Figure 4 shows a revised interpretation of the lava stratigraphy at 63ºN based on new constraints from ⁴⁰Ar-³⁹Ar dating, magnetostratigraphy, and geochemistry from Leg 163, in addition to the previously established lava stratigraphy for Leg 152 (summarized in fig. 5 of Larsen and Saunders, 1998). The oldest radiometrically dated volcanic rocks are the ~61-60 Ma Lower and Middle Series continental type lavas at Site 917 (Sinton and Duncan, 1998; Werner et al., 1998). The next-oldest dated rocks are the ~57-55 Ma lavas from Holes 989B and 990A. The presence of normally polarized lava units from Holes 989B and 990A, the only normally polarized lavas recovered during Legs 152 and 163, and indistinguishable radiometric ages within analytical error (57.1 ± 1.3 and 55.6 ± 0.6 Ma), strongly suggest correlation with a common normal polarity interval, C25n (56.4-55.9 Ma). In the results section we discussed the possible correlation of the normally polarized lavas at Site 990 with cryptochron C24r-11, but we note that this correlation is outside the 1 age range for the normally polarized lavas at Site 989. The correlation of the normally polarized lavas at Sites 989 and 990 to a common magnetochron is supported by their close compositional relation (both belong to the oceanic type lavas) (Larsen et al., Chap. 7, this volume; Saunders et al., Chap. 8, this volume) and their apparent identical magnetic properties (Duncan, Larsen, Allan, et al., 1996; Hooper et al., Chap. 10, this volume). Admittedly, however, we can not strictly rule out the possibility that the normally polarized lavas at Site 989 correlate with C26n or C25n, whereas the normally polarized lavas at Site 990 correlate with a cryptochron in C24r (e.g., C24r-11).

The Upper Series of Hole 917A unfortunately remains undated because of a lack of suitable material (Sinton and Duncan, 1998) and a lack of stratigraphic (Fig. 2) and major element compositional overlap with the presumed overlying lavas at Site 990A (Duncan, Larsen, Allan, et al., 1996). Compositionally, the Upper Series belongs to the transitional-to-oceanic type lavas (Fitton et al., 1998b; Saunders et al., 1998) consistent with eruption prior to the oceanic type lavas at Site 990A. Evidence from the geochemistry of the two lava successions, in particular the detailed trace element study of Fram et al. (1998), shows that one of two interfingering geochemical groups within the upper portion of the Upper Series is continuous with the lavas at Holes 915A and 918D, linked through temporal changes in mantle-melting conditions. This inference suggests the lavas of the Upper Series erupted shortly before those at Holes 915A/990A, most likely during C25r (57.6-56.4 Ma). We thus agree with Larsen et al. (1994), Sinton and Duncan (1998), and Larsen and Saunders (1998) that there seems to be a major hiatus in volcanism between the Middle and Upper Series of Hole 917A. This view is supported by the presence of a sedimentary unit at this level and a fundamental compositional change from continental type lavas in the Middle Series to transitional-to-oceanic type basalts in the Upper Series (Fitton et al., 1998b; Fram et al., 1998; Larsen, Saunders, Clift, et al., 1994; Saunders et al., 1998).

The age (57.1 ± 1.3 Ma) and composition (oceanic type) of the lavas recovered from Hole 989B was unexpected and poses new questions about the structure and evolution of the rifted margin. The current seismic interpretation suggests the lavas at Site 989 erupted prior to the 61-60 Ma Lower and Middle Series at Site 917 (Larsen and Duncan, 1996; Larsen and Saunders, 1998). If Hole 989B bottomed shortly above what is interpreted as non-volcanic basement (Fig. 2) (Larsen and Duncan, 1996), it implies a volcanic succession equivalent to Hole 917A is missing below Hole 989B. One possible explanation is that a steep paleo-relief (riftward-facing escarpment) at ~61-60 Ma prevented the emplacement of older continental type lavas at Site 989, located only ~5 km west of Site 917. Alternatively, following Larsen and Saunders (1998), uplift and erosion may have removed the older succession at Site 989 prior to the emplacement of the oceanic type lavas. This scenario would imply a feasible erosion rate of ~0.12 mm/yr between 61 and 56 Ma. If so, the basalts at Site 989 may represent the occasional off-axis lava flow that unconformably spilled over older, seaward-tilted and eroded lavas, consistent with the likely correlation with the normally polarized lavas at the top of Hole 990A.

