S. A. Pustilnik^1,4 - A. Y. Kniazev^1,2,4 - J. Masegosa^3, - I. Márquez^3* - A. G. Pramskij^1,4 - A. V. Ugryumov^1,4

1 - Special Astrophysical Observatory RAS, Nizhnij Arkhyz, Karachai-Circassia, 369167 Russia
2 - Max Planck Institut für Astronomie, Königstuhl 17, 69117, Heidelberg, Germany
3 - Instituto de Astrofísica de Andalucía, Granada, Spain
4 - Isaac Newton Institute of Chile, SAO Branch

Abstract
We present the results of new spectrophotometry and heavy element abundance determinations for 3 dwarf galaxies UM 151, UM 408 and A 1228+12 (RMB 132). These galaxies have been claimed in the literature to have very low metallicities, corresponding to $\rm log(O/H)+12 \le$ 7.65, that are in the metallicity range of some candidate local young galaxies. We present higher S/N data for these three galaxies. UM 151 and UM 408 have significantly larger metallicities: $\rm log(O/H)+12 =$ 8.5 and 7.93, respectively. For A 1228+12 our new $\rm log(O/H)+12 =$ 7.73 is close to that recalculated from earlier data (7.68). Thus, the rederived metallicities allow us to remove these objects from the list of galaxies with $Z \lesssim 1/20~Z$ $_{\odot}$ .

Key words: galaxies: dwarf - galaxies: stellar content - galaxies: abundances - galaxies: individual: UM 151=Mkn 1169, UM 408, A 1228+12=RMB 132

1 Introduction

Extremely metal-deficient gas-rich galaxies (commonly taken as objects with an ionized gas metallicity $Z \lesssim 1/20~Z_{\odot}$ ) have received much attention recently. The discovery of an unusually low metallicity of the ionized gas in the blue compact galaxy (BCG) I Zw 18 (Mkn 116) (Searle & Sargent 1972) led to the claim that such objects can be considered as the best candidates for truly local young galaxies. However, the search for similar galaxies conducted by several groups (e.g., Sargent & Kunth 1986; Terlevich et al. 1991) has failed to find a large population with such extremely low metallicities. In their review, Kunth & Östlin (2000) summarize and discuss many problems related to very metal-poor galaxies and they present a compilation of all the known gas-rich galaxies with $Z \lesssim 1/20~Z_{\odot}$ . From the analysis of the ratios of C/O and N/O abundances Izotov & Thuan (1999) suggested that the lowest metallicity blue compact galaxies should experience their first episode of star formation (SF) and therefore the hypothesis of youth for these galaxies needs to be carefully studied. In any case the properties of such galaxies best approximate those of young low-mass galaxies formed at the epoch of galaxy formation.

Therefore, very metal-poor galaxies in the Local Universe attract much attention, and multiwavelength studies of such objects have been performed to better identify their nature. However not all such galaxies found in the early studies as very metal-deficient (Kunth & Östlin 2000) have spectroscopic data of sufficient quality for an accurate determination of the metallicity. Since detailed studies of individual galaxies require significant effort and observational time, thus, one should be confident about the very low metallicity of the selected galaxies.

In Paper I (Kniazev et al. 2001a) we reported a revision of 3 BCGs from the list of Kunth & Östlin (2000). For one, the low metallicity value was confirmed while the other two showed an underestimation of 0.4 dex. In this paper we present high S/N spectroscopy of two more BCGs (UM 408 and A 1228+12) from the list by Kunth & Östlin (2000), for which the metallicity determination raised some concerns, and one galaxy (UM 151) from the work by Telles (1996). All these galaxies were claimed as $Z \sim 1/20~Z_{\odot}$ objects based on the results of earlier spectroscopy. As a result of the previous and present work the number of extremely metal-deficient BCGs from the Kunth & Östlin list has decreased by $\sim$ 20%. The remaining objects are in fact representative of very metal-deficient BCGs and deserve detailed multiwavelength studies.

