Abstract

The North Atlantic igneous province offers an unrivalled opportunity to study mantle sources contributing to flood basalt magmatism, and melting dynamics associated with active and passive upwelling of hot mantle beneath the lithosphere. In this study, Palaeogene basalts sampled at localities across the British Isles (from the Hebrides in the north to Lundy Island in the south) are shown to have concentrations of Nb, Zr and Y consistent with derivation from two mantle sources: ‘Icelandic’ (plume) mantle and hot N-MORB-like mantle forming an outer envelope to the plume. These sources were sampled over the period 61–58 Ma (chrons 26R–26N). Values of ΔNb—an expression of the deficiency or excess of Nb relative to the lower bound of the data array for Icelandic basalt in Nb–Zr–Y space—indicate that, with time, the proportion of ‘Icelandic’ material entering the melting regime below the British Isles (up to 1300 km from the plume axis) increased and then decreased relative to the contribution from the N-MORB source. Within the British Isles, subsidence data and basalt compositions suggest that melting to generate the parent magmas of the bulk of Palaeocene basalts occurred beneath intact lithosphere. Melting began at depths well in excess of 100 km, made possible by the high temperature (∼1550°C) of the ancestral Iceland plume. In the final stages of magmatism, depleted melt fractions were generated beneath the Rockall Trough and other basins to the NW of the British Isles, at depths as shallow as 55 km. These melt fractions were extracted rapidly from the mantle, without undergoing significant mixing with melt generated deeper in the melting column. The result is a distinctive magma type (Preshal More or Central Mull Tholeiite) not observed in the more southerly parts of the British Palaeogene igneous province.

INTRODUCTION

Flood basalt provinces are formed by eruption at the Earth’s surface of prodigious quantities (millions of cubic kilometres) of basaltic magma. During or subsequent to these eruptions, equally large volumes of igneous rock are added to the crust by intrusion. The most widely accepted explanation for such phenomena is melting of hot mantle welling up beneath the lithosphere (e.g. White & McKenzie, 1989). Melting may occur by both passive and buoyant upwelling and is enhanced by temperatures in the plume that are up to 300°C higher than the temperature of ambient asthenosphere (White & McKenzie, 1989; Campbell & Griffiths, 1990). Melt volumes are greatest where near-instantaneous rifting of the lithosphere accompanies volcanism (Pedersen & Ro, 1992).

To fully understand the melting processes that occur during flood basalt magmatism, it is necessary to place constraints on the timing, magnitude and duration of rifting, and key aspects of melt generation such as duration of volcanism, melt volume and melt composition. Unfortunately, few flood basalt provinces are sufficiently well known that all of these aspects are adequately constrained. Commonly, a significant part of the province is submerged below sea-level, making study difficult. Even when exposure is good, evidence relating magmatism to rifting may be obscured by lavas erupted after the initial burst of volcanic activity (for example, in East Greenland and in the Deccan and Paraná igneous provinces), by weathering, or by erosion.

The North Atlantic igneous province (NAIP, Fig. 1) is perhaps the most promising place in which to understand flood basalt melting processes. After a century of field mapping, complemented in recent years by the collection of extensive seismic and borehole data, it has become one of the world’s best-known flood basalt provinces. Recent work in the NAIP has focused on three main aspects: reconstructing the early Tertiary subsidence history of the East Greenland and NW European continental shelves, elucidating the geochemistry of basaltic rocks sampled in the onshore and offshore basins, and identifying mantle sources for the lavas. In this study, we explore mantle melting in the eastern NAIP, with particular emphasis on evidence provided by the Palaeogene basaltic igneous rocks of the British Isles. New and published geochemical data are used to address some outstanding questions; namely, the nature of mantle sources contributing to the volcanism, petrogenetic relationships between successive basaltic magma types, and variations in the mean depth of mantle melting (and melt segregation) beneath the European continental shelf.

Fig. 1.

Map of the North Atlantic igneous province, showing bathymetric features (500 and 2000 m contours), the location of the ancestral Iceland plume at 60 Ma (White & McKenzie, 1989) and at the present day, the distribution of Palaeogene igneous rocks and localities mentioned in the main text [modified from Eldholm & Grue (1994) and Saunders et al. (1997)]. We use White & McKenzie’s (1989) estimate of the position of the ancestral Iceland plume axis because of the close match between this location and the thickest sequences of Palaeogene lavas in East Greenland. Offshore flood basalts are more extensive than the seaward-dipping reflector sequence shown, but are omitted for clarity. Inset shows location of Fig. 2. FSB, Faeroes–Shetland Basin; IoM, Isle of Man.

Fig. 1.

Map of the North Atlantic igneous province, showing bathymetric features (500 and 2000 m contours), the location of the ancestral Iceland plume at 60 Ma (White & McKenzie, 1989) and at the present day, the distribution of Palaeogene igneous rocks and localities mentioned in the main text [modified from Eldholm & Grue (1994) and Saunders et al. (1997)]. We use White & McKenzie’s (1989) estimate of the position of the ancestral Iceland plume axis because of the close match between this location and the thickest sequences of Palaeogene lavas in East Greenland. Offshore flood basalts are more extensive than the seaward-dipping reflector sequence shown, but are omitted for clarity. Inset shows location of Fig. 2. FSB, Faeroes–Shetland Basin; IoM, Isle of Man.

GEOLOGICAL BACKGROUND

The Palaeogene igneous rocks of the British Isles form part of a flood basalt province [(6–10) × 106 km3; Eldholm & Grue, 1994; Saunders et al., 1997] emplaced along the continental margins of Greenland and NW Europe before, and during, sea-floor spreading to form the eastern North Atlantic Ocean (Fig. 1). In the British Isles, Tertiary volcanism commenced at ∼61 Ma (chron 26R; timescale of Berggren et al., 1995) with eruption of basaltic lavas and tuffs in the Sea of the Hebrides basin (Fig. 2; Chambers et al., 1999) and in Antrim, Northern Ireland. Basaltic lava eruption is estimated to have occurred over a period of 1·6 ± 0·2 my on Skye and ∼3 my on Mull (Hamilton et al., 1998; Chambers & Fitton, 2000). Concomitant with this volcanism, intrusive igneous complexes were emplaced in Antrim and on Lundy, Arran, Ardnamurchan, Skye and Mull (see, e.g. Kerr et al., 1999). Each of these igneous complexes has an attendant basic dyke swarm, which extends away from the complex for several tens to several hundreds of kilometres (Speight et al., 1982). The Antrim dyke swarm, for example, can be traced beneath the Irish Sea and across the Isle of Man into North Wales and central England (Kirton & Donato, 1985; Ford, 1993; Thompson & Winchester, 1995).

Fig. 2.

Map of western Scotland, showing main Palaeogene lava fields and central igneous complexes, sedimentary basins and extensional faults (ticks indicate downthrow side) and sample areas for Palaeogene basaltic dykes. Key: 1, North Mull; 2, Canna–Sanday; 3, North Uist; 4, North and South Harris; 5, Scalpay.

Fig. 2.

Map of western Scotland, showing main Palaeogene lava fields and central igneous complexes, sedimentary basins and extensional faults (ticks indicate downthrow side) and sample areas for Palaeogene basaltic dykes. Key: 1, North Mull; 2, Canna–Sanday; 3, North Uist; 4, North and South Harris; 5, Scalpay.

The bulk of Palaeogene basic volcanism in the British Isles was confined to the period 61–59 Ma (Chambers et al., 1999; Chambers & Fitton, 2000), whereas granite and late-stage dyke emplacement associated with the development of central complexes continued until ∼58 Ma [chron 26N; for a recent summary, see Saunders et al. (1997, fig. 1)]. A small number of Ar–Ar and fission-track ages (e.g. Mussett et al., 1988; Lewis et al., 1992), supported by an extensive palaeomagnetic dataset (e.g. Dagley & Mussett, 1981), suggest that the British Tertiary dyke swarms also were emplaced in the period 61–58 Ma. Only 18% of the ∼1240 Palaeogene dykes investigated thus far show normal magnetic polarity, consistent with the intrusion of the bulk of the British dykes during chron 26R and the remainder during chron 26N (note that chron 26N may include cryptochrons; Chambers et al., 1999).

The span of igneous activity observed in the British Isles is similar to that in West and SE Greenland, but volcanism in NE Greenland, the Faeroes, and on the Rockall and Vøring plateaux is generally younger (59–52 Ma; Saunders et al., 1997). This Palaeogene magmatism is generally attributed to the arrival beneath Greenland of the ancestral Iceland plume (e.g. Vink, 1984; White & McKenzie, 1989).

EARLY TERTIARY TECTONIC HISTORY OF THE NE ATLANTIC REGION

Before final opening of the eastern North Atlantic at 56–53·5 Ma (chrons 25N–24R; Vogt & Avery, 1974), the entire region was subjected to rapid surface uplift with a magnitude of up to 500 m (e.g. Nadin et al., 1995; Clift et al., 1998; Dam et al., 1998). This uplift commenced at ∼63 Ma, i.e. slightly before igneous activity commenced in Baffin Island, West and SE Greenland and the British Isles (Fig. 1). The size of the area affected—some 2000 km across on pre-rift plate reconstructions—and decay of the uplift after volcanism and final continental break-up are consistent with dynamic support provided by arrival and dissipation of a mantle plume head beneath the North Atlantic region (White & McKenzie, 1989).

