On the palaeogeography of Baltica during the Palaeozoic: new palaeomagnetic data from the Scandinavian Caledonides

SUMMARY Based on new palaeomagnetic results from the North Norwegian Caledonides, we propose new apparent polar wander paths for Baltica during the Early-Mid Palaeozoic and discuss their palaeogeographic implications. In Cambrian and Early Ordovician times, Baltica occupied southerly latitudes of the order of 30" to SO", but was 'inverted' with respect to its present orientation. Consequently, the Russian Platform faced Avalonia and Gondwana, the latter continent occupying high southerly latitudes. Closure of the Tornquist Sea was then accompanied by continental scale, anticlockwise rotation of Baltica relative to Avalonia. This rotation probably occurred during mid-Ordovician times, although as yet, the timing of final suturing is poorly constrained by available palaeomagnetic data. At this time Laurentia occupied an equatorial position. Baltica then moved northwards in Late Ordovician and Silurian times, and subsequently collided obliquely with Laurentia to produce the Mid-Silurian to Early Devonian Scandian Orogeny . Oblique convergence set up sinistral orogen-parallel shear zones, on which major movements ceased by Late Silurian times. After amalgamation, Baltica and Laurentia occupied equatorial to tropical southerly latitudes. Reconstructions for the Siluro-Devonian boundary are now relatively straight-forward. Euramerica was assembled by that time, and occupied equatorial (N. Baltica) to high (c. 60") southerly latitudes (S. Laurentia) prior to northerly movement and the final assembly of Pangea.


INTRODUCTION
; Van der Voo 1988;Torsvik et al. 1990; Palaeozoic reconstructions for the circum-Atlantic region recount-the former movements of the bordering continents and the opening and closing of the intervening oceans. Faunal distributions, petrotectonic assemblages, sedimentary facies and palaeomagnetic data concur in suggesting the existence of the Iapetus Ocean (Harland & Gayer 1972;Roberts & Gale 1978), separating Laurentia* from Baltica and Avaloniat during Ordovician times. Data discrepancies exist, however, and have resulted in a number of contrasting palaeogeographic models (Cocks & Fortey 1982;Perroud, Van der VOO & Bonhommet 1984;McKerrow 1988;Miller Scotese & McKerrow 1990). The position of Laurentia is now reasonably well constrained by palaeomagnetic data. However, a sparsity of reliable Early Palaeozoic results from Baltica makes its Ordovician-Silurian position uncertain . Similarly, the movement history of Eastern Avalonia based on palaeomagnetic data is contentious (Briden & Mullan 1984;McCabe & Channel1 1990;Torsvik et al. 1990), and thus the significance of the intervening Tornquist Sea (Cocks & Fortey 1982) is unclear. In an attempt to address these problems, we have initiated palaeomagnetic studies of Palaeozoic rocks from N. Norway, S. Sweden and S. Britain * N. America, Greenland, W. Newfoundland and N. Britain. (Wales). This account concerns the palaeomagnetic

REGIONAL GEOLOGY A N D SAMPLING
Two major orogenic events have been postulated for the development of the northern Scandinavian Caledonides, i.e. a Late-Cambrian-Early-Ordovician event, known as the 'Finnmarkian' and a Silurian to Early Devonian, 'Scandian' event (Sturt, Miller & Fitch 1967;Pringle & Sturt 1969;Sturt, Pringle & Ramsay 1978;Ramsay et al. 1985;Roberts 1988a). Western Finnmark (Fig. 1) is characterized by a series of nappes and thrust sheets. The uppermost M a g e r~y Nappe was emplaced by Scandian thrusting, with the syn-orogenic but pre-thrusting Finnvik granite on Magereiy dated at 411 f 7 Ma (Rb/Sr whole rock; Andersen et al. 1982). The underlying Kalak Nappe Complex (KNC; Fig. 1) comprises a variety of rock units of different ages. A psammitic sequence (Klubben Group) predominates, passing up into pelites, limestones and finally turbidites (see reviews in Ramsay et al. 1985;Gayer et al. 1987). A Late Proterozoic to Cambrian age has been assumed for this succession, but some parts may be older than c. 800 Ma (Aitcheson, Daly & Cliff 1989). The rocks of the KNC have undergone polyphasal deformation and amphibolite-facies metamorphism. Garnet-biotite geothermometry indicates postmetamorphic peak temperatures in the range of 551 f 78" to 6 4 9 f 5 7 " C . Both Finnmarkian and Scandian tectonothermai events have been recorded (Dallmeyer 1988). The KNC is intruded by a complex series of assumed syn-tectonic intrusions, the Seiland Igneous Province (SIP), which have yielded Rb-Sr isotopic age-dates in the range of 540-490Ma (Sturt et al. 1967(Sturt et al. , 1978Ramsay et al. 1985). More recent U-Pb zircon dating has placed a younger age limit of 523 f 2 Ma on the SIP magmatism (Pedersen, Dunning & Robins 1989).
A total of 27 sites (310 drill-cores) from three major areas, SBrdy, aksnes and Altenes-Alta, were investigated (Figs 1-3 and Table 1; site numbers are not sequentially listed). A separate study of gabbros from M a g e r~y [area IV in Fig. l(a)] wili be reported elsewhere (Torsvik et al., in preparation).
