New geologic and paleomagnetic data from Knight Inlet in the southwestern Coast Mountains Batholith, British Columbia, support significant revision to the paleogeography of the Insular and Intermontane terranes. Recompilation of radiometric ages confirms that after 100 Ma, a magmatic arc migrated northeastward across the Coast Mountains Batholith at ∼2 km/m.y. Magmatic age patterns suggest that plutons older than 100 Ma intruded the Intermontane terrane, not the expected Insular terrane. The distribution of brittle faults along Knight Inlet defines a structurally intact central domain, ∼45 km wide, flanked to the SW and NE by faulted domains, with no evidence of the widespread Tertiary extension affecting the batholith farther north. Al-in-hornblende geobarometry yields emplacement depths of ∼2.5–4 kbar and does not reveal systematic postemplacement tilting. Plutons in the central structural domain yield a consistently oriented paleomagnetic remanence presumably acquired as the Late Cretaceous arc cooled from ca. 110 to 85 Ma. In the absence of recognizable tilting, this result indicates ∼1700 km of northward translation since ca. 85 Ma, which is significantly less than predicted for the Insular terrane in the “Baja British Columbia” model but similar to results from the Intermontane terrane. The pluton ages and the paleomagnetic results suggest that the Intermontane terrane, not the Insular terrane, underlies the southwestern flank of the Coast Mountains Batholith. This conclusion is compatible with a paleogeographic model in which the Vancouver Island fragment of Wrangellia was juxtaposed against the Intermontane terrane prior to ca. 120–100 Ma and emplaced in southern British Columbia after ca. 75 Ma.


The Coast Mountains Batholith, or Coast plutonic complex, is one of the largest batholithic belts on Earth, yet much of its evolution, especially its Mesozoic paleogeography, is poorly understood. Unlike the better-known Mesozoic Sierra Nevada Batholith of California, paleomagnetic studies of the plutonic rocks and adjacent stratified rocks suggest that much of western British Columbia was located 1000–3000 km farther south relative to cratonic North America in the Late Cretaceous (Beck and Noson, 1972; Beck et al., 1981; Bogue et al., 1995; Bogue and Grommé, 2004; Enkin, 2006; Enkin et al., 2006; Haskin et al., 2003; Housen et al., 2003; Irving et al., 1996, 1985; Kim and Kodama, 2004; Ward et al., 1997; Wynne et al., 1995). This interpretation, referred to as the “Baja British Columbia” hypothesis (Irving, 1985), has fueled controversy for decades, (e.g., Cowan et al., 1997; Butler et al., 2001b). The crux of the debate is that the northward translation called for by the paleomagnetic results is larger than has been accounted for by geologic studies and seems to require differential displacement of crustal blocks that appear geologically intact. Despite this discrepancy, geological and paleomagnetic research in the last decade has irrefutably shown that rocks in British Columbia and Alaska lay south of their present location in Late Cretaceous and earliest Tertiary time (summarized in Enkin, 2006; Wyld et al., 2006). Geologic estimates call for up to ∼900 km of northward translation since mid-Cretaceous time; paleomagnetic estimates vary from ∼1000 to 3000 km.

Beyond the fundamental mismatch in paleolatitude, there is also disagreement on the size and distribution of translated blocks. Early models calling for larger displacement (∼2500 km) of a single Baja British Columbia block juxtaposed against an “interior domain” of rocks displaced ∼1500 km (e.g., Irving et al., 1985, 1996; Cowan et al., 1997) were followed by reconstructions envisioning differential displacement, with the more western blocks displaced farther than those in the east (Wyld et al., 2006; Umhoefer, 2003). A return to the single-block model was suggested by work on stratified rocks in central British Columbia (around Mount Tatlow and Spences Bridge on Fig. 1; Enkin, 2006; Enkin et al., 2003, 2006; Haskin et al., 2003; Kent and Irving, 2010), but in this scenario, the far-traveled block is much larger, encompassing much of British Columbia. These studies show clearly that parts of British Columbia were translated ∼2000–2500 km northward with respect to cratonic North America between ca. 100 and 50 Ma, but the extent, terrane affiliation, and travel history remain uncertain for much of British Columbia. Insight into the paleogeographic reconstruction of the entire Cordillera hinges on understanding the nature of these crustal blocks now lodged in western British Columbia.

In this study, we investigate the Late Cretaceous paleolatitude and relevant geologic features of plutons on the western flank of the southern Coast Mountains Batholith (Fig. 1) to better define the amount of displacement and size of blocks within the batholith. These plutons are comparable in age, size, and lithology to others in the Coast Mountains Batholith (the Porteau and Spuzzum plutons, 300 km to the SE, and the Captains Cove, Ecstall, and Stephens Island plutons, 400 km to the NW) that have figured prominently in the Baja British Columbia debate.


The Coast Mountains Batholith is composed of Jurassic–Tertiary plutons that define a belt more than 1600 km along the present coastline of British Columbia and southeastern Alaska. Arc magmatism was driven by subduction of oceanic plates from the Jurassic until Early Tertiary time (Armstrong, 1988; Engebretson et al., 1985; Lonsdale, 1988; Monger et al., 1982; Stock and Molnar, 1988; van der Heyden, 1992), when contact with the Pacific plate began to convert the margin from convergent to dextral transform (Engebretson et al., 1985; Haeussler et al., 2003; Hyndman and Hamilton, 1991; Madsen et al., 2006; Stock and Molnar, 1988). Significant Early Tertiary crustal extension disrupted the central part of the orogen, adjacent to the transform boundary, causing opening of the Queen Charlotte basin and tilting of crustal blocks within the batholith (Butler et al., 2001a; Butzer et al., 2004; Dehler et al., 1997; Rohr and Dietrich, 1991; Rohr and Currie, 1997; Rusmore et al., 2010). A convergent margin persisted along the southern part of the orogen and produces an active volcanic arc in the Pacific Northwest.