2 Observations and reduction

2.1 Observations

**Table 1:** Journal of observations.
Object	Date	Exposure	Wavelength	Dispersion	Seeing	Airmass	PA
		time [s]	Range [Å]	[Å/pixel]	[arcsec]		[degree]
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)

UM 151	1.02.2000	2 $\times$ 1800	$3700\div7000$	0.81/0.54	1.8	1.50	0
UM 408	2.02.2000	2 $\times$ 1800	$3700\div7000$	0.81/0.54	1.4	1.30	6
A 1228+12	20.01.2001	2 $\times$ 1800	$3700\div7400$	2.4	1.7	1.18	74

**Table 2:** Line intensities of the studied galaxies.
	A 1228+12		UM 151		UM 408
$\lambda_{0}$ (Å) Ion^a	F( $\lambda$ )/F(H $\beta$ )^b	I( $\lambda$ )/I(H $\beta$ )^c	F( $\lambda$ )/F(H $\beta$ )^b	I( $\lambda$ )/I(H $\beta$ )^c	F( $\lambda$ )/F(H $\beta$ )^b	I( $\lambda$ )/I(H $\beta$ )^c
3727 [O II]	0.9503 $\pm$ 0.0722	0.9259 $\pm$ 0.0778	2.4257 $\pm$ 0.1997	2.6919 $\pm$ 0.2466	1.4502 $\pm$ 0.1280	2.1700 $\pm$ 0.2045
3835 H9	--	--	--	--	0.0590 $\pm$ 0.0187	0.0860 $\pm$ 0.0380
3868 [Ne III]	0.3791 $\pm$ 0.0308	0.3693 $\pm$ 0.0324	--	--	0.3812 $\pm$ 0.0409	0.5390 $\pm$ 0.0601
3889 He I + H8	0.1634 $\pm$ 0.0170	0.1958 $\pm$ 0.0275	0.1262 $\pm$ 0.0259	0.1815 $\pm$ 0.0528	0.1474 $\pm$ 0.0246	0.2082 $\pm$ 0.0434
3967 [Ne III] + H7	0.2254 $\pm$ 0.0196	0.2555 $\pm$ 0.0288	0.1020 $\pm$ 0.0285	0.1470 $\pm$ 0.0527	0.2077 $\pm$ 0.0294	0.2842 $\pm$ 0.0507
4101 H $\delta$	0.2210 $\pm$ 0.0196	0.2492 $\pm$ 0.0279	0.2409 $\pm$ 0.0320	0.2947 $\pm$ 0.0518	0.2113 $\pm$ 0.0230	0.2749 $\pm$ 0.0425
4340 H $\gamma$	0.4605 $\pm$ 0.0360	0.4798 $\pm$ 0.0416	0.4086 $\pm$ 0.0389	0.4588 $\pm$ 0.0553	0.4251 $\pm$ 0.0362	0.5056 $\pm$ 0.0492
4363 [O III]	0.1055 $\pm$ 0.0130	0.1028 $\pm$ 0.0131	--	--	0.0851 $\pm$ 0.0170	0.1002 $\pm$ 0.0202
4471 He I	0.0321 $\pm$ 0.0073	0.0313 $\pm$ 0.0073	0.0271 $\pm$ 0.0144	0.0273 $\pm$ 0.0151	0.0266 $\pm$ 0.0113	0.0302 $\pm$ 0.0129
4686 He II	0.0403 $\pm$ 0.0078	0.0392 $\pm$ 0.0078	--	--	--	--
4713 [Ar IV] + He I	0.0208 $\pm$ 0.0073	0.0203 $\pm$ 0.0073	--	--	--	--
4713 [Ar IV]	0.0163 $\pm$ 0.0080	0.0159 $\pm$ 0.0080	--	--	--	--
4861 H $\beta$	1.0000 $\pm$ 0.0771	1.0000 $\pm$ 0.0800	1.0000 $\pm$ 0.0829	1.0000 $\pm$ 0.0899	1.0000 $\pm$ 0.0758	1.0000 $\pm$ 0.0789
4959 [O III]	1.5651 $\pm$ 0.1213	1.5251 $\pm$ 0.1213	0.5509 $\pm$ 0.0493	0.5270 $\pm$ 0.0490	1.7714 $\pm$ 0.1318	1.7174 $\pm$ 0.1291
5007 [O III]	4.6938 $\pm$ 0.3588	4.5736 $\pm$ 0.3590	1.6881 $\pm$ 0.1343	1.6069 $\pm$ 0.1328	5.6848 $\pm$ 0.4445	5.4322 $\pm$ 0.4294
5876 He I	0.0992 $\pm$ 0.0107	0.0967 $\pm$ 0.0109	0.1227 $\pm$ 0.0211	0.1077 $\pm$ 0.0194	0.1538 $\pm$ 0.0311	0.1161 $\pm$ 0.0238
6300 [O I]	0.0230 $\pm$ 0.0096	0.0224 $\pm$ 0.0096	0.0636 $\pm$ 0.0171	0.0540 $\pm$ 0.0152	0.0584 $\pm$ 0.0232	0.0399 $\pm$ 0.0161
6312 [S III]	0.0146 $\pm$ 0.0087	0.0143 $\pm$ 0.0087	0.0161 $\pm$ 0.0119	0.0137 $\pm$ 0.0105	0.0309 $\pm$ 0.0231	0.0211 $\pm$ 0.0159
6364 [O I]	0.0085 $\pm$ 0.0055	0.0083 $\pm$ 0.0055	0.0203 $\pm$ 0.0145	0.0171 $\pm$ 0.0128	--	--
6548 [N II]	0.0102 $\pm$ 0.0068	0.0100 $\pm$ 0.0068	0.1455 $\pm$ 0.0234	0.1212 $\pm$ 0.0206	0.0342 $\pm$ 0.0306	0.0222 $\pm$ 0.0201
6563 H $\alpha$	2.6763 $\pm$ 0.2002	2.6217 $\pm$ 0.2185	3.4333 $\pm$ 0.2673	2.8819 $\pm$ 0.2533	4.3736 $\pm$ 0.3226	2.8264 $\pm$ 0.2289
6584 [N II]	0.0310 $\pm$ 0.0229	0.0302 $\pm$ 0.0229	0.4404 $\pm$ 0.0383	0.3661 $\pm$ 0.0353	0.0957 $\pm$ 0.0245	0.0615 $\pm$ 0.0160
6678 He I	0.0250 $\pm$ 0.0071	0.0244 $\pm$ 0.0071	0.0379 $\pm$ 0.0144	0.0313 $\pm$ 0.0124	0.0506 $\pm$ 0.0220	0.0319 $\pm$ 0.0140
6717 [S II]	0.0936 $\pm$ 0.0109	0.0912 $\pm$ 0.0113	0.4766 $\pm$ 0.0409	0.3926 $\pm$ 0.0378	0.2467 $\pm$ 0.0342	0.1545 $\pm$ 0.0222
6731 [S II]	0.0645 $\pm$ 0.0096	0.0628 $\pm$ 0.0099	0.3423 $\pm$ 0.0328	0.2816 $\pm$ 0.0298	0.1635 $\pm$ 0.0260	0.1021 $\pm$ 0.0168
7065 He I	0.0173 $\pm$ 0.0062	0.0168 $\pm$ 0.0063	--	--	--	--
7136 [Ar III]	0.0430 $\pm$ 0.0083	0.0419 $\pm$ 0.0085	--	--	--	--

