (Côté et al. 2000), the $M_{\rm HI}/L_B$ ratio is normal for its luminosity compared to typical dIs (e.g., Roberts & Haynes 1994; Skillman 1996; Pildis et al. 1997).

6.1.5 A1346-358 (ESO383-G087)

Webster et al. (1983) obtained spectroscopy of three H II regions (A, B, and C), but did not detect [O III] $\lambda 4363$ at all. They estimated the oxygen abundance to be $\rm 12{+}log(O/H) \approx 8.3$. In our spectrum of H II region #A, [O III] $\lambda 4363$ was detected and the subsequent oxygen abundance is 12+log(O/H) = $8.19 \pm 0.06$, or about one-fifth of the solar value. The nitrogen- and neon-to-oxygen ratios are: $\rm log(N/O) = -1.37 \pm 0.08$ and $\rm log(Ne/O) = -0.46 \pm 0.05$. The latter is about 0.2 dex larger than values obtained for blue compact dwarf galaxies at similar oxygen abundance (Izotov & Thuan 1999). This dwarf lies (in projection) approximately between the elliptical Cen A and the spiral M 83. The proximity to these two giant galaxies may explain why the $M_{\rm HI}/L_B$ value for the dwarf is several times lower than typical values for dIs at comparable luminosities (e.g., Staveley-Smith et al. 1992).

6.1.6 AM0106-382

Aside from the properties listed in Table 1, and the inclusion of this galaxy in the H I catalog by Côté et al. (1997), there is little in the literature describing further this dwarf irregular in the Sculptor group. The adopted oxygen abundance is $\rm 12{+}log(O/H) = 7.60$ and the nitrogen-to-oxygen ratio is $\rm log(N/O) = -1.42$. These values are the first ever published for this galaxy.

6.1.7 DDO 161

This Cen A group dwarf galaxy was studied in some detail by Karachentsev et al. (1981), but the oxygen abundance here is believed to be the first ever published for this galaxy. The H I extent is almost five times larger than the optical extent (Côté et al. 2000). The spectrum presented here was taken with the east-west slit placed so that the most northern star-forming clump was observed. The adopted oxygen abundance is 12+log(O/H) = 8.06 and the nitrogen-to-oxygen ratio is log(N/O) = -1.91.

6.1.8 ESO302-G014 (AM0349-383)

Little is described of this galaxy in the literature. Because only [O III] $\lambda 5007$ and H$\alpha$ were detected in the spectrum, there is no subsequent analysis.

6.1.9 ESO347-G017, ESO348-G009

Aside from the properties listed in Table 1, and the inclusion of this galaxy in the H I catalog by Côté et al. (1997), there is little in the literature describing further these two dwarf galaxies in the Sculptor group. The adopted (bright-line) oxygen abundances, respectively, are: 12+log(O/H) = 7.84, and 7.98. The nitrogen-to-oxygen ratios, respectively, are: log(N/O) = -1.49, and -1.60. The neon-to-oxygen ratio for ESO347-G017 is log(Ne/O) = -0.53.

   
6.1.10 ESO358-G060 (FCC 302)

This dwarf is a confirmed member of the Fornax Cluster (FCC 302; Drinkwater et al. 2001; Schröder et al. 2001). With an assumed distance modulus of 31.5 (Mould et al. 2000), the absolute magnitude is $M_B \simeq -15.7$. The derived H I mass to blue luminosity ratio is $\simeq$4, which is the largest in the present sample.

The derived oxygen abundance (12+log(O/H) = 7.32; $\sim$2.8% of solar) makes this galaxy the most metal-poor dwarf in the present study. The abundance is comparable to the second most metal-poor galaxy known, i.e., the blue compact dwarf galaxy SBS 0335-052 with 12+log(O/H) = 7.33 (Melnick et al. 1992; Izotov et al. 1997a). If the abundance is truly low, additional spectroscopy of this dI with a larger telescope should reveal [O III] $\lambda 4363$, which should still be observable even at the distance of $\sim$20 Mpc; see Lee et al. (2003b) for [O III] $\lambda 4363$ detections in dwarf galaxies at the distance of the Virgo Cluster. The adopted nitrogen-to-oxygen ratio is log(N/O) = -1.24.

