Relationships between ductile and brittle fabrics in the Waite composite fault zone are complex and vary somewhat along strike, but some features are constant within our study area. The northwestern margin of the Waite composite fault zone is typically a gradational contact between undeformed Fredericton belt metasedimentary rocks and their ductilely deformed products — generally phyllonite and sheared metasandstone characterized by penetrative pressure solution cleavage in which clasts are strongly elongated. Biotite and actinolite formed in the highest-strain part of the Waite ductile shear zone.

Brittle effects are most pronounced along the southeastern margin of the composite fault zone in the area between Waite and the Maine-New Brunswick border and in the Nicatous Lake area. In both locations, the early-stage phyllonite is juxtaposed against post-Acadian redbeds or metasedimentary rocks of the Fredericton belt that exhibit little evidence of ductile strain. Details of brittle strain zonation will be presented below. Anastomosing brittle fault strands carve the phyllonite into lenses, convert it to cataclasite, and retrograde the biotite and actinolite to chlorite. Relict phyllonitic foliation in adjacent lenses is typically discordant, suggesting that they have rotated relative to one another during the brittle faulting. The originally near-vertical phyllonitic foliation is locally folded and flattened to sub-horizontal attitudes.

Near-vertical outcrop-scale conjugate faults have been identified within the sheared Deblois granite along the Kellyland ductile shear zone that exhibit a combination of ductile and brittle features. These small faults are superimposed on granitic mylonite generated by early ductile shearing and are cut by all other brittle structures and fractures. In the Wabassus Mountain area, they also cut and offset fine-grained granitic dikes. One fault set of this brittle/ductile conjugate system strikes at around 010° and the other at around 050°. All displacement indicators show that the former set is sinistral and the latter dextral, demonstrating that the maximum principal stress is in the direction of about 120°. Obviously the maximum principal stress orientation bisects the obtuse angle between two sets, suggesting ductile deformational mechanisms consistent with the following observations.

This conjugate system generated outcrop-scale ductile-brittle faults 2 cm — 1.5 m-wide, most commonly several centimetres. The faulted rocks are partly mylonitic and partly cataclastic. The ductile-brittle faults are fundamentally ductile, meaning that the feldspar crystals are elongated and the quartz grains plastically deformed and ribboned to become banded or laminated mylonite, and that abundant features of plastic deformation have been seen within quartz grains. However, at least one side of each mylonite zone has a sharp and straight fault contact, and the change from this mylonite to unaffected rocks is so abrupt that only brittle fracturing is likely to have been responsible for this relationship. In fact, these contacts are generally delineated by very narrow bands of cataclasite. It is also common for the offset fine-grained granitic dikes to show drag folding, but they are abruptly broken at the contacts. According to Ramsay and Huber (1983) and Davis and Reynolds (1996), such small-scale fault zones are best defined as brittle-ductile faults. They most likely began as continuous ductile shear zones but developed sharp, fault-like features along their margins when the physical conditions changed during deformation.

The maximum principal stress orientation (about 120°) deduced from this brittle-ductile conjugate system is about the same as that from the episode 1 small-scale conjugate system discussed next, implying that they were generated under similar dynamic environments. We ascribe the ductile-brittle transitional event to the precursor or the early phase of episode 1 brittle reactivation.

A broad, continuous zone of cataclasite and breccia crops out along the Waite composite fault zone, indicating that the older Waite ductile shear zone was reactivated during later brittle activity (Fig. 6a, 6b). Two other sets of brittle structures in the Waite composite fault zone overprint this cataclasite and breccia zone (see below). Therefore it is believed that this zone represents the earliest episode of brittle faulting, here termed episode 1. In contrast, only minor cataclasis and brecciation were found in granite, granitic mylonite, and phyllonite in the Kellyland ductile shear zone.

For the most part, the episode 1 brittle fault zone followed the pre-existing Waite ductile shear zone (Fig. 1, 2), affecting the Flume Ridge Formation, granite of the Bottle Lake pluton, and ductilely-faulted rocks produced during early ductile shearing. Faults associated with episode 1 brittle activity have not been recognized in the redbeds. In the Nicatous Lake-Gassabias Lake area, the Waite ductile shear zone has an anastomosing nature and branches locally into two major sub-zones (Fig. 2). Both sub-zones were reactivated during the episode 1 event, with mappable cataclasite and breccia belts. Brittle fabrics dip steeply (65-75°) to the southeast. The more competent siltstone layers and quartz veins in Flume Ridge phyllonite and metasedimentary rocks are boudinaged, and the boudin arrays also dip southeast at about 65-75°, as do the fault displacement surfaces. These features suggest that the master brittle episode 1 fault dips steeply to the southeast.