Following uplift, the European and East Greenland shelves did not subside in accordance with the predictions of theoretical models such as that of Jarvis & McKenzie (1980) (see, e.g. Hall & White, 1994; Clift et al., 1998). The ‘missing’ subsidence in the shelf areas, notably the Blosseville Kyst of East Greenland and offshore UK, is consistent with permanent uplift of the crust as a result of igneous underplating (Brodie & White, 1995). In the British Isles, permanent uplift appears to increase towards the NW, in agreement with the distribution of Palaeogene lavas and intrusive centres (Fig. 2; Brodie & White, 1995). Brodie & White showed that the addition of 4–5 km of basalt to the lower crust of NW Britain would generate ∼600 m of permanent isostatic uplift, which in turn could result in ∼2·5 km of denudation (the amount suggested by fission-track data from the Hebrides and northern Highlands of Scotland; Lewis et al., 1992; Clift et al., 1998).

While the region occupied by the British Isles was undergoing uplift and exhumation, the northern Rockall Trough and the Faeroes–Shetland basin (Fig. 1) experienced a period of rapid subsidence and deposition of thick sedimentary sequences. In these basins, backstripping of seismic data and subsidence analyses suggest late Cretaceous to Danian (early Palaeocene) β values of ≥1·2 and ≤1·3, respectively (Brodie, 1995). Closer to the British Isles, in the Sea of the Hebrides basin (a Mesozoic structure lying on Archaean–Proterozoic crust; Fig. 2), Cretaceous and Tertiary sediments are thin or absent. This could imply no subsidence (no stretching) or else deposition of sediments followed by erosion. The former explanation is favoured by seismic and field evidence: fault systems that offset Triassic–Jurassic strata in the basin by up to 1 km appear to have had little or no effect on the thin sequence of Upper Cretaceous sediments and overlying Palaeocene lavas (Morton, 1987). Evidence for lithospheric extension after the cessation of volcanism is largely obscured as a result of exhumation of the Sea of the Hebrides basin. However, our field studies indicate that the lava pile in several parts of the basin (e.g. Mull, Skye) was subjected to minor Eocene–Oligocene(?) subsidence and faulting (throws of <100 m). This amounts to little more than a ‘creaking’ of the old Mesozoic faults, most probably in response to flexural loading of the lithosphere.

In summary, the available data suggest that, unlike the northern Rockall Trough and Faeroes–Shetland basin, the Sea of the Hebrides basin was not stretched immediately before, or during, the Palaeogene. This is an important observation, in so far as it implies that the lithosphere in this area was largely intact at the onset and close of Palaeogene magmatism.

GEOCHEMISTRY OF THE EARLY TERTIARY BASALTS

Previous work

Basalts preserved on the Atlantic margins are well exposed, generally fresh and hence amenable to geochemical analysis. The composition of Palaeogene basalt emplaced in the onshore UK region and, to a lesser extent, offshore UK, is therefore reasonably well known as a result of geochemical and Pb–Nd–Sr isotopic studies (e.g. Carter et al., 1979; Dickin, 1981; Thompson & Morrison, 1988; Thorpe & Tindle, 1992; Wallace et al., 1994; Kerr, 1995a, 1995b; Barrat & Nesbitt, 1996). In addition, comprehensive mineralogical and petrographic data for Palaeocene basalt samples from Mull have been presented by Kerr (1998). After correcting for the many and various complications introduced by alteration, fractionation and crustal contamination, it is apparent that three Palaeogene magma types occur in the Hebrides and Antrim (Mattey et al., 1977; Lyle, 1980; Kerr, 1995a, 1995b). These magma types appear to be successive on Mull and Skye [note that the Skye Fairy Bridge ‘lava flow’ identified by Mattey et al. (1977) is actually a sill intruding rocks of the Skye Main Lava Series; thus, the Skye lava pile shows a succession equivalent to that seen on Mull]. The Antrim magma types, although broadly similar to those in the Hebrides, are not successive (e.g. Lyle, 1980; Barrat & Nesbitt, 1996). A fourth magma type is preserved as dykes and plugs on Skye, Mull, Morvern and Arran (Thompson & Morrison, 1988; Kerr et al., 1999; G. F. Marriner & M. J. Norry, unpublished data, 1995). These dykes and plugs are volumetrically insignificant. On Mull they appear to have been intruded at ∼58 Ma (Chambers et al., 1999).

Distinguishing geochemical features of the four British Palaeogene magma types, here termed M1–M4, are shown in Table 1. The element ratios Y/Zr, Ti/Zr and Ce/Y were chosen because they are not affected significantly by crustal contamination (although high Ce/Y coupled with high SiO2 may indicate addition of a silica-rich contaminant) and low-pressure fractional crystallization. Rather, variations in these ratios mainly reflect different degrees and depths of partial melting, and/or differences in source composition.

Table 1:

Key characteristics of Palaeogene magma types found in the British Isles

Magma Local name (Mull, Skye) Classification Thickness of Y/Zr range (mean) Ti/Zr range (mean) Ce/Y range (mean) 
type 
 

 
lavas, Mull (m)
 

 

 

 

 

 

 
M1 Mull Plateau Group, Skye Main Lava Series Transitional (tholeiitic–alkalic) 730 0·1–0·3 0·2 43–93  69 0·4–4·8 1·5 
M2 Coire Gorm, Fairy Bridge Transitional (tholeiitic–alkalic) 250 0·2–0·4 0·3 84–143 100 0·3–1·2 0·7 
M3 Central Mull Tholeiites, Preshal More Tholeiitic 900 0·2–1·0 0·5 82–139 113 0·01–1·6 0·2 
M4 Late Mull type, Beinn Dearg More Transitional (tholeiitic–mildly alkalic) No lavas present* 0·3–0·4 0·3 80–119 100 0·1–0·4 0·2 
Magma Local name (Mull, Skye) Classification Thickness of Y/Zr range (mean) Ti/Zr range (mean) Ce/Y range (mean) 
type 
 

 
lavas, Mull (m)
 

 

 

 

 

 

 
M1 Mull Plateau Group, Skye Main Lava Series Transitional (tholeiitic–alkalic) 730 0·1–0·3 0·2 43–93  69 0·4–4·8 1·5 
M2 Coire Gorm, Fairy Bridge Transitional (tholeiitic–alkalic) 250 0·2–0·4 0·3 84–143 100 0·3–1·2 0·7 
M3 Central Mull Tholeiites, Preshal More Tholeiitic 900 0·2–1·0 0·5 82–139 113 0·01–1·6 0·2 
M4 Late Mull type, Beinn Dearg More Transitional (tholeiitic–mildly alkalic) No lavas present* 0·3–0·4 0·3 80–119 100 0·1–0·4 0·2 

Values for Y/Zr, Ti/Zr and Ce/Y were calculated using Edinburgh XRF data only [samples from this study, Kent (1995) and Kerr et al. (1999, table 7)].

*M4 basalts occur only as dykes and plugs.

New geochemical results

Consideration of available geochemical datasets shows that there are few analyses of Palaeogene basalts from the British Isles that include precise measurements of the concentrations of Sc, V and the middle to heavy rare-earth elements (MREE to HREE), combined with Nb, Zr and Y. These elements are usually considered to be immobile during weathering and alteration of basaltic rock (e.g. Fitton et al., 1998). Furthermore, ratios of Nb, Zr and Y are not greatly affected by low-pressure crystal fractionation, and abundances of Sc and V are not affected by crustal contamination. Thus, these elements can prove particularly helpful in deciphering melting processes and source compositions. In light of this, a decision was made to produce a comprehensive set of data for a suite of British Palaeogene basalts. We chose to sample dykes, rather than lavas, to provide the best possible geographical coverage of the British Palaeogene igneous province. Selected dykes were sampled on the Hebridean islands of Canna–Sanday, Harris, Mull, Scalpay and the Uists (Fig. 2), on the Isle of Man and on the island of Lundy in the Bristol Channel (locations shown in Fig. 1). In contrast to the Palaeocene lavas, few geochemical data exist for these dykes. Samples were chosen on the basis of freshness, with no alteration visible in hand specimens. However, when viewed in thin section, olivine frequently is altered to iddingsite along cracks and grain boundaries. More rarely, plagioclase may show signs of alteration.

The samples were analysed for major and trace elements by X-ray fluorescence (XRF) spectrometry at the University of Edinburgh, and for REE by inductively coupled plasma–mass spectrometry (ICP–MS) at the Scottish Universities Environmental Research Centre (SUERC), East Kilbride. Analytical techniques and operating conditions for XRF spectrometry in Edinburgh have been described by Fitton et al. (1998); techniques for ICP–MS are similar to those described by Jarvis (1997). As with analyses reported by Fitton et al. (1997, 1998) and Chambers & Fitton (2000), very long count times were used to determine Nb, Zr and Y abundances in the basalts. Samples were analysed at least three times and averages calculated. Precision is estimated to be ±0·1 ppm (2σ) for samples with <2 ppm Nb, and ±0·2 ppm (2σ) for rocks with >2 ppm Nb. For Zr and Y, precision is ±0·4 ppm and ±0·6 ppm, respectively (both quoted to 2σ). For other elements, precision is similar to that reported by Fitton et al. (1998, tables 2 and 3). Accuracy can be gauged by comparing the average standard concentrations reported by Kent (1995, table 1) with those given by Govindaraju (1994).

It should be noted that all XRF or ICP–MS data discussed below and plotted on the diagrams were obtained by the same methods as those used for the analysis of Icelandic basalt samples (Fitton et al., 1997, 1998). The results are thus directly comparable. A representative set of the new geochemical analyses is shown in Table 2 and normalized trace element abundance patterns typical of M1–M3 basalts are shown in Fig. 3. The complete dataset for Hebridean dykes analysed during this study is available on the Journal of Petrology Web site (http://www.petrology.oupjournals.org). Data for the Isle of Man and Lundy samples will be presented in full elsewhere. Analyses of M4-type dykes obtained as part of this study were presented by Kerr et al. (1999, table 7). The reader is referred to Kerr et al. (1999) for a discussion of the composition of M4 basalts.