From S~reiy, the Breivikbotn Gabbro and mafic dykes (Sites 20,21,22a,22b,(24)(25) were sampled in addition to the Hasvik Gabbro and cross-cutting dykes (sites 10a, b and 26). The Klubben Group psammites and cross-cutting mafic dykes were studied at sites 11, 12, 15 and 23 (Fig. 2). Folding and regional foliation development on S Q~Q~ is essentially related to two principal deformation periods, D, and D2 of Sturt & Ramsay (1965). These authors suggested that the Breivikbotn gabbro was emplaced during a late stage of D,. Aplitic granodiorite veins yield ages from 533 f 17 (Rb/Sr whole rock isochron) to 420-390 Ma (K/Ar biotite; Sturt et al. 1978). An attempt to date a nepheline syenite cutting the Breivikbotn gabbro by means of Ar39/Ar40 (nepheline) was undertaken by Dallmeyer (1988). Although a plateau age was not obtained, a total gas age of 425 f 8.4 Ma was reported.
The Breivikbotn Gabbro is cut by a younger alkaline complex from which Sturt et al. (1967) report a Rb/Sr whole-rock age of 483 f 27 Ma, considered to represent an emplacement age.  Gayer et al. 1985) of the northern Norwegian Caledonides. This account reports on study area 1-111 (shaded). Sampling area IV is detailed in Torsvik et af. (in preparation). BD = BBtsfjord Dykes, TK = Trollfjord-Komagelv Fault. Inset: N = Norway; S = Sweden; F = Finland. (b) Interpreted cross-section from Seruiy to Vargsund (from Gayer et al. 1987). Cross-section is shown in (a) as a stippled line. The Hasvik Gabbro was considered to have intruded during the peak of Finnmarkian metamorphism (Sturt & Ramsay 1965), and was thus considered of critical importance for dating the Finnmarkian metamorphism. Sturt et af. (1978) reported a Rb/Sr age of 521 f 27 Ma from an anatectic dyke from the Hasvik Gabbro. New Sm-Nd mineral isochron ages for the Hasvik Gabbro, however, suggest a pre-700 Ma emplacement age (Aitcheson et al. 1989; see later discussion).

THE MAGNETIC FABRICS-CONTACT RELATIONSHIPS ON S 0 R 0 Y
The anisotropy of magnetic susceptibility (AMS) was measured on a low-field induction bridge (KLY-2). AMS and anisotropy of remanence data are described in a separate paper (Torsvik & Walderhaug, in preparation), and only some pertinent general observations from Serey are outlined here.
The magnetic foliations (Kmax -Ki,J from the Klubben Group psammites are closely parallel to the observed megascopic foliation (&). On Serey, D1 and D, fold-axes are almost coaxial, and low-angle magnetic lineations (Kmm), N-S or NE-SW, are fold-axis parallel to both (Figs 2 and 4a-c). Sites 11-12,15 and 23 also include sbeared and partly amphibolitized dykes which intrude the Klubben Group psammites. The dyke margins from sites 12 and 15 parallel the psammitic foliation . Both dyke and psammite samples from site 12 show well-defined oblate ellipsoids (Fig. 4d), display a similar degree of anisotropy, and have comparable Kmin directions and NE-directed lineations (Figs 2 and 4b). The dyke of site 15 is amphibolitized and boudinaged. Both dyke and psammite samples show high anisotropies, 30-50 per cent of strongly developed prolate ellipsoids and a southerly directed K, , ( Fig. 4c and d). Sites 12 and 15 provide clear evidence for the dykes having a pre-or possible syntectonic origin.
Site 11 embraces two generations of basic dykes which cut obliquely across foliated and partly migmatized Klubben Group sediments (see Figs 2 and 4a). Psammites are characterized by a steeply dipping magnetic foliation which coincides with the megascopic foliation, and lineations plunge due south. A steeply inclined foliation is also recorded in the oldest dykes (denoted D1 in Figs 2 and 4a), but with a different orientation, and no lineation is developed. Conversely, the magnetic fabrics obtained from the D2 dyke, cutting and offsetting D1 dykes, show no correspondence with Klubben Group foliations or D1 dyke foliations. In addition, the D2 dykes show a low degree of anisotropy (<5 per cent), whereas D1 dykes show a wider range which converges toward psammite values of around 10 per cent of anisotropy (Fig. 4d). Krill & Zwaan (1987) have suggested that all dykes on %ray are pre-fold and hence pre-tectonic. Although magnetic competence contrasts might explain the 'strain' variations as exemplified from site 11, both field evidence and magnetic fabric data clearly indicate that some of the mafic dykes on S0r0y post-date the regional foliation. These relationships have recently been highlighted by  and Roberts (1988b).