C(H $\beta$ ) dex	0.00 $\pm$ 0.10		0.20 $\pm$ 0.10		0.57 $\pm$ 0.10
EW(abs) Å	2.45 $\pm$ 1.08		0.90 $\pm$ 0.67		0.05 $\pm$ 0.94
F(H $\beta$ )^d	232 $\pm$ 13		68 $\pm$ 4		33 $\pm$ 2
EW(H $\beta$ ,emis) Å	93 $\pm$ 5		20 $\pm$ 2		50 $\pm$ 3

^a The rest-frame wavelength in Å; ^b observed flux ratio; ^c corrected flux ratio; ^d in units of 10^-16 ergs s^-1 cm^-2.

The spectra of UM 151 and UM 408 were obtained with the TWIN spectrograph attached to the Cassegrain focus of the 3.5 m telescope at the Calar Alto Observatory (Spain) as supplementary objects to the main program devoted to the detailed spectroscopy of the HSS (Hamburg/SAO Survey, Ugryumov et al. 2001, and references therein) blue compact galaxies. Parameters of these observations are shown in Table 1. The setup used for TWIN was the T07 grating in second order for the blue and T06 in first order for the red arm, that provided dispersions of 54 Å mm^-1 and 36 Å mm^-1 respectively. We have used the CCD detectors SITE12a-11 and SITe6a-11 for the blue and red arms with the 5500 Å beam splitter and a slit width of $2\farcs1$ for UM 151 and $1\farcs2$ for UM 408. The resulting FWHM spectral resolution measured on strong lines were 3.1 Å and 2.5 Å in the blue and red, for UM 151, and 2.9 and 2.6 Å, for UM 408. The scale along the slit was $0\farcs56$ pix^-1.