The Coast Mountains Batholith straddles the boundary between allochthonous terranes, grouped into the eastern Intermontane terrane, with Stikinia and the Yukon-Tanana terrane flanking the Coast Mountains Batholith, and the western Insular terrane, composed primarily of Wrangellia and the Alexander terrane (Monger et al., 1982; Wheeler and McFeely, 1991). Recent work shows the central part of the batholith (between latitudes ∼52°N and 55°N) is composed of two magmatic arcs formed prior to ca. 100 Ma, with the western arc intruding the Insular terrane and eastern arc intruding the Intermontane terrane (Gehrels et al., 2009). A younger eastward-migrating arc was built across the two older arcs beginning around 100 Ma (Armstrong, 1988; Friedman and Armstrong, 1995; Gehrels et al., 2009; van der Heyden, 1992; Woodsworth et al., 1991). Mid-Cretaceous thrust belts affected the entire orogen (e.g., Crawford et al., 1987; Gehrels et al., 1992; Journeay and Friedman, 1993; Monger et al., 1982; Rubin et al., 1990; Rusmore and Woodsworth, 1991, 1994).

The configuration of the batholith prior to development of the mid-Cretaceous arc has been debated. Early models proposed a single west-facing Andean-style arc (van der Heyden 1992; Crawford et al., 1987; Armstrong, 1988; McClelland et al., 1992). Alternative models call for two Jurassic–Early Cretaceous arcs, with the Insular terrane and a west-facing arc separated from the Intermontane terrane by an ocean basin (Monger et al., 1982; Friedman and Armstrong, 1995). In this scenario, subduction beneath the Intermontane terrane drove magmatism in the eastern arc and caused closure of the ocean basin and accretion of the Insular terrane in mid-Cretaceous time. This model predicts an oceanic suture zone between the arcs, which has not been observed. Rather than a mid-Cretaceous suturing to accrete the Insular terrane, Monger et al. (1994) proposed that the Insular terrane was accreted in Late Jurassic time and doubled by sinistral displacement of the northernmost part of the batholith before ca. 100 Ma. This model is compatible with the presence of early–mid-Cretaceous sinistral shear zones and two magmatic arcs in the central part of the batholith (Chardon et al., 1999; Gehrels et al., 2009; Nelson et al., 2012; Wolf et al., 2010).

In all of these models, the Insular terrane is stitched to the Intermontane terrane by ca. 100 Ma. Subsequent dextral transpression and dextral faulting, however, have affected the batholith. Wyld et al. (2006) pointed out that documented dextral faulting has moved the Coast Mountains Batholith and the terranes it intruded at least 450–900 km northward relative to North America since ca. 100 Ma. Within the batholith, additional northward translation is accommodated by NE-SW shortening related to mid-Cretaceous thrust faulting (Journeay and Friedman, 1993; Brown et al., 2000; Rubin et al., 1990; Rusmore and Woodsworth, 1991, 1994; Rusmore et al., 2000) and dextral and dextral reverse faults active within the batholith from ca. 90 to 50 Ma (Andronicos et al., 1999, 2003; Brown et al., 2000; Hollister and Andronicos, 1997). These results help reconcile an unknown amount of the missing displacement called for by paleomagnetic data.

Our study area, on the southwestern end of the Coast Mountains Batholith (Figs. 1 and 2), lies inboard of the long-recognized Wrangellia terrane on Vancouver Island. Tectonic maps show the sparse pendants in the Coast Mountains Batholith as part of the Insular terrane and place the boundary with the Intermontane terrane to the east, in the core of the batholith (Wheeler and McFeely, 1991). This interpretation is consistent with the presence of two magmatic arcs ∼300 km along strike to the northwest (Gehrels et al., 2009), and interpretations based on geochronologic data to the south of our study area (Friedman and Armstrong, 1995). This previous work suggests that the plutons on Knight Inlet should belong to the western arc, and thereby would have intruded into Wrangellia. We chose this study area in order to examine a part of the Insular terrane where no significant crustal extension was expected. In the absence of documented tilting, we would expect results compatible with the ∼2500 km northward latitudinal shift inferred for coeval plutons in the Insular terrane farther north, stratified rocks east of the Coast Mountains Batholith, and the now iconic Mount Stewart pluton in the North Cascades.

Our work focuses on an orogen-perpendicular transect extending from islands fringing the mainland to the northeast up Knight Inlet (Fig. 2). Jurassic–Cretaceous plutonic rocks underlie most of the area, together with minor pendants of volcanic and metamorphic rocks (Roddick and Woodsworth, 1977, 2006). Pendants near Vancouver Island are part of Wrangellia, but the tectonic affinity of small metavolcanic and metasedimentary pendants farther up the channel is unknown. The plutons are mostly tonalite and granodiorite with subordinate amounts of diorite and granite. As mapped by Roddick and Woodsworth (1977, 2006), most plutons are elongate, striking northwest, and are a few kilometers to ∼15 km across. Magmatic foliations are absent to moderately developed in the plutons; these foliations strike generally northwest and dip steeply. Most plutons locally contain abundant mafic enclaves and are commonly cut by basaltic to andesitic dikes of uncertain ages. No through-going ductile shear zones were found in the area considered in our interpretation of the paleomagnetic results, although an ∼50-m-wide mylonite zone is present. Field relations suggest that this shear zone is intruded by mid-Cretaceous plutons and thus predates assembly of the mid-Cretaceous batholith interpreted in this study.


Age Compilation

All published K-Ar, Ar-Ar, and U-Pb data for the region were compiled, locations were verified, pre-1977 K-Ar ages were recalculated, and K-Ar ages were culled based on our assessment of reliability. Consequently, the resulting compilation differs from that of Breitsprecher and Mortensen (2004), in which all data were reported directly and users were warned to check all original sources. Data for our compilation are from Friedman and Armstrong (1995); Green et al. (1988); Hunt and Roddick (1990, 1993, 1996); Isachsen et al. (1985); Monger and McNicoll (1993); Nelson (1979); Roddick (1996); Roddick and Tipper (1985); Roddick and Woodsworth (1977, 2006); Rusmore et al. (2000, 2001); Stevens et al. (1982); Wanless et al. (1974, 1978, 1979); and Woodsworth et al. (2000). In cases where we suspected that location errors had crept into successive compilations, locations were verified from original field maps and notes available from the Geological Survey of Canada. K-Ar dates done before 1977 were recalculated using the modern constants of Steiger and Jäger (1977). All whole-rock K-Ar dates on plutonic rocks (mostly done in the 1960s) were discarded as unreliable, given that unknown amounts of the K in such samples may be contained in K-feldspar with an Ar closure temperature of <300 °C. Seven pairs of K-Ar dates in which the biotite dates were older than hornblende dates were also discarded as unreliable. One K-Ar date was removed because it was >20 m.y. older than a U-Pb zircon age on the same rock, and laboratory notes report possible excess Ar. Because they are not relevant to this paper, Ar-Ar dates from Pliocene–Pleistocene volcanic complexes were excluded. In total, this compilation includes 23 U-Pb zircon ages and 72 Ar-Ar and K-Ar dates on hornblende and biotite.

