Significance of dextral reactivation of an E-W transfer fault in the formation of the Pennsylvania orocline, central Appalachians

TECTONICS, VOL. 23,, doi:10.1029/2003tc001593, 2004 Significance of dextral reactivation of an E-W transfer fault in the formation of the Pennsylvania orocline, central Appalachians M.-W. Yeh 1 and T. H. Bell School of Earth Sciences, James Cook University, Townsville, Queensland, Australia Received 21 October 2003; revised 19 March 2004; accepted 28 June 2004; published 5 October 2004. [1] Three hypotheses for orocline development including (1) primary plate boundary shape, (2) arc rotation, and (3) arc overprinting were evaluated for the formation of the Pennsylvania orocline, Maryland, USA, using foliation intersection/inflection axes preserved within porphyroblasts (FIA). The distribution and the timing of sequences of FIA data measured from garnet porphyroblasts within the Loch Raven formation are statistically the same across the Pennsylvania orocline, suggesting that all subregions of the arc have experienced the same deformation history. This may suggest that the geometry of the paleocontinental margin controlled the basic shape of this orocline. However, if the orocline is treated as three separate regions, defined by the SW-WE-NE trending portions of its slightly staircase-shaped outline, a much higher proportion of the NNE-SSW trending FIAs (set II) are preserved in the W-E trending region than those to either side. This geometric relationship could have resulted from WNW-ESE directed bulk stress causing a zone of dextral shear along what is now the W-E trending portion of this orogen. It appears that what could have been an early formed W-E trending sinistral transform shear zone was preferentially dextrally reactivated during the Taconian orogeny. INDEX TERMS: 8015 Structural Geology: Local crustal structure; 8102 Tectonophysics: Continental contractional orogenic belts; 8110 Tectonophysics: Continental tectonics general (0905); 8030 Structural Geology: Microstructures; KEYWORDS: FIA, orocline, Appalachian Piedmont, microstructure, strike-slip reactivation. Citation: Yeh, M.-W., and T. H. Bell (2004), Significance of dextral reactivation of an E-W transfer fault in the formation of the Pennsylvania orocline, central Appalachians, Tectonics, 23,, doi:10.1029/ 2003TC001593. 1 Now at Department of Earth Sciences, National Taiwan Normal University, Taipei, Taiwan. Copyright 2004 by the American Geophysical Union. 0278-7407/04/2003TC001593 1. Introduction [2] Oroclinal arcs have intrigued geologists for many years because of their size, spectacular appearance from space or on geological maps, and possible tectonic significance. Do they (1) reflect the preexisting curvature on the boundary of ancient plate margins as the primary plate boundary model suggests [Carey, 1955, 1958; Lefort, 1984, 1988; Marshak, 1988; Hindle et al., 2000], (2) result from deformation (such as shearing and/or rotation) of a previously developed orogenic belt, as suggested for the Variscan arc, northeastern Spain and called the arc deformation/rotation model [Dana, 1866; Ries et al., 1980; Marshak, 1988; Schwartz and van der Voo, 1983; Valentino et al., 1994; Gates et al., 1999], or (3) form by overprinting of two noncoaxial fold belts, called the arc overprinting model [Marshak and Tabor, 1989], as appears to be the case for the Kimberley arc, NW Australia [Bell and Mares, 1999]? The Appalachian orogen, extending from Newfoundland to Alabama along the eastern side of North America, contains four distinct zones of curvature (Figure 1). From NE to SW along the orogen, the portions of these bends that are concave toward the craton are called the St. Lawrence, New York, Roanoke (or Virginia) and Alabama promontories (or recesses [Williams, 1978]). The portions that are convex toward the North American craton are called the Newfoundland, Quebec, Pennsylvania (or Baltimore) and Tennessee reentrants (or arcs [Williams, 1978]). All of the above models have been suggested as possibilities for development of these Appalachian oroclines. The primary plate boundary model school emphasized evidence that no changes in the position of the paleomagnetically reconstructed north pole have been found across each of the arcs [Irving and Opdyke, 1965; Roy et al., 1967; Schwartz and van der Voo, 1983; van der Voo, 1983]. This is supported by the distribution of geosynclinal sedimentary units that match the irregularities (salients and reentrants) along the Appalachian orogenic belt. Consequently, Rodgers [1968], Fleming and Sumner [1975], Rankin [1976], Thomas [1977] and Schwartz and van der Voo [1983] attributed the prominence of curvatures of the Appalachian orogenic belt to inheritance from the shape of an earlier cratonic margin, making them primary features. More recently, Alterman [1981, 1983, 1984] suggested that late clockwise rotation of the southern Appalachians has occurred between Virginia and South Carolina. Miller and Kent [1986], and Kent and Opdyke [1983] argued for the same type of rotation using paleomagnetic correlations 1of17

Figure 1. Outline structural geology map of Appalachian orogen along the east coast of the United States, and the interpreted outline of the late Precambrian-Palaeozoic continental margin as bounded by rift and transform faults during opening of the Iapetus Sea prior to compression, modified from Thomas [1977]. This outline of the paleocontinental margin coincides with curves in the Appalachian orogen. Such a pattern would suggest the Appalachian oroclines are a primary feature that reflects the geometry of a paleocontinental margin. The outlined box shows the location of Figure 2. between the arms of the Pennsylvania salient. Stamatakos and Hirt [1994] noted difference in paleomagnetic results measured from the Silurian Bloomsburg Formation, and suggested the occurrence of preorogenic clockwise rotation during the Alleghanian orogeny followed by later remagnetization during the Late Palaeozoic. A similar rotation of the Pennsylvania salient was suggested by Valentino et al. [1994] and Valentino and Gates [1995]. However, they used structural and metamorphic criteria and they argued that the Pennsylvania orocline resulted from original rifting with later modification because of overprinting deformation. [3] The enormous scale of oroclines makes correlation of structures and metamorphism around them almost impossible using normal structural and metamorphic approaches. Dating of metamorphic events from arm to arm of the oroclines provides one means of correlation of metamorphic and structural events, but the lack of such data plus the assumptions and errors involved make this approach problematical as well. A paleomagnetic approach [Eldredge et al., 1985], based on the assumption that the formation of an orocline is due to rotation of rocks in the orogen around a vertical axis, fails to recognize oroclines formed by rearrangement of fault-bounded blocks that are displaced rather than rotated. Besides, Late Paleozoic remagnetizations are well documented for the central Appalachian, it would be very difficult to distinguish the primary from the reset paleopoles [Kent and Opdyke, 1983; Stamatakos and Hirt, 1994]. To evaluate how the Pennsylvania orocline formed, we performed a new quantitative approach that utilizes foliation inflexion/intersection axes preserved within porphyroblasts (called FIAs). The FIAs were measured using garnet porphyroblasts from oriented samples collected all around the Pennsylvania orocline in Maryland (Figure 2) to see if this aided understanding of how this structure developed. 2. Geological Setting [4] The 250 km long, 30 km wide Baltimore-Washington anticlinorium within the Baltimore terrane of the Maryland 2of17