The spectroscopic data for A 1228+12 (RMB 132) were obtained with the 6 m telescope of the Special Astrophysical Observatory of the Russian Academy of Sciences (SAO RAS). Parameters of these observations are shown in Table 1. The long-slit spectrograph (LSS) (Afanasiev et al. 1995) was used with a Photometrics CCD detector of 1024 $\times$ 1024 pixels with a $24\times24~\mu$ m pixel size. Observations were conducted mainly with the software package NICE in MIDAS, as described by Kniazev & Shergin (1995). The scale along the slit was $0\farcs39$ pix^-1. A grating with 651 grooves mm^-1 and a slit width of $2\arcsec$ were used, giving a FWHM spectral resolution of 8 Å. Two 0.5-hour spectra were obtained, one after the other, each in the spectral range of 3700-6100 Å and 5000-7400 Å, with the same pointing and long slit orientation. Spectra were extracted from the same region and the two spectra were combined to get the full spectrum of the object for analysis.

For each night we obtained biases, flat-fields and illumination correction images. Comparison lamps of Fe-Ne and Ar-Ne-He were used for wavelength calibration for the Calar Alto and SAO data, respectively. For flux calibrations, spectrophotometric standard stars from Bohlin (1996) for the 6 m telescope observations and Oke (1990) for the 3.5 m telescope were used. Average sensitivity curves were produced for each night with rms deviations of $\sim$ 5% in the whole spectral blue + red range.

2.2 Reduction

Standard reduction procedures were used with the IRAF package. Once 2D spectra were wavelength calibrated and sky subtracted, flux calibration was performed by using the average sensitivity curves.

The 1D spectra were extracted with the apertures of $3.6\times2.0\arcsec$ , $6.2\times2.1\arcsec$ and $6.2\times2.1\arcsec$ , respectively for the galaxies A 1228+12, UM 151 and UM 408. The 1D final spectra are shown in Fig. 1. The continuum determination and the measurements of the flux and equivalent width (EW) of spectral lines were performed with MIDAS (for details, see e.g., Kniazev et al. 2000). EWs for individual emission lines were measured with the standard MIDAS procedure INTEGRATE/LINE. The flux and equivalent width of the blended lines were measured using Gaussian decomposition fitting. In both cases the background was drawn by two methods: manually and with the use of the automatic procedure, with the help of the algorithm, described in detail by Shergin et al. (1996). The results of both cases were compared. The errors of the sensitivity curve and those of the line intensities have been combined in quadrature and propagated to calculate element abundances.

In particular, for the A 1228+12 spectrum with $\sim$ 7 Å resolution, deblending was performed for H $\gamma$ /[O III] $\lambda$ 4363, H $\alpha$ /[N II] $\lambda$ 6548, 6584, [O I] $\lambda$ 6300/[S II] $\lambda$ 6312. With these procedures the redshift and the line width were derived first for the stronger line of the blend. For the Gaussian fitting of the fainter blend components, these parameters have been fixed with the values derived for the stronger component. For the fitting of [N II] $\lambda$ 6548, 6584 lines we also fixed the intensity ratio of the two lines as 1:3, expected from theory (e.g. Aller 1984). The uncertainties of these fitted values were determined from the residual noise of the spectrum near the lines under analysis. These uncertainties were combined, as well as for all other measured line intensities, with the other error components (see below). Therefore, the derived errors of [N II] lines can be large, and their intensity ratio in the table is theoretical. For the TWIN spectra the spectral resolution is sufficient to measure each line separately. While in the spectrum of UM 408 [N II]-lines are rather faint, their line ratio is occasionally close to the theoretical value. In addition to the noise of the underlying continuum, quoted errors in the line intensities include two more components: one comes from Poisson statistics of photon flux, the other comes from the uncertainties of the sensitivity curve, contributing a few per cent to all lines. The line intensity errors presented in Table 2 incorporate all three components, and thus should be reliable estimators for other derived physical parameters and chemical abundances in the H II regions considered. An independent check of the reliability of the cited errors is the good consistency between our results and the results of Kinman & Davidson (1981) for A 1228+12 (see Sect. 4.3). Both sets of line intensities are consistent within the cited errors, if their extinction correction is accounted for. The latter can be overestimated, since Kinman & Davidson indicate a mismatch in the continuum level for independent red and blue spectra. Another factor leading to small differences is that Kinman & Davidson did not account for the underlying Balmer absorption. In the present work we determined underlying Balmer absorption at H $\beta$ of EW(H $\beta$ ) $\sim$ 2.5 Å. We derived from the spectrum of A 1228+12 the value of C(H $\beta$ ) = 0. This is consistent within the cited uncertainties ( $\sigma_{C({\rm H}\beta)}=$ 0.10) with the minimum value of C(H $\beta$ ) = 0.043 following from the Galaxy extinction in this direction, A_B= 0.12 mag (see Table 4). We have checked the effect of the change of C(H $\beta$ ) from zero to 0.043 on the derived element abundances. The O/H value does not change at all. The values of log (N/O), (Ne/O), (S/O) and (Ar/O) change by only 0.02-0.03 dex, which is many times smaller than their cited uncertainties.

