Magnetic field models showing the location of the BPs and the intersections with the photosphere of their associated separatrices from 21 to 23 October 2003. a) Corresponds the MDI map at 20:47 UT on 21 October, b) to the MDI map at 14:23 UT on 22 October, and c) to that at 01:35 UT on 23 October. The latter is the closest in time to the events we analyze in detail in this article. In panel c) we labeled the BPs as BP1 and BP2 and overlaid the RHESSI contours shown in Fig. 4d (light green and yellow contours correspond to the 50–100 keV range and 10–15 keV range, respectively). The three panels are shown from the point of view of the observer. The inclination of the sides of the boxes (defined in the local frame) indicates that the AR was not at CM. BPs are shown as the thickest green continuous lines, while the magenta thick continuous lines correspond to the photospheric trace of the BP separatrices. The contours correspond to ±50 G, ±100 G, and ±500 G (magenta (cyan) for positive (negative) field values) and the axes are in Mm.