Figure 2. Detailed geologic map of Baltimore-Washington anticlinorium, Maryland, modified from Cleaves et al. [1986] and Muller and Chapin [1984]. PH is Phoenix, TX is Texas, TW is Towson, CH is Chattolanee, WD is Woodstock, MF is Mayfield, and CL is Clarksville. The light gray line A-A 0 marks the boundary where the arc is divided into two arms. The region to the right of the line is defined as Pennsylvania nappe, and the one to the left is defined as the Maryland nappe. The light gray lines B-B 0 and C-C 0 mark the boundaries between the west arm, central portion, and the east arm, with the arc divided as if it formed a staircase shape. Piedmont is bounded by the dextral strike-slip fault systems of the Gunpowder fault zone to the southeast, and the Pleasant Grove shear zone to the northwest (Figure 1 [Gates et al., 1999]). This region is characterized by gneiss-cored domes and doubly plunging antiforms flanked by metasedimentary rocks ranging from Cambrian to Early Ordovician in age, and affected by Grenvillian, Taconian, Acadian and Alleghanian tectonism [Higgins, 1973; Crowley, 1976; Fisher et al., 1979]. The oldest unit is the Precambrian Baltimore Gneiss (Grenville in age), which is unconform- 3of17

ably overlain by the Cambrian to Ordovician Setters Formation (quartzite), Cockeysville Formation (marble and calcsilicate schist), and Loch Raven Formation (metapelites, Figure 2 [Crowley, 1976; Fisher et al., 1979; Gates et al., 1991, 1999; Valentino, 1999]. All rock units within the Baltimore terrane were multiply folded and foliated to kyanite-staurolite grade metamorphic conditions [Muller and Chapin, 1984; Lang, 1990; Gates et al., 1991, 1999]. This deformation and metamorphism extended from the Taconic (510 to 460 Ma) to the early Acadian (408 to 360 Ma) [Tilton et al., 1970; Fisher, 1971; Grauert, 1973a, 1973b, 1974; Lang, 1990; Sinha et al., 1997]. [5] Bosbyshell et al. [1998, 1999] suggested the Loch Raven Formation experienced two main stages of Palaeozoic metamorphism; an earlier andalusite-sillimanite facies zone (4 6 kbars; 480 Ma) was variably overprinted by a kyanite bearing, Barrovian sequence (550 650 C, 7 9 Kbars; 380 Ma). Generalized garnet, staurolite and kyanite isograds have been mapped around the entire Baltimore area [Doe et al., 1965], and the southwestern part of Harford county [Southwick and Owens, 1968]. The isograd pattern reflects the outline of the gneiss domes (Figure 1), and generally increases in grade from the garnet to the kyanite zones moving from west to east (Figure 1a). Plank [1989] suggested this increase was due to differential uplift with deeper crustal levels exposed in the east. It has been proposed that this systematic areal variation in estimated pressure indicates that near peak metamorphism was attained prior to refolding of a gneiss-cored nappe, and continued through the refolding event [Fisher et al., 1979; Hall, 1988; Lang, 1990]. The cooling ages are all Alleghanian (about 300 Ma [Muller and Chapin, 1984; Lang, 1990; Gates et al., 1991, 1999]) and appear to be associated with orogen-parallel thrusting accompanied by dextral strike-slip motion along the Pleasant Grove and Gunpowder shear zones [Gates et al., 1999]. [6] The northeastern portions of the anticlinorium, which include the Phoenix, Texas, Towson, and Chattolanee domes (Figure 2), belong to a westward directed Taconic nappe system that rooted under the Towson dome, and were later refolded during the late Acadian or Alleghanian orogenies [Kodama and Chapin, 1984; Muller and Chapin, 1984; Gates et al., 1999]. The Woodstock, Mayfield, and Clarksville domes in the southwestern portion of the anticlinorum experienced a similar deformation history in terms of the matrix foliations, but are regarded as belonging to another nappe system with an unknown root [Kodama and Chapin, 1984]. Muller and Chapin [1984] suggested that at least four stages or phases of ductile deformation have affected the Pennsylvania nappe system. They suggested that a younger folding event, with a NW-SE striking axial plane, was responsible for the formation of the large stair case geometry (in plan view) with steeply dipping but variably striking axial planes (NNE-SSW western arm, E-W hinge, and NE-SW eastern arm; Figure 2). 3. Methods [7] Jones [1994] and Johnson [1999] have shown that the microstructures preserved as inclusion trails in porphyroblasts provide a record of the deformation history to compliment P T calculations. Bell and Welch [2002] have found an age progression, using monazite grains that lie within foliations that define successive FIA within porphyroblasts, which indicate that the last four of five Acadian FIA sets in Vermont began to form prior to 424 ± 3, 405 ± 6, 386 ± 6 and 366 ± 4 Ma, respectively. Undoubtedly, porphyroblasts record progressive kinematic and metamorphic conditions during orogenesis, and provide useful tools to investigate many geological processes including fold mechanisms, deformation and metamorphic history, and large-scale bulk shortening directions that can be used in tectonic reconstructions. Correlation of these microstructural features with deformation and metamorphic events and tectonic movements across a geological terrain was not possible prior to development of techniques for measuring FIAs within porphyroblasts [Powell and Treagus, 1967, 1970; Rosenfeld, 1968] (for earlier methods); [Hayward, 1990; Bell et al., 1995, 1998, 2003; Yeh, 2003]. Measurement of FIAs provides quantitative data that time and link foliations, overprinting foliation asymmetries, successive phases of mineral growth, metamorphism, deformation and bulk tectonic movement directions [e.g., Bell and Mares, 1999; Bell et al., 2003, 2004]. 3.1. FIA Measurements [8] FIA trends are generally measured by observing the asymmetry switch of equivalent inclusion trail curvatures from arrays of differently oriented vertical thin sections that fan about the compass of an oriented sample viewing from the same direction (Figures 3a 3b [Hayward, 1990; Bell et al., 1995]. The geometry observed (S or Z; clockwise or anticlockwise) within porphyroblasts represents the intersection of the 3-D shape of the included foliations with the differently oriented thin sections. The FIA is located between the sections with two opposing inclusion trail geometries [Hayward, 1990; Bell et al., 1995]. The FIA orientation is recorded as the midpoint of between those two oriented thin sections with a 10 range (e.g., FIA trend = 95 ranging between 90 100 ). Bell and Hickey [1997] determined the total accumulated error from using a compass to orient and reorient the sample in determining the trend of FIA to be ±8. These errors are all random and of the same order of magnitude, suggesting that the total accumulated error probably has a normal distribution. Where more than one FIA trend is preserved from core to rim (Figures 3c 3d), relative errors between trends are only a function of cutting horizontal slabs of rock, and then the thin section blocks from these slabs, as they are obtained from one sample; they should have a relative precision of ±4 [Bell and Hickey, 1997]. [9] The spread of FIA trends within a defined FIA set has been discussed in several studies [Hayward, 1990; Johnson, 1993; Bell et al., 1995; Bell and Wang, 1999; Timms, 2003]. Other than the systematic errors described above, this spread has been attributed to the anastomosing of foliations around euhedral porphyroblast crystal faces [Hayward, 1990], and the natural anastomosing of foliations due to a bulk shortening component during deformation [Bell, 4of17