6.1.11 IC 1613

Located in the Local Group, IC 1613 is a well-studied dI whose position is near the celestial equator. The $M_{\rm HI}/L_B$ value for IC 1613 is typical for its luminosity compared to other dIs. Its H II regions have been catalogued by Sandage (1971), Lequeux et al. (1987), Price et al. (1990), and Hodge et al. (1990). The brightest H II region was identified by Sandage as #3 (S3), which has been labelled H II #37 by Hodge et al. (1990). Talent (1980), D'Odorico & Rosa (1982), and Davidson & Kinman (1982) obtained photoelectric spectrophotometry of S3, and all three studies showed that the ionization source was a single Wolf-Rayet (W-R) star. Cole et al. (1999) and Dolphin et al. (2001b) obtained resolved stellar photometry and derived distances from the tip of the red giant branch and the red clump, respectively; these distance measurements are in agreement (730 kpc). Rosado et al. (2001) and Valdez-Gutiérrez et al. (2001) obtained Fabry-Perot interferometry at H$\alpha$ and [S II] $\lambda\lambda 6716,6731$ to study the kinematics of S3 and S8 and found that the nebular emission exhibits two-lobe structure with superbubbles covering the entire galaxy.

Talent (1980) and Davidson & Kinman (1982) obtained [O III] $\lambda 4363$ measurements of H II #37, and derived 12+log(O/H) = 7.86 and 7.87, respectively, and $\rm log(N/O) = -1.21$ (upper limit) and -0.99, respectively. Hunter & Gallagher (1985) and Peimbert et al. (1988) measured additional spectrophotometry for other H II regions, of which one was identified as a supernova remnant (H II #49, Sandage #8 or S8; D'Odorico et al. 1980; Rosado et al. 2001). For H II #49, D'Odorico & Dopita (1983) derived $\rm 12{+}log(O/H) = 7.60$ and $\rm log(N/O) = -0.90$, while Peimbert et al. (1988) derived $\rm 12{+}log(O/H) = 7.83$ and $\rm log(N/O) = -1.15$.

Our spectrum of H II #37 (Fig. 2) also exhibits broad Wolf-Rayet features near 4471, 4686, and 5800 Å. Due to a strong [O III] $\lambda 4363$ detection, an oxygen abundance of $\rm 12{+}log(O/H) = 7.62 \pm 0.05$ was derived, which is the adopted value here, and 1.6$\sigma$ lower than the [O III] $\lambda 4363$ abundance (7.70) obtained most recently by Kingsburgh & Barlow (1995). Our log(N/O) = $-1.13 \pm 0.18$ value is in agreement with the value (-0.96) obtained by Kingsburgh & Barlow (1995). Our adopted neon-to-oxygen ratio is log(Ne/O) = $-0.60 \pm 0.05$, which is in rough agreement with determinations by Izotov & Thuan (1999) for other dwarf galaxies at similar oxygen abundance. A new spectrum for H II #13 did not reveal [O III] $\lambda 4363$. Oxygen abundances and N/O values derived using the bright-line method for H II regions #13 and #37 are in agreement.

6.1.12 IC 2032

Not much is known about this dwarf, although it may be a member of the Dorado group; see Carrasco et al. (2001) and references therein. The H I properties of the galaxy have been studied and listed in the survey by Huchtmeier et al. (2000a). There is a bright "shell'' of star formation, giving the galaxy a "cometary'' appearance. The spectrum presented here was taken with the east-west slit placed along the northern part of the shell, but [O III] $\lambda 4363$ was not detected. The resulting (bright-line) oxygen abundance (12+log(O/H) = 7.97) is the first ever published for this dwarf. The adopted nitrogen- and neon-to-oxygen ratios are: $\rm log(N/O) = -1.37$, and $\rm log(Ne/O) = -0.21$, respectively. The derived Ne/O ratio is somewhat large compared to known values. However, the $I(\hbox{[Ne III]$\lambda 3869$ })/I(\hbox{H$\beta$ })$ ratio is comparable to that observed in the nearby dI IC 4662 in the Local Volume (Heydari-Malayeri et al. 1990). Taking their published intensity ratios and temperatures for IC 4662, the derived value of log(Ne/O) would be about zero.

6.1.13 IC 5152

Recent attention has been paid to this field dI, although a very bright foreground star in the northwest corner of the galaxy prevents deep imaging from taking place. Talent (1980) and Webster et al. (1983) obtained spectroscopy of the brightest H II region. Zijlstra & Minniti (1999) obtained deep stellar photometry and constructed colour-magnitude diagrams; their field included the southeast quadrant of the galaxy to avoid the bright foreground star. A young population is indicated by the presence of H II regions and ultraviolet bright stars. A distance was inferred after comparison with colour-magnitude diagrams of other similar dIs and with theoretical isochrones. For the intermediate-age stars a metallicity of one-tenth solar was inferred. With resolved photometry from the HST, Karachentsev et al. (2002c) have recently derived a TRGB distance of 2.07 Mpc.