Table 3: Abundances in the studied galaxies.
Value A 1228+12 UM 408

$T_{\rm e}$ (OIII)(K) 16100 $\pm$ 1100   14800 $\pm$ 1400

$T_{\rm e}$ (OII)(K) 14300 $\pm$ 900    13600 $\pm$ 1200

$T_{\rm e}$ (SIII)(K) 15100 $\pm$ 900    14000 $\pm$ 1200

$N_{\rm e}$ (SII)(cm^-3) 10 $\pm$ 10    10 $\pm$ 10

O⁺/H⁺( $\times$ 10⁵) 0.936 $\pm$ 0.186   2.518 $\pm$ 0.689

O⁺⁺/H⁺( $\times$ 10⁵) 4.198 $\pm$ 0.770   6.050 $\pm$ 1.546

O⁺⁺⁺/H⁺( $\times$ 10⁵) 0.254 $\pm$ 0.098   -

O/H( $\times$ 10⁵) 5.388 $\pm$ 0.798   8.568 $\pm$ 1.693

12+log(O/H) 7.73 $\pm$ 0.06   7.93 $\pm$ 0.09

N⁺/H⁺( $\times$ 10⁷) 2.470 $\pm$ 1.472   5.491 $\pm$ 1.439

ICF(N) 5.758 3.402

log(N/O) -1.58 $\pm$ 0.27   -1.66 $\pm$ 0.14

Ne⁺⁺/H⁺( $\times$ 10⁵) 0.737 $\pm$ 0.143   1.369 $\pm$ 0.381

ICF(Ne) 1.284 1.416

log(Ne/O) -0.76 $\pm$ 0.11   -0.65 $\pm$ 0.15

S⁺/H⁺( $\times$ 10⁷) 1.658 $\pm$ 0.232   2.990 $\pm$ 0.537

S⁺⁺/H⁺( $\times$ 10⁷) 7.212 $\pm$ 4.560   13.490 $\pm$ 10.660

ICF(S) 1.709 1.357

log(S/O) -1.55 $\pm$ 0.23   -1.58 $\pm$ 0.29

Ar⁺⁺/H⁺( $\times$ 10⁷) 1.500 $\pm$ 0.327   --

Ar⁺⁺⁺/H⁺( $\times$ 10⁷) 1.957 $\pm$ 1.023   --

ICF(Ar) 1.026 --

log(Ar/O) -2.18 $\pm$ 0.15   --

**Table 3:** Abundances in the studied galaxies.
Value	A 1228+12	UM 408
$T_{\rm e}$ (OIII)(K)	16100 $\pm$ 1100	14800 $\pm$ 1400
$T_{\rm e}$ (OII)(K)	14300 $\pm$ 900	13600 $\pm$ 1200
$T_{\rm e}$ (SIII)(K)	15100 $\pm$ 900	14000 $\pm$ 1200
$N_{\rm e}$ (SII)(cm^-3)	10 $\pm$ 10	10 $\pm$ 10

O⁺/H⁺( $\times$ 10⁵)	0.936 $\pm$ 0.186	2.518 $\pm$ 0.689
O⁺⁺/H⁺( $\times$ 10⁵)	4.198 $\pm$ 0.770	6.050 $\pm$ 1.546
O⁺⁺⁺/H⁺( $\times$ 10⁵)	0.254 $\pm$ 0.098	-
O/H( $\times$ 10⁵)	5.388 $\pm$ 0.798	8.568 $\pm$ 1.693
12+log(O/H)	7.73 $\pm$ 0.06	7.93 $\pm$ 0.09

N⁺/H⁺( $\times$ 10⁷)	2.470 $\pm$ 1.472	5.491 $\pm$ 1.439
ICF(N)	5.758	3.402
log(N/O)	-1.58 $\pm$ 0.27	-1.66 $\pm$ 0.14

Ne⁺⁺/H⁺( $\times$ 10⁵)	0.737 $\pm$ 0.143	1.369 $\pm$ 0.381
ICF(Ne)	1.284	1.416
log(Ne/O)	-0.76 $\pm$ 0.11	-0.65 $\pm$ 0.15

S⁺/H⁺( $\times$ 10⁷)	1.658 $\pm$ 0.232	2.990 $\pm$ 0.537
S⁺⁺/H⁺( $\times$ 10⁷)	7.212 $\pm$ 4.560	13.490 $\pm$ 10.660
ICF(S)	1.709	1.357
log(S/O)	-1.55 $\pm$ 0.23	-1.58 $\pm$ 0.29