The magnetic fabric relationship between dykes intruding the Breivikbotn and Hasvik gabbros is less definitive. Some sites are also directionally isotropic (sites 1 0 a d y k e ; 22a-gabbro, 22b--gabbro), i.e. individual samples show a 'random' directional pattern, thus complicating direct comparisons. Most commonly, however, dykes appearing fresh in hand-specimen show little if any microfabric relationship to the host gabbros (compare e.g. sites 21, 22a and 25; Fig. 2). Site-mean degree of anisotropy in gabbroic samples vaned from 6 to 25 per cent, commonly an order of magnitude higher than in dyke specimens (Fig. 5). Only sites 10a (gabbro) and 10b (gabbro & dyke) provided clear evidence for compatible magnetic fabrics (Fig. 2).

PALAEOMAGNETIC RESULTS
The natural remanent magnetization (NRM) was measured on a two-axis SQUID magnetometer and a Molspin magnetometer. The stability of NRM was tested by thermal demagnetization and to a lesser extent alternating field (AF) demagnetization. Characteristic remanence components were calculated by least-squares analysis.

Oksnes
All sites from Bksnes are dominated by magnetite, but minor amounts of accessory pyrrhotite (rare) were observed from thermomagnetic analysis.  low-coercivity component during A F demagnetization when co-existing with the B component, but unblocking temperature spectra analysis reveals a more complex pattern. At some sites (sites 42, 44 and 4 3 , component C occupies the higher unblocking spectra (Fig. 7a), whereas the opposite relationship is observed from sites 40 and 41. A more complicated directional relationship was observed in a few samples from site 45. Thermal demagnetization identifies the C component in the NRM-510 "C range, then the B component is randomized between 510" and 545"C, and finally the C component is once more identified above 545°C (Fig. 7b). Thus, B appears to intervene between or split the C component blocking spectra. The intervening components varied in direction, probably all B in origin, but proved difficult to identify due to overlapping blocking spectra.
Site 46 gave normal-polarity magnetizations due NE and upward-dipping inclinations, a magnetization component commonly observed from Scir~y (Figs 6a and b; Table 2). This component is designated component A.
The Hasvik and Breivikbotn Gabbros are dominated by components C and B. When co-existing at sample-level, component C always displays a higher unblocking temperature spectra than B (site 26). Magnetically stable dyke samples, all appearing fresh in hand-specimen (hereafter referred to as fresh dykes), always reveal component A. From site 22a, samples from dykes intruding the Breivikbotn Gabbro gave almost single component A remanences with very discrete unblocking spectra above 540°-555"C ( Fig. 7c). Contact gabbro samples are also dominated by the A component, clearly identified above 20 mT (Fig. 7d). Unfortunately, non-baked gabbroic samples at site 22 proved viscous. Both gabbroic and dyke samples from site 22a are dominated by variable amounts of magnetite and pyrrhotite.
Site 21 embraces a 20cm wide dyke intruding the Breivikbotn Gabbro (Fig. 8). Thermal and A F demagnetization of dyke specimens show exemplary A magnetizations (Figs. 8a,b and d). Baked gabbro samples are also dominated by the A component. On the other hand, non-baked gabbroic samples are influenced by component C, usually identified above 350"-450 "C after randomizing a somewhat more NE-directed component ( Fig. 8c and d). The NRM of dyke samples from site 21 is dominated by magnetite, whereas gabbroic samples show a mixture of pyrrhotite and magnetite. Gabbroic samples commonly showed a somewhat 'noisy' irregular directional pattern below 350°C, attendant on 70-90 per cent intensity reduction. This we relate to the influence of pyrrhotite. The contrasting magnetic fabrics between the dyke and the host gabbro are also portrayed in Fig. 8(e) (see also Fig. 2). The host gabbro is dominated by a shallow eastward-dipping foliation with lineations plunging due north, whereas the dyke is dominated by a more steeply inclined foliation dipping toward the NNW and no lineations are developed. This observation is consistent with the remanence data, and signifies that the dyke has intruded after foliation of the gabbro.

Altenes-Alta
Characteristic remanence components from the Altenes-Aka area ( Fig. 6c and Table 2) generally display components B and C (Fig. 9a). The palaeomagnetic data from site 70, however, are ambiguous. Most samples from this site show an anti-parallel and dual-polarity interplay of a southerly, downward-dipping, low-blocking component and a northerly, upward-pointing, high-blocking component (Fig. 9b). Due to this dual-polarity at sample level, and the declination discordance with the NW-directed A component, we have designated the northerly component as the normal polarity counterpart of component B. The majority of tested samples are dominated by magnetite, but site 71 shows evidence of haematite with blocking temperatures and directional stability up to 675 "C.

REMANENCE A N D TRM DEVIATION ANALYSIS
The comparison of area mean directions ( Fig. 6d; Tables 2 and 3) suggests that remanence acquisition post-dates local folding and/or rotational thrusting, and components B and C most likely relate to post-metamorphic cooling/uplift or phases of thermal overprinting. Thermo-remanent magnetization (TRM) acquisition in anisotropic rocks may involve considerable remanence deviation. Remanence deviation depends on grain-size, the degree of remanence anisotropy and more importantly on the angular relationship between the inducing ambient field and the pre-existing magnetic foliation. In order to explore the effect of TRM deviation, the anisotropy of remanence (AR) was tested from S~r 0 y and Bksnes (detailed in Torsvik & Walderhaug, in preparation) Anisotropy of low-field (4 mT) isothermal remanence (IRM) was examined since TRM and IRM anisotropy tensors are intrinsically identical (Stephenson, Sadikun & Potter 1986).