Dikes and Brittle Faults

Detailed mapping of brittle faults and dikes followed the NE-trending transect used to collect paleomagnetic samples (Fig. 3). Most of the transect is underlain by Mesozoic granitoid rocks in which dikes and brittle faults are easily visible. Nearly continuous coverage along the transect was built from the best exposed areas; poorly exposed parts of shoreline or areas with very rough water were avoided, and mapping was conducted on parallel coastlines. This approach allowed us to map at scales of ∼1:5000–1:10,000. Less-detailed mapping was conducted northeast of the northeasternmost paleomagnetic sites, where metamorphic rocks predominate and exposures are discontinuous, to the head of Knight Inlet. Nonetheless, it is unlikely that significant brittle faults were overlooked in the upper end of the inlet.

To help compare faults on Knight Inlet with those previously mapped on Douglas Channel, all brittle faults were classified according to the scheme used by Rusmore et al. (2010). This ranking is based on the width of the fault zone and the development of fault-related structures within it. Grade I faults are the largest, consisting of many meters of cataclastic and ultracataclastic rocks. No faults this large were observed in the Knight Inlet area. Faults classified as grades II through IV consist of multiple discrete fault surfaces and variably developed cataclastic and ultracataclastic zones. Individual fault surfaces within fault zones II–IV are considered to be kinematically linked. A network of faults wider than a meter characterizes grades II and III; grade II is distinguished from grade III by the common presence of cataclastic rocks between individual fault surfaces and faults marked by cataclastic and ultracataclastic zones greater than 5 cm wide. Grade IV and V zones are less than a meter wide, with cataclastic rocks not present between individual fault surfaces and few to no ultracataclastic rocks. Grade V zones are marked by a single to a few isolated faults lacking intervening cataclastic rocks. These faults appear quite insignificant in the field, although Rusmore et al. (2010) documented 1–2 m of slip on a few grade V faults. The orientation and distribution of dikes and brittle faults were investigated in ArcGIS 9.3 and Spheristat.

Hornblende Geobarometry

Most granitoid samples from the Knight Inlet area are not suitable for hornblende geobarometry; they either lack the required assemblage or are heavily altered to subgreenschist-facies minerals. After screening for the presence of the full buffering assemblage Hbl + Pl + Bt + Kfs + Qz + Ttn + Fe-Ti oxide and checking for obvious alteration to chlorite, sericite, or actinolite, or the presence of subsolidus fabrics, we selected seven samples for analysis.

For each sample, we analyzed 10–12 hornblende grains for 3–4 areas of the thin section. We picked grains with relatively straight grain boundaries in contact with plagioclase, and we analyzed the rims of these grains, usually ∼20 μm from the rim. We avoided altered areas and areas where backscattered electron images suggested albitic alteration rims on plagioclase or chloritic or actinolitic cracks or rims on hornblende. Analyses were done on polished thin sections by wavelength dispersion methods. We used a Jeol Superprobe 733 electron microprobe at the California Institute of Technology, an accelerating potential of 15 kV, a focused beam, and counting times of 40 s, and well-characterized natural and synthetic minerals as standards. Data reduction was done using conventional ZAF methods and the program CITZAF 3.03.

Hornblende structural formulae were calculated using the method of Holland and Blundy (1994), because their temperature calculations, which we used, are based on their method. We discarded analyses with abnormally low total alkalis, which we take to indicate incipient alteration to actinolite, and analyses with abnormally high or low totals. Pressures were calculated using the calibration of Schmidt (1992) as modified for temperature by Anderson and Smith (1995). We calculated temperatures using the plagioclase-hornblende exchange thermometry reaction B of Holland and Blundy (1994) and iteratively solved the equations from Holland and Blundy (1994) and Anderson and Smith(1995). Estimated 2σ error limits are based solely on within-grain and within-sample heterogeneity and on analytical uncertainty and do not incorporate uncertainties in calibration of the barometer. Several comparisons of pressures derived from Al-in-hornblende with those derived from contact aureoles (e.g., Hollister et al., 1987; Ague and Brandon, 1997) suggest that the accuracy of the barometer is about ±1.5 kbar.


Forty-one paleomagnetic sites were sampled in tonalitic to quartz dioritic plutons (Figs. 2 and 3). Most sites were tidal exposures along the walls of the fjord accessible by boat. Typically, seven oriented 2.5-cm-diameter core samples were collected from a continuous outcrop over a distance of ∼10 m. Guided by the experiences of others working in similar rock (e.g., Grommé and Merrill, 1965), we drilled most core samples from rounded, sub-meter-size enclaves, markedly darker and finer grained than the surrounding intrusive rock, which were plentiful at most sites. Site locations were spaced at ∼3–4 km after projection to a transect line perpendicular to the trend of the orogen. At 14 of the sites, cores were oriented using a sun compass. At the remaining sites, cores were oriented using a magnetic compass with one or more sightings to distant landmarks to correct for local magnetic declination.

The remanent magnetizations of short (2.5-cm-long) specimens cut from the core samples were analyzed using commercially available cryogenic magnetometers and demagnetization devices at the Caltech and Occidental College paleomagnetism laboratories. At sites for which designations begin with “1K”, at least 4 of 7 samples underwent stepwise alternating field (AF) demagnetization to peak field strengths of 70 mT. At most (18/24) of these sites, a detailed (typically 14 step) thermal demagnetization to 580 °C was performed on specimens from all seven samples. For the remaining six sites, three samples underwent detailed thermal demagnetizations, and the remaining samples were thermally demagnetized at 150 °C. For 12 of the 17 sites for which designations begin with “3K”, detailed stepwise AF demagnetizations were performed on 5–7 samples per site. At 6 of the 17 sites, specimens from all samples were subjected to detailed thermal demagnetization to temperatures greater than 600 °C. At the remaining sites, specimens from two samples underwent detailed thermal demagnetization for comparison to the AF results. In all thermal demagnetization experiments on specimens from the 3K sites, the specimens were first subjected to light AF demagnetization (peak fields of 2.5 and 5 mT) to reduce the remanence in low-coercivity grains prior to the thermal treatments.


To facilitate comparison of the various data sets, the results are projected to a common transect line with an azimuth of 050° with 0 km at 50.46°N and 126.37° W, shown as line A–B on Figure 2.