Ar⁺⁺/H⁺( $\times$ 10⁷)	1.500 $\pm$ 0.327	--
Ar⁺⁺⁺/H⁺( $\times$ 10⁷)	1.957 $\pm$ 1.023	--
ICF(Ar)	1.026	--
log(Ar/O)	-2.18 $\pm$ 0.15	--

**Table 4:** Main parameters of studied galaxies.
Parameter	UM 151	UM 408	A 1228+12
$\alpha_{\rm 2000}$	01 57 38.87	02 11 23.55	12 30 48.52
$\delta_{\rm 2000}$	+02 25 23.9	+02 20 31.0	+12 02 42.1
A_B^N	0.12	0.15	0.12
$B_{\rm tot}$	16.21⁽¹⁾	17.74⁽¹⁾	17.15⁽³⁾
$V_{\rm Hel}$ (km s $^{\rm -1}$ )	4851⁽²⁾	3507⁽⁴⁾	1263⁽⁵⁾
Dist (Mpc)	64.7	46.8	17.0^V
M_B⁰ ⁽⁶⁾	-17.96	-15.76	-14.10
Opt. size ( $\arcsec$ )⁽⁷⁾	35 $\times$ 15.3⁽¹⁾	15.6 $\times$ 10.6⁽⁴⁾	12 $\times$ 9⁽³⁾
Opt. size (kpc)	11.0 $\times$ 2.4⁽²⁾	3.5 $\times$ 2.4⁽²⁾	1.0 $\times$ 0.75⁽²⁾
12+log(O/H)	8.5⁽²⁾	7.93⁽²⁾	7.73⁽²⁾
H I flux⁽⁸⁾	<1.2⁽⁹⁾	1.5⁽⁴⁾	1.4⁽⁵⁾
$W_{\rm 50}$ (km s $^{\rm -1}$ )	--	77⁽⁴⁾	84⁽⁵⁾
M(H I) (10 $^{(8)} M_{\odot}$ )	<11.9⁽²⁾	7.8^(2,4)	0.95^(2,5)
M(H I)/L $_{\rm B}$ ⁽¹⁰⁾	<2.0⁽²⁾	2.5⁽²⁾	1.5⁽²⁾

3 Results of the abundance determination

Relative intensities of all emission lines together with the equivalent width EW(H $\beta$ , emis), extinction coefficient C(H $\beta$ ) and the equivalent width of the hydrogen absorption lines are given in Table 2. The extinction coefficient C(H $\beta$ ) was derived from the hydrogen Balmer emission decrement using the self-consistent method described by Izotov et al. (1994). For both UM 408 and A 1228+12 the continuum was drawn as a running mean without accounting for possible absorption features. For UM 151, before obtaining emission line intensities, the underlying continuum was drawn including Balmer absorption lines and other apparent absorption features. Their measured equivalent widths were used for age estimates (see Table 5). The quoted $EW({\rm abs})$ for this galaxy in Table 2 is the residual value that is derived after the underlying continuum was drawn, including strong Balmer absorptions. The EW of emission H $\beta$ is calculated on a running mean continuum that will be compared with the model value below. The derived extinction coefficients are in the range from zero for A 1228+12, 0.2 for UM 151, to $\sim$ 0.6 for UM 408. The latter value is somewhat larger than usually is observed in this type of galaxies. For A 1228+12 and UM 408 the chemical abundances and physical parameters have been obtained with the method outlined in the paper of Izotov et al. (1997). The resulting values are given in Table 3.

For UM 151, since no measurable [O III] $\lambda$ 4363 Å line has been detected, the metallicity has been estimated by means of empirical methods (see more details in Sect. 4.1).

The EW of emission H $\beta$ presented in Table 2 were used to derive starburst ages according to the Starburst99 model (Leitherer et al. 1999) in Sects. 4.1 and 4.3. It was assumed that extinction values for the ionized gas and the young stellar clusters are similar.

4 Discussion and summary

We discuss below the individual galaxies in more detail, based mainly on the new spectral data, and the observational data from the literature, appropriate for the present discussion. The main parameters of the studied galaxies are given in Table 4.