Following Stephenson et al. (1986) and Cogne (1988), the Cartesian coordinates Ji of TRM as a function of field Hi through the TRM anisotropy tensor Mij can be denoted as Ji = MijH, (i, j = 1,2,3). A site-mean TRM tensor Mi,. was determined from measurement of IRM induced in three orthogonal axes (x, y, z), typically based on 2-5 samples. The true TRM or ambient magnetic field direction from each site was subsequently calculated as Hi = Mij-'Jj.
The eigenvectors of the AR ellipsoids proved similar to AMS eigenvectors, but the A R eigenvalues were an order of magnitude higher. Percentage anisotropy of remanence varied between 5 and 123 per cent (Fig. 1Oc). Site-mean angular deviation, however, is typically less than lo" (Fig.  lob), but values as high as 42" were predicted from site 41. The effect on the overall statistics, however, is not significant (Table 3). This relates to the variable orientation of the magnetic foliation planes (Figs 2 and 3). Thus, TRM deviation analysis is of considerably more importance in deformed regions with a 'simple' and consistent deformation pattern. In such cases, TRM deviation would produce systematic errors in the average field-direction. In the more complex case presented here, mean directions are fairly correct, but increased between-site scatter is to expected.
The highest degree of anisotropy of remanence (and anisotropy of susceptibility; Fig. 5) is recorded from the Bksnes gabbros, and it is evident that component B and C from Bksnes show a directional overlap caused by the SE position of the C component from site 40 (Fig. 6b). This site has a high degree of anisotropy of remanence (123 per cent). The 'in situ' mean direction converges toward the other site-mean C components when accounting for TRM deviation, and the directional overlap with the mean B components is also reduced (cf . Figs 10a and 6b). Only component C displays a lower dispersion after correcting for TRM deviation (Table 3). Thus, in Fig. lO(a) component B is left 'in situ', whereas the C components have been corrected for TRM deviation.
The technique outlined above only applies to the acquisition of a TRM, and in the case of chemical remanent magnetization (CRM) or TCRM the technique may have no relevance or would result in an overcorrection of the remanence data. As demonstrated earlier, a somewhat complex and intriguing thermal blocking-spectra relationship between components B and C (Bksnes gabbros) suggests that one of these components has a CRM or TCRM origin. We propose that component C has a TRM origin as indicated by the reduced scatter after TRM deviation analysis, and further corroborated by laboratory TRM experiments (Torsvik & Walderhaug , in preparation).
Component B has a younger TCRM origin, possibly acquired by inversion of primary pyrrhotite to form magnetite, but also as a PTRM in magnetite. The former process would be expected to produce the somewhat complicated blocking-temperature spectra as were, for example, observed from the Bksnes gabbros.
Component A recovered from dykes was not notably changed after TRM deviation analysis. The percentage of remanence anisotropy is small, however (<5-8 per cent), and the predicted TRM deviations are consequently less than 3-4 degrees of arc. A primary TRM origin, however, is indicated by the positive contact test (Figs 7c,d and 8).

REMANENCE AGES A N D ASPECTS OF APPARENT POLAR WANDER
The nature and timing of the Finnmarkian Orogenic event is a matter of some dispute (see Townsend ). Krill, Rodgers & Sundvoll (1989) reject a dichotomous Finnmarkian-Scandian model, and suggest that these represent a single orogenic event, implying the stacking of the KNC to be entirely Scandian.
A considerable amount of isotopic age-data are available from igneous (SIP) and basement rocks within the KNC which have been compiled in Fig. 11 (updated from Pedersen et al. 1989). New Ar39/Ar40 plateau ages (Dallmeyer 1988), Sm/Nd, U/Pb zircon and a few Rb/Sr whole-rock data (including the Breivikbotn and Hasvik Gabbros) yield ages of around 530Ma. According to Pedersen et al. (1989) this age records the crystallization of late alkaline rocks emplaced at shallow crustal levels. Since these rocks cut deformed rocks of the SIP, 530Ma must sj up a \ NRM represent a younger age limit for early Caledonian metamorphism (Pedersen et al. 1989). As the alkaline rocks were subsequently deformed amd metamorphosed, there is evidence of both pre-and post-530 Ma tectonometamorphic events (see also Roberts 1988b). Based on a recent Sm-Nd mineral age of c. 700Ma (not included in Fig. 11)     Proterozoic orogenic event overprinted by the younger Caledonian deformations. Some Rb/Sr whole-rock ages and Ar39/Ar40 hornblende mineral ages cluster around 490 Ma; these were ascribed by Dallmeyer (1988) to Early Ordovician post-metamorphic uplift. Preliminary Rb-Sr dating of ductile mylonites from the base of KNC in one area has yielded an isochron age of 479 f 15 Ma (Roberts & Sundvoll 1989). Finally, Rb/Sr biotite, K/Ar biotite, Ar39/Ar40 nepheline-hornblende and Ar39/Ar40 muscovite ages cluster around 430-410 Ma. This youngest peak suggests that Scandian metamorphism in the Magerey Nappe (411 f 7 Ma) was broadly synchronous with an internal tectonothermal event within the KNC (Dallmeyer 1988), and that the final translation of the KNC was

Compilation of isotopic age data from the Seiland Igneous Complex and the Kalak Nappe. Updated from Pedersen et al. (1989) with new Ar3"/ArW reported by Dallmeyer (1988). Ages (small numbers) represent mean-values for the various age groups. Note that the vertical length of each 'age' bar embraces the maximum error of confidence within each 'age' group. Large numbers displayed at the top of each 'age' bar refer to age method listed in the right part of the diagram. Note that 'age' bars 1 and 2 and 4-6 are from the Seiland Igneous Complex, whereas 3 and 5 are from basement rocks located to the Kalak Nappe.