Age Compilation

The compiled ages are shown in Figures 2, 4, and 5, and all pre-Pliocene ages, including those that were culled from the final compilation, are listed in supplementary material Table DR1.1 This compilation of 23 U-Pb zircon dates and 72 Ar-Ar and K-Ar dates on biotite and hornblende confirms that plutons in the region are Jurassic to Early Tertiary in age (Friedman and Armstrong, 1995), and it shows that in these three geochronologic systems, the youngest ages occur in the northeast (Figs. 2, 4, and 5). U-Pb zircon ages of plutons range from 165 to 56 Ma. Despite the small number of U-Pb zircon dates, there are no gaps >10 m.y. except between 114 and 97 Ma. There is no discernible relation between location and U-Pb zircon ages older than 114 Ma; these older ages occur throughout the area. The 11 U-Pb zircon ages younger than 100 Ma, however, become younger toward the northeast (Fig. 4).

Ar-Ar and K-Ar ages of biotite and hornblende show a similar trend of ages decreasing to the northeast (Fig. 4). Using only the modern Ar-Ar ages, this trend is well defined (Fig. 5); from southwest to northeast along the transect, hornblende ages decrease ∼0.44 m.y./km, and biotite Ar-Ar ages decrease ∼0.5 m.y./km. Comparison of Ar-Ar ages to U-Pb ages younger than 100 Ma shows that at any location, the Ar-Ar ages are similar to the U-Pb zircon ages (Fig. 4). These ages gradually get younger to the northeast at about the same rate, although the U-Pb zircon trend is less well defined than the trend in Ar-Ar ages. Few Ar ages are older than 100 Ma (and all modern Ar-Ar ages are younger than ca. 120 Ma), so gaps of >10 m.y. are typical between Ar ages and the older (pre–114 Ma) group of zircon ages, despite the presence of these older zircon ages across the transect.


Dikes are very common on Knight Inlet, but unlike areas previously studied farther north in the batholith (e.g., Rusmore et al., 2010), they have a wide range of appearances, compositions, orientations, and presumably ages. Many are greenish gray to greenish black with abundant plagioclase phenocrysts, but colors range from light gray to dark greenish black. A low-grade chlorite alteration is prevalent. Dike margins vary from planar and chilled to irregular and poorly defined. We found no reliable way to distinguish any unique dike populations. For example, a stereonet plot of 78 dikes confirms that they dip steeply but have no preferred strike direction, and we were unable to discern correlations among location, width, composition, or orientation. The lack of recognizable populations prompted us to stop collecting dike data early in the study, but we continued to observe the composition of dikes along the length Knight Inlet. The most notable finding is the near absence of distinctive Miocene mafic dikes, which are abundant in the Coast Mountains Batholith farther north, where they are related to extension of the continental margin after initiation of the transform (Irving et al., 1992; Lindline et al., 2004; Rusmore et al., 2010; Souther and Jessop, 1991).

Brittle Faults

In total, 257 brittle faults were observed along the shoreline of Knight Inlet and nearby islands (Fig. 3). Most of these faults are small. Of these, 177 are isolated faults a few millimeters wide (classified as Grade V), and 65 more are small grade IV faults zones. Larger fault zones, those 1–10 m wide, are rare; only four grade II and nine grade III fault zones were found, and no large faults (grade I) were observed. Faults are not evenly distributed along the transect; they are most abundant in the southwestern end, sparse along the central part of the transect, and more common in the northeastern end (Fig. 3). Similarly, most of the largest faults occur in the west, with none in the central part and two in the northeast. Based on this distribution, we recognize three domains along the transect: a southwestern, a central, and a northeastern domain (Fig. 3). The scarcity of significant faults in the central domain leads us to consider it as a coherent structural block when interpreting other data sets.

Regardless of size or location, most faults are steeply dipping and strike either northeast or northwest (Figs. 6A and 6B). Contouring of the orientation data from all three domains shows two nearly vertical populations: the larger one strikes NE-SW, with an average orientation of 216, 86NW, and the smaller population of faults strikes NW-SE, ∼75° from the other population, with an average orientation of 293, 89NE. This pattern is unaffected by removing isolated grade V faults and forming six fault zones by grouping closely spaced grade IV and V faults (Fig. 6C). Of the largest faults (grade II and III), most strike northeast (Fig. 6D), reinforcing the view that NE-striking faults are the most significant faults along the transect. Northeast-striking faults occur in all domains, but NW-striking faults gradually become less prevalent to the northeast, with the northeastern domain containing only NE-striking faults.

Only sparse kinematic data are available for the faults; heavy wave erosion has removed many of the small structures, which could have been used to determine slip. Only 10 faults preserved striae and mineral fibers, which are scattered in orientation (Fig. 6E). Fortunately, the generally steep dip of the faults and the abundance of steeply dipping dikes allowed interpretation of slip direction on 20 other faults (Fig. 6F). Of the 30 faults with kinematic information, 25 are strike slip, divided nearly evenly between dextral and sinistral slip, and dextral and sinistral faults occur in both the NE- and NW-striking fault populations (Fig. 6F).

Dextral faults are more common in the southwestern domain, where eight of the 15 faults with known slip are dextral and two are sinistral. Conversely, four of the six faults in the northeastern domain are sinistral or sinistral oblique, and none has dextral slip. Faults in this domain strike northeast, perhaps indicating that sinistral motion is associated with the NE-striking population. The paucity of kinematic data weakens the significance of this pattern.

The presence of two fault populations at 70°–75° and a mix of sinistral and dextral faults could signal development of a conjugate fault system. This interpretation, however, is not consistent with the mixed sense of shear present within each fault population, nor the spatial variation in orientation. It is more likely that the brittle faults represent minor deformation of the upper plate of the subduction zone. The NW-striking dextral faults may accommodate minor trench-parallel motion, and the NE-striking faults may be small transfer faults, taking up variations in contraction of the upper plate. Regardless of origin, the structural data provide an important context for the paleomagnetic data in this study by showing that: (1) the central domain represents a structurally coherent block, ∼45 km wide; (2) the batholith has experienced less faulting overall than farther north; and (3) the widespread normal faulting and crustal extension affecting the batholith farther north is not present here.

