TRACE white-light images present a $\delta$ spot which is formed by rapid coalescence of two magnetic features in a pre-existing $\delta$ -configuration in NOAA 9077. The origin of this $\delta$ spot is rather different from the regular patterns described by Zirin & Liggett (1987). Particularities of this $\delta$ -spot have been shown in some papers. This $\delta$ -configuration is found to be the position of one end of a rising magnetic flux rope system (Yan et al. 2001), or of an activated filament (Zhang et al. 2001). Kosovichev & Zharkova (2001) find that it is a magnetic transient region during the major flare.

In the present work, we find that $\delta$ -configuration P6-F4 is important for the formation of the highly sheared magnetic neutral line of NOAA 9077, and, hence for the trigger of the filament activation and an associated CME. Almost all of the obvious changes (motions, velocities, and current helicity) of the $\delta$ -configuration developed about two hours before the major flare.

$\begin{figure} \par\includegraphics[width=4.8cm,clip]{MS1766f8.ps} \end{figure}$

Figure 8: Magnetic flux loop model. Two such systems could co-exist in a spot. No reconnection happens between them. Note that the induced currents are opposite to each other.

Undoubtedly, magnetic chirality (handedness) plays an important role in reconnection. In the corona, shorter X-ray loops with the same chirality tend to coalesce and form longer forward or backward sigmoids (Canfield & Pevtsov 1996; Pevtsov & Canfield 1996). In the chromosphere, two close filaments may unite into a single one if their axial fields are in the same direction (Martin et al. 1994). In the photosphere, two spots of the same chirality can form a $\delta$ -spot under the third pattern mentioned in Sect. 1. Certainly, if the helicity of one spot is reversed for some reason then it is hard for the $\delta$ -configuration to keep its compact configuration. Therefore, separation will be seen.

Based on the observational results, we think that the disintegration of P6-F4 is caused by the sign reversal of the current helicity of F4, while the major flare accelerates the separation. To explain the reversal of helicity, we introduce a magnetic flux loop model for spots. In Fig. 8, two twisted loop systems co-exist in the magnetic system of a sunspot, which is not in contradiction to observations and theory (Pevtsov et al. 1994; Stenflo et al. 1984). The currents induced by the loops shown in Fig. 8 should be in opposite directions. There will be no reconnection at the interface between these loops because of the identical directions of their field lines. Possible reconnection between twisted loop systems are discussed in Sakai & Koide (1992).

If this is the case, let us consider the process of current reversal for F4. Originally, a larger portion of the left-handed helicity loop system (left one in Fig. 8) exists in F4. Two hours before the major flare, some disturbance, which cannot be derived from our observation, produces the dominance of the right-hand helicity (right one in Fig. 8). Thus, the average current of F4 is reversed and is now negative.

In contrast to our previous work, we study for the first time the relation between helicity evolution and $\delta$ -configurations. In future work, it is essential to study more complex $\delta$ -configurations and examine whether it is a universal rule that $\delta$ -spots can be disintegrated by helicity reversal.

6 Discussion and conclusions