Scandian rather than Finnmarkian. This appears to be confirmed by Rb-Sr dating work by Roberts & Sundvoll (1989), late phase shear-banded mylonites yielding a Devonian isochron age.
Given the ages outlined above, the maximum unblocking temperatures of magnetite (TbTmax = 580 "C) and its timedependence, remanence acquisitions should range from Late Cambrian to Late Silurian or even Early Devonian times, unless younger low-temperature chemical alterations occurred below the isotopic blocking-temperature of, for example biotite, c. 200 "C. Seen in conjunction with Ar39/Ara mineral ages, remanence ages may straddle the 490-410Ma range (Fig. 11).
In view of the clear deficiency of reliable pre-Devonian Palaeozoic palaeomagnetic data from Scandinavia (Pesonen et af. 1989;Torsvik et al. 1990) the reported palaeomagnetic data are compared with British Apparent Polar Wander Paths (APWP). All British APWPs show a marked counterclockwise loop during the Palaeozoic (Fig. 12), and they are characterized by a Siluro-Devonian (mid-Silurian to mid-Devonian) 'corner' where they converge to produce a common path. This 'corner' is also recognized from data from North America and Greenland when rotated into a Bullard, Everett & Smith (1965) reconstruction (Fig. 13c). Moreover, Smethurst & Khramov (in preparation) have reported that palaeomagnetic data from the Lower ORS facies sediments of the western Ukraine, Russian Platform, also plot in this 'corner'.
The relative south-poles for component A and B plot within the Ordovician-Silurian section of the APWP for Southern Britain (Fig. 12). This apparent temporal match, however, should be treated with caution since the APWP for S. Britain is poorly time-constrained. McCabe & Channel1 (1990) have also challenged the reliability of the Ordovician poles which were used to constrain the path of Fig. 12  .
Component A approaches the 'corner' in the British APWPs. Contact tests clearly favour A as the youngest recorded remanence, and viewed in relation to a range of isotopic age-data averaging around 415-430 (Fig. ll), a mid-Silurian age is most likely. Whilst pole A (and B) apparently compares with Ordovician-Silurian poles from Southern Britain, a comparison with the only 'reliable' Silurian palaeomagnetic pole from Norway, the Wenlock-Ludlow Ringerike Sandstone pole (Douglas 1988), appears enigmatic (Fig. 13). If the Ringerike pole is older than the A pole, this implies that the Baltic APWP joins the Laurentian APWP by mid-Silurian time (Fig. 13), followed by convergence toward the 'corner' through pole A (see discussion).
Component C is clearly pre-Silurian or pre-Scandian in age, unless subsequent large ( > 90") tectonic rotations of  the KNC have occurred (Fig. 12). In a recent summary of palaeomagnetic data from Fennoscandia, Pesonen et al. (1989) considered the Lower Cambrian Nexg Sandstone pole (Prasad & Sharma 1978), the 565-603Ma (Rb/Sr biotite) Fen Complex pole (Poorter 1972) (Siedlecka & Siedlecki 1967;Roberts 1972;Harland & Gayer 1972;Johnson, Level1 & Siedlecka 1978;Rice et af. 1989), but the amount of dextral movement indicated by K j~d e et al. (1978) is questionable as the Bltsfjord pole was compared with Sveconorwegian and Torridonian (Scotland) poles, i.e. poles some 100-150 Ma older. A more realistic comparison (Pesonen et al. 1989) suggests more moderate differences, but detailed future work is required to test the postulates of movements on the Trollfjord Komagelv Fault. The C pole and the Bhtsfjord dyke pole partly overlap (Fig. 13a), but we do not consider the C pole to be Late Precambrian (640Ma) in age, given the isotopic constraints (Fig. 11). A Cambrian (c. 530ma) age for pole C combined with insignificant APW during the Late Precambrian-Cambrian presents a plausible solution, but we may equally well dispute the magnetic and perhaps also the isotopic age (Rb/Sr biotite) for the Bltsfjord pole. The Bltsfjord dykes are situated immediately beneath the overthrusted KNC, and the dykes have a magnetic structure closely similar to e.g. the 0ksnes gabbros. Indeed, fig. 5 of K j~d e (1980) shows a mixture of an easterly component (comparable with C) and a SW component (comparable with B). Thus, we would argue that the C component, and perhaps also the B component, is compatible with palaeomagnetic data from the Bitsfjord dykes.

