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.