Geobarometry Results

The calculated pressures and temperatures for samples are summarized in Table 1. Overall, the pressures are moderate, ranging from 1.8 kbar to 3.8 kbar, and temperatures are ∼650–750 °C. Pressures and temperatures from all but one sample indicate that the hornblende-plagioclase assemblages crystallized above the water-saturated solidus for tonalite magma (Anderson and Smith, 1995), consistent with the magmatic origin of the hornblende based on petrographic characteristics. The sample that plots below the water-saturated solidus (03MR-23) is one of the visibly freshest samples, and the reason for this anomalous result is not obvious.

Detailed petrographic study done after the microprobe work shows textural evidence of at least minor alteration along grain boundaries in all samples. This alteration is lightly developed in samples 01MR-76, 01MR-84, and 03MR-23, moderate in 01MR-37 and 01MR-83, and highest in 01MR-66P and 03MR-20. In the two most-altered samples, the presence of tiny actinolite crystals on hornblende grain boundaries and a patchy coloration within the grains cause us to regard the geobarometric results from these samples as unreliable. Excluding these two samples, the calculated pressure and temperatures do not appear to be correlated with the visible alteration, hornblende chemistry, or within-sample variation in chemistry and so likely represent crystallization pressures of the magmatic hornblende. These inferred crystallization pressures show no correlation to location on the transect line (Fig. 7).

Paleomagnetic Results

Remanence Carrier and Direction

As illustrated in Figure 8, thermal demagnetization to 580 °C almost completely removed the remanence of all samples. AF demagnetization was also effective at removing the remanence, with median destructive fields of a few tens of milliteslas being typical. These observations strongly suggest that low-Ti titanomagnetite carries all the remanence that we were able to characterize by the laboratory demagnetization experiments.

For most samples, thermal and AF demagnetization both revealed that the samples contained two remanence components (Fig. 8). The first, removed at low unblocking temperatures (Tub) or peak AF field strengths, was always a steeply positive inclination component, likely a Brunhes-age viscous remanent magnetization (VRM). We interpret the second component, well defined on orthogonal vector demagnetization diagrams (Fig. 8) by series of points trending toward the origin, as the characteristic remanent magnetization (ChRM) of the samples, likely a primary thermoremanent magnetization (TRM).

In almost all cases, we estimated the ChRM direction using the line fitting procedure of Kirschvink (1980) on either the thermal or AF demagnetization result, although we preferred components defined by thermal demagnetization. In cases where there was no thermal demagnetization data available, or where AF demagnetization better defined a ChRM component than thermal demagnetization, we used the ChRM direction defined by AF demagnetization in calculating site-mean directions. In a few cases, (e.g., where the entire ChRM resided in a very narrow Tub range, or where its coercivity exceeded the reach of the AF demagnetizer), we took the remanence remaining after removal of VRM or other secondary components as an estimate of the ChRM direction. We only included directions estimated this way in site means when consistent with well-defined demagnetization trends on other samples from the same site.

Classification of Site-Mean Directions

Samples from about half (20/41) the sites, labeled as type A in Table 2, exhibited straightforward demagnetization behavior and consistently oriented ChRM directions. At four sites, labeled type B in Table 2, the demagnetization behavior of samples appeared straightforward, but the ChRM directions were not consistent within the site. As a result, the site-mean directions for type B sites have large uncertainty, in some cases because they are based on a small subset of samples from the site. Samples from another nine sites, designated as type C in the table, exhibited complicated behavior during demagnetization, making it difficult to precisely define magnetization component directions. Using the terminology of Kirschvink (1980), the maximum angles of deviation (MAD) for line fits attempted on these experiments were excessively large. In spite of the uncertainty associated with each sample direction, the type C sites showed relatively good within-site consistency, meaning that the site-mean direction was associated with a small α95 value (see Table 2). Finally, there were six sites with samples that displayed complex demagnetization behavior and poorly defined site-mean directions (type D) and two sites where the sample demagnetization behavior was so poor that it was impossible to define a site-mean direction.

As discussed in more detail later herein, we base all geologic interpretations on the sites with best-defined site-mean directions, using an arbitrary value of 7.5° as the maximum for the α95 value associated with an acceptable site-mean direction. Of the 41 paleomagnetic sites, 20 met this criterion. All but one (1K701) were type A sites. With two exceptions (1 each at the 3K181 and 1K271), the MAD value for line fits to demagnetization trajectories (AF and thermal) for the 131 samples from these sites was less than 15°, and all but seven were less than 10°. The great majority of ChRM directions were defined very precisely.

As shown on Figure 9A, all but one of the 20 best sites in this study yielded north and steeply down site-mean directions, consistent with the normal polarity of the geomagnetic field between 83 Ma and 121 Ma (polarity time scale of Gee and Kent, 2007). The distribution of site-mean directions is roughly circular (or “Fisherian”); the data are not streaked, as would be expected if sites had rotated different amounts about similar axes. Site 1K401, at the NE extreme of the study area, is the only one magnetized south and up (D = 149.3, I = –73.3). All samples from this site carried nearly ideal single-component remanences (see Fig. 8) that were largely (90%–97%) removed by heating to 580 °C and had median destructive fields ranging from 30 to >70 mT. In contrast, the characteristic remanence of samples from site 3K211, located 0.8 km to the NW, was oriented almost exactly 180° away (D = 333.2, I = 74.5). There was evidence in these samples of a second, antiparallel component (i.e., in the direction of the 1K401 remanence) that unblocked at intermediate temperatures. In the example shown in Figure 8, this south and up component is evident over the course of four thermal demagnetization steps at temperatures between 450 °C and 540 °C.

Selection of Data for Paleolatitude Estimate

Figure 9 shows how the site-mean inclination varies with the uncertainty in the site-mean direction (i.e., the α95 of Fisher, 1953). It is apparent on the plot that there is much less variation in site-mean inclination among the 20 sites with the best-defined site-mean directions, i.e., those with α95 ≤ 7.5°. The pattern is even more distinct for the 13 of the 20 sites in the relatively unfaulted central structural domain of the transect; the angular dispersion (6.7°) of their site-mean directions is less than half that (15.3°) of the other 25 site means. Because they are from a relatively large and minimally deformed structural block, we derive all geological inferences in the discussion that follows from the 13 sites with α95 <7.5° in the central structural domain of the transect. None of the conclusions, however, is significantly affected by this choice. The paleomagnetic pole of the central subset (long = 350.2, lat = 81.6, α95 = 5.1) differs from the pole for all sites (long = 350.1, lat = 85.0, α95 = 4.7) by 3.4°, and the mean directions corresponding to these poles differ by 2.2°. Evidently, whatever is causing the within-site scatter does not also impart a significant bias to the site-mean directions.