The brightest H II region is located at the northeast corner of the dwarf and is labelled H II region #A by Webster et al. (1983); this H II region is believed to be the same one measured by Talent (1980). Our spectrum of H II region #A is the most up-to-date since the work of Talent (1980) and Webster et al. (1983). Talent (1980) measured [O III] $\lambda 4363$ and obtained an oxygen abundance of 12+log(O/H) = 8.36 and $\rm log(N/O) = -1.52$. Webster et al. (1983) derived a bright-line abundance of 12+log(O/H) = 8.35. We obtained an [O III] $\lambda 4363$ measurement of H II region #A and the resulting oxygen abundance is 12+log(O/H) = $7.92 \pm 0.07$. Our value of the oxygen abundance is about one-tenth of the solar value, which is consistent with the stellar metallicity inferred by Zijlstra & Minniti (1999), but 2.7 times lower than the values reported by Talent and Webster et al. Our reported value of log(N/O) = $-1.05 \pm 0.12$ is three times larger than the value given by Talent (1980). Also, our adopted neon-to-oxygen ratio is log(Ne/O) = $-0.69 \pm 0.08$, which is in rough agreement with determinations by Izotov & Thuan (1999) for other dwarf galaxies at similar oxygen abundance.

Additional spectra have been obtained of an H II region in the southwest region of the galaxy by Hidalgo-Gámez & Olofsson (2002). They measured [O III] $\lambda 4363$ for which they derived an oxygen abundance of 12+log(O/H) = 8.2, although their corrected $I(\hbox{H$\alpha$ })/I(\hbox{H$\beta$ })$ratio is over a factor of two lower than the theoretical value for typical H II regions. Future observations of this and other H II regions would be very helpful to confirm the nature of the unusual H II region observed by Hidalgo-Gámez & Olofsson, the relatively large N/O, and the homogeneity of oxygen abundances in this dwarf galaxy. The ground-based stellar photometry obtained by Zijlstra & Minniti (1999) includes the centre and the southeast portion of the dwarf galaxy, which unfortunately avoids the part of the galaxy where H II region #A is located. Deeper resolved photometry would be very useful in gaining further clues about the underlying stellar populations and determining the history of star formation.

6.1.14 NGC 2915

This galaxy is likely a relatively nearby blue compact dwarf galaxy. Sérsic et al. (1977) presented one of the first comprehensive studies using photoelectric photometry, spectrograms, and radio observations. With updated broadband photometry, Meurer et al. (1994) found that the dwarf contained two dominant stellar populations. Current star formation is occurring at the centre where most of the ionized gas and a bright blue population of stars are present. A red diffuse population with an exponential surface brightness profile lies outside of the central region. Meurer et al. (1994) label this galaxy as an amorphous blue compact dwarf galaxy with properties of a dwarf elliptical at large galactocentric radii. Meurer et al. (1996) obtained H I synthesis observations and found that the H I is five times larger in spatial extent than the optical extent defined by the Holmberg radius. The galaxy also has a central bar and spiral arms with a maximum rotation speed of $v_{\rm max} \simeq 85$ km s^-1. Their subsequent modelling shows that the dark matter content is dominant at all radii. Bureau et al. (1999) studied the barred spiral arm structure and found that a rotating triaxial dark matter halo can best explain the H I observations.

The observed spectrum presented here was taken with the east-west slit placement through the centre of the galaxy. Our adopted (average) value of the oxygen abundance is 12+log(O/H) = 8.30. The adopted (average) nitrogen-to-oxygen ratio is log(N/O) = -1.48. From the spectrum of the galaxy nucleus measured by Sérsic et al. (1977), they found I(H$\alpha$)/I([O III] $\lambda\lambda 4959,5007$) = 0.25 and [I(H$\alpha$) + I([N II] $\lambda 6583$)]/I([S II] $\lambda\lambda 6716,6731$) = 5, which they claimed as evidence for a low excitation H II region. Our corresponding ratios are: I(H$\alpha$)/I([O III] $\lambda\lambda 4959,5007$) = 1.2, and [I(H$\alpha$) + I([N II] $\lambda 6583$)]/I([S II] $\lambda\lambda 6716,6731$) = 5.1. Meurer et al. (1994) also obtained a nuclear spectrum and derived an upper limit to the oxygen abundance, 12+log(O/H) $\la$ 8.5, which is consistent with our measurements.