Figure 3. (a) Three-dimensional sketches showing the method of using oriented thin sections to determine the FIA orientation. The asymmetry of inclusion trail curvature switches when viewed in the same direction on thin sections across a FIA. (b) Schematic diagrams showing the location of FIA relative to foliations within a porphyroblast with multiple growth episodes containing three internal foliations (Si1, Si2 and Si3) and one matrix foliation (Se; modified from Bell and Hickey [1997] and Bell et al. [1998]). Correlation of inclusion trail patterns from vertical thin sections with various strikes indicated three FIAs of 15 (core), 155 (median), and 85 (rim). Black solid line indicates the core inclusion trail; black dashed line indicates the median inclusion trail, and gray line indicates the rim inclusion trail. 1981]. We estimate that the effect of anastomosing foliations on the variation in FIA trend within a single FIA set ranges from 30 to a maximum of 50, on the basis of the rose diagram distributions of detailed measurements made in several regions (Figure 11 of Bell and Mares [1999]; Figure 8 of Bell and Chen [2002]; Figure 12 of Bell et al., 2004]). It is noteworthy that for the first phase of porphyroblast growth during an event that produced a horizontal foliation, heterogeneous rotation of the crenulation hinge toward a developing stretching lineation prior to porphyroblast growth could cause additional variation [e.g., Bell and Wang, 1999]. [10] The complex curvilinear S i (included schistosity) geometries preserved within porphyroblasts (Figure 3d) are commonly the result of successive overprinting matrix crenulations against porphyroblast rims that have been subsequently included within the growing porphyroblast, with or without porphyroblast rotation [Bell et al., 1995, 1998, 2003]. This pattern can be interpreted as reflecting successive horizontal compression and gravitational loading at the macroscopic scale during orogenesis [Bell and Johnson, 1989; Hayward, 1992; Johnson, 1999]. If this is correct, FIAs should be mostly subhorizontal and form perpendicular to the direction of bulk shortening [Bell et 5of17

al., 1995, 1998]. FIA plunges measured around the Kimberly Arc, NW Australia [Bell and Mares, 1999], the Spring Hill [Bell and Hickey, 1997] and the Bolton Synforms [Hickey and Bell, 1999] are all subhorizontal with a maximum plunge of 30. A FIA trend is bidirectional (110 = 290 ) if the plunge has not been determined. It has been impossible to differentiate bidirectional trending FIAs into different FIA sets in the regions where the plunges have been measured. Consequently, FIA plunges were not measured in this study, and bidirectional FIA trends are treated as one set. 3.2. Correlation of FIAs [11] The geological usefulness of FIA data depends upon being able to distinguish sets of trends and the temporal correlation of related sets. Systematic correlation between inclusion trail curvatures within a sample for determining the FIA orientations, and correlating FIAs between samples to distinguish temporally related sets are two major components for FIA data interpretation. Relative timing of FIA trends from the core to rims of porphyroblasts provide the best criterion for correlation and can be tested by dating using monazite grains preserved within the foliations defining each FIA for the resulting succession [e.g., Bell and Welch, 2002]. In this study, we applied (1) the relative timing of FIA orientations measured from core to medium to rim and (2) changing FIA orientations evaluated by statistical tests of the different FIA sets to establish the relative timing of these FIA sets. Textural criteria were never used for FIA set distinction or for temporal relationship evaluation. 3.3. Statistical Evaluation [12] The most prominent feature is the nonrandom multimodal nature of the FIA trends in the rose diagrams of FIA data (Figure 4). Bell et al. [1998] applied the Watson s U 2 test, modified for grouped axial data by Freedman [1981] and Upton and Fingleton [1989], to determine the significance of the multimodal nature of FIA data (Appendix A). They showed that the U 2 value for FIA data measured from Spring Hill, Vermont greatly exceeds the upper 0.5% critical value (0.187), and that multimodal FIA data are not a population with random trends. That is, the polymodal FIA trends observed in rose diagrams are meaningful and can be separated into groups (FIA sets). Some fluctuations seen in rose diagrams may not allow easy interpretation of the boundaries for each peak or mode. In these cases a moving average can be applied to smooth the FIA data and this may allow the discernment of modes that represent real properties of the FIA populations [King, 1994]. Moving averages of t-he order N = 5 were applied for FIA frequencies with a 10 class intervals (Appendix B [King, 1994]). The temporal succession of differently trending FIAs from the core to rims of porphyroblasts provides the most reliable correlation criteria. Consequently, each distinct FIA population and its range were determined by correlating the microstructural relationships between S i (included schistosity) and S e (matrix schistosity), the temporal succession, the modal peaks suggested by the rose diagrams, and the moving average test. Other than the U 2 and moving average tests, a third statistics test, the X 2 test (Appendix C) was also used. This determines the similarities and differences between the tested FIA populations. If the result greatly exceeds the 0.05 critical value, the null hypothesis is rejected, that the tested FIA populations are different. If the result does not exceed the 0.05 critical value, the null hypothesis is accepted, that the tested FIA populations are the same. 4. Results and Interpretation [13] A total of 221 FIA trends were measured from garnet porphyroblasts in 140 oriented samples collected from amphibolite facies (garnet ± staurolite ± kyanite bearing) biotite-oligoclase-muscovite-quartz schist (Loch Raven formation; Figure 2) in the eastern Maryland piedmont. 70 of the samples yielded 151 multi-fias. The other 70 samples yielded single FIAs (Figure 3b). Of the multi-fia samples, 11 contain core-median-rim FIA successions, while the rest show core and rim FIAs (Table 1). The U 2 value of all FIAs around the Pennsylvania orocline (0.501) greatly exceeds the critical value (0.187; p < 0.05) rejecting the null hypothesis (H 0 ), that this is a randomly distributed population. That is, the measured FIA populations are not randomly distributed, and each peak seen in the rose diagrams (Figure 4) is a meaningful cluster. With similar clusters indicated by the moving averages (Figure 4), six sets of FIA populations with trends at 20 30, 40 50, 70 80, 100 110, 130 140 and 170 180 can be distinguished. Since the isotopic ages for peak metamorphism, determined using U-Pb systematics from the Baltimore Gneiss (1180 to 1080 Ma and 455 to 421 Ma [Grauert, 1972, 1974; Pavlides, 1974; Higgins et al., 1977], are similar across the arc, correlation and comparison of FIA populations from different domains should provide solutions that enable resolution of the model for formation of the Pennsylvania orocline. 4.1. FIA Sets and Their Timing Sequence Along the Pennsylvania Orocline [14] The succession of differently trending FIAs preserved from the core to median to rim of porphyroblasts in 70 multi-fia samples provides a means of distinguishing distinct populations of FIAs as well as timing them. The X 2 value for core and rim populations (5.576) is much larger than the 0.05 level of significance, suggesting that the core FIAs define a statistically different population than those in the rim. Therefore the different trends in the FIA population appear to be due to a succession of different foliations trapped as inclusion trails within the garnet porphyroblasts. [15] The 30 50 trending FIAs (set I) only occur within cores and are followed by the other sets, strongly suggesting that this is the earliest formed set. In contrast, the 130 160 trending FIAs, which are completely missing from the core FIA distribution (Figure 4), are the only samples showing porphyroblast inclusion trails 6of17