Table 2:

Representative geochemical analyses of Palaeogene basaltic dykes from the British Isles

Sample: D155-1B D200-2A D240-1A T503-1A NUD2 SHD36 SHD42 SYD4 T140E 
Locality: N Mull N Mull N Mull Canna N Uist S Harris S Harris Scalpay Lundy 
Magma type: M1
 
M1
 
M1
 
M1
 
M1
 
M1
 
M1
 
M1
 
M1
 
SiO2  44·05  44·09  51·35  44·91   46·74  48·47  51·49  45·12  44·28 
TiO2   3·39   2·65   1·17   1·80    3·45   3·43   2·35   2·33   2·76 
Al2O3  15·86  15·28  14·24  15·91   11·31  12·59  13·91  16·04  15·17 
Fe2O3*  14·77  15·00  11·58  14·85   11·59  11·80  10·04  16·02  14·98 
MgO   5·22   7·40   6·35   6·04    8·11   7·15   5·62   4·47   6·89 
MnO   0·22   0·24   0·19   0·22    0·16   0·17   0·12   0·27   0·35 
CaO   7·37   7·54  10·08   7·93    8·73   7·07   7·59   7·22   8·82 
Na2  4·04   3·35   2·64   3·51    2·60   3·66   3·31   4·32   3·48 
K2  0·46   0·39   0·96   0·59    2·28   2·07   1·02   0·53   0·36 
P2O5   0·47   0·32   0·15   0·22    0·84   0·49   0·55   0·29   0·43 
LOI   3·68   3·57   0·91   3·68    3·57   2·73   3·86   3·17   3·15 
Total  99·52  99·83  99·62  99·65   99·38  99·63  99·86  99·78 100·67 
Nb  14·4   6·2   7·1   4·1   45·5  34·4  34·1  16·3  13·5 
Zr 256·8 202·3 144·7 117·6  389·3 278·3 311·6 175·7 358·3 
 36·7  26·8  36·8  34·2   30·6  28·9  29·0  42·1  51·0 
Sr 463·6 338·7 156·0 230·5  712·8 372·0 686·4 309·3 263·0 
Rb   4·3   3·5  26·3   7·8   40·0  42·6  13·7  18·8   7·5 
Th   5·1   3·9   4·7   3·7    6·5   4·8   7·0   5·0   4·6 
Pb   3·0   0·4   5·0   1·9    4·0   3·3   4·8   0·7   2·6 
Zn  79 107  99 105  128 122 108  91 122 
Cu  53  49 139  48   64  66  45  66  91 
Ni  37  60  83  66  205 171 135  59  76 
Cr b.d.  54 214  38  288 274 192  10 126 
174 277 303 283  234 260 174 121 391 
Ba 152 112 236 221 1275 469 414 313 155 
Sc  14  22  46  29   20  25  22  21  39 
La  14·3   7·3   9·3  10·9   61·4  28·0  44·3   5·6  13·3 
Ce  39·9  20·3  36·8  25·1  150·6  78·6 105·8  24·6  43·7 
Pr   6·3   3·5    4·0      
Nd  31·1  17·5  20·0  17·7   82·1  46·5  53·7  18·4  32·1 
Sm   7·99   4·88    4·88      
Eu   2·80   1·64    1·81      
Gd   8·41   5·15    5·94      
Tb   1·17   0·73    0·96      
Dy   6·76   3·73    6·15      
Ho   1·24   0·67    1·25      
Er   3·34   1·76    3·59      
Tm   0·49   0·22    0·49      
Yb   2·99   1·41    3·26      
Lu   0·48   0·23    0·55      
ΔNb  −0·29  −0·58  −0·12  −0·21   −0·21  −0·07  −0·17   0·14  −0·47 
Ba/Zr   0·6   0·6   1·6   1·9    3·3   1·7   1·3   1·8   0·4 
Ti/Zr  79  78  48  91   53  73  45  79  46 
Zr/Nb  17·8  32·6  20·4  28·7    8·6   8·1   9·1  10·8  26·5 
Zr/Sc  18·2   9·2   3·1   4·1   19·2  11·1  14·2   8·6   9·1 
Sample: D155-1B D200-2A D240-1A T503-1A NUD2 SHD36 SHD42 SYD4 T140E 
Locality: N Mull N Mull N Mull Canna N Uist S Harris S Harris Scalpay Lundy 
Magma type: M1
 
M1
 
M1
 
M1
 
M1
 
M1
 
M1
 
M1
 
M1
 
SiO2  44·05  44·09  51·35  44·91   46·74  48·47  51·49  45·12  44·28 
TiO2   3·39   2·65   1·17   1·80    3·45   3·43   2·35   2·33   2·76 
Al2O3  15·86  15·28  14·24  15·91   11·31  12·59  13·91  16·04  15·17 
Fe2O3*  14·77  15·00  11·58  14·85   11·59  11·80  10·04  16·02  14·98 
MgO   5·22   7·40   6·35   6·04    8·11   7·15   5·62   4·47   6·89 
MnO   0·22   0·24   0·19   0·22    0·16   0·17   0·12   0·27   0·35 
CaO   7·37   7·54  10·08   7·93    8·73   7·07   7·59   7·22   8·82 
Na2  4·04   3·35   2·64   3·51    2·60   3·66   3·31   4·32   3·48 
K2  0·46   0·39   0·96   0·59    2·28   2·07   1·02   0·53   0·36 
P2O5   0·47   0·32   0·15   0·22    0·84   0·49   0·55   0·29   0·43 
LOI   3·68   3·57   0·91   3·68    3·57   2·73   3·86   3·17   3·15 
Total  99·52  99·83  99·62  99·65   99·38  99·63  99·86  99·78 100·67 
Nb  14·4   6·2   7·1   4·1   45·5  34·4  34·1  16·3  13·5 
Zr 256·8 202·3 144·7 117·6  389·3 278·3 311·6 175·7 358·3 
 36·7  26·8  36·8  34·2   30·6  28·9  29·0  42·1  51·0 
Sr 463·6 338·7 156·0 230·5  712·8 372·0 686·4 309·3 263·0 
Rb   4·3   3·5  26·3   7·8   40·0  42·6  13·7  18·8   7·5 
Th   5·1   3·9   4·7   3·7    6·5   4·8   7·0   5·0   4·6 
Pb   3·0   0·4   5·0   1·9    4·0   3·3   4·8   0·7   2·6 
Zn  79 107  99 105  128 122 108  91 122 
Cu  53  49 139  48   64  66  45  66  91 
Ni  37  60  83  66  205 171 135  59  76 
Cr b.d.  54 214  38  288 274 192  10 126 
174 277 303 283  234 260 174 121 391 
Ba 152 112 236 221 1275 469 414 313 155 
Sc  14  22  46  29   20  25  22  21  39 
La  14·3   7·3   9·3  10·9   61·4  28·0  44·3   5·6  13·3 
Ce  39·9  20·3  36·8  25·1  150·6  78·6 105·8  24·6  43·7 
Pr   6·3   3·5    4·0      
Nd  31·1  17·5  20·0  17·7   82·1  46·5  53·7  18·4  32·1 
Sm   7·99   4·88    4·88      
Eu   2·80   1·64    1·81      
Gd   8·41   5·15    5·94      
Tb   1·17   0·73    0·96      
Dy   6·76   3·73    6·15      
Ho   1·24   0·67    1·25      
Er   3·34   1·76    3·59      
Tm   0·49   0·22    0·49      
Yb   2·99   1·41    3·26      
Lu   0·48   0·23    0·55      
ΔNb  −0·29  −0·58  −0·12  −0·21   −0·21  −0·07  −0·17   0·14  −0·47 
Ba/Zr   0·6   0·6   1·6   1·9    3·3   1·7   1·3   1·8   0·4 
Ti/Zr  79  78  48  91   53  73  45  79  46 
Zr/Nb  17·8  32·6  20·4  28·7    8·6   8·1   9·1  10·8  26·5 
Zr/Sc  18·2   9·2   3·1   4·1   19·2  11·1  14·2   8·6   9·1 
Sample: D140-100 D190-35A D290-8A T501-1A T504-1A T506-1A T507-1A T515-1A SHD40 SYD9 
Locality: N Mull N Mull N Mull Canna Canna Canna Canna Canna S Harris Scalpay 
Magma type: M2
 