   
6.1.15 NGC 3109

NGC 3109 is the most massive galaxy in the nearby, extremely poor Antlia-Sextans group (van den Bergh 1999; Tully et al. 2002). While considered as dI by some workers, this galaxy may be better described as a Magellanic spiral or, perhaps, even a dwarf spiral (Grebel 2001a,b). Studies of the H I gas and stellar content have been well documented for this galaxy. For example, Carignan (1985) and Jobin & Carignan (1990) obtained optical and H I observations and showed that: (1) compared to similar galaxies (e.g., SMC), the total galaxy luminosity is comparable, but the optical extent is roughly two times larger; (2) the outer parts of the H I gas is warped, likely due to an interaction with the Antlia dwarf galaxy; (3) model fits to the rotation profile show that there is roughly ten times more dark matter than luminous matter; and (4) that the gas is a good tracer of the dark matter distribution. Musella et al. (1997) and Minniti et al. (1999) have obtained distances of 1.36 Mpc and 1.33 Mpc using observations of Cepheid variable stars and the tip of the red giant branch, respectively. Also, Karachentsev et al. (2002c) obtained a distance of 1.33 Mpc from the magnitude of the tip of the red giant branch using HST WFPC2 observations. Grebel et al. (2003) have obtained ${\rm [Fe/H]} \simeq -1.7$ dex for the metallicity of the red giant branch, which points to the metal-poor nature of this galaxy.

Richer & McCall (1992) identified a number of H II regions and planetary nebulae in the eastern section of the galaxy. M. McCall and C. Stevenson carried out spectroscopic measurements at the Steward Observatory of H II region "#5''; see Richer & McCall (1992) for their labelling. An [O III] $\lambda 4363$ detection was reported and a lower branch oxygen abundance was obtained (7.73) by Lee et al. (2003a). For the present work, measurements were carried out for H II region "#6'' (as labelled by Richer & McCall 1992). Using $I(\hbox {[N II]$\lambda 6583$ })/I(\hbox {H$\alpha $ })$ and the calibrations by van Zee et al. (1998b) and Denicoló et al. (2002), the resulting oxygen abundances are consistent with lower branch values, but are $\approx$0.4 dex higher than the values determined from either the McGaugh or the Pilyugin calibration. We adopt here the McGaugh and Pilyugin calibrations, as most of the remaining bright-line oxygen abundances in the southern sample are derived in a similar manner. So, taking the average of the ten values listed in Table 9, the adopted oxygen abundance is 12+log(O/H) = 7.63, which is similar to the value (7.73) determined by Lee et al. (2003a).

   
6.1.16 NGC 5264

NGC 5264 is a Cen A dwarf galaxy with a TRGB distance of 4.53 Mpc (Karachentsev et al. 2002b). Compared to other H II spectra, our spectrum shows relatively low $I(\hbox{[O III]$\lambda 5007$ })/I(\hbox{H$\beta$ })$, whereas $I(\hbox{[O II]$\lambda 3727$ })/I(\hbox{H$\beta$ })$, $I(\hbox {[N II]$\lambda 6583$ })/I(\hbox {H$\alpha $ })$, and $I(\hbox{[S II]$\lambda\lambda 6716,6731$ })/I(\hbox{H$\alpha$ })$ are high. The [N II]/[O II] discriminant indicates an upper branch abundance. Because of the low negative value (-0.599) for $\log~O_{32}$, the oxygen abundance derived using the Pilyugin bright-line method is somewhat uncertain, as his method is calibrated to observations of H II regions with larger values of $\log~O_{32}$. Using $I(\hbox {[N II]$\lambda 6583$ })/I(\hbox {H$\alpha $ })$, the calibrations of van Zee et al. (1998b) and Denicoló et al. (2002) give $\rm 12{+}log(O/H) = 8.68$ and 8.55, respectively, which agree with the upper branch abundances from the McGaugh and the Pilyugin calibrations. Taking an average of the values from each calibration, the adopted oxygen abundance is $\rm 12{+}log(O/H) = 8.61$, which is similar to measured abundances in spiral galaxies. Using the Thurston et al. (1996) method for H II regions in spiral galaxies, the nitrogen-to-oxygen abundance ratio is $\rm log(N/O) = -0.57$.