Figure 4. (a) Equal-area (radii = square root of frequency) rose diagrams, with 10 class intervals, of all FIA trends measured in garnet porphyroblasts in all samples around the Pennsylvania orocline. Diagrams show FIAs for all samples, samples containing multi or single FIAs, and the distribution of core versus median versus rim FIAs. (b) Smoothed frequency distribution of the fractionated moving averages with the order N = 5 for total garnet FIAs measured with a 10 class interval. The six distinct peaks occurring between 10 20, 40 50, 70 80, 90 110, 130 150 and around 170 are indicated with the solid arrows. that are continuous with the matrix foliation (Figure 5), suggesting that they form the youngest set. The general change from 30 50 trending FIAs in the core, to 10 30 trending FIAs and then 160 190 trending FIAs in the rim, suggests a time sequence of set I to set II and set III (Table 1). Further successions with 160 190 trending FIAs in the core, to 50 100, 100 130, and finally the youngest 130 160 trending FIAs in the rim suggest an additional time sequence of set III to set IV, set V and VI (Table 1). Although 73% of the multi-fia samples suggest the relative timing sequence set I through VI, 27% (19 multi-fia samples: M9A, M18, M48B, M49, M53, M60, M83, M87, M93, M158, M164, M174, M180, M205, M206, M207, M223; Table 1) suggest a reversal in the swing about the compass from 160 190 to 160 200 to 70 110 trending FIAs. The change from a 100 130 (FIA V) core to a 130 160 (FIA VI) median to a broad N-S (160 200 ) rim FIA in sample M180 may indicate the presence of a younger succession of FIAs after set VI. Since most of these samples lie within the staurolite and kyanite zones (Figure 6), the swing of FIA orientation from 130 160 (set VI) to 160 200 (set VII), and then 70 110 (set VIII) FIAs appears to record a younger generation of garnet growth. From the general core to rim correlations preserved within the multi-fia samples, a time sequence of eight FIA sets that change in the mean FIA trend from NE-SW (set I, 30 50 ) to NNE-SSW (set II, 10 30 ), NNW-SSE (set III, 150 190 ), ENE-WSW (set IV, 50 100 ), ESE-WNW (set V, 90 120 ), NW-SE (set VI, 7of17

Table 1. Samples With Multi-FIA Measured From Garnet Porphroblasts Around the Baltimore-Washington Anticlinorium of the Pennsylvania Orocline a Core Median Rim ArmM Sample Longitude Latitude Porphyroblast Porph FIA Set FIA Set FIA Set P/H 1 76.4857 39.2921 Gt-St-Ky 175 3 70 4 P/H 3 76.5116 39.2645 Gt 45 1 5 3 P/H 4 76.5874 39.2733 Gt-St 25 2 155 6 P/H 5A 76.6386 39.2553 Gt-St-Ky 150 180 3 130 150 6 P/H 5B 76.6386 39.2553 Gt-St 150 180 3 95 4 P/H 9A 76.4852 39.2914 Gt-St-Pl 165 3 175 7 P/H 12 76.4856 39.295 Gt-St 35 1 175 3 M/SW 14 76.782 39.1191 Gt 60 90 4 90 120 5 M/SW 16 76.7801 39.1198 Gt 150 180 3 120 150 6 M/SW 17 76.7582 39.1114 Gt 5 3 115 5 85 8 M/SW 18 76.758 39.1123 Gt-St 165 3 175 3/7? M/SW 21 76.7587 39.1143 Gt 150 210 3 90 5 120 150 6 M/SW 22A 76.6915 39.0976 Gt-St 20 2 85 4 M/SW 23A 76.6894 39.0984 Gt-St 170 3 155 6 M/SW 32 76.5393 39.0788 Gt 60 90 4 150 6 M/SW 34 76.5471 39.0805 Gt 25 2 95 4 P/H 48A 76.4856 39.2942 Gt-St-Ky-Sil 10 2 95 4 P/H 48B 76.4874 39.2934 Gt-St-Ky 150 6 150 170 7 P/H 49 76.4875 39.2925 Gt-St 5 3 135 6 20 7 P/H 50 76.4877 39.2921 Gt-St-Ky 0 30 2 175 3 P/H 51 76.4887 39.2917 Gt-St 165 3 175 3 P/H 53 76.4888 39.2901 Gt-St-Ky-Pl 155 6 165 7 P/H 54B 76.4893 39.2889 Gt-St-Ky 15 2 60 4 P/H 57 76.4694 39.3002 Gt-St-Ky 25 2 55 4 P/H 59 76.4717 39.3025 Gt-St-Ky 0 30 2 95 4 P/H 60 76.4843 39.2954 Gt-St 90 110 5 85 8 P/H 61 76.4843 39.2954 Gt-St-Sil 0 20 2 120 150 6 P/H 62 76.4866 39.2937 Gt-St-Pl 25 2 5 3 120 5 P/H 63 76.4864 39.2939 Gt-St-Ky 175 3 125 5 P/H 64 76.4993 39.2784 Gt-St-Ky 30 60 1 60 90 4 M/SW 70A 76.7651 39.2026 Gt-St-Ky 150 210 3 120 5 M/SW 71 76.7642 39.201 Gt-St 30 60 1 150 180 3 M/SW 82 76.75 39.211 Gt 175 3 150 6 M/SW 83 76.7469 39.2107 Gt 90 120 5 175 7 M/H 86 76.5442 39.1787 Gt 170 3 90 4 M/SW 87 76.7551 39.2123 Gt 90 120 5 150 180 7 M/SW 89 76.7503 39.2129 Gt 30 60 1 130 6 M/SW 91 76.7461 39.2129 Gt 175 3 70 4 M/H 93 76.5583 39.1826 Gt-Sil 75 4 85 8 P/H 101 76.8148 39.3293 Gt 60 90 4 90 120 5 P/H 105 76.6593 39.248 Gt 10 20 2 115 5 130 6 M/SW 112 76.8084 39.287 Gt 30 60 1 0 30 2 M/SW 114 76.8098 39.2888 Gt 150 3 65 4 M/SW 115 76.8115 39.2876 Gt 150 180 3 70 4 M/SW 120 76.7394 39.2446 Gt 150 180 3 70 4 P/H 130 76.6503 39.32 Gt 20 2 175 3 P/H 132 76.5897 39.3614 Gt 150 180 3 55 4 P/H 145 76.4172 39.3008 Gt-St-Ky-Pl-Sil 55 1 0 3 P/H 146 76.4141 39.3024 Gt-St-Ky-Sil 15 2 70 4 P/H 149 76.4881 39.293 Gt-St-Ky 40 1 120 150 6 P/NE 155 76.2681 39.2528 Gt 150 180 3 60 4 P/NE 158 76.246 39.3042 Gt 175 3 0 30 7 80 8 P/NE 160 76.2398 39.3045 Gt-St-Ky-Sil 5 2 150 170 3 P/NE 164 76.3081 39.3367 Gt 150 180 3 90 120 5 25 7 P/NE 166 76.3221 39.3398 Gt 150 180 3 95 4 P/NE 167 76.3198 39.3393 Gt-Sil 5 2 160 180 3 120 160 6 P/NE 174 76.243 39.3046 Gt 60 1 45 7 P/NE 178 76.2215 39.327 Gt 80 4 85 8 P/NE 180 76.2195 39.3267 Gt 125 5 140 6 175 7 P/NE 183 76.2172 39.3277 Gt-St 177 3 62 92 4 M/SW 185 76.6258 39.0507 Gt-St-Ky 50 1 30 2 P/NE 201 76.2174 39.3239 Gt-St-Ky-Sil 150 180 3 0 40 2 60 90 4 P/NE 203 76.2255 39.3259 Gt-St-Ky-Sil 5 3 10 3 P/NE 204 76.2257 39.3267 Gt-St 100 5 145 6 P/NE 205 76.2234 39.3274 Gt-St-Ky-Sil 5 3 5 3 170 190 7 8of17