M2
 
M2
 
M2
 
M2
 
M2
 
M2
 
M2
 
M2
 
M2
 
SiO2  44·64  46·33  45·36  47·47  45·20  46·29  47·51  45·70  49·33   44·51 
TiO2   1·55   2·61   1·93   1·07   2·10   1·54   1·03   2·03   1·41    1·40 
Al2O3  17·32  13·10  17·69  16·79  15·08  17·00  17·15  15·54  13·48   13·40 
Fe2O3*  14·23  16·73  14·73  11·89  14·30  14·20  11·58  14·38  14·73   13·17 
MgO   6·29   5·10   4·76   8·59   8·07   7·30   8·46   7·71   6·16   13·19 
MnO   0·29   0·24   0·15   0·18   0·20   0·19   0·18   0·20   0·22    0·22 
CaO   8·08   9·54   8·90  10·61   9·30   8·39  10·68   8·62  10·96    8·67 
Na2  3·50   2·94   3·16   2·52   2·93   3·54   2·48   2·90   2·41    1·87 
K2  0·19   0·89   0·30   0·19   0·43   0·14   0·25   0·48   0·42    0·15 
P2O5   0·18   0·39   0·14   0·09   0·21   0·16   0·09   0·21   0·11    0·11 
LOI   3·43   1·66   2·79   0·42   2·00   1·22   0·27   2·08   0·81    3·09 
Total  99·71  99·52  99·92  99·83  99·84  99·96  99·65  99·84 100·04   99·77 
Nb   5·3   6·9   2·9   1·8   2·1   1·3   1·3   3·8   3·6    2·7 
Zr  96·1 166·0 120·0  63·2  87·6  97·5  61·7 114·8  91·4   86·5 
 33·8  43·2  28·4  24·6  28·9  33·9  23·9  31·3  35·3   18·5 
Sr 223·0 305·3 294·3 273·3 270·8 218·8 281·8 318·0 146·4  320·9 
Rb   2·8   9·4   2·9   1·9   3·5   1·8   2·3   4·0  10·2    4·4 
Th   4·5   7·1   3·4   1·4   3·9   3·8   2·9   3·9   4·2    2·6 
Pb   1·3   5·0   1·9   0·6   1·4   1·4   2·0   1·4   1·4  b.d. 
Zn  76 130  99  82  92  82  77 106 100   77 
Cu  90 108  46 145  91  89 139 120 233  116 
Ni  83  22  57 165 123 115 169 371  63  480 
Cr b.d.  46  29 255 169  32 251 124  36  1026 
193 519 328 268 350 176 260 161 411  267 
Ba  85 440  45 110 273  53 118 230 104   96 
Sc  27  49  26  37  36  26  32  33  49   24 
La   5·2  18·2   4·6   4·1   4·2   3·1   3·6   7·0   2·9    1·6 
Ce  13·0  40·6  13·5   9·9  10·3   8·9   8·6  26·2  16·6    4·1 
Pr   2·4   5·6   2·7   2·2   2·3   2·3   1·9      1·1 
Nd  10·9  24·8  13·8   8·2   9·5  10·1   6·8  19·0  12·5    3·7 
Sm   3·76   6·18   4·43   2·70   3·27   3·65   2·21   8·67     1·19 
Eu   1·44   2·19   1·63   1·10   1·31   1·53   0·88   2·73     0·43 
Gd   4·83   7·43   5·09   3·57   4·09   5·15   2·97  12·68     1·38 
Tb   0·82   1·09   0·79   0·63   0·64   0·86   0·51   1·14     0·22 
Dy   5·36   7·12   4·99   3·96   4·15   5·78   3·35   6·00     1·33 
Ho   1·11   1·40   0·97   0·83   0·82   1·19   0·68   1·07     0·25 
Er   3·24   4·08   2·62   2·40   2·26   3·34   2·04   2·88     0·67 
Tm   0·48   0·57   0·36   0·35   0·31   0·48   0·28   0·38     0·10 
Yb   3·26   3·84   2·32   2·33   2·07   3·18   1·88   2·46     0·60 
Lu   0·51   0·60   0·39   0·37   0·33   0·53   0·31   0·40     0·12 
ΔNb   0·06  −0·18  −0·45  −0·18  −0·32  −0·56  −0·32  −0·26  −0·04   −0·38 
Ba/Zr   0·9   2·6   0·4   1·7   3·1   0·5   1·9   2·0   1·1    1·1 
Ti/Zr  96  94  96 101 143  94  99 105  92   96 
Zr/Nb  18·1  24·1  41·4  35·1  41·7  75·0  47·5  30·2  25·4   32·0 
Zr/Sc   3·6   3·4   4·5   1·7   2·4   3·7   1·9   3·5   1·9    3·7 
Sample: D140-100 D190-35A D290-8A T501-1A T504-1A T506-1A T507-1A T515-1A SHD40 SYD9 
Locality: N Mull N Mull N Mull Canna Canna Canna Canna Canna S Harris Scalpay 
Magma type: M2
 
M2
 
M2
 
M2
 
M2
 
M2
 
M2
 
M2
 
M2
 
M2
 
SiO2  44·64  46·33  45·36  47·47  45·20  46·29  47·51  45·70  49·33   44·51 
TiO2   1·55   2·61   1·93   1·07   2·10   1·54   1·03   2·03   1·41    1·40 
Al2O3  17·32  13·10  17·69  16·79  15·08  17·00  17·15  15·54  13·48   13·40 
Fe2O3*  14·23  16·73  14·73  11·89  14·30  14·20  11·58  14·38  14·73   13·17 
MgO   6·29   5·10   4·76   8·59   8·07   7·30   8·46   7·71   6·16   13·19 
MnO   0·29   0·24   0·15   0·18   0·20   0·19   0·18   0·20   0·22    0·22 
CaO   8·08   9·54   8·90  10·61   9·30   8·39  10·68   8·62  10·96    8·67 
Na2  3·50   2·94   3·16   2·52   2·93   3·54   2·48   2·90   2·41    1·87 
K2  0·19   0·89   0·30   0·19   0·43   0·14   0·25   0·48   0·42    0·15 
P2O5   0·18   0·39   0·14   0·09   0·21   0·16   0·09   0·21   0·11    0·11 
LOI   3·43   1·66   2·79   0·42   2·00   1·22   0·27   2·08   0·81    3·09 
Total  99·71  99·52  99·92  99·83  99·84  99·96  99·65  99·84 100·04   99·77 
Nb   5·3   6·9   2·9   1·8   2·1   1·3   1·3   3·8   3·6    2·7 
Zr  96·1 166·0 120·0  63·2  87·6  97·5  61·7 114·8  91·4   86·5 
 33·8  43·2  28·4  24·6  28·9  33·9  23·9  31·3  35·3   18·5 
Sr 223·0 305·3 294·3 273·3 270·8 218·8 281·8 318·0 146·4  320·9 
Rb   2·8   9·4   2·9   1·9   3·5   1·8   2·3   4·0  10·2    4·4 
Th   4·5   7·1   3·4   1·4   3·9   3·8   2·9   3·9   4·2    2·6 
Pb   1·3   5·0   1·9   0·6   1·4   1·4   2·0   1·4   1·4  b.d. 
Zn  76 130  99  82  92  82  77 106 100   77 
Cu  90 108  46 145  91  89 139 120 233  116 
Ni  83  22  57 165 123 115 169 371  63  480 
Cr b.d.  46  29 255 169  32 251 124  36  1026 
193 519 328 268 350 176 260 161 411  267 
Ba  85 440  45 110 273  53 118 230 104   96 
Sc  27  49  26  37  36  26  32  33  49   24 
La   5·2  18·2   4·6   4·1   4·2   3·1   3·6   7·0   2·9    1·6 
Ce  13·0  40·6  13·5   9·9  10·3   8·9   8·6  26·2  16·6    4·1 
Pr   2·4   5·6   2·7   2·2   2·3   2·3   1·9      1·1 
Nd  10·9  24·8  13·8   8·2   9·5  10·1   6·8  19·0  12·5    3·7 
Sm   3·76   6·18   4·43   2·70   3·27   3·65   2·21   8·67     1·19 
Eu   1·44   2·19   1·63   1·10   1·31   1·53   0·88   2·73     0·43 
Gd   4·83   7·43   5·09   3·57   4·09   5·15   2·97  12·68     1·38 
Tb   0·82   1·09   0·79   0·63   0·64   0·86   0·51   1·14     0·22 
Dy   5·36   7·12   4·99   3·96   4·15   5·78   3·35   6·00     1·33 
Ho   1·11   1·40   0·97   0·83   0·82   1·19   0·68   1·07     0·25 
Er   3·24   4·08   2·62   2·40   2·26   3·34   2·04   2·88     0·67 
Tm   0·48   0·57   0·36   0·35   0·31   0·48   0·28   0·38     0·10 
Yb   3·26   3·84   2·32   2·33   2·07   3·18   1·88   2·46     0·60 
Lu   0·51   0·60   0·39   0·37   0·33   0·53   0·31   0·40     0·12 
ΔNb   0·06  −0·18  −0·45  −0·18  −0·32  −0·56  −0·32  −0·26  −0·04   −0·38 
Ba/Zr   0·9   2·6   0·4   1·7   3·1   0·5   1·9   2·0   1·1    1·1 
Ti/Zr  96  94  96 101 143  94  99 105  92   96 
Zr/Nb  18·1  24·1  41·4  35·1  41·7  75·0  47·5  30·2  25·4   32·0 
Zr/Sc   3·6   3·4   4·5   1·7   2·4   3·7   1·9   3·5   1·9    3·7 
Sample: D175-3A D205-2B T508-1A T512-1B NHD10 NHD13 SYD8 SYD11 SYD12 SYD13 SYD16 
Locality: N Mull N Mull Canna Canna N Harris N Harris Scalpay Scalpay Scalpay Scalpay Scalpay 
Magma 
type: M3
 