Figure 10 shows how the paleomagnetic results vary in a direction approximately normal to the trend of the orogen. On the figure, the paleomagnetic sites have been projected to the transect line A–B (Fig. 2). If the transect line is centered on the sites, the average distance between the transect line and a site is 12.8 km. There are six paleomagnetic sites in the heavily faulted southwestern structural domain, 13 in the relatively unfaulted central structural domain, and one in the heavily faulted northeastern structural domain.

Compared to sites in the central structural domain, the southwestern sites exhibit more variation in paleomagnetic direction, likely the result of deformation associated with the faults. In the relatively unfaulted central structural domain, declinations are near 015°, except the NE end (km 50 through km 60), where four sites have declinations near 355°. The data allow the possibility that declination is nearly constant from km 20 to km 50, but a firmer conclusion is not possible because of the absence of acceptable sites between km 30 and 50. The paleomagnetic inclinations at all sites in the central structural domain are within 6° of 64°, with no significant trend from SW to NE.

The paleomagnetic pole and 95% confidence region corresponding to the 13 site-mean paleomagnetic results from the central structural domain of the Knight Inlet transect are plotted on Figure 11. The mean pole lies 3.3° from the spin axis, and the two are statistically distinct at the 95% confidence level. This evidence suggests that the characteristic remanence we have isolated from the 13 sites is a primary TRM rather than a secondary, Brunhes-age VRM.

Age of Magnetization

The plutons from which the 13 sites were obtained have not been dated, but compiled U-Pb zircon ages suggest they are mid-Jurassic to Cretaceous (Figs. 2 and 3). The age of the magnetization, however, is best estimated from the Ar-system hornblende ages, which have blocking temperatures close to that of the magnetization. Three K-Ar ages on hornblende are located on our transect and near the 13 sites used our analysis (Fig. 2). These ages are 108 ± 5 Ma, located near the middle of the central structural domain, and 89.5 ± 4.1 and 87 ± 3.9 Ma, near the northeasternmost paleomagnetic site, within the northern structural domain. Projecting Ar-Ar hornblende ages from ∼125 km NW along strike of the orogen onto our transect line (Figs. 2 and 5) gives similar, but slightly younger, ages: Hornblende cooling ages at the southwesternmost of the 13 sites would be ca. 100–105 Ma, and the northeasternmost of the 13 sites would be ca. 85 Ma. The general concordance between the two methods suggests that magnetization of the 13 sites occurred between ca. 110 and 85 Ma, gradually decreasing to the northeast. The magnetization was acquired when the North America pole was nearly stationary from 130 Ma to 60 Ma (e.g., Enkin, 2006; Kent and Irving, 2010); therefore, defining the age of magnetization more precisely than 110–85 Ma would not affect comparisons to the North America pole. The age of magnetization of these 13 sites is within the Cretaceous normal superchron, compatible with the normal magnetization of the samples. Interestingly, the only reverse polarity site we encountered is northeast of the central stable block. Its polarity is consistent with its inferred age (ca. 82 Ma as projected from Ar-Ar sites), which was just after the end of the Cretaceous normal superchron.

Paleolatitude: Tilt or Translation?

A comparison of the site-mean inclinations recorded at Knight Inlet sites to the North America pole from 125 to 85 Ma shows that, similar to most studies on the western fringe of North America, the inclinations are anomalously shallow. Figure 9 shows how the grand mean inclination of the 13 sites from the relatively unfaulted central structural domain compares to various expectations; these comparisons are summarized numerically in Table 2. The observed paleomagnetic inclination is between 13.4° and 15.9° shallow, depending on which North America pole it is compared to. The 95% confidence interval (shaded in the figure) on the mean inclination is ∼5° after correction by the method of Demarest (1983), and it is much smaller than the observed inclination anomaly.

Substantial NE-side-up tilting could have produced the anomalously shallow inclinations present in the Knight Inlet sites. If tilted as a single, ∼40-km-wide crustal block, the paleomagnetic results imply ∼17° tilting about a subhorizontal, SE-trending axis. (The minimum rotation is 17.2° counterclockwise about an axis with trend = 132.3° and plunge = 13.9°.) This axis is parallel to the regional structural grain of the orogen, a geologically reasonable orientation. The differential exhumation produced by tilting this block would be ∼12 km (or nearly 400 MPa [4 kbar] of pressure difference); the expected trend for this amount of exhumation is shown as a sloping line in Figure 7. Inferred pressures from this structural domain, however, differ by <1.5 kbar and show no significant spatial trend (Fig. 7). The available geobarometric data provide no support for the hypothesis that block tilting has significantly affected the paleomagnetic inclination of the central domain.

Tilting could be signaled by the Ar-system cooling ages (which are younger to the NE), if these ages were produced as NE-side-up tilting progressively exhumed and cooled the 40-km-wide block, as suggested for the batholith farther north near Prince Rupert (Butler et al., 2006). The decreasing ages along Knight Inlet, however, closely mimic the migration rate of the post–100 Ma arc: The U-Pb zircon ages show that the active magmatic front migrated to the northeast at ∼2 km/m.y., a rate indistinguishable from the 2.3 km/m.y. migration in the hornblende ages and 1.9 km/m.y. in biotite. The similarity in rates and ages in these systems suggests that the cooling ages reflect the evolving thermal structure of the arc, not tilting of a wide crustal block. Recent studies (Brownlee and Renne, 2010; Denyszyn et al., 2011) of the batholith in the Prince Rupert area suggest eastward-younging arc plutonism reset cooling ages within the batholith, reducing the evidence for tilting there.

Overall, we find no evidence of the differential exhumation that would result from block tilting of a 40-km-wide block. An alternative possibility is that tilting occurred on much smaller-scale blocks (separated by unobserved faults), as suggested for the Prince Rupert area by Butler et al. (2006). If so, the tilting would have to have been remarkably uniform in magnitude and direction to be undetected by the methods employed in this study. Differential tilts between kilometer-scale blocks were detected in the paleomagnetism of the Paleocene Quottoon intrusion (Butler et al., 2001a; Rusmore et al., 2010) near Prince Rupert, demonstrating that kilometer-scale tilting is likely to be evident in paleomagnetic results. Unlike the Prince Rupert area, our structural results suggest that Knight Inlet lacks the regionally developed crustal extension that could cause this style of tilting.