Faults attributed to a second reactivation of the Waite zone (brittle episode 2) offset those of episode 1 and are themselves overprinted by faults assigned to episode 3 (Fig. 8). Episode 2 structures are concentrated geographically along the episode 1 master faults, but occur in much narrower zones only a few centimetres to metres wide. Like brittle faults of episode 1, they affected metasedimentary, granitic, and ductilely sheared rocks but have not been observed in the redbeds.

Small-scale faults of episode 3 cut metasedimentary, granitic, and ductile fault rocks, and also overprinted episode 1 and episode 2 faults and cataclasite zones (Fig. 7, 8). In contrast to features associated with episodes 1 and 2, episode 3 faults also cut conglomerate, sandstone, and mudstone in the post-Acadian redbed slivers. Mapping during this study and by previous workers showed that the redbeds are everywhere in fault contact with granite, metasedimentary rocks, mylonite, or phyllonite, and that these rocks are steeply tilted with variable dip directions along those faults. These faults are considered to be map-scale examples of episode 3 brittle structures.

The oldest shearing recognized in the Norumbega fault system in eastern Maine occurred in the Middle Devonian (~380 Ma), and is attributed to late-stage activity associated with the Acadian orogeny (West and Hubbard 1997; Ludman et al. 1999). West et al. (1992) demonstrated that the climax of Acadian deformation in the Fredericton belt was late Silurian. Bradley et al. (1998) showed that the Acadian event was regionally diachronous, beginning in eastern Maine in late Silurian time and becoming progressively younger (to Middle Devonian) in central and north-central Maine. The Early Devonian phase in easternmost Maine was explained by Bradley et al. (1998) as the result of a "tectonic backstop". Thermal adjustments to Acadian tectonism led to melting and emplacement of numerous Devonian plutons such as the Bottle Lake, Deblois and Lucerne, and continued adjustment of plates whose oblique collision caused the Acadian orogeny led to the dextral ductile shearing.

The timing and dextral sense of motion associated with the ductile shearing are thus consistent with the timing and kinematics of plate collision generally interpreted for the Acadian orogeny (e.g., Bradley et al. 1998; Gates et al. 1988).

Paleostress analyses described earlier yielded different major compressive stress directions for the three post-380 Ma, brittle Norumbega faulting episodes in eastern Maine. ఼₁ for the earliest (episode 1) and most intense event was oriented approximately WNW — ESE, but rotated to NE — SW for episode 2 and changed drastically to N-S for episode 3. These paleostress results share some similarities with those from other segments of the Appalachian orogen but also contain significant differences, both in orientation and time.

In Quebec and northern New Brunswick, ductile fabrics comparable to those of the Kellyland and Waite fault zones are absent, but similar small-scale conjugate sets of reverse and strike-slip brittle faults are abundant (Faure et al. 1996). A numerical paleostress tensor analysis of striated fault planes by Faure et al. (1996) identified stress fields associated with three distinct brittle compressional events. The earliest and most intense had ఼₁ oriented NNW — SSE, ఼₁ for the second was NNE — SSW, and ఼₁ for the last was WNW — ESE. Interestingly, the third compressive event in Quebec, like episode 3 in the study area, is also mostly recorded in Carboniferous rocks although its maximum stress direction is oriented differently from that inferred along the Waite composite zone.

Faure et al. (1996) described a counterclockwise rotation of compressional axes in the Quebec and northern New Brunswick Appalachians. A counterclockwise rotation of compressional axes from episode 1 to episode 2 in eastern Maine is in this respect consistent with their results. The ages of faulting, however, are not. Faure et al. (1996) suggested that the first two compressional events in the Quebec Appalachians took place during Late Carboniferous to Early Permian and the third from Early to Late Permian time, and are the principal results of the Alleghanian orogeny in the region. Because episodes 1 and 2 in eastern Maine predated deposition of Pennsylvanian (?) redbeds, they cannot be related to Alleghanian compression and most likely predate all three events described by Faure et al. (1996). Only episode 3 is potentially young enough to correlate with the features reported by Faure et al. (1996). Based on similar stress field orientations, episode 3 appears to correlate best with the first Alleghanian compressional event in Quebec and northern New Brunswick.