Table 1. (continued) Core Median Rim ArmM Sample Longitude Latitude Porphyroblast Porph FIA Set FIA Set FIA Set P/NE 206 76.2248 39.3273 Gt-St-Ky 165 3 10 7 P/NE 207 76.2264 39.3256 Gt-St-Ky 150 6 10 7 P/NE 215 76.3396 39.2658 Gt-St 65 1 75 4 P/NE 223 76.2715 39.3083 Gt 90 120 5 25 7 P/NE 225 76.2806 39.3099 Gt 170 3 115 5 a The two subgroupings within the orocline are the nappe system (abbreviations are as follows: P, Pennsylvania nappe; M, Maryland nappe) and staircase system (NE, northeast limb, H, hinge area, SW, south west limb). 120 150 ), N-S (set VII, 160 200 ), and finally to E-W (set VIII, 70 110 ), is proposed. 4.2. Comparison of FIA Distribution Between Different Subregions Across the Pennsylvania Orocline [16] To test the orocline hypothesis, we subdivided the region in two ways. [17] 1. The arc was divided into two according to the two nappes described in the geological setting, with the boundary between them aligned along the maximum curvature of the Pennsylvania orocline (line A-A 0 in Figure 2). [18] 2. The arc was divided into three domains consisting of two outer NE-SW trending arms and a central W-E trending zone that define a regional staircase shape (line B-B 0 and C-C 0 in Figure 2). [19] A comparison of all FIA data measured from garnet porphyroblasts, conducted using a X 2 test, revealed no significant differences between the FIA populations from the different domains of the Pennsylvania orocline regardless of the manner in which the region was subdivided (Tables 2 and 3). The X 2 value is 5.447 for the comparison of the Pennsylvania nappe and Maryland nappes (Table 2). For the 3 domain split the X 2 values are 6.138 for the NE arm versus the SW arm, 2.043 for the hinge versus the SW arm, 2.755 for the hinge versus the NE arm and 6.879 for the NE arm versus the hinge versus the SW arm (Table 3). These are all much smaller than the critical value (11.07 with 5 degrees of freedom) suggesting acceptance of the null hypothesis (H 0 ) that statistically, no significant difference in the FIA population can be distinguished between the subdomains of the orocline. [20] Further comparisons were conducted by superimposing the distributions of moving averages of FIA populations across different regions (Figure 7). All FIA populations for different subregions other than the NE arm show the same peaks and troughs on the basis of moving averages (Figures 7c and 7d). A total of six peaks around 20 30, 40 50, 70 80, 100 110, 130 140 and 170 180 Figure 5. Microphotograph and accompanying line diagram showing garnet porphyroblasts with gently curved rim inclusion trails continuous with S 1 in the matrix and overprinted by S 2. Vertical thin section, arrow indicates strike, plane polarized light, sample M5A. Weakly developed subhorizontal S 3 can also be observed. The core inclusion trails are defined by fine-grained ilmenite and quartz. The rim portions are defined by graphite, coarse ilmenite, and quartz inclusions. Staurolite and kyanite porphyroblasts are also present. 9of17

Figure 6. Map showing the staurolite and kyanite zones and the locations of samples containing the younger FIA succession relative to these. were observed through most sub domains of the orocline. The NE arm, has peaks around 0 10, 60 70, 80 90, 100 110, 140 150 and 170 180 based on the moving averages, but no significant differences were revealed by the X 2 test. 4.3. Comparison of FIA Set Sequences Between Different Subregions Across the Pennsylvania Orocline [21] Even if the FIA populations between different regions belong to the same population statistically, the temporal succession of FIAs might vary between the three regions. Consequently, the data were treated according to the same two systems of subdivisions described above, to evaluate whether the core/rim FIAs record the same succession throughout the subregions. An identical comparison was conducted for all core/rim FIAs between the different regions to see if there is any time variance in the growth of garnet porphyroblasts. The X 2 values for both core and rim FIAs between different regions (Tables 2 and 4) are all much smaller than the 0.05 level of significance, suggesting acceptance of the null hypothesis (H 0 ) that the core/rim FIAs between different regions belong to the same population regardless of the division of the system. That is, all core/rim FIAs throughout the orocline represent the same period of history. 4.4. Implications of FIA Data for Orocline Development [22] The statistical results described above indicate that the measured FIA trends and the same relative time relationships remain consistent across the Pennsylvania Orocline. The similar isotopic ages for the peak metamorphism and polydeformation style (indicated by both matrix foliations and FIAs [Tilton et al., 1970; Grauert, 1973a, 1973b, 1974; Glover et al., 1983; Muller and Chapin, 1984; Sinha et al., 1997] suggest similar histories were recorded by the FIA succession along the orocline. This neither supports nor detracts from formation of the orocline as a primary feature, such as a plate boundary (model I) that might be affected by later deformation such as a strike-slip fault (model II). 10 of 17