Because the inferred non-volcanic basement below Hole 989B was not drilled, it leaves open the possibility that the lava succession beneath Hole 989B could be thicker than indicated by the current seismic interpretation (Larsen et al., 1998; Larsen and Duncan, 1996). It is thus possible that the volcanic piles beneath each of the Sites (Sites 989, 917, and 915/990), in fact, represent seawardly rotated fault blocks of an originally correlative volcanic succession analogous to the onland flood basalt province exposed to the north of the drilling transect (Pedersen et al., 1997). This view is consistent with the landward dip of the normal fault zone recognized between Sites 917 and 915/990 (Fig. 2) (Larsen and Duncan, 1996; Larsen et al., 1994). A similar landward downthrow of the lava pile beneath Hole 989B may have taken place at the landward-dipping zone marked "possible faulting" by Larsen and Duncan (1996). If so, the faulting and seaward rotation of the upper crustal fault blocks took place after the eruption of the lavas retrieved from Holes 989B and 990A. This could be analogous to the structure of the seawardly rotated flood basalt province exposed on land (Nielsen and Brooks, 1981; Pedersen et al., 1997). Moreover, this interpretation is entirely consistent with the crustal evolution model depicted in figure 6 of Larsen et al. (1998). Their model invokes early seaward rotation of continental crust during lithospheric thinning followed by flood volcanism and later seaward rotation and sagging during thermal relaxation of the margin. Following this tenet, the faulting of the volcanic succession at the featheredge of the transect at 63ºN would reflect late rotation and sagging. At present we cannot, however, distinguish between these alternative scenarios; a deepening of Hole 989B would undoubtedly throw more light on this subject.

The previously established chronology for the volcanism associated with the breakup of the Southeast Greenland volcanic rifted margin (summary in Larsen and Saunders, 1998) is significantly extended by the new argon ages from Leg 163. Our data, in particular, allow more firm constraints to be placed on the timing of oceanic type volcanism erupted through attenuated continental crust immediately prior to the onset of steady-state seafloor spreading and the timing of postbreakup volcanism. In Figure 5 we have summarized available results of ⁴⁰Ar-³⁹Ar dating and paleomagnetic data from Legs 152 and 163, which comprises a virtually complete time frame for ~12 m.y. of pre-, syn-, and postbreakup volcanism.

Prebreakup volcanism is apparently restricted to a short time window at ~61-60 Ma (Sinton and Duncan, 1998; Werner et al., 1998) and, most likely, is correlative with the earliest portion of C26r (Fig. 5) (see Larsen and Saunders, 1998, and summary above for further details). This volcanic sequence, as sampled in the Lower and Middle Series of Hole 917A, is at least 600 m thick, consists of crustally contaminated continental type basalts and dacites, and erupted through thick continental lithosphere (Fitton et al., 1998a; Fitton et al., 1998b; Larsen, Saunders, Clift, et al., 1994; Saunders et al., 1998).

Volcanism resumed at ~57 Ma. Following Larsen and Saunders (1998) and the discussion above, the Upper Series at Hole 919A most likely erupted first and was followed shortly after by the lavas at Sites 989 and 990. This period of volcanism, which can be correlated to C25 (57.6-55.9 Ma) and records the transition from transitional-to-oceanic type to oceanic type lava compositions, constrains the timing of complete plate separation between Greenland and northwest Europe to ~56 Ma. Some of the oceanic-type lavas from Holes 989B, 990A, and 915A, however, still show evidence for slight crustal contamination that most likely originate from fragments of continental crust, or the edge of the continents (Saunders et al., Chap. 8, this volume; Larsen et al., Chap. 7, this volume). Following continental separation, the main SDRS of the Southeast Greenland margin was developed until the formation of steady-state oceanic crust was under way in C24n.3n (53.3-52.9 Ma), the age of the most landward sea-floor spreading anomaly along Southeast Greenland (Larsen and Jakobsdóttir, 1988; Larsen and Saunders, 1998).

The dating of Leg 163 material has also documented the occurrence of postbreakup basaltic magmatism at the Southeast Greenland volcanic rifted margin. The one fresh lava flow drilled at the 66ºN transect (Hole 988A) is dated at 49.6 ± 0.2 Ma. This age is remarkable since it postdates plate separation by ~6 m.y. and marine sedimentation at Site 918 by ~2-3 m.y. (Ali and Vandamme, 1998; Jolley, 1998; Wei, 1998), despite the emplacement of this lava flow onto the extended edge of the Greenland craton. The distance between Hole 988A and seafloor spreading anomaly 21 (~49 Ma) is ~150 km (H.C. Larsen, pers. comm., 1997) and is a measure of the distance between Hole 988A and the spreading ridge at the time of flow emplacement. This sort of distance is occasionally seen for individual lava flows in continental flood basalt provinces (Hooper, 1997; Pedersen et al., 1997), but it would seem less likely in an oceanic setting. Below we discuss alternative causes of postbreakup volcanism at the rifted margin.