Farther south, analysis of joint distributions and layer-parallel shortening fabrics in the Appalachian foreland sequence reveals two main directions of compression; an older one oriented north-south and a younger oriented WNW — ESE (Engelder and Geiser 1980; Dean et al. 1988; Evans 1994). In the northern part of the Appalachian Plateau, two sets of joints oriented NNW — SSE and NNE — SSW were attributed, respectively, to the Lackawana phase and to the main phase of the Alleghanian orogeny (Engelder and Geiser 1980). In the central Appalachian foreland, both north-south and WNW — ESE shortening directions have been reported (Dean et al. 1988; Evans 1994). Further south in the central Appalachian Plateau in Virginia, Dean et al. (1988) also found an earlier compression axis oriented between 330° and 350° that was followed by a progressive counterclockwise rotation of the compressional axis to 275°. Limestone of the Appalachian Plateau and of the Hudson River Valley also records NS- and WNW — ESE-oriented compressive directions (Craddock et al. 1993; Sierra et al. 1993).

If the ages inferred for episodes 1 and 2 are correct, it appears that these events record a previously poorly-known part of northern Appalachian history — the transition from Acadian stress regimes related to one set of plate collisions to Alleghanian stresses associated with very different plate interactions. Effects of Alleghanian deformation in the study area are probably restricted to episode 3.

The origin of the nine narrow redbed slivers along the length of the Waite composite fault zone has seldom been mentioned by previous workers and is problematic, with three likely scenarios:

• 1) The redbeds were originally part of an extensive post-Acadian molasse sheet, possibly connected to similar rocks in New Brunswick. Post-depositional dip-slip faulting could have down-dropped and therefore preserved the slivers, while uplifting the remainder of the sheet and leading to its erosion. Larrabee (1964) mentioned this possibility but without any further discussion.

• 2) An episode of brittle faulting following episode 1 could have formed small pull-apart basins in which the rocks were deposited. Pull-apart basins comparable to those envisaged in this second model are common along major strike-slip faults throughout the world (Burchfiel and Stewart 1966). Bradley (1982) mentioned the possible pull-apart mechanism for the "Norumbega basin" (Fig. 1) but without a detailed explanation.

• 3) The third possibility, not mentioned by previous workers, is that the redbeds were deposited in grabens or half grabens generated by extensional deformation along the fault system.

This study strongly favours the second scenario, the strike-slip pull-apart model, and suggests that all the redbeds were deposited in pull-apart basins generated by episode 2 dextral strike-slip faulting. This conclusion is based on the following observations and arguments:

• 1) Rocks in the nine individual redbed slivers do not fit readily into a simple facies model, and each sliver possesses its own, slightly to distinctly different, stratigraphic package. This difference does not disprove the molasse sheet model as rapid facies changes would be expected in a complex intermontane fluvial system, but does raise some questions and keeps the second model in contention.

• 2) All the redbed slivers now have elongate, narrow shapes and are bounded by brittle faults of the Waite composite fault zone. No other similar rocks have been found outside of the Norumbega fault system in east-central and eastern Maine. This observation implies that the redbeds were associated with brittle faulting. If the first model were valid, the redbed slivers would have to have been down-dropped in the Waite fault zone while the rest of the broad molasse sheet would have been uplifted and completely eroded all around the Waite fault zone. This situation is very unlikely as there is no evidence of post-episode-1 dip-slip fault events with appropriate kinematics to have down-dropped the redbed slivers anywhere along the Norumbega fault system in eastern Maine. Brittle episode 3, the only brittle faulting event that affected the redbeds, was compressional and oblique according to this study. It deformed the redbeds but did not down-drop them. The Fredericton fault in New Brunswick offsets the Carboniferous red-beds, but no significant vertical motion has been reported there either, therefore the first model should be discarded.

• 3) Previous studies of post-Acadian molasse in New Brunswick, Nova Scotia, and Prince Edward Island show that these rocks were deposited in large Carboniferous continental redbed basins that initiated as independent pull-apart basins (e.g. Bradley 1982, 1988). Bradley (1982, 1988) proposed that these basins, all associated with orogen-parallel transcurrent fault systems in the northern Appalachian orogen, formed as a result of pull-apart activity. The Norumbega fault system is one of these transcurrent fault systems. The late Devonian-early Carboniferous rhomboidal Magdalen pull-apart basin, located in the Gulf of St. Lawrence, was believed to have been generated by dextral strike-slip movement along the Fredericton fault (Bradley 1982, 1988), the northeastern extension of the Norumbega fault system in New Brunswick. Therefore the same pull-apart mechanism could also have occurred farther to the southwest, along the Maine segment of the fault system.