Table 2. X 2 Test of Independence of the Null Hypothesis, H 0, That the Distributions of FIA Trends Between the Pennsylvania Nappe and Maryland Nappe of the Baltimore Orocline Are the Same a Pennsylvania Nappe Maryland Nappe Total FIAs 0 19 25 6 20 69 24 14 70 89 20 15 90 119 18 13 120 159 20 13 160 179 36 17 Total 143 78 Core FIAs 0 69 22 9 70 159 11 6 160 179 13 8 Total 46 23 Rim FIAs 0 79 20 6 80 149 17 10 150 179 10 7 Total 47 23 a Tests results assessed at the 0.05 level of significance with the degrees of freedom defined by the number of groupings minus 1. The null hypothesis is rejected when the X 2 result value exceeds the critical value. For total FIAs, X 2 2 = 5.447; d.f. 5, X 0.05 = 11.07, not significant. For Core FIAs, X 2 = 2 0.502; d.f. 2, X 0.05 = 5.99, not significant. For Rim FIAs, X 2 = 1.874; d.f. 2, X 2 0.05 = 5.99, not significant. However, it diminishes the possibility that the orocline formed by the overprinting of two orogens with different trends. If this had occurred a significant difference in the distribution and perhaps the temporal succession of FIAs from arm to arm would be expected as they might preserve a different deformation history [e.g., Bell and Mares, 1999]. The lack of variation in the progression of FIA trends suggests that the porphyroblasts containing the successive sets of FIAs were not rotated or altered during younger periods of ductile deformation. [23] To test models I and II further, the proportions of samples containing each FIA set were compared across the orocline. If the orocline was modified by later deformation, a difference in the proportion of the FIA sets responsible for such a development might be seen between the different regions. Figure 8 shows the distribution of FIA sets for the twofold versus threefold system of dividing the arc mentioned above. No significant differences are apparent for division based on the two nappes (Figure 8a). However, a strikingly higher concentration of set II FIAs (10 30 ) were measured from the W-E trending central region compared to the NE and SW arms for the threefold subdivision (Figure 8b). A higher proportion of set I FIAs (30 50 ) versus sets III (160 190 ) and VII FIAs (160 200 ) are present on the SW and NE arms, respectively (Figure 8b). [24] The similarity in distributions of the FIA sets, for the division into two nappes, suggests that the orocline shape reflects a primary paleocontinental margin, and that the variation of the trends of the regional folds were formed at a very early stage of orogenesis. These variations may have remained essentially unaltered as orogenesis continued. However, the distribution and orientation of sets I, II and III FIAs for the threefold division suggests that the orocline resulted from later overprinting deformation. Most set I FIAs were measured from the Maryland nappe, or the SW arm. The similar NE-SW trends of the NE and SW arms may indicate that this was the primary orientation of the Pennsylvania orocline. If a FIA trend reflects the direction of bulk shortening [Bell et al., 1995, 1998], then the set II FIA resulted from WNW-ESE directed compression. The higher proportion of these FIA in the hinge region (Figure 8b) may indicate that the orocline is not a bent orogen, but rather a product of Table 3. X 2 Test of Independence of the Null Hypothesis, H 0, That the Distributions of FIA Trends Between Different Regions of the Baltimore Orocline Are the Same a Hinge Area NE Limb 0 19 13 12 20 69 16 8 70 89 16 7 90 119 11 8 120 159 14 7 160 179 21 17 Total 91 59 Hinge Area SW Limb 0 19 13 6 20 69 16 14 70 89 16 12 90 119 11 12 120 159 14 12 160 179 21 15 Total 91 71 SW Limb NE Limb 0 19 6 12 20 69 14 8 70 89 12 7 90 119 12 8 120 159 12 7 160 179 15 17 Total 71 59 SW Limb Hinge Area NE Limb 0 19 6 13 12 20 69 14 16 8 70 89 12 16 7 90 119 12 11 8 120 159 12 14 7 160 179 15 21 17 Total 71 91 59 a Tests results assessed at the 0.05 level of significance with the degrees of freedom defined by the number of groupings minus 1. The null hypothesis is rejected when the X2 result value exceeds the critical value. For hinge area versus NE limb, X 2 = 2.755; d.f. 5, X 0.05 2 = 11.07, not significant. For hinge area versus SW limb, X 2 = 2.043; d.f. 5, X 0.05 2 = 11.07, not significant. For SW limb versus NE limb, X 2 = 6.138; d.f. 5, X 0.05 2 = 11.07, not significant. For SW limb versus hinge area versus NE limb, X 2 = 6.879; d.f. 5, X 0.05 2 = 11.07, not significant. 11 of 17

