At present, the peridotite bodies are severely thinned, ranging from kilometric thickness in the most important massifs to decametric or even metric thickness in the nearby serpentinite sheets (Figs. 3 and 5). Several brittle and ductile fault systems are responsible for this thinning. The oldest fault system recognized is contemporaneous with the formation of the granitoids (Sánchez-Gómez et al., 1995), which is 20-22 M.a. in age (Priem et al., 1979, Zeck et al., 1992, Monié et al., 1994). Later fault systems completed the exhumation of the peridotites during the middle Miocene (García-Dueñas and Balanyá, 1991, García-Dueñas et al., 1992).

The first Miocene extensional episode is characterized by a distinctive N-S to NNE-SSW directed stretching lineation that developed in shear zones preferentially situated close to the lower boundary of the peridotite massifs. A comparable stretching lineation is more weakly represented at some points in the Jubrique sequence. The shear zones are retrograde, and developed over the granite simultaneously with or just after its intrusion (Sánchez-Gómez, 1997). Broken feldspars, elongated in the direction of the lineation, are transformed to white micas, and chlorites grow at the expense of ferromagnesian minerals where the deformation is most intense. Quartz, however, sustains its ductile behaviour, showing ribbons and cross-hatched mosaic microstructures. Therefore, a great deal of the observed structures developed in brittle-ductile conditions. Likewise, vertical penetrative E-W joints, occasionally with quartz filling, are frequent even when the stretching lineation is not well developed.

The transformation of peridotite into serpentinite (now chrysotile) occurs when the shear zones affect the peridotite slab; S-C structures and stretching lineation (Hoogerduijn Strating and Vissers, 1994) develop in these shear zones. Marbles and dark schists, close to or surrounded by the granite, include similar shear zones, their lineation respectively marked by large calcite crystals (1-5mm) and andalusite porphyroblasts that grow over chlorite and biotite aggregates.

The transport sense of the hanging-wall is variable, and conjugate senses within a single outcrop are frequent, but N to NNE transport directions are constant, even on both sides of the Gibraltar strait. In Ceuta (Figs.1 and 2), an extensional shear zone that encloses a 50m-thick layer of serpentinite and peridotite has a predominant southward sense of transport. In the northern part of the Gibraltar arc, transport senses to the north are more common.

The brittle-ductile shear zones of the first extensional episode dismembered the former peridotite slab. Much of the present N-S separation is likely due to this episode. Nonetheless, the fault systems developing later throughout the Miocene accentuated the initial separation or fragmented the first longitudinal extensional bodies, depending on the superimposed extensional direction. This would have caused the layers of serpentinite to be stretched and become locally discontinuous. The distribution of the peridotite bodies (Figs. 2, 3 and 5) is congruent with the superposition of the almost transverse extensional systems, as mentioned earlier. The shape of the peridotite bodies, probably all connected by thin serpentinite sheets, would conform well with the Miocene extensional patterns, as has been shown in the western Betics (García-Dueñas et al., 1992). The distribution of the highest gravity anomalies (Fig. 2) is also compatible with the established directions of extension.

The serpentinite layer in Ceuta represents a link between the Ronda peridotites (~90mGal anomaly) and the Beni Bousera massif (~70mGal anomaly) through a hypothetical peridotite body near Cabo Negro that would correspond to a minor gravity anomaly (~30mGa anomaly) (Figs. 2 and 5).

In order to illustrate the first extensional episode described above, Figure 6 shows the initial effects of N-S extension on a major peridotite slab during the primary stages of its dismemberment. Throughout the extensional event, the peridotites exhibit brittle behaviour, though serpentinization would result from the circulation of fluids promoted by fracturing. As serpentinite minerals remain ductile until much shallower conditions, above 1.5 kbar and 350°C (Raleigh and Paterson, 1965, Wicks, 1984), it is reasonable to postulate that the brittle individualization of the bodies was followed by their separation under ductile conditions from the serpentinite sheets and crustal rocks. Later, the thin serpentinite layers would have undergone widespread extension while the behaviour of the quartzfeldspathic crustal rocks evolved from ductile to brittle.