4.1 UM 151 = Mkn 1169

This galaxy does not look like a bona fide BCG. Its appearance resembles that of a face-on disk with somewhat disturbed outermost parts, but clearly without spiral arms. Salzer (1989a) classified it as a Dwarf Amorphous Nuclear Starburst (DANS). Its absolute magnitude quoted here (Table 4, $M_B=-17\fm96$ ) is within the range for DANS. A bright knot is seen near the galaxy center, but does not change the generally regular appearance of this galaxy. The low metallicity value from Telles (1996) was derived from a very low S/N spectrum (see Terlevich et al. 1991).

For the metallicity estimation both the Pilyugin (2001) calibration based on the strong oxygen lines and the recently reported calibration based on the [N II] line (Denicoló et al. 2002) give very close values of $\rm 12+log(O/H)$ , 8.50 and 8.47, respectively.

Balmer absorption lines and other features with reliable detections can be used to estimate the age of star-formation episodes.

The Equivalent Widths (EW) of Balmer absorption lines from the underlying continuum have been calculated following the prescription in González-Delgado et al. (1999) and the results are presented in Table 5. The comparison with the González-Delgado et al. (1999) models gives an age for the starburst consistent with an instantaneous starburst $\sim$ 10 Myr old. The EW of H $\beta$ emission is also consistent with the age of instantaneous starburst of $\sim$ 10 Myr (Leitherer et al. 1999). Following Raimann et al. (2000), we also measured the EW of the Ca II K-line and the G-band as well as the continuum flux ratios in the 3 bands (see Table 5). All but one of the parameters are consistent with a mixture of two starbursts with ages $\sim$ 10 Myr and a $\sim$ 50 Myr. The rather high EW of the G-band suggests an additional contribution of a stellar population with an age of a few hundred Myr.

**Table 5:** The absorption lines EW and continuum ratios of UM 151.
Abs. line	Value	Band	Value
EW(H $_{\beta}$ )	5.4	CaII-K	2.2
EW(H $_{\gamma}$ )	4.0	G band	3.2
EW(H $_{\delta}$ )	6.1	F₃₆₆₀/F₄₀₂₀	0.71
EW(H₈)	7.4	F₃₇₈₀/F₄₀₂₀	0.88
EW(H₉)	7.9	F₄₅₁₀/F₄₀₂₀	0.83
EW(H₁₀)	4.0

The known upper limit on H I flux of this galaxy (see Table 4) is rather high, so the upper limit on the ratio $M(H{\sc i})/L_B$ is consistent with the range typical of gas-rich starbursting galaxies.

4.2 UM 408

Appearing like a typical blue compact dwarf, this galaxy was classified by Salzer (1989a) as a Dwarf H II Hotspot Galaxy. The absolute B-band luminosity is M_B=-15.16 (Campos-Aguilar et al. 1993), almost at the lower end of the BCG luminosity distribution, with a quoted diameter of 2.1 kpc. The calculated metallicity reported by Masegosa et al. (1994) was 12 + log(O/H) = 7.63 using the data from the Spectroscopic Catalogue of H II Galaxies (Terlevich et al. 1991). Using the same set of data Telles (1996) reported a value of 12 + log(O/H) = 7.66. The new estimation with the present data suggests a significantly higher metallicity with a difference of 0.2 to 0.3 dex. This difference cannot be attributed to the reddening estimation or differences in slit positioning. Comparing the present data with Masegosa et al. (1994) and Telles (1996), the reddening coefficients are similar within the uncertainties and the values of both EW(H $\beta$ , emis) and integrated H $\beta$ flux are the same. Therefore the main reason for the discrepancy must be a poor estimation of the [O III] $\lambda$ 4363 due to low S/N. The comparison between the line intensities of both sets of data shows that a large difference is found not only in the faint [O III] line, but also in [O II] being larger by a factor of 2 for the TWIN data. The same is also true for the measured [N II] line. The low S/N on the continuum for this faint galaxy can account for the large difference in oxygen abundance.

The measured H I flux of this BCG (see Table 4) corresponds to a very high value of the parameter $M({\rm HI})/L_B$ . The $M({\rm HI})/L_B$ is comparable to values derived for the most extreme objects in the sample of dwarf galaxies with extended H I (van Zee et al. 1995).

4.3 A 1228+12 = RMB 132 = VCC 1313

This galaxy is one of the most compact, almost starlike in appearance, of the BCG family (Drinkwater & Hardy 1991). It resides in the Virgo cluster, and because of the surrounding environment, its properties are probably somewhat affected by more frequent interactions with surrounding galaxies and the hot intracluster medium (ICM). There is a number of sufficiently massive candidate galaxies in the vicinity of this BCG, which could trigger its current SF burst, including NGC 4478 at the projected distance of $18.7\arcmin$ ( $\sim$ 90 kpc) and M 87 at $20.8\arcmin$ ( $\sim$ 100 kpc), whose relative radial velocities are lower than 100 km s $^{\rm -1}$ .