Figure 7. (a) FIA trends measured in garnet porphyroblasts in samples from different subregions based on the two-arm division of the arc shown in Figure 2. Equal-area (radii = square root of frequency) rose diagrams with a 10 class interval. (b) Smoothed frequency distribution of the fractionated moving averages with the order N = 5. Both populations show similar fluctuations with six peaks between 10 20, 40 50, 70 80, 90 110, 130 150 and around 170 indicated with the solid arrows. The peak between 40 50 is more strongly indicated with the Maryland nappe population, while the peak between 10 20 is more strongly indicated with the Pennsylvania nappe population. (c) FIA trends for different subregions based on the staircase division of the arc shown in Figure 2. Equal-area (radii = square root of frequency) rose diagrams with a 10 class interval. (d) Smoothed frequency distribution of the fractionated moving averages with the order N = 5. All subregion populations other than that for the east arm show similar fluctuations. Six distinct peaks between 10 20, 40 50, 70 80, 90 110, 130 150 and around 170 are present indicated by black arrows. Although the east arm population shows similar fluctuations to the other regions, the peaks are about 10 further to the east when as indicated by the gray arrows. heterogeneous W-E directed dextral shear along the axis of Chattolanee dome (Figure 1). [25] Dextral shear along a transcurrent system through the central Appalachian piedmont has previously been suggested as occurring during the Alleghanian [Vauchez et al., 1987; Krol et al., 1990, 1999; Gates et al., 1999; Valentino, 1999]. However, an Alleghanian timing is unlikely for development of the set II FIA. The metamorphism is considered to be late Taconian and Acadian [Tilton et al., 1970; Grauert, 1973a, 1973b, 1974; Muller and Chapin, 1984] and this is one of the earliest FIA sets that formed. The inclusion trails within the garnet porphyroblasts that define this FIA set are also truncated by the matrix foliation. Perhaps dextral shearing began in the late Taconian but reoccurred at lower metamorphic grade, unaccompanied by garnet porphyroblast growth, during the Alleghanian orogeny. Such later Alleghanian dextral shearing is potentially supported by the FIA data as all moving average peaks from the NE arm show a uniform 10 clockwise shift compared to the hinge region and the SW arm (Figure 7). This coincides with a 10 variation in the strike of the NE arm compared to the SW arm (Figure 2) and may result from some component of rigid block rotation due to faulting post amphibolite facies metamorphism. [26] If compression along the length of the orogen had occurred to form the arc, higher proportions of NW-SE trending set VI FIAs resulting from SW-NE directed compression would be expected instead of the set II FIAs observed. If some form of block rotation had occurred, a near 70 of clockwise shift of the older FIA sets should occur in the central W-E trending region. However, the same peaks in FIA populations were recorded in the latter region as in the NE and SW arms (Figure 7). The higher proportion of set VII FIAs recorded on the NE arm (Figure 8b) may reflect the metamorphic conditions rather than orocline development. As described above, set VII and VIII FIAs were mainly recorded from samples located within the kyanite zone. A higher proportion of these FIA sets are predicted for the NE arm as it covers more kyanite zone than other regions (Figure 6). 5. Discussion [27] Following the advent of plate tectonics, most pre 170 Ma orogens were reinterpreted in terms of plate tectonics Table 4. X 2 Test of Independence of the Null Hypothesis, H 0, That the Rim and Core FIA Trend Distributions Between Different Regions of the Baltimore Orocline Are the Same a Regions Rim FIAs Core FIAs Hinge area versus NE limb 3.747 2.004 Hinge area versus SW limb 0.012 0.728 SW limb versus NE limb 3.510 0.371 SW limb versus Hinge area versus NE limb 4.821 2.107 Degree of freedom 1 2 2 X 0.05 3.84 5.99 a Tests results are assessed at the 0.05 level of significance with the degrees of freedom defined by the number of groupings minus 1. The null hypothesis is rejected when the X 2 result value exceeds the critical value. 12 of 17

Figure 8. Histograms showing the percentage distribution of each FIA set recorded from the different subregions around the Pennsylvania orocline. (a) Arc divided into two arms; (b) arc divided into three portions with a staircase shape. Standard error (SE) = 3.79% for two-arm division, and 2.61% for two long arms and one central hinge. SE = {sum of Yjs2/[(ny 1)ny]} 2 for nongrouped independent random populations. Both ways of dividing the arc show similar distributions for each FIA set across the orocline. No significant differences can be observed between the subregions except for set II, III, and VII for the two long arms and one central hinge. processes. During this period, Lefort [1984, 1988] proposed that the Pennsylvania orocline was an arcuate megastructure caused by the rigid indentation of the Reguibat uplift with Laurentia during the Carboniferous collision of North America with Africa. Vauchez et al. [1987] expanded Lefort s model by having this zone form the northern boundary of a region where there was southward extrusion of the Appalachians. Valentino et al. [1994, 1995] proposed an alternative theory with no indenter and Valentino and Gates [1995] showed that the stratigraphy of the area favored original rifting as the mechanism that produced the orocline. Our study appears to preclude indentation during the Carboniferous as a model for the bulk of formation of the Pennsylvania orocline. It does so because garnet porphyroblasts preserving FIA set II, which grew early during amphibolite facies metamorphism extending from the Taconic (510 to 460 Ma) to the early Acadian (408 to 360 Ma [Tilton et al., 1970; Fisher, 1971; Grauert, 1973a, 1973b, 1974; Lang, 1990; Sinha et al., 1997]), are preferentially distributed along the W-E trending portion of the arc. 5.1. Primary Plate Boundary Model [28] The consistent succession of FIA trends across the Pennsylvania orocline suggests this arc formed as a primary feature, at a very early stage in the orogenic history of these rocks. This is supported by a lack of significant variation in the palaeomagnetic data across the orocline [Schwartz and van der Voo, 1983]. It is also supported by sedimentary units of the Appalachian-Ouachita geosyncline being distributed in a manner coincident with the arc, potentially reflecting the continental margin shape during the convergence of North America and the Avalon micro plate 13 of 17

transform-rift junctions creating the primary curvature in the plate boundary. Figure 9. Map showing the localization of orogen flexure in Pennsylvania to the north, and near Baltimore in Maryland to the south, both marked with a shaded zone. The distribution of the 20 trending set II FIA is shown. The flexure localization near Baltimore contains all samples but one with the set II FIA. These FIA are distributed along the regionally localized zone of flexure in a W-E trending zone. However, the FIA themselves trend SSW-NNE. This suggests that the direction of bulk shortening that produced this FIA set resulted in a zone of shearing and rotation of the orogen along a preexisting zone of crustal weakness. [Thomas, 1977; Dalziel et al., 1994]. Consequently, several studies [e.g., Fleming and Sumner, 1975; Rankin, 1976; Schwartz and van der Voo, 1983; Gates and Valentino, 1991; Valentino et al., 1994; Valentino and Gates, 1995; Gates et al., 1999; Valentino, 1999] have suggested that the shape of the initial breakup of the precursor continental mass was the origin of an arcuate Palaeocontinental margin. Using gravity and magnetic anomalies at the southeastern end of Pennsylvania, Fleming and Sumner [1975] proposed that an embayment of ancient Iapetus oceanic crust into the continental margin was present. Rankin [1976] interpreted occurrences of rhyolite in the arcs as the location of failedarm troughs. Thomas [1977] interpreted that the orthogonally zigzag continental margin shape originated from several transform faults along a Late Precambrian rift during initial spreading of the Iapetus oceanic lithosphere, the 5.2. Arc Deformation/Rotation and Arc Overprinting Models [29] Thrusts and strike-slip faults are common at terrane boundaries [Gates et al., 1991; Gates and Valentino, 1991]. The possibility of an Iapetan rift-related W-E trending transform fault around 40 N latitude that crosscuts the Maryland-Pennsylvania Piedmont has been suggested by several researchers [Woodward, 1964; Thomas, 1977, 1983; Fisher et al., 1979; Krol et al., 1990; Gates and Valentino, 1991; Valentino and Gates, 1995]. It has been argued that this unseen wrench fault was a major contributor to the meandering nature of the Pennsylvanian orocline. Transform faults that accompanied the late Proterozoic-Cambrian rifting that formed the Iapetus Sea may have been a major factor in shaping the continental margin before the orogen formed. This possibility is supported by the sediment distribution and palaeomagnetism [Thomas, 1977; Valentino and Gates, 1995]. [30] The FIA results from this study indicate that deformation during the development of FIA set II was preferentially partitioned into a W-E trending zone centered on the W-E trending portion of the arc (Figure 9). This FIA set trends SSW-NNE, suggesting that bulk shortening was directed WNW-ESE (Figure 9). The preferential development of SSW-NNE trending foliations in a near-orthogonal W-E trending zone is an unusual geometric relationship and would only be likely to occur if a preexisting zone of crustal weakness with such a trend was present that could be reused. Therefore we interpret that the distribution of set II FIA data in Figure 9 suggest a component of W-E dextral shear occurred along a W-E trending zone when the porphyroblasts containing these FIA grew. This occurred during the Taconic orogeny as this FIA set is the second of eight that formed from the Taconian through into the Acadian according to the dates on the timing of metamorphism [Tilton et al., 1970; Fisher, 1971; Grauert, 1973a, 1973b, 1974; Lang, 1990; Sinha et al., 1997]. Such deformation should have contributed significantly to the development of the oroclinal arc and supports the suggestion that a zone of crustal weakness, produced by a primary W-E trending sinistral transform fault zone, resulted in later dextral shear that enhanced the arc shape [e.g., Muller and Edwards, 1985]. [31] Truncation of the Cambrian-Ordovician shelf [Rodgers, 1968] along the Pleasant Grove-Huntingdon Valley shear zone northwest of Philadelphia and Baltimore has been used to suggest that 150 km of dextral displacement placed the Baltimore-Philadelphia crustal block into the New York promontory during the late Palaeozoic after the Taconic orogeny [Valentino et al., 1994]. Reactivation of a bounding sinistral transform fault as a dextral strikeslip fault along the central Appalachian piedmont during the Alleghanian has been suggested by several studies [Vauchez et al., 1987; Krol et al., 1990, 1999; Gates et al., 1999; Valentino, 1999]. If the curvature of the orocline resulted from strike-slip movement without block rotation, 14 of 17