M3
 
M3
 
M3
 
M3
 
M3
 
M3
 
M3
 
M3
 
M3
 
M3
 
SiO2  45·00  46·65  46·22  47·67  48·41  46·56  47·58  48·27  45·91  48·18  47·38 
TiO2   0·76   0·96   0·82   0·96   0·70   0·80   1·07   0·61   1·01   0·62   0·98 
Al2O3  16·49  17·73  15·94  15·50  17·00  16·68  15·88  18·98  16·62  18·58  15·79 
Fe2O3*  10·43  11·09  11·03  12·17   9·97  10·62  11·43   8·63  11·56   8·87  12·21 
MgO  10·79   6·72   9·52   8·61   8·05   9·28   7·89   7·76   8·97   8·39   7·52 
MnO   0·15   0·17   0·18   0·19   0·14   0·17  12·16  12·45   0·17   0·14  12·96 
CaO  11·35  11·97  11·99  12·97  11·74  12·38   2·23   1·65  10·50  12·31   2·02 
Na2  1·87   2·36   2·12   1·92   1·90   1·67   0·20   0·27   2·57   1·62   0·09 
K2  0·13   0·09   0·21   0·06   0·27   0·12   0·17   0·14   0·21   0·28   0·20 
P2O5   0·07   0·07   0·05   0·06   0·07   0·07   0·09   0·04   0·06   0·04   0·07 
LOI   2·60   2·40   1·78   0·05   1·36   1·51   1·09   1·29   2·32   1·12   0·83 
Total  99·64 100·22  99·85 100·17  99·61  99·85  99·80 100·09  99·90 100·14 100·04 
Nb   1·3   1·8   1·3   1·9   1·2   1·5   2·8   0·8   5   1·7   1·4 
Zr  43·1  45·1  40·8  48·7  47·0  41·9  67·7  44·4  52·2  44·1  49·4 
 18·4  21·1  20·3  25·3  19·1  20·0  23·7  16·2  20·9  16·8  48·2 
Sr 172·6 183·8 150·5 105·1 178·9 123·5 221·2 132·1 202·7 135·4 111·4 
Rb   2·3   1·3   2·1   1·0   8·1   3·2   4·0   8·4   6·0   7·6   2·3 
Th   0·8   3·0   3·0   3·3   1·7   2·0   2·6   1·9   3·3   1·6   2·7 
Pb   0·9   1·4   1·3 b.d.   1·6   0·7   2·8   1·0 b.d.   2·6 b.d. 
Zn  67  70  67  80  63  61  80  54  70  54  71 
Cu 138 137 147 161 122 132 148 103  95 107 163 
Ni 301 196 152 121 140 151 152 143 165 137  88 
Cr 292 238 357 366 342 327 406 327 241 329 293 
215 276 274 320 234 252 304 194 206 197 307 
Ba  59  35  12  16  74  44 123  74 102  80  22 
Sc  31  40  41  45  33  36  42  29  30  32  40 
La   1·0   2·4   1·5  13·3   1·0 b.d.   1·9   0·9   2·0 b.d.   1·1 
Ce   2·1   5·7   3·9   8·4  11·4   7·4  15·8   9·6   9·5   9·5   7·9 
Pr   0·6   1·2   1·4         
Nd   1·5   5·5   4·2   9·5   6·1   7·5   9·2   6·9   6·5   6·6   6·6 
Sm   0·52   1·98   1·75   2·37        
Eu   0·18   0·88   0·73   0·89        
Gd   0·68   2·84   2·61   4·13        
Tb   0·12   0·48   0·47   0·56        
Dy   0·73   3·21   3·06   3·57        
Ho   0·16   0·64   0·63   0·76        
Er   0·46   1·93   1·79   2·13        
Tm   0·09   0·27   0·26   0·32        
Yb   0·45   1·79   1·78   2·14        
Lu   0·08   0·29   0·29   0·33        
ΔNb  −0·12   0·04  −0·04   0·07  −0·21   0·00  −0·06  −0·41   0·36  −0·06   0·18 
Ba/Zr   1·4   0·8   0·3   0·3   1·6   1·0   1·8   1·7   1·9   1·8   0·4 
Ti/Zr 105 126 119 118  89 113  94  82 115  84 118 
Zr/Nb  33·2  25·1  31·4  25·6  39·2  27·9  24·2  55·5  10·4  25·9  35·3 
Zr/Sc   1·4   1·1   1·0   1·1   1·4   1·2   1·6   1·5   1·7   1·4   1·2 
Sample: D175-3A D205-2B T508-1A T512-1B NHD10 NHD13 SYD8 SYD11 SYD12 SYD13 SYD16 
Locality: N Mull N Mull Canna Canna N Harris N Harris Scalpay Scalpay Scalpay Scalpay Scalpay 
Magma 
type: M3
 
M3
 
M3
 
M3
 
M3
 
M3
 
M3
 
M3
 
M3
 
M3
 
M3
 
SiO2  45·00  46·65  46·22  47·67  48·41  46·56  47·58  48·27  45·91  48·18  47·38 
TiO2   0·76   0·96   0·82   0·96   0·70   0·80   1·07   0·61   1·01   0·62   0·98 
Al2O3  16·49  17·73  15·94  15·50  17·00  16·68  15·88  18·98  16·62  18·58  15·79 
Fe2O3*  10·43  11·09  11·03  12·17   9·97  10·62  11·43   8·63  11·56   8·87  12·21 
MgO  10·79   6·72   9·52   8·61   8·05   9·28   7·89   7·76   8·97   8·39   7·52 
MnO   0·15   0·17   0·18   0·19   0·14   0·17  12·16  12·45   0·17   0·14  12·96 
CaO  11·35  11·97  11·99  12·97  11·74  12·38   2·23   1·65  10·50  12·31   2·02 
Na2  1·87   2·36   2·12   1·92   1·90   1·67   0·20   0·27   2·57   1·62   0·09 
K2  0·13   0·09   0·21   0·06   0·27   0·12   0·17   0·14   0·21   0·28   0·20 
P2O5   0·07   0·07   0·05   0·06   0·07   0·07   0·09   0·04   0·06   0·04   0·07 
LOI   2·60   2·40   1·78   0·05   1·36   1·51   1·09   1·29   2·32   1·12   0·83 
Total  99·64 100·22  99·85 100·17  99·61  99·85  99·80 100·09  99·90 100·14 100·04 
Nb   1·3   1·8   1·3   1·9   1·2   1·5   2·8   0·8   5   1·7   1·4 
Zr  43·1  45·1  40·8  48·7  47·0  41·9  67·7  44·4  52·2  44·1  49·4 
 18·4  21·1  20·3  25·3  19·1  20·0  23·7  16·2  20·9  16·8  48·2 
Sr 172·6 183·8 150·5 105·1 178·9 123·5 221·2 132·1 202·7 135·4 111·4 
Rb   2·3   1·3   2·1   1·0   8·1   3·2   4·0   8·4   6·0   7·6   2·3 
Th   0·8   3·0   3·0   3·3   1·7   2·0   2·6   1·9   3·3   1·6   2·7 
Pb   0·9   1·4   1·3 b.d.   1·6   0·7   2·8   1·0 b.d.   2·6 b.d. 
Zn  67  70  67  80  63  61  80  54  70  54  71 
Cu 138 137 147 161 122 132 148 103  95 107 163 
Ni 301 196 152 121 140 151 152 143 165 137  88 
Cr 292 238 357 366 342 327 406 327 241 329 293 
215 276 274 320 234 252 304 194 206 197 307 
Ba  59  35  12  16  74  44 123  74 102  80  22 
Sc  31  40  41  45  33  36  42  29  30  32  40 
La   1·0   2·4   1·5  13·3   1·0 b.d.   1·9   0·9   2·0 b.d.   1·1 
Ce   2·1   5·7   3·9   8·4  11·4   7·4  15·8   9·6   9·5   9·5   7·9 
Pr   0·6   1·2   1·4         
Nd   1·5   5·5   4·2   9·5   6·1   7·5   9·2   6·9   6·5   6·6   6·6 
Sm   0·52   1·98   1·75   2·37        
Eu   0·18   0·88   0·73   0·89        
Gd   0·68   2·84   2·61   4·13        
Tb   0·12   0·48   0·47   0·56        
Dy   0·73   3·21   3·06   3·57        
Ho   0·16   0·64   0·63   0·76        
Er   0·46   1·93   1·79   2·13        
Tm   0·09   0·27   0·26   0·32        
Yb   0·45   1·79   1·78   2·14        
Lu   0·08   0·29   0·29   0·33        
ΔNb  −0·12   0·04  −0·04   0·07  −0·21   0·00  −0·06  −0·41   0·36  −0·06   0·18 
Ba/Zr   1·4   0·8   0·3   0·3   1·6   1·0   1·8   1·7   1·9   1·8   0·4 
Ti/Zr 105 126 119 118  89 113  94  82 115  84 118 
Zr/Nb  33·2  25·1  31·4  25·6  39·2  27·9  24·2  55·5  10·4  25·9  35·3 
Zr/Sc   1·4   1·1   1·0   1·1   1·4   1·2   1·6   1·5   1·7   1·4   1·2 

Major elements in weight percent, trace elements in ppm. LOI, loss on ignition; b.d., below detection limit. ΔNb is equal to 1·74 + log(Nb/Y) − 1·92 log(Zr/Y), and is the deficiency or excess of Nb in a given basalt sample, relative to the lower bound of the Icelandic basalt array on a plot of Nb/Y vs Zr/Y [see main text and Fitton et al. (1997, 1998)].

*Total Fe reported as Fe2O3.

Fig. 3.

Trace element patterns of Palaeogene basaltic dykes representative of magma types M1–M3 (see Table 1), normalized to normal mid-ocean ridge basalt (N-MORB) (Pearce, 1983) and chondritic abundances (Nakamura, 1974). The data plotted are Edinburgh XRF and SUERC ICP–MS analyses of samples from the Canna–Sanday dyke swarm, Inner Hebrides (this study).

Fig. 3.