• 4) No regional scale extensional structures have been observed along the Norumbega fault system in eastern Maine, suggesting that third model cited above is not appropriate. This study suggests that the kinematics of episode 2 are consistent with the pull-apart fault basin model. A strike-slip movement regime with local extensional environments is indicated if the redbeds were initially deposited in fault-bounded pull-apart basins. Episode 1 could not have been the basin-forming event, because of its oblique compressional nature described earlier. Small-scale faults associated with episode 2 structures have not been identified in any of the redbed slivers, suggesting that it predated or perhaps was synchronous with their deposition. Episode 2 also experienced the dextral strike-slip motion appropriate for the pull-apart model. Episode 3 post-dated the redbeds, leaving episode 2 as the best candidate for a basin-forming event.

Pull-apart basins (Burchfiel and Stewart 1966) are commonly developed along regional strike-slip fault systems. Studies of these natural pull-apart basins and experiments (e.g. Aydin and Nur 1982; Mann 1997; Mann et al. 1983; Peacock 1991; Peacock and Sanderson 1995; McClay and Dooley 1995) have proposed two models to explain the formation and evolution of large, transtensional basins along strike-slip faults: one model of transtensional and rhombohedral pull-apart basins at step-overs or releasing bends, and a second model of more elongate and rift-like basins produced by fault-normal extension coeval with strike-slip motion (Ben-Avraham and Zoback 1992). In addition, Mann (1997) proposed a third model for such basins involving tectonic escape.

The first pull-apart model is the most common one. The extensive New Brunswick Carboniferous redbed basin, into which the Waite fault zone continues northeastward as the Fredericton fault, was associated with the late Devonian-early Carboniferous rhomboidal Magdalen pull-apart basin in the Gulf of St. Lawrence (Bradley 1982). This pull-apart basin was generated at a releasing bend of a northeast-trending dextral strike-slip system, and the subsequent thermal subsidence affected a large area including most of New Brunswick during middle and late Carboniferous time (Bradley 1982). The Moncton basin in southern New Brunswick has also been viewed as a pull-apart basin associated with the same strike-slip system, and smaller pull-aparts created by similar strike-slip systems in the northern Appalachian orogen are possible (Bradley 1982).

The mechanisms for redbed basin development along the Norumbega fault system in eastern Maine have not been adequately addressed by previous workers. This study suggests that all the redbed basins in the Norumbega fault system in eastern Maine are pull-apart basins formed at releasing bends during a brittle phase of activity along the Waite fault zone (Fig. 1 and 2). Episode 1 master faults are not always straight but curve, bend, and anastomose locally, and episode 2 faulting mostly reactivated the episode 1 faults (e.g. Fig. 2), so that some releasing bends would have been expected to form. The pull-apart mechanism at the releasing bends fits the continuum model proposed by Mann et al. (1983). In this mechanism, pull-apart basins nucleate at releasing bend fault segments along the fault system, and initial opening across the bends produce spindle-shaped basins. Increased displacement on the strike-slip faults results in rhomboidal pull-aparts. The small redbed basins on Figure 1 appear to be examples of such rhomboidal pull-apart basins. With continued strike-slip displacement and a constant regional horizontal maximum principal stress orientation, the rhomboidal pull-apart basins would evolve into narrow elongate basins whose basin width might remain fixed by the width of the releasing bends or the initial fault separation if the shear strain is plane strain (Mann et al. 1983). The Norumbega (Fig. 1; Wones 1980; Fig. 2 of Bradley 1982), middle Nicatous Lake, and lower Nicatous Lake basins mapped in this study are examples of such long, narrow basins (Fig. 1 and 2). These basins were clearly synkinematic, and sedimentation may have accompanied their progressive opening.

Most of the redbed basins within the Norumbega fault system are small and narrow and characteristically contain pebble- and cobble-size sediments. These sediments are thick bedded (greater than 1 m), generally poorly sorted, and range in shape from well rounded to angular and sub-angular. Only scarce outcrops of finer-grained redbeds such as red sandstone have been observed in the middle of Norumbega Basin and basins next to the Maine/ New Brunswick border on Fig. 1. This observation further supports the proposed dynamic synkinematic sedimentation environment. Because conglomerate predominates in almost all the redbed basins and the redbeds are uniformly steeply tilted, it is hard to reconstruct the distribution pattern of sedimentary facies and facies variations.