DISCUSSION A N D PALAEOGEOGRAPHIC CONSIDERATIONS
Palaeomagnetic studies of igneous rocks within the Kalak Nappe Complex and of metagabbros and basalts within the tectonic windows of Karelian rocks reveal three major remanence components (termed A, B and C). The consistency of these components over a wide geographic (c. 3500 km2) and complex geological area (cf. the suggested cross-section in Fig. lb) suggests that they all post-date folding and local rotational thrusting. Some of the tested rocks possess high anisotropies of remanence, but TRM deviation analysis predicts only minor modification of the overall mean directions. The magnetic fabrics from mafic  (Table 4) is based on a smooth path fit (Jupp & Kent 1987; see also Torsvik er al. 1990) of poles listed in Kent & Van der Voo (1990) with the inclusion of a recently reported Middle-Ordovician pole for North America (Farr & Sprowl 1989). British Paths from Torsvik et al. (1990). The two Baltic paths (X and Y) are constructed by using all the poles shown in Fig. 13(a and c). Path X, however, ignores the Ringerike Sandstone pole (RS), whereas path Y is anchored by the RS pole (cf. text). Time-calibration of the two Baltic paths is approximate (Table 4) dykes from S0r0y provide evidence for pre-, syn-and post-tectonic origins. Only the post-tectonic dykes provided stable palaeomagnetic directions.
Component A is identified in fresh and undeformed dykes, and a positive contact test proves a primary TRM origin. Two gabbroic sites were also overprinted by this component. The normal polarity A component is the only magnetic signature as yet identified in the Honningsvag Gabbro (Torsvik et #I., in preparation; pole MG in Fig.   13c), which is situated within the overlying Scandian Magerdy Nappe (Fig. la). The studied area is remarkably unaffected by Late Palaeozoic-Mesozoic magnetic overprinting, which presents a major problem in Southern and Central Norway (Torsvik et al. 1986(Torsvik et al. , 1987(Torsvik et al. , 1988. According to Gayer et al. (1987), early Caledonian peak metamorphism within the KNC occurred prior to initial thrust stacking of the nappes. An early phase of SE-thrusting occurred under ductile conditions, whereas later thrust movements took place under brittle conditions (chlorite grade) accompanied by an anticlockwise change in transport direction toward the ESE . Component C may relate to uplift following the early ductile Downloaded from https://academic.oup.com/gji/article-abstract/103/1/261/567406 by guest on 29 July 2018 deformation, or may have been acquired as forelandpropagating thrust units ascended to shallower more brittle conditions. Any associated rotations, however, appear to pre-date acquisition of component C.
Aitcheson et af. (1989) have recently suggested that some of the SIP gabbros are older than 700Ma, and that a Late Proterozoic orogenic event ( > 800 Ma) is represented in parts of the KNC. These authors have suggested that Caledonian deformation of the KNC was restricted to movements along major nappe boundaries and internal shear zones after 520-530Ma, but this is difficult to reconcile with Ar39/Ar40 data. Seen in relation to Ar39/Ar40 hornblende ages (Fig. l l ) , with blocking temperatures of about 500"C, it is suggested that C was acquired during post-metamorphic cooling in Early Ordovician time, approximately 490 Ma, although a Cambrian age (c. 530 Ma) cannot be excluded. We propose that component C 'dates' the initial emplacement of the KNC onto the margin of continent Baltica.
A pole from the Swedish Orthoceras Limestones (Claesson 1978), previously described as 'anomalous' in the literature, may provide a lower age limit for the C pole. These Arenig to Llanvirn limestones record steep downward-dipping magnetizations due ESE. Although Claesson (1978) had some reservations concerning the reliability of her results, the site-mean directions form a very well-defined group (based on 73 sites), clearly removed from the present-day field (see also Piper 1987, p. 265), and the relative pole-position shows no similarity with poles from younger times (Fig. 13c). We have assigned a primary Arenig-Llanvirn age (> = 470 Ma) for this pole, although uncertainties remain concerning the relative importance of diagenetic/depositional controls on remanence aquisition. The relative pole position for component C, plotting between Late Precambrian-Cambrian poles and the Arenig-Llanvirn pole described above (Fig. 13a and c), and isotopic age data concur in suggesting an Late-Cambrian-Early-Ordovician (530-490 Ma) age for pole C.
Component B reflects a younger, possibly Late Ordovician or Silurian (?) thermo-chemical overprint, facilitated by inversion of primary pyrrhotite to form magnetite and as a PTRM in magnetite. As previously described, component A is related to a subordinate mafic dyke phase and localized thermal enhancement in the Silurian, probably in post-Wenlock times (see later). Whether this dyke activity relates to a phase of crustal extension following the Scandian collision and crustal thickening remains unclear. A regional gravity anomaly 30km to the NW of S6r6y can be interpreted in terms of a thinned crust (Olesen et al. 1990). Superimposed on this positive regional gravity anomaly there is a high-frequency negative anomaly interpreted as a sedimentary basin of Late Palaeozoic age.
