3 Data reduction and analysis

The frames were processed using IRAF. They were de-biased and flat-fielded using the ING WFC data-reduction pipeline (Irwin & Lewis 2001). Subsequently, they were corrected for geometrical distortions and aligned to the [O  III] frames. The sky background was determined in areas far from the galaxy and then subtracted from all images.

Emission-line objects were identified using a standard on-band/off-band technique (Magrini et al. 2000). Briefly, the stellar continuum of Sextans B (Strömgren y images, properly scaled) was subtracted from the narrow-band images in order to highlight emission-line objects. Aperture photometry was applied using the IRAF task APPHOT in the continuum-subtracted frames. Photometric errors were derived from photon statistics for both sources and background with APPHOT. They are of few per cents both for [O  III] and for H$\alpha$+[N  II] fluxes.

The instrumental magnitudes were calibrated by convolving the spectrum of the standards used (Oke 1990) with the response curve of each filters. An interstellar extinction of E(B-V)=0.02 was used (de Vaucouleurs et al. 1991), following the Seaton (1979) prescription. The fluxes in the [O  III] line at $\lambda=500.7$ nm were corrected for the contribution of the companion oxygen line at $\lambda$ 495.9 nm, which amounts to 20% of the total flux in the relatively broad filter used. Including the uncertainties in all steps, we estimate that the total absolute error on the absolute fluxes quoted in this paper is $\le$20$\%$.

The astrometric solution for the [O  III] frames was also obtained through the WFC pipeline using the USNO A2.0 catalogue (Monet et al. 1998). The solution has an rms accuracy of 0".3.