Trace element patterns of Palaeogene basaltic dykes representative of magma types M1–M3 (see Table 1), normalized to normal mid-ocean ridge basalt (N-MORB) (Pearce, 1983) and chondritic abundances (Nakamura, 1974). The data plotted are Edinburgh XRF and SUERC ICP–MS analyses of samples from the Canna–Sanday dyke swarm, Inner Hebrides (this study).

The new data indicate that basaltic dykes collected from the Outer Hebridean islands of North and South Harris, Scalpay and the Uists (60 samples in total) are dominated by compositions corresponding to M3. More rarely, M1 and M2 compositions are observed. The dykes from the Inner Hebridean islands of Canna–Sanday and Mull (63 samples) include examples from each of M1, M2 and M3, with M2 being dominant on Canna. Farther south, the Isle of Man and Lundy dykes (two and 34 samples, respectively; our unpublished data) belong to M1 or M2. The absence of M3 dykes on the Isle of Man requires confirmation, as our small sample set need not be representative. It is noteworthy, however, that M3 samples are absent from a suite of Palaeogene dykes from central England (Lancashire, Staffordshire, Shropshire) and North Wales (Anglesey) studied by Thompson & Winchester (1995). This strongly suggests that, within the British Isles, M3 is confined to the Hebrides and Antrim, i.e. those areas closest to the rifted margin. If confirmed, this ‘gradient’ would be similar to that pertaining in the Faeroe Islands, where M3-like dykes and sills are most common in the northern part of the Islands, closest to the line of opening of the North Atlantic (Hald & Waagstein, 1991).

MANTLE SOURCES DURING PALAEOGENE MAGMATISM

Chemical and Pb–Nd–Sr isotopic analyses of Miocene–Recent (<16 Ma) Icelandic basalts and Palaeogene basic lavas from the NAIP (e.g. Carter et al., 1979; Hémond et al., 1993; Hardarson et al., 1997; Kempton et al., 2000) show that the mantle source of both suites includes a depleted end-member similar to the mantle source of normal mid-ocean ridge basalt (N-MORB). This depleted component could be asthenospheric mantle heated and/or entrained by the rising Iceland plume, or else could be intrinsic to the plume.

In the British Isles, the Nd–Sr isotopic ratios of Palaeogene basalts are known to be sensitive to even small amounts of crustal contamination (e.g. Moorbath & Thompson, 1980; Thorpe & Tindle, 1992; Wallace et al., 1994). This problem also extends to Pb isotopes, ratios of which in British Palaeogene basalts invariably provide insights into the nature of the contaminant but not the mantle source (e.g. Dickin, 1981). Even in samples with low Ba/Zr and Sr/Zr, the identity of the mantle source is unclear. For example, the least-contaminated M3 dykes have 87Sr/86Sr(t) ≤ 0·70320 and 143Nd/144Nd(t) ≥0·51305 (R. W. Kent, G. Rogers & R. M. Ellam, unpublished data, 1995) and could have sampled either an N-MORB source or the depleted end-member in the Iceland plume.

Tracking mantle sources using ΔNb

To overcome the above problem, Fitton et al. (1997, 1998) and Chambers & Fitton (2000) have used the abundance of Nb relative to Zr and Y (ΔNb) to distinguish between the mantle sources of NAIP basalts, noting that N-MORB is deficient in Nb when compared with Icelandic basalt and primitive mantle. Thus, on a Nb/Y vs Zr/Y diagram (e.g. Fig. 4, showing data for samples from this study), N-MORB and Icelandic basalt define distinct, parallel arrays, which cannot reflect mixing of the N-MORB mantle source with the plume source. The lower bound of the Iceland array can be used as a reference line (Fitton et al., 1997):  

\[\displaylines{\Delta \hbox{Nb} = 1\cdot 74 + \log(\hbox{Nb/Y}) - 1\cdot 92 \log(\hbox{Zr/Y}).\hfill (1)}\]
Values of ΔNb > 0 imply an Icelandic mantle source whereas ΔNb < 0 indicates a source in the depleted upper mantle. Importantly, assimilation of continental crust has little effect on ΔNb values because crustal rocks plot on or below the lower bound of the Iceland array, i.e. contamination by crust serves only to lower ΔNb and can never make a sample from an N-MORB source appear to be ‘Icelandic’ (Fitton et al., 1997; Chambers & Fitton, 2000). To limit the effects of crustal contamination, it is possible to filter a given dataset to remove samples with Ba/Zr > 2 (compare Fig. 4b with Fig. 4c). With this additional safeguard, ΔNb is considered to reflect only the mantle source composition.

Fig. 4.

(a) Nb/Y and Zr/Y variation for Palaeogene basaltic dykes from the Hebrides and the southern part of the British Palaeogene igneous province. Parallel lines show the lower and upper bounds of the array for Icelandic basalt (Fitton et al., 1997, 1998). M, N-MORB average (Hofmann, 1988); PM, primitive mantle (McDonough & Sun, 1995). The dyke data are Edinburgh XRF analyses from this study. (b) Variation in Nb/Y and Zr/Y for the same set of samples, grouped according to magma type (for explanation, see main text and Table 1). (c) Nb/Y and Zr/Y variation in our dyke dataset, filtered to remove crustally contaminated samples with Ba/Zr > 2.

Fig. 4.

(a) Nb/Y and Zr/Y variation for Palaeogene basaltic dykes from the Hebrides and the southern part of the British Palaeogene igneous province. Parallel lines show the lower and upper bounds of the array for Icelandic basalt (Fitton et al., 1997, 1998). M, N-MORB average (Hofmann, 1988); PM, primitive mantle (McDonough & Sun, 1995). The dyke data are Edinburgh XRF analyses from this study. (b) Variation in Nb/Y and Zr/Y for the same set of samples, grouped according to magma type (for explanation, see main text and Table 1). (c) Nb/Y and Zr/Y variation in our dyke dataset, filtered to remove crustally contaminated samples with Ba/Zr > 2.

Using this rationale, Fitton et al. (1997, 1998) noted that many of the Palaeogene lavas erupted in SE Greenland and in the lower and middle portions of the Faeroese lava sequence have ΔNb > 0, signifying an Iceland plume source [but see Tegner et al. (1998)]. During the Palaeogene, these areas lay close to the presumed centre of the ancestral Iceland plume (Fig. 1). Basalts from Baffin Island and West Greenland show a bimodal (positive and negative) distribution of ΔNb, indicating the availability of ‘Icelandic’ and N-MORB sources up to 600 km away from the plume axis. In contrast, almost all Palaeocene lavas from the British Isles analysed by Fitton et al. (1997) have ΔNb < 0, implying an N-MORB source. The latter area lay on the fringes of the plume head at 61 Ma (Fig. 1). Fitton et al. (1997, 1998) interpreted these findings to suggest that the ancestral Iceland plume was compositionally zoned, with a central core of lower-mantle material (part of which was depleted in incompatible elements but not relatively deficient in Nb) and an outer rim of heated, N-MORB-type mantle.

The hypothesis of a zoned plume head is supported by a recent study of the Mull Palaeogene lava succession. Using ΔNb values, Chambers & Fitton (2000) showed that beneath western Mull, an N-MORB source was tapped initially (∼60·6 Ma), followed some 2 my later by the ‘Icelandic’ mantle source. At ∼700 m above the base of the Mull lava pile, values of ΔNb fluctuate briefly between positive and negative. Towards the top of the lava pile at the present day, the basalts again show negative ΔNb, indicating a return to an N-MORB source. This source appears to have been sampled for ∼1 my until the close of magmatism on Mull at ∼58 Ma. This is confirmed by the fact that M4 dykes cutting the youngest part of the Mull central complex have ΔNb values in the range 0·00 to −0·08 [Edinburgh XRF data of Kerr et al. (1999, table 7)]. Thus, on Mull, an N-MORB source and the ‘Icelandic’ mantle source appear to have been tapped sequentially. In addition, there is a strong correlation between magma type and mantle source: the Mull M1 basalts have negative ΔNb, M2 basalts mostly have positive ΔNb and M3 basalts have negative ΔNb. The occurrence of basalts on Mull with an ‘Icelandic’ source is remarkable in that the Mull lavas were emplaced at least 700 km from the axis of the ancestral Iceland plume (Fig. 1). At the present day, the compositional effects of the Iceland plume are limited to a distance of 670 km from the plume centre (Fitton et al., 1997).

Does an equivalent picture emerge from the new data for British Palaeogene dykes, which provide excellent geographical coverage of the British Isles but which are not stratigraphically well controlled? Values of ΔNb for the Hebridean dykes range from +0·36 to −0·58, with a mean value of −0·14. The values of ΔNb for the two Isle of Man dykes are −0·33 and −0·43, respectively (our unpublished data). For Lundy dykes, ΔNb varies from +0·09 to −0·54, with a mean of −0·21. These values extend considerably the range in ΔNb reported by Fitton et al. (1997, fig. 7) and Chambers & Fitton (2000) for Palaeocene basalts from the Inner Hebrides, and indicate that throughout the British Palaeogene igneous province, both the ‘Icelandic’ source and an N-MORB source were tapped during volcanism. Values of ΔNb for the British dykes also suggest that the thermal and chemical structure of the ancestral Iceland plume varied with increasing distance from the plume centre (Fig. 5). Thus, on Scalpay in the Outer Hebrides, 58% of analysed dyke samples have positive ΔNb (‘Icelandic’ source), whereas on Lundy, ∼1300 km from the plume axis, the proportion is 3%.

Fig. 5.