The shallow paleomagnetic inclinations can be alternatively interpreted as the result of significant northward translation relative to North America. To evaluate this shift, we calculated an expected inclination using the 125–85 Ma North America pole (191.2°E, 70.1°N) of Housen et al. (2003) and then adjusted it for transport relative to North America by amounts up to 3000 km. Use of an alternative reference pole (199.3°E, 75.2°N, the average of 125 Ma to 85 Ma poles for North America from Besse and Courtillot, 2002) changes these estimates by less than 1°. As can be seen on Figure 9, this analysis of the Knight Inlet paleomagnetic results suggests that this part of the Coast Mountains orogen has been transported poleward some 1700 km relative to North America; amounts between 1300 km and 2100 km are allowed by the data. This amount of translation implies that in the Late Cretaceous (ca. 90 Ma), the Knight Inlet study area would have been at the same latitude as central California. Given the apparent lack of tilting, we favor this interpretation of the paleomagnetic results. Comparison to other studies with similar results and some implications of this finding are discussed next.


The structural data show that the southern Coast Mountains Batholith at Knight Inlet, in contrast to the central Coast Mountains Batholith at the latitude of Prince Rupert, is an unextended Late Cretaceous magmatic arc. The batholith at Knight Inlet lacks the widespread normal faults and mafic dikes that signal Tertiary crustal extension and development of the transform margin adjacent to the central Coast Mountains Batholith (Dehler et al., 1997; Hollister et al., 2008; Rohr and Dietrich, 1991; Rohr and Currie, 1997; Rusmore et al., 2010). The absence of significant crustal extension on Knight Inlet, inboard of a convergent margin, strengthens the linkage of Tertiary extension with formation of the transform margin in the central Coast Mountains Batholith. Interpretations of our geologic and paleomagnetic results from Knight Inlet are aided by lack of this extensional overprint.

The established tectonic framework for southern British Columbia shows Wrangellia extending from Vancouver Island east to the core of the Coast Mountains, through our study area (e.g., Cowan et al., 1997; Gagnon et al., 2012; Gehrels et al., 2009; Monger et al., 1982; Wheeler et al., 1991). Major Late Cretaceous shear zones in the core of the Coast Mountains, such as the Harrison Lake shear zone (Brown and McClelland, 2000) and the eastern Waddington thrust belt and related structures (Journeay and Friedman, 1993; Rusmore and Woodsworth, 1994), are inferred to separate Wrangellia from Stikinia and related terranes of the Intermontane terrane. Two aspects of our study, the paleomagnetic results and compilation of U-Pb zircon ages, call into question this framework and imply that paleographic interpretations for southwestern British Columbia should be reconsidered.

Assignment of the southwest Coast Mountains to Wrangellia has been based primarily on proximity to the well-established terrane on Vancouver Island, the presence of greenstone pendants in the batholith (Roddick and Tipper, 1985; Roddick and Woodsworth, 1977, 2006), and an absence of mapped faults separating Wrangellia from the batholith (Nelson, 1979). The rarity of distinctive pendants in most of the southern batholith has made testing terrane affinities difficult, and few modern studies have attempted to investigate the nature of the eastern boundary of Wrangellia here.

Close examination of the magmatic patterns in the northern part of the Coast Mountains Batholith (Gehrels et al., 2009; Mahoney et al., 2009) offers a new framework in which to interpret the tectonic affinity of the southern batholith, and our paleomagnetic results can be compared to a substantial body of paleomagnetic studies from the Insular and Intermontane terranes. The central Coast Mountains Batholith consists of two Late Jurassic–Early Cretaceous arcs distinguished by their magmatic histories (Gehrels et al., 2009). The eastern arc was active from ca. 180 to 110 Ma (Gehrels et al., 2009; Mahoney et al., 2009), whereas the western arc is characterized by episodic magmatism with active periods at 177–162 Ma, 157–142 Ma, and 118–100 Ma (Gehrels et al., 2009). The magmatic gap between 142 Ma and 118 Ma is especially distinctive. Because the western arc intruded the Insular terrane (Alexander terrane and Wrangellia) and the eastern arc intruded the Intermontane terrane, these distinctive magmatic histories can be used as a guide to the terrane affinity of the magmatic arc and sparse pendants in the southwestern Coast Mountains. Projecting the mapped patterns of the two arcs along the batholith suggests that the western arc should underlie our study area on the western flank of the batholith (Fig. 1). The U-Pb ages compiled here, however, do not show the long magmatic gap (142–118 Ma) that characterizes the western arc. Of the 23 ages compiled, six fall into this gap, making them far more common than in the western arc, where, along 400 km of the batholith, only one age falls in this gap (Gehrels et al., 2009). If the framework established farther north does indeed represent two separate arcs that intruded different basement terranes, then plutons in the Knight Inlet area belong to the eastern arc rather than the western arc. Furthermore, they have intruded terranes of the Intermontane terrane, not Wrangellia. While surprising, this interpretation is consistent with our paleomagnetic results.

Most paleomagnetic studies of the Intermontane terrane imply ∼1600 km of post–mid-Cretaceous translation, as opposed to ∼2500 km commonly inferred for the Insular terrane (Bogue et al., 1995; Bogue and Grommé, 2004; Enkin, 2006; Enkin et al., 2006; Housen et al., 2003; Irving et al., 1985, 1996; Ward et al., 1997; Wynne et al., 1995). The 1700 km distance of translation suggested by our paleomagnetic results matches well with the lesser amount of post–mid-Cretaceous northward translation inferred for the Intermontane terrane. This translation is also similar to one interpretation of the nearest pluton studied paleomagnetically: the Porteau pluton near Vancouver, British Columbia (Fig. 1). Irving et al. (1995) documented a streaked pattern of site-mean remanence directions acquired as the pluton tilted during cooling. If the tilting was up to the west, then the paleomagnetism of the pluton implies 1600–1900 km of poleward transport relative to North America. (Much greater transport—3200 km—is implied if the tilting was down to the east.) Similar to the Knight Inlet plutons, the Porteau pluton lies west of mid-Cretaceous faults presumed to mark the western edge of the Intermontane terrane (Harrison Lake shear zone and southern Coast Mountains thrust belt; Brown et al., 2000; Journeay and Friedman, 1993). Another relevant result comes from rocks of the Mitchell Inlier in central Oregon. These rocks are separated from units to the east affiliated with the Intermontane terrane by the Salmon River and Western Idaho shear zones (Giorgis et al., 2008; McClelland and Oldow, 2007; Wyld et al., 2006). Paleomagnetic data from the Mitchell Inlier suggest 1200−1700 km of Late Cretaceous–Early Tertiary northward translation (Housen and Dorsey, 2005). Based on the paleomagnetism and lithological similarities, Housen and Dorsey (2005) correlated the Mitchell Inlier and the larger Blue Mountains terrane with the Intermontane terrane and suggested that the Blue Mountains terrane was emplaced in Late Cretaceous or Early Tertiary time. Together, these studies hint that parts of the Intermontane terrane lie farther west than previously recognized, perhaps because Late Cretaceous and Early Tertiary northward displacements have modified its western margin.