NGC 5264 has a higher oxygen abundance and a nitrogen-to-oxygen ratio than expected for its galaxy luminosity. This galaxy may be an example of a dwarf spiral (e.g., Schombert et al. 1995), whose luminosity is similar to the SMC, but whose oxygen abundance is a factor of four larger than the SMC. A systematic spectroscopic survey of H II regions in NGC 5264 should show whether or not the galaxy has a radial gradient in oxygen abundance.

Compared to typical dIs, the $M_{\rm HI}/L_B$ value for NGC 5264 is somewhat low compared to other late-type galaxies of similar luminosity. NGC 5264 is located one degree to the east of the luminous spiral NGC 5236 (M 83), where an interaction with the latter could remove H I gas from NGC 5264. Interestingly, there is in the vicinity another Cen A group member dI, UGCA 365 (ESO444-G078), which appears to be even closer in projection to M 83 (see Fig. 1 in Karachentsev et al. 2002b). There may be an interaction between M 83, UGCA 365, and NGC 5264, which could explain the comparatively lower H I content in the latter galaxy.

6.1.17 Sag DIG

The Sagittarius dwarf irregular (Sag DIG) galaxy was first discovered by Cesarsky et al. (1977) and later confirmed by Longmore et al. (1978). Strobel et al. (1991) detected what appeared to be three H II regions; only one appears to be a true H II region, whereas the other two sources are likely stars in H$\alpha$ emission. Young & Lo (1997) found that the H I component was more spatially extended than the stellar component. An H I clump of relatively high density is nearly spatially coincident with the H II region. There is no clear rotational motion for the H I gas. Rather, the gas appears to be supported by random motions with a broad- and a narrow-velocity component. Using resolved stellar photometry, Karachentsev et al. (1999) and Lee & Kim (2000) independently obtained TRGB distances of 1.11 Mpc, which places this dwarf at the outer boundaries of the Local Group and makes the membership to the Local Group uncertain. In addition, photometry has revealed that the underlying stellar population is very metal-poor with [Fe/H] $\la -2$, which was also noted by Momany et al. (2002).

Skillman et al. (1989b) attempted deep spectroscopy, but failed to detect [O III] $\lambda 4363$ or [N II] $\lambda 6583$. Saviane et al. (2002) have also recently obtained deep emission-line spectroscopy of the H II region. Despite three hours of total exposure, [O III] $\lambda 4363$ was not detected. Although Saviane et al. (2002) used a larger telescope, they did not detect [O III] $\lambda 4363$. The oxygen abundance derived here (12+log(O/H) = 7.39) agrees with the bright-line abundance obtained by Skillman et al. (1989b) and with the upper end of the range of bright-line abundances derived by Saviane et al. (2002). The oxygen-poor interstellar medium is consistent with the very low iron abundance seen in the stars. Although $\approx$0.2 dex lower than the value obtained by Saviane et al. (2002), our derived $\rm log(N/O) = -1.63$ is almost identical with the near-constant value for blue compact dwarf galaxies (Izotov & Thuan 1999), even though Sag DIG is a very low-luminosity dI with fairly quiescent star formation. However, the $M_{\rm HI}/L_B$ value and the relatively low N/O may be indicative of a very recent (and, perhaps, small) burst, which is borne out by the young stars seen in resolved photometry (Lee & Kim 2000).

6.1.18 UGCA 442

Miller (1996) obtained H$\alpha$ imaging of this edge-on galaxy in the Scl group and found five H II regions. Spectroscopy for two H II regions in the southwest part of the disk (#2 and #4 as labelled by Miller 1996) did not reveal [O III] $\lambda 4363$; he derived an average bright-line oxygen abundance of $\rm 12{+}log(O/H) = 7.90$. These two H II regions appear to be in a part of the galaxy where Côté et al. (2000) found a local maximum in H I; the H I extent is about a factor of five larger than the optical extent. Our spectrum of H II region #2 (labelling by Miller 1996) also did not reveal [O III] $\lambda 4363$, and our subsequent bright-line oxygen abundance (12+log(O/H) = 7.85) agrees with the value obtained by Miller (1996). The adopted nitrogen-to-oxygen ratio is log(N/O) = -1.41. The $M_{\rm HI}/L_B$ value for this dwarf is comparable to typical dIs at the given luminosity.