very little variation in the paleomagnetic pole, and in the trends of FIA sets across the orocline would be observed. However, displacement on narrow shear or fault zones during the Alleghanian, may have resulted in some fault block like rotation of the NE arm of the orocline and produced the 10 shift in FIA trend that we have noted. 5.3. Significance and Application of FIAs [32] This study has revealed a striking consistency in orientation and relative timing of the FIA sets across a large region that has undergone multiple deformation events. A similar consistency has been observed across the Kimberley Arc, NW Australia [Bell and Mares, 1999] and the Vermont Appalachians [Bell et al., 1998, 2004]. Many of the matrix foliations were completely obliterated or reoriented by intensely developed, highly partitioned deformation. If each developing generation of porphyroblasts had formed by rotation, previously developed porphyroblasts would also have been rotated creating a complex pattern and random succession of relative timing of porphyroblast growth versus FIA trend. This, however, is not the case and consistent trends and relative timing of FIA sets are preserved along the length of the portion of the orogen examined. [33] Correlation of structural history is the biggest problem faced by those seeking to resolve how oroclines form. FIAs in the Central Appalachians and the Kimberley Arc, NW Australia have enabled correlation of structural histories between rocks that have potentially underwent very different P-T-t paths and which have been affected by younger shear zones. The regional distribution of FIAs in both cases has provided strong indications as to the mechanism by which each orocline formed, and showed that the FIAs can be used to effectively solve regional tectonic problems and should be more regularly employed. 6. Conclusions [34] The striking consistency in orientation and relative succession of FIA sets across the Pennsylvania orocline indicates that successive FIA trends have not been rotated by subsequent phases of overprinting ductile deformation and that the same deformation events have affected the bulk of the Pennsylvania orocline. The latter relationship appears to support the suggestion that the orocline reflects the primary geometry of the continental boundary as indicated by sedimentary and palaeomagnetic data [Fleming and Sumner, 1975; Rodgers, 1975; Rankin, 1976; Thomas, 1977; Schwartz and van der Voo, 1983]. However, the second developed SSW-NNE trending FIA set, which probably formed during the Taconic orogeny, is preserved in significantly more samples along the central W-E trending portion of the orogen than to either side. This FIA trend requires approximately WNW-ESE directed bulk stress, which in combination with a W-E zone of SSW-NNE trending FIAs suggests orogenesis at this time resulted in a zone of dextral shear along what is now the W-E trending portion of this orogen. This may have involved reactivation of a primary W-E trending transform fault and the trend of this portion of the orogen could be predominantly the result of the deformation that occurred as this FIA set developed, possibly with some Alleghanian dextral shear superimposed. Appendix A: Watson s U 2 Test for Grouped Data [35] The Watson s U 2 test for uniformity is based on the variability of the squared differences of the observed frequencies associated with directions observed with those expected according to the null hypothesis. To adapt this test for use with grouped axial data, changes of the periodicity from 180 to 360 by doubling the recorded directions and scaling into 10 groups is required [Freedman, 1981; Upton and Fingleton, 1989; Bell et al., 1998]. [36] With n i observations in the ith groups after K groups are subdivided, a probability p i is specified according to the null hypothesis for the ith group. The test is based on the difference between observed and expected cumulative frequencies, thus it is necessary to compute S j ¼ X ðn i np i Þ; j ¼ 1; 2; 3::k: ða1þ The grouped version of Watson s U 2 test is then: U 2 g ¼ 1=nk X S 2 j 1=k X 2 S j : ða2þ Because the Watson s U 2 test uses cumulative frequencies, the ordered nature of the successive groups are explored, whereas the chi-squared test will have less power for detecting such ordering when it occurs. Appendix B: Moving Averages [37] The frequency distributions of tests measurements often contain random fluctuations as well as more meaningful ones. It is always difficult to interpret some true meaning for these random fluctuations. Moving averages provide a mean to smooth the frequencies to see more meaningful peaks or modes in the graph. By applying a moving average, it may be possible to discern modes, which may represent real properties of the sampled populations. [38] For any set of frequencies (F 1, F 2, F 3... F n ), a moving average of order N has the following sequence of arithmetic means: ðf 1 þ F 2 þ...þ F N Þ= N ; ðf 2 þ F 3 þ...þ F Nþ1 Þ= N and so on ðb1þ [39] In other words, the smoothed frequencies are calculated as the mean of N consecutive frequencies. The choice of smoothing factor (value of N) is largely a matter of judgment [King, 1994]. [40] The order N = 5 is applied for the FIA data as each FIA class has a 10 interval and the natural spread of each FIA set ranges between 40. Thus, to eliminate the effects of the natural spread, the order of 5 (=50 ) is applied. Furthermore, since FIAs are axial data, the moving average 15 of 17