If the Knight Inlet plutons intruded part of the Intermontane terrane, what is the relation between this fragment of the Intermontane terrane and Wrangellia, exposed just to the west? Brittle faults are abundant west of the plutons we studied and may underlie Johnston Strait, but the absence of regionally recognized, through-going brittle faults makes it unlikely that tectonically significant displacement has taken place along this zone. Similarly, our work did not reveal any significant ductile shear zones that could have juxtaposed the terranes after ca. 80 Ma. The most straightforward interpretation of our results is that Wrangellia and this fragment of the Intermontane terrane were adjacent prior to intrusion of the plutons studied, that is, prior to ca. 100–120 Ma. This interpretation is consistent with several models positing pre–mid-Cretaceous amalgamation of the Insular and Intermontane terranes (e.g., McClelland et al., 1992; Monger et al., 1994; van der Heyden, 1989; Gehrels et al., 2009), but it is inconsistent with paleomagnetic results calling for Wrangellia (as part of the Insular terrane) to be ∼2500 km farther south in the mid-Cretaceous (Bogue et al., 1995; Bogue and Grommé, 2004; Enkin, 2006; Enkin et al., 2006; Housen et al., 2003; Irving et al., 1985, 1996; Ward et al., 1997; Wynne et al., 1995).

This discrepancy lies at the heart of the Baja British Columbia controversy, and it cannot be resolved on the basis of our findings. Two points, however, are relevant: (1) The mid-Cretaceous latitude of Wrangellia on Vancouver Island is not constrained by data from Vancouver Island, but by plutons of the Coast Mountains Batholith near Prince Rupert, where the results are controversial (Bogue et al., 1995; Bogue and Grommé, 2004; Butler et al., 2006; Brownlee and Renne, 2010); and (2) the paleolatitude of the Late Cretaceous Nanaimo Group that overlies Wrangellia on Vancouver Island requires ∼1600 km of northward translation after latest Cretaceous time (Kim and Kodama, 2004; Kodama and Ward, 2001). Collectively, these findings can be reconciled with a paleogeographic model that places a fragment of Wrangellia against the Intermontane terrane by ca. 100 Ma, ∼1600 km south of the present location. Translation of these terranes then took place after deposition of the Nanaimo Group, after ca. 75 Ma. This model does not require significant differential transport between the Intermontane terrane and southernmost Wrangellia after mid-Cretaceous time. This scenario, however, places a fragment of moderately translated (∼1700 km) crust outboard of rocks with well-documented post–100 Ma translations of ∼2300 km. A few hundred kilometers of dextral slip on Tertiary faults east of the batholith (e.g., Wyld et al., 2006) could have produced this arrangement by truncating the composite Intermontane terrane and translating a sliver of it to the northwest, outboard of the far-traveled, previously accreted classic Baja British Columbia block. Regardless of the terrane affiliations of the plutons in Knight Inlet, the similarly of their paleolatitude to that of the Nanaimo Group, Mitchell Inlier, and parts of the Intermontane terrane lends support to models based on variable amounts of translation of smaller crustal blocks and those calling for moderate northward translation of western British Columbia in latest Cretaceous to Early Tertiary time (Umhoefer, 2003; Umhoefer and Blakley, 2006; Wyld et al., 2006).


Two surprising results emerge from this study: (1) Compilation of the U-Pb ages suggests that the plutons in the Knight Inlet area belong to the Late Jurassic–Early Cretaceous eastern magmatic arc, rather than the western arc as would be expected from regional magmatic patterns, and (2) northward translation inferred from interpretation of paleomagnetic inclinations is less than predicted by the Baja British Columbia model, but it is similar to results from the Porteau pluton, the Mitchell Inlier in southeastern Oregon, and the Late Cretaceous Nanaimo Group on Vancouver Island. These findings are compatible with a paleogeographic model in which the Vancouver Island fragment of Wrangellia was juxtaposed against the Intermontane terrane prior to ca. 120–100 Ma and emplaced in southern British Columbia after deposition of the Nanaimo Group around 75 Ma. Paleomagnetically detectable differential motion between this block in the southern Coast Mountains Batholith and the Intermontane terrane is not required, but a few hundred kilometers of translation, as documented on faults between the terranes (summarized in Wyld et al., 2006), is needed to place the Knight Inlet and Nanaimo rocks outboard of sites with greater post–100 Ma northward translation. Overall, the results suggest that, at least in southern British Columbia, the Baja British Columbia block is more fragmented than previously recognized, with some fragments, including Vancouver Island and the adjacent western flank of the batholith, emplaced after ca. 75 Ma and requiring only modest northward translation.

This study also reinforces the link between late Eocene and younger extension in the central batholith and development of the transform margin. Extension along the transform margin is marked by widespread mafic dikes and normal faulting. The absence of this signature along Knight Inlet suggests that significant crustal extension has not occurred here, where a convergent plate boundary has persisted.

Funding was provided by the National Science Foundation (NSF) award EAR-001147, and the Occidental College Center for Undergraduate Research. Assistants Bill de Boer, Robert Bogue, Phoung Chau, Erik Day, Tina Dura, Ron Karpilo, Cesar Larios, and Sam McManus are thanked for their contributions in the field and laboratory. We are grateful to Don Wilson of Silverking Ventures for excellent field and logistical support and to Joe Kirschvink for use of the Caltech paleomagnetism laboratory. Regional mapping by J.A. Roddick of the Geological Survey of Canada developed the critical regional framework, without which our research would not have been possible. We appreciate the careful reviews by Bernie Housen, Ken Kodama, an anonymous reviewer, and Lithosphere Editor John Goodge.

1GSA Data Repository Item 2013320, table of compiled ages and table of hornblende chemistry, is available at www.geosociety.org/pubs/ft2013.htm, or on request from editing@geosociety.org, Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301-9140, USA.