Variation of ΔNb with distance from the centre of the ancestral Iceland plume at 60 Ma [location from White & McKenzie (1989)] for basaltic dykes of the British Palaeogene igneous province. Filled symbols represent mean values of ΔNb; bars indicate upper and lower bounds of the data. The limit of the present-day compositional anomaly associated with the Iceland plume (at 670 km) is taken from Fitton et al. (1997). Key: H, Harris; S, Scalpay; U, Uists; C, Canna–Sanday; NM, North Mull; CM, Central Mull; IoM, Isle of Man; L, Lundy. Data plotted are from this study.

Fig. 5.

Variation of ΔNb with distance from the centre of the ancestral Iceland plume at 60 Ma [location from White & McKenzie (1989)] for basaltic dykes of the British Palaeogene igneous province. Filled symbols represent mean values of ΔNb; bars indicate upper and lower bounds of the data. The limit of the present-day compositional anomaly associated with the Iceland plume (at 670 km) is taken from Fitton et al. (1997). Key: H, Harris; S, Scalpay; U, Uists; C, Canna–Sanday; NM, North Mull; CM, Central Mull; IoM, Isle of Man; L, Lundy. Data plotted are from this study.

Is the sequence in which the two mantle sources were tapped across the British Palaeogene igneous province the same as that on Mull? Samples of M1 basalt from our Canna–Sanday, Mull and Lundy datasets show reversed or normal polarity, which we equate with chrons 26R and 26N [see above and investigations by Ade-Hall et al. (1972), Mussett et al. (1976) and Dagley & Mussett (1981)]. With two exceptions—both of which are samples from the Outer Hebrides—the M1 dykes have negative ΔNb. Samples of M2 basalt from this study show reversed, intermediate or normal polarity, with reversed polarity being dominant. We equate this with emplacement of these dykes during chron 26R or 26N. The majority (76%) of the M2 basalts have negative ΔNb. The M3 dykes from our dataset show reversed or normal polarity, again consistent with their intrusion during chron 26R or 26N. Some 75% of these samples have negative ΔNb. Comparing these results with those obtained by Chambers & Fitton (2000), it is apparent that whereas almost all analysed M1 basalts from the British Isles have negative ΔNb, the analysed M2 population consists of roughly equal numbers of samples with negative ΔNb and samples with positive ΔNb (most analysed M2 lavas have positive ΔNb whereas most M2 dykes have negative ΔNb). M3 is dominated by samples with negative ΔNb, but includes a substantial cohort of basalts with positive ΔNb.

The dyke data broadly support the hypothesis of Chambers & Fitton (2000), that there was an overall shift in mean ΔNb with time from N-MORB values towards ‘Icelandic’ values and back again (Fig. 6). However, our data necessitate some important modifications to the Fitton et al. (1997) hypothesis of a plume head consisting of an ‘Icelandic’ core and an outer carapace of N-MORB-source mantle. To explain the existence of a small number of M1 dykes with positive ΔNb, the outer shell of the plume head must initially (∼61–59 Ma) have been heterogeneous, i.e. it contained blebs of ‘Icelandic’ material within a MORB-source matrix. To account for the occurrence of two distinct populations of M2 basalts (one with positive ΔNb and one with negative ΔNb) we suggest that the heterogeneous margin to the plume head was subsequently displaced by ‘Icelandic’ material spreading from the core of the plume. ΔNb and Ar–Ar data for Mull lavas (Chambers & Fitton, 2000) indicate that in the Hebrides, this displacement was abrupt and occurred at ∼58·7 Ma. After the ‘wave’ of plume material had receded (∼58 Ma), there was a return to dominance of an N-MORB mantle source. To account for the small cohort of M3 basalts with positive ΔNb, we propose that this source was heterogeneous, i.e. similar to that which was tapped to produce the parent magmas of M1 basalts.

Fig. 6.

Variation of ΔNb with time for magma types M1–M4 (Table 1) from the British Palaeogene igneous province. Filled symbols represent mean values of ΔNb. Bars indicate upper and lower bounds of the data. Data are Edinburgh XRF analyses from this study, Kent (1995) and Kerr et al. (1999, table 7).

Fig. 6.

Variation of ΔNb with time for magma types M1–M4 (Table 1) from the British Palaeogene igneous province. Filled symbols represent mean values of ΔNb. Bars indicate upper and lower bounds of the data. Data are Edinburgh XRF analyses from this study, Kent (1995) and Kerr et al. (1999, table 7).

The apparent ebb and flow of the ancestral Iceland plume over a ∼3 my period suggested by our data could reflect cooling of the plume head (Tegner et al., 1998) or a decline in output leading to a reduced flux of material through the melting region. Alternatively, incipient opening of the NE Atlantic Ocean may have led to channelling of plume material into the main rift zone and consequent diminution of the plume flux in marginal areas, such as the British Isles.

MANTLE MELTING DYNAMICS BENEATH THE NE ATLANTIC REGION

The above findings indicate that geochemical differences between the main British Palaeogene magma types are not due solely to differences in source composition. It was also noted above that the different compositions of M1, M2 and M3 basalts do not reflect alteration, fractional crystallization or crustal assimilation. Therefore, in addition to source influences, the differences must relate in some way to the melting processes that produced the basalts.

Depth and degree of melting

Previous workers (Thompson, 1974; Mattey et al., 1977; Wood, 1979; Ellam, 1992; Kerr, 1995a; Barrat & Nesbitt, 1996) have used geochemical and experimental data to model M1 parent magmas as small-degree (7–11%), point-and-depth average fractional melts of a garnet–spinel lherzolite source. This lherzolite composition is assumed to be depleted in trace elements with respect to primitive mantle, i.e. it had previously undergone a melt extraction event. M2 and M3 parent magmas were considered to be products of a further 7–20% melting of the garnet-poor residue remaining after extraction of M1 (e.g. Mattey et al., 1977). This explanation suggests a progressive shallowing in the mean depth of melt extraction over time, with the bulk of melting taking place within the garnet–spinel transition zone. For a mantle potential temperature of 1550°C, appropriate to a plume environment, this transition corresponds to a depth range of 85–95 km (Watson, 1993). Final melt segregation within, or just above, this depth range is consistent with a ‘lithospheric lid effect’, assuming an average thickness for the thermal lithosphere of 100 km (Stein & Stein, 1992) and a corresponding mechanical lithosphere thickness of 75 km.

Polybaric, near-fractional melting of a mixed garnet–spinel source to produce M1–M3 parent magmas is in keeping with the Zr/Sc ratios of samples from this study (Table 2 and supplementary data). Because of retention by garnet in the source, concentrations of Sc are low in primitive basalts formed by small degrees of melting at high pressure (e.g. the more-magnesian M1 samples). At lower pressures, under increasingly garnet-poor melting conditions, Sc concentrations in basaltic magmas become progressively higher (e.g. MgO-rich M2 and M3 samples). Figure 7 illustrates these effects by showing the variation of Zr and Sc for each magma type. Data points for M1 basalts generally lie above the melting curves, suggesting low degrees of melting (<10%) of a source depleted in Sc and enriched in Zr relative to that used in the modelling calculations. The M2 data cluster about the garnet lherzolite and garnet–spinel (equal mix) lherzolite melting curves, implying ∼6–10% melting of a garnet-rich source followed by 10–30% crystallization of olivine and plagioclase. Accumulation of olivine in certain samples is suggested by data points plotting to the left of the melting curves [see also Kent (1995)]. The more-magnesian M3 samples have Sc contents that are higher than those of M2 basalts. This is suggestive of 10–20% melting of a spinel lherzolite source, followed by up to 50% fractional crystallization of olivine and plagioclase. The melting curves are, of course, highly sensitive to uncertainties in the various parameters used. However, major element data also support the contention of progressively higher degrees of melting across the garnet–spinel transition zone. For example, fractionation-corrected values of FeO, Na2O and TiO2 for the Mull and Skye basalts differ in a manner that is consonant with a reduction in the mean pressure of melting over time (Kerr, 1995a; Scarrow & Cox, 1995).

Fig. 7.

Variation of Zr and Sc in Palaeogene basaltic dykes from the British Isles. Our dataset has been filtered to exclude samples with <6 wt % MgO and/or Ba/Zr > 2. Pooled fractional melting curves for garnet lherzolite, an equal mix garnet–spinel (gt–sp) lherzolite and spinel lherzolite are shown for reference. Crosses represent percentage of melt generated. The arrow shows the effect of 40% fractional crystallization of olivine and plagioclase. Mineral–melt partition coefficients (D) used to construct the melting curves are from McKenzie & O’Nions (1995). Values of Dspinel–liquid have not been determined for Zr and Sc, and the effects of residual spinel have not been taken into account. Source concentrations of 7·2 ppm for Zr and 12 ppm for Sc, appropriate to depleted mantle, are from McKenzie & O’Nions (1995); mantle phase proportions are taken from Brodie (1995).

Fig. 7.

Variation of Zr and Sc in Palaeogene basaltic dykes from the British Isles. Our dataset has been filtered to exclude samples with <6 wt % MgO and/or Ba/Zr > 2. Pooled fractional melting curves for garnet lherzolite, an equal mix garnet–spinel (gt–sp) lherzolite and spinel lherzolite are shown for reference. Crosses represent percentage of melt generated. The arrow shows the effect of 40% fractional crystallization of olivine and plagioclase. Mineral–melt partition coefficients (D) used to construct the melting curves are from McKenzie & O’Nions (1995). Values of Dspinel–liquid have not been determined for Zr and Sc, and the effects of residual spinel have not been taken into account. Source concentrations of 7·2 ppm for Zr and 12 ppm for Sc, appropriate to depleted mantle, are from McKenzie & O’Nions (1995); mantle phase proportions are taken from Brodie (1995).