The time windows established for eruption of the SDRS of the Southeast Greenland volcanic rifted margin coincide with distinct periods of basaltic magmatism in the Tertiary Igneous Province exposed along the eastern coastline of Greenland between 63ºN and 70ºN. The onland rocks comprise a large continental flood basalt province, numerous layered gabbro intrusions, a large sill complex, and a coast-parallel basaltic dike complex (Brooks and Nielsen, 1982; Myers, 1980; Pedersen et al., 1997). Recent ⁴⁰Ar-³⁹Ar geochronology has established three distinct magmatic periods for these rocks (Fig. 5) (Hansen et al., 1995; Hirschmann et al., 1997; Storey et al., 1996; Tegner et al., 1998a). Prebreakup magmatism of the Tertiary Igneous Province includes a basaltic dike exposed on land immediately landward of the drilling transect at 63ºN dated at 60.7 ± 0.4 Ma (M. Storey/Danish Lithosphere Centre, unpubl. data), and crustally contaminated lavas of the lower lava series erupted in the Kangerlussuaq area (70ºN) at ~60-59 Ma (Storey et al., 1996). Synbreakup lavas, sills, and some gabbro intrusions are the dominant onshore Tertiary rock types and formed between ~57 and ~54 Ma (Hansen et al., 1995; Hirschmann et al., 1997; Storey et al., 1996; Tegner et al., 1998a). Postbreakup tholeiitic magmatism, forming several layered gabbro intrusions and rare lavas, apparently defines a distinct time window at ~50-47 Ma (Tegner et al., 1998a).

Many researchers acknowledge a causal link between the mantle plume residing beneath Southeast Iceland today and early Tertiary volcanism along the borderlands of the northeast Atlantic. The aseismic Greenland-Iceland-Faeroes ridge, for example, is the record of persistent anomalous volcanic productivity of the plume axis (hot spot track) through the Tertiary (Larsen and Jakobsdóttir, 1988). The details of the timing, location, structure, and composition of the plume in the early Tertiary, however, remain controversial. Some researchers, for example, have assumed the axis of the ancestral Iceland plume was centered under the incipient northeast Atlantic rift close to the Kangerlussuaq area (Brooks, 1973; White and McKenzie, 1989), whereas others suggest it was centered under central Greenland at the time of breakup 57-54 m.y. ago and subsequently crossed the East Greenland rifted margin in consequence of north-westward continental drift (Fig. 6) (Duncan and Richards, 1991; Lawver and Müller, 1994; Bernstein et al., 1998; Tegner et al., 1998a).

The evidence for distinct periods of pre- and postbreakup magmatism at the Southeast Greenland volcanic rifted margin, in particular the abundant postbreakup tholeiitic magmatism at ~50-47 Ma on land and at the 66ºN drilling transect (this study; Tegner et al., 1998a), supports the crossing of a focused plume tail ~5-8 m.y. after continental breakup. The apparent absence of ~50-47 Ma tholeiitic magmatism in the drilling transect at 63ºN may indicate that this later magmatic event is restricted to a ~400- to 500-km-wide geographic zone (~66º-70ºN) that coincides with the Greenland-Iceland-Faeroes ridge (Fig. 1) (Larsen and Jakobsdóttir, 1988). This, together with compositional evidence (e.g., elevated ³He/⁴He) from the onshore ~50-47 Ma layered gabbros (Bernstein et al., 1998), is consistent with melting caused by focused mantle upwelling in a deep-seated plume. We thus conclude, following Tegner et al. (1998) and Larsen and Saunders (1998), that the seemingly episodic volcanic history for the East Greenland volcanic rifted margin (onshore and offshore) may reflect brief and distinct mantle-melting events triggered by plume impact (~62-60 Ma) under central Greenland (Larsen and Saunders, 1998; Saunders et al., 1997; Sinton and Duncan, 1998; Storey et al., 1998), continental breakup (~57-54 Ma), and later passage of the plume axis beneath the East Greenland volcanic rifted margin (~50-47 Ma). This scenario is consistent with geochemical evidence for declining mantle temperatures during eruption at the onshore (57-54 Ma) flood basalt succession in east Greenland (Tegner et al., 1998b).