6.2 Metallicity-luminosity relation

Skillman et al. (1989a) and Richer & McCall (1995) showed for dIs that metallicity in the form of oxygen abundances increase with the galaxy luminosity in B. This relationship has been interpreted as being representative of a relationship between metallicity and mass, at least where stellar mass is concerned. The metallicity-luminosity diagram is shown in Fig. 6. Plotted are the set of dwarfs in the control sample from Lee et al. (2003a). A new fit to galaxies with [O III] $\lambda 4363$ abundances and well-measured distances (i.e., control sample excluding NGC 3109, and new data for IC 1613 and IC 5152) leaves the slope and intercept essentially unchanged within the stated errors compared to the values obtained in the fits by Richer & McCall (1995) and Lee et al. (2003a).

Figure 6: Oxygen abundance versus galaxy luminosity in B. Filled circles mark the control sample of dIs (Lee et al. 2003a). The best-fit to the control sample is shown as a solid line. For the present sample of southern dwarfs, crosses and open squares mark galaxies for which distances are measured and estimated, respectively. In addition, galaxies for which [O III] $\lambda 4363$ has been measured are marked with open circles. For galaxies with [O III] $\lambda 4363$ measurements, the oxygen abundance has an uncertainty of at most 0.1 dex. The remaining galaxies without [O III] $\lambda 4363$ measurements have an uncertainty of 0.2 dex in oxygen abundance. A typical uncertainty of 0.2 mag in absolute magnitude accounts for the various methods used to determine distances for dIs in both samples.

Plotted also are dwarfs from the southern sample, combining oxygen abundances derived using the bright-line method and both measured and estimated distances. Most dIs in the present sample have oxygen abundances consistent with the metallicity-luminosity relation defined by dIs in the control sample. Because of the excellent agreement with the metallicity-luminosity relation, IC 5152 will in the future be added to the growing list of galaxies in the control sample with [O III] $\lambda 4363$ measurements and distances derived from resolved stellar photometry (see Sect. 2.2).

There may be an impression that the metallicity-luminosity relation exhibits greater scatter at the range of luminosities shown. At lower galaxy luminosities, Sag DIG and A1334-277 have measured distances, but one should keep in mind that their oxygen abundances were obtained using the bright-line method. ESO358-G060 is an outlier, which appears to be too bright in Bluminosity for its low oxygen abundance. If the oxygen abundance is confirmed, this galaxy would resemble a blue compact dwarf galaxy, as galaxies representative of this type lie mostly below the metallicity-luminosity relation (e.g., Kunth & Östlin 2000). At $M_B \approx -16$, the adopted oxygen abundance for NGC 5264 (Sect. 6.1.16) is approximately 0.2 to 0.3 dex higher than expected at the given luminosity. Additional H II region spectra would be useful in confirming this result.

6.3 Relative nitrogen to oxygen abundances

A plot of log(N/O) versus log(O/H) is shown for star-forming dwarf galaxies in Fig. 7. In general, the present data overlaps with the loci defined by dwarfs in the control sample (Lee et al. 2003a) and by other dwarf galaxies (Garnett 1990; Kobulnicky & Skillman 1996; van Zee et al. 1997; Izotov & Thuan 1999), although there are a few H II regions with elevated values of log(N/O) for their oxygen abundance. There still appears to be a great deal of dispersion in N/O values at a given oxygen abundance. For IC 1613 H II#37 and IC 5152 H II#A, their log(N/O) values are somewhat elevated compared to the mean value for H II from dIs in the control sample. For NGC 5264, the oxygen and nitrogen abundances are in better agreement with values found in spiral galaxies; see the compilation by Henry et al. (2000).

Figure 7: Nitrogen-to-oxygen abundance ratio versus oxygen abundance. Filled circles indicate H II regions from the control sample of dIs, whose oxygen abundances were obtained directly from measurements of the [O III] $\lambda 4363$ emission line. Open triangles represent other star-forming dwarf galaxies from Garnett (1990), Kobulnicky & Skillman (1996), van Zee et al. (1997), and Izotov & Thuan (1999). Crosses mark H II regions from dwarfs in the southern sample; open circles denote galaxies where [O III] $\lambda 4363$ was detected. Labelled are IC 1613 H II#37 and IC 5152 H II#A for their unusually high N/O ratios. Two models for the production of nitrogen are shown (Vila-Costas & Edmunds 1993); the solid line is a model with both primary and secondary production, whereas the dashed line is a model for secondary production. The solar value of N/O (Grevesse et al. 1996) is also indicated. Typical errors in log(O/H) and log(N/O) are shown at the lower right.