One can shed further light on the relationship between M1, M2 and M3 by considering the Ti, Zr and Y concentrations of selected basalts from this study. Figure 8 shows that data points for British Palaeogene dyke samples with MgO > 6 wt % and Ba/Zr < 2 form en echelon arrays in Ti/Y vs Zr/Y space, where Zr/Y is an index of the degree and depth of partial melting (Nicholson & Latin, 1992). How should these arrays be interpreted? Data for M1 dykes (e.g. Table 2) are consistent with derivation of M1 parent magmas by up to 10% melting of a garnet lherzolite matrix previously depleted in Ti (e.g. the residue remaining after production of the Faeroes Lower Series basalts). For a given value of Zr/Y, M2 basalts have values of Ti/Y that in some cases are substantially (up to two or three times) higher than those of M1 basalts. This implies that the portion of mantle that melted to produce M2 parent magmas was less depleted in Ti than in Zr, when compared with the lherzolitic residue remaining after extraction of M1 parent magmas. In other words, Ti abundances in M2 basalts indicate that their parent melts could not have been produced by further incremental melting of the residue that produced M1 parent magmas. The same relationship, or lack thereof, appears to be true for M2 and M3 (Fig. 8), although the differences in Ti/Y at a given value of Zr/Y are not as pronounced.

Fig. 8.

Ti/Y variation with respect to Zr/Y for samples of British Palaeogene basalt. The dyke dataset (Table 2 and supplementary data) has been filtered to remove samples with <6 wt % MgO and/or Ba/Zr > 2. Zirconium and Y are incompatible in fractionating phases (olivine, plagioclase ± clinopyroxene), so the ratio of these elements will not be affected by moderate amounts of fractional crystallization. However, Zr is more incompatible in mantle phases than Y, resulting in high values of Zr/Y at small degrees of melting (particularly in the presence of garnet). Variation of Ti/Y and Zr/Y in the samples reflects depth and degree of melting, including source depletion through melt extraction.

Fig. 8.

Ti/Y variation with respect to Zr/Y for samples of British Palaeogene basalt. The dyke dataset (Table 2 and supplementary data) has been filtered to remove samples with <6 wt % MgO and/or Ba/Zr > 2. Zirconium and Y are incompatible in fractionating phases (olivine, plagioclase ± clinopyroxene), so the ratio of these elements will not be affected by moderate amounts of fractional crystallization. However, Zr is more incompatible in mantle phases than Y, resulting in high values of Zr/Y at small degrees of melting (particularly in the presence of garnet). Variation of Ti/Y and Zr/Y in the samples reflects depth and degree of melting, including source depletion through melt extraction.

Solving the ‘problem’ of M3 basalts

Changes in the average depth of melting and melt segregation during the period 61–58 Ma (the likely ‘emplacement window’ for M1–M3 basalts; see above) could, in theory, be explained by rapid stretching and thinning of the lithosphere, allowing hot mantle to well up in response to adiabatic decompression (Ellam, 1992). A serious difficulty with this explanation is that it is contrary to the Cenozoic subsidence history of the Sea of the Hebrides basin, as deduced from stratigraphic studies (see above). To circumvent this, White (1992) argued that the parent magmas of M1 and M2 basalts were produced beneath the Inner Hebrides, whereas magmas parental to M3 were generated below rapidly thinned lithosphere to the NW of the British Isles (i.e. in the northern Rockall Trough; Fig. 1). This explanation avoids the need to invoke thinning of the lithosphere close to the site of eruption. White proposed that the M3 parent magmas traversed the crust in linked dyke–sill complexes, eventually reaching the surface in the Hebrides. His hypothesis is based on two observations: (1) M3 compositions are known to occur in wells drilled in the northern Rockall Trough and on Hatton Bank (Merriman et al., 1988; Morton et al., 1988; Brodie & Fitton, 1998); (2) basaltic dyke–sill complexes occur locally on Skye and elsewhere in the Sea of the Hebrides basin (England, 1992), and appear to be common in other sedimentary basins around the British Isles. It therefore is conceivable that these dyke–sill systems acted as conduits for magma generated to the west of the British Isles.

An alternative hypothesis was proposed by Brodie (1995), who argued that M3 parent magmas do not represent point-and-depth average (integrated) melt compositions. Brodie noted that similarities exist between the REE concentrations of the least-contaminated M3 basalts and those of ultra-depleted melt inclusions in olivines from MORB and Icelandic lavas described by Sobolev & Shimizu (1993) and Sobolev et al. (1994). These inclusions have been interpreted as aliquots of melt generated near the top of a melting column and preserved within individual olivine grains. Likewise, Brodie (1995) made the assumption that M3 parent melts were not integrated with melt from lower in the column (>102 km) before extraction. He used inversion modelling of the REE abundances of M1 basalts to calculate a partial point-and-depth average composition over the depth range 102–55 km [for details and a critique of the modelling technique, see Brodie et al. (1994)]. This composition mimics the measured HREE abundances of M3 basalts reasonably well, but provides a poor fit for the LREE and large ion lithophile elements. To explain these discrepancies, Brodie (1995) appealed to contamination of M3 basalts by amphibolite facies crust. On the basis of the limited amount of Pb isotopic data available for M3 basalts (Dickin, 1981), this suggestion is not unreasonable; hence, one may concur with Brodie (1995) that M3 parent magmas do not represent well-mixed polybaric partial melts, but are a ‘snapshot’ of the melting column at depths of <102 km. In contrast, M1 parent magmas would be candidates for well-mixed melts. M2 parent magmas would then lie somewhere close to M1 on the spectrum from perfectly to imperfectly mixed fractional melts. Brodie’s hypothesis is (necessarily) speculative, but works in that it provides a mechanism to explain M3 compositions without the necessity for lithospheric thinning in the Sea of the Hebrides basin. However, M3 parent magmas are still required to have been generated outside the British Isles and one must provide a rationale for the lateral transport of M3 magmas through the crust from an area of thin lithosphere (the basins outboard of the British Isles) to one of thick lithosphere (the western part of the British Isles). Such a rationale was not given by White (1992).

Arguments similar to those presented by Brodie (1995) could be applied to other M3-like basalts in the NAIP, notably those in the Faeroes (Hald & Waagstein, 1991) and East Greenland (Fitton et al., 1998; Tegner et al., 1998). Precisely why variations in the efficiency of melt integration should have occurred during production of these basalts is not clear. Moreover, the whereabouts of the ‘deep’ melts complementary to the M3 basalts is uncertain. Did these deep melts reach the surface? Might they be represented in the British Isles by M4 basalts (Table 1)?

A model for the Palaeogene melting regime beneath the British Isles

The above observations are compatible with a model of upwelling within a strongly convecting mantle plume, part of which was located beneath rifted lithosphere (the incipient North Atlantic rift zone, lying to the west of the European continental shelf) and part beneath intact lithosphere (e.g. the British Isles) (Fig. 9). At ∼61 Ma, active upwelling and melting of mantle close to the centre of the ancestral Iceland plume may have been accompanied by lateral displacement of basalt-depleted matrices and replenishment of the melting column by fertile material drawn in from below. Residues swept out from the deepest parts of the column continued their ascent, undergoing a second episode of partial melting on the periphery of the plume (offshore UK and the British Isles). The products of this second-stage melting were M1 parent magmas, which were derived from a source consisting of hot, heterogeneous N-MORB-like mantle forming the outer envelope of the ancestral Iceland plume. M2 parent magmas were produced in a similar fashion, with a change in the mantle source from heterogeneous N-MORB-source to ‘Icelandic’ mantle over time. This change, which possibly was diachronous across the British Isles, reflected the outward spread of material from the core of the plume head. Melting beneath the intact lithosphere of the Sea of the Hebrides basin and that of basins farther south (Irish Sea, central England, Lundy) was made possible by the high potential temperature of the ancestral Iceland plume, estimated from olivine compositions in Hebridean basalts to have been in excess of 1540°C at a pressure of 2 GPa (Kent, 1995). Subsequently, M3 parent magmas were generated below the thinned lithosphere of the Rockall Trough and other basins outboard of the European continental shelf. These magmas were derived from heterogeneous N-MORB-source mantle similar to that which was tapped earlier in the history of the British Palaeogene igneous province.

Fig. 9.

Schematic section from NW to SE across the North Atlantic igneous province at 61 Ma, showing melt streamlines (indicated by ‘melt’) at the centre and periphery of the ancestral Iceland plume and matrix streamlines (‘matrix’) diverging from the deeper portions of the melting regime beneath East Greenland (EG) and the Faeroes. Melt fractions generated beneath the British Isles are second-stage melts, produced from matrices that were variably depleted in incompatible elements by melt extraction close to the plume axis. BI/WG, Baffin Island–West Greenland; EG, East Greenland.

Fig. 9.

Schematic section from NW to SE across the North Atlantic igneous province at 61 Ma, showing melt streamlines (indicated by ‘melt’) at the centre and periphery of the ancestral Iceland plume and matrix streamlines (‘matrix’) diverging from the deeper portions of the melting regime beneath East Greenland (EG) and the Faeroes. Melt fractions generated beneath the British Isles are second-stage melts, produced from matrices that were variably depleted in incompatible elements by melt extraction close to the plume axis. BI/WG, Baffin Island–West Greenland; EG, East Greenland.

The scenario suggested here for the British Isles is akin to that proposed by Watson & McKenzie (1991) and Watson (1993) for melting below Hawaii at the present day. They estimated the mechanical boundary layer beneath Hawaii