Any relative rotations between the tested areas is below the resolution of the present palaeomagnetic data. It is important, however, to consider large-scale rotations of the entire Caledonian nappe pile and/or continental scale tilting of the margin of Baltica. The former suggestion implies that the tectonic windows, including the Komagfjord Antiformal Stack of Gayer et af. (1987) are allocthonous. As an example, the effect of anticlockwise rotations on a vertical axis is portrayed in Fig. 12, a sense of rotation which corresponds to the observed change of thrust transport direction from SE (ductile) to ESE (brittle). Such a rotation would bring the C pole closer to the A and B poles, and remove it from its apparent match with Late Precambrian-Cambrian poles from Scandinavia. If C was 'Scandian' in origin, more than 90" of anticlockwise rotation would be required. Aeromagnetic and gravity data however, indicate that the Komagfjord Antiformal Stack becomes less allochthnous towards the southwest, eventually becoming 'autochthonous' near Kvreenangen (Olesen et al. 1990). The matching of components B and C from the tectonic windows, the KNC and probably also the BBtsfjord area precludes significant relative rotations.
While unaffected by rotations on a vertical axis, rotations on inclined or horizontal axes will influence these palaeolatitude estimates. In terms of rotations on a horizontal axis, we can consider the isostatic response to loading/unloading associated with thrusting (Gretener 1981), continental scale tilting caused by subduction (Mitrovicia, Beaumont & Jarvis 1989) or post-Caledonian erosion and differential uplift. The envelope of the Kalak Thrust front strikes NE-SW, orthogonal to the early SE directed thrusting phase , and the rocks of the SIP define a gently NW-dipping disc-shaped body in the central part of the KNC (Olesen et af. 1990). Using an orogen-parallel axis (045") as the rotation axis we can tentatively model the affect of tilting. Since the declinations for components A and B are close to this tilt-axis, tilting would most seriously effect the C component (Fig. 12). Indeed, moderate SE tilts of the order of 15-20" (assuming initially steeper NW dips) can bring the C pole in correspondence with palaeolatitude estimates based on the B or A component.
Given the tectonic uncertainties outlined above, the precise pole positions of B and C should be treated with some caution. The youngest pole, however, pole A, shows a clear convergence toward the Siluro-Devonian 'corner' seen in all the British APWP's, and we assign the highest reliability to this pole. The convergence of pole C toward Late Precambrian-Cambrian poles from Scandinavia is also noteworthy.
The APWPs for Laurentia, based on data from Scotland and North AmericaIGreenland (Table 4) viewed in a Bullard et af. (1965) fit, and S. Britain (Avalonia) show convergence in Silurian-Early Devonian time. There is still, however, a minor discordance in the Ordovician-Silurian section of the Scottish and North American paths (Fig. 13c).  If correct, Scotland (based on data from the Northern and Grampian Highlands and the Midland Valley) was positioned to the south or possibly to the SW of the North American craton. The latter position requires considerable sinistral strike-slip faulting during Ordovician times, most of which must have taken place offshore Northern Scotland. To a first approximation, however, North American and Scottish data now form a reasonably consistent path, and in the reconstructions given below we have positioned North America and Scotland as a coherent unit in a Bullard et al. (1965) fit.

Cambro-Ordovician reconstruction
A reconstruction of Early Ordovician or possibly Late Cambrian times (490-530 Ma), depending on the age calibration of the C pole, is shown in Fig. 14. During Early Ordovician times Gondwana occupied high southerly latitudes with Avalonia probably plotting close to NW Africa (Van der Voo 1988). In contrast, Laurentia occupied equatorial latitudes with Scotland confined to latitudes around 15"-20°S (Fig. 14). Pole C implies that Baltica was 'inverted' with respect to its present orientation in pre-Scandian times. This is also indicated by Late Precambrian-Cambrian palaeomagnetic poles from Scandinavia (Fig. 13a). Recently, Pesonen et al. (1989) presented the alternative view, i.e. Baltica in its present orientation at 30"N during the Late Precambrian-Cambrian. We now prefer Baltica 'inverted' in southerly latitudes during Late Precambrian-Cambrian to possibly Early Ordovician times. This configuration has important implications for Early Palaeozoic reconstructions and implies that the Russian Platform was facing Avalonia/Gondwana during Cambrian and possibly Early Ordovician times. Avalonia has been placed marginal to Gondwana on faunal grounds (Cocks & Fortey 1982) given a lack of reliable Lower Ordovician palaeomagnetic data from Southern Britain. With Baltica 'inverted', the Lower Ordovician limestones in Southern Scandinavia would have faced the equator (Fig. 14) rather than the South Pole as portrayed in traditional reconstructions (e.g. Scotese & McKerrow 1990). This scenario implies that the closure of the Tornquist Sea (Cocks & Fortey 1982) was accompanied by continental scale anticlockwise rotation of Baltica. The anticlockwise rotation relative to Avalonia was reduced by Arenig-Llanvirn times (cf. pole SL in Fig. 13c), and probably ceased by Late Ordovician times. It i s possible that component B correlates with the amalgamation of Baltica and Avalonia (?), A southerly latitude for Baltica during Cambrian and Early Ordovician times resembles the palaeogeographic scenario of Scotese & McKerrow (1990), but with Baltica now 'inverted'. A northern hemisphere position during Cambrian times (Pesonen et al. 1989) requires 50"-60" southerly latitudinal drift during Ordovician times.

Wenlock-Ludlow reconstruction
Using our A and B poles, a Late Ordovician pole recently reported by Behm et al. (1989), and a preliminary pole from CAMBRO-ORDOVICIAN Figure 14. A Cambro-Ordovician to Lower Ordovician palaeoreconstruction showing Baltica in an 'inverted' position at intermediate southerly latitudes. Baltica is positioned according to pole C. Note that the distribution of Lower Ordovician limestones, brick ornament, faces the equator in this reconstruction rather than the South Pole. Laurentia is positioned according to a pole of 13"s and 29"E [European coordinates; Bullard er al. (1965) fit], representing an average of data from North America and Scotland (see Fig. 13). Gondwana is positioned using a Lower Ordovician pole of 28"N and 1"E (African coordinates) reported by Bachtadse, Van der Voo & Haelbich (1987). For simplicity, Avalonia is shown as an integral part of Gondwana and positioned NW of Africa (see Van der Voo 1988). Symbols are as follows: LA = Laurentia, B = Baltica, G = Gondwana, WA = Western Avalonia and EA = Eastern Avalonia.
Avalonia is latitudinally constrained using a pole of 2"N and 348"E [420 Ma pole for S. Britain listed in table 7 of Torsvik et al. (1990)l. Baltica progressed in a NW direction relative to Laurentia during the Silurian, and during the collision stage Baltica/Laurentia occupied equatorial to tropical southerly latitudes (Fig. 15). The oblique convergence of Baltica and Laurentia led to sinistral orogen-parallel shear-zones, on which major movements took place prior to Late Silurian times. The consistency of Late Silurian-Devonian poles from the various Scottish terranes precludes any subsequent large-scale movement between them (Trench et al. 1989;Torsvik et al. 1990;Trench & Haughton 1990).
The position of Avalonia with respect to Baltica (using Path Y) and Laurentia during the Silurian is enigmatic.
Following Cocks & Fortey (1982), the similarity of Baltic and S. Britain fauna suggests effective closure of the intervening Tornquist Sea by Late Ordovician times. In marked contrast, the reconstruction of Fig. 15 suggests a wide Tornquist Sea between Avalonia and Baltica in Silurian times (c. 2500 km). Given the geological constraints we find *his palaeogeography somewhat unreasonable. Path X for Baltica (Fig. 13c), however, ignoring the Ringerike Pole, presents a solution to this problem (cf. earlier discussion). If so, the Wenlock-Ludlow reconstruction ( Fig.  15) is identical to the Silurian-Lower Devonian boundary reconstruction given below.
the HonnigsvHg Gabbro (Torsvik et al., in preparation) it is feasible to connect the paths for Avalonia and Baltica (Path X in Fig. 13c; Table 4) by Late Ordovician time. This would imply effective closure of the Tornquist Sea by that time and a termination of the relative rotation between Baltica and Avalonia. Furthermore, given that some British 'corner' poles are as old as mid-Silurian (e.g. the mid-Silurian Salrock Formation pole of Smethurst & Briden 1988), one might argue that this 'corner' represents the continental docking of Laurentia and Baltica. Path X, however, ignores the Ringerike Sandstone pole of Douglass (1988), a pole which is constrained by fold-tests and a stratigraphically related magnetic polarity pattern. If one considers the Ringerike Sandstone pole to be reliable, the Baltic path (Path Y in Fig. 13c; Table 4) joins the Laurentian path by Mid-Late Silurian times. Alternatively, this convergence may mark the Scandian collision of Laurentia and Baltica, either as a collision of Scandinavia-Greenland or Scandinavia-Scotland (McKerrow 1988). Path Y for Baltica (Fig. 14) assigns the highest reliability to the Ringerike Sandstone pole, and the older section of the path is tentatively drawn between the B pole and the Late Ordovician pole reported by B~h m et nl. (1989), followed by convergence toward the C pole. Path Y takes pole A to be younger than the Ringerike Sandstone pole.
A Mid-Silurian scenario (c. 425 Ma) marking the convergence of Laurentia and Baltica is portrayed in Fig. 15. We use a common palaeomagnetic pole of 16"S, 337"E

Silurian-Devonian boundary reconstruction
Whereas the Mid-Silurian scenario is enigmatic with respect to Avalonia and the Tornquist Sea, the palaeogeographic setting for the Siluro-Devonian boundary is now relatively straightforward. The APWPs for Laurentia, Baltica and Avalonia all converge during Late-Silurian-Early-Devonian time (Fig. 13c), and Euramerica retains its most southerly position prior to northward drift and the final assembly of Pangea. The Baltic resemblance with the other paths is justified in view of the A pole reported here and new data from the Russian Platform (Smethurst & Khramov, in preparation). The Siluro-Devonian reconstruction (Fig. 15) is based on a common palaeomagnetic pole of 2"s and 319"E (European coordinates  (1990), but we position Laurentia and Baltica in slightly higher southerly latitudes.