Of the three galaxies studied in this paper, this is the faintest system with an absolute B magnitude of -14.1. Taking into account its compactness and luminosity, this is the type of galaxy classified as a Searle-Sargent object by Salzer (1989a), or H II galaxies by Campos-Aguilar et al. (1993). This galaxy was one of the first BCGs studied for metallicity purposes (Kinman & Davidson 1981, hereafter KD81). From this study the metallicity reported of $\rm 12+log(O/H)=7.64\pm0.07$ is in reasonable agreement with the present value of $7.73\pm0.06$ , based on the new, higher S/N data from the 6 m telescope. The agreement can be considered as a good one, taking into account that the areas sampled are 29 arcsec² for KD81 and 7.2 arcsec² in this work.

However, it is evident that the abundance calculations by KD81 could have some systematic differences, since these authors did not account for temperature gradients. They used one temperature for all zones, and noticed that this would lead to slight underestimation of oxygen and neon abundances. Therefore we recalculated oxygen and neon abundances from their relative intensities and the cited errors using our methodology. This resulted in an upward shift of 0.04 dex for the KD81 oxygen abundance, giving $\rm 12+log(O/H)=7.68\pm0.10$ and $\rm log(Ne/O)=-$ 0.75. The difference between the KD81 O/H abundance and the recalculated one appears to be from a somewhat larger abundance of ${\rm O}^{+}$ ions: $\rm 12+log(O^{+}/H)$ = 6.87 from KD81, and 7.05 in the model with the lower $T({\rm O}^{+})$ .

One of the interesting features in the spectrum of A 1228+12 is the appearance of an emission line centered at $\lambda$ 4591 Å, with flux $\sim$ (2.5 $\pm$ 1.2)% of H $\beta$ and FWHM= 28 Å. The only reasonable identification is as the Si III $\lambda$ 4565 line, characteristic of WR stars. The implied radial velocity of this feature is then $\sim$ 450 km s $^{\rm -1}$ higher than the system velocity of the galaxy found in other (narrow) emission lines. Given the large line width and low S/N ratio, the velocity shift is likely not significant. However the appearance of WR stars in the starburst region of this BCG would not be unexpected. Indeed, the observed EW of H $\beta$ (93 Å) according to Fig. 85e (corresponding to metallicity z= 0.001, the closest to the oxygen abundance of A 1228+12) from Starburst99 (Leitherer et al. 1999), corresponds to an instantaneous starburst of age 3-3.5 Myr. We adopted the Salpeter IMF with $M_{\rm low}=0.1~M$ $_{\odot}$ for this estimate. For the metallicity of z=0.001 this is exactly the age range where the models predict significant numbers of WR-stars (0.5 to 2.5% of the number of O-stars, see Schaerer & Vacca 1998). Since we did not detect other characteristic WR features (N III/N IV $\lambda$ 4640 and broad component of He II $\lambda$ 4686), that should have comparable EW, doubt about the reality of the Si III $\lambda$ 4565 line remains. Deeper spectroscopy is necessary in order to measure the strength of the probable WR features in this BCG.

The $M({\rm HI})/L_B$ ratio for this BCG is quite large (see Table 4). Accounting for very large EW of strong emission lines and respective significant brightening ( $\Delta B \gtrsim ~1\fm0$ ), this implies a large gas mass-fraction. The latter is difficult to understand for a low-mass galaxy affected both by the ICM ram pressure (unless the BCG is only now entering the hot gas) and tidal interactions from massive neighbours.

In fact, the very existence of such a low-metallicity BCG in a dense environment such as the Virgo Cluster (despite the opposite tendency of additional enrichment by heavy elements for the Virgo cluster BCGs, noticed by Izotov & Guseva 1989), poses interesting questions on the evolutionary history of this and similar objects. It is worth mentioning that there are two more BCGs with well determined low metallicities ( $Z\sim Z_{\odot}/20$ ) probably belonging to the Virgo Cluster or its outskirts: the Optical Counterpart of HI 1225+01 (Salzer et al. 1991) and VV 432=IC 3105=VCC 241 (Zasov et al. 2000). Less intriguing, but also not well understood, is the appearance of SBS 0335-052 (E+W), the pair of extremely metal-deficient BCGs, situated at the outskirts of the loose galaxy group LGG 103 (Pustilnik et al. 2001; Peebles 2001).

The metallicities of UM 151, UM 408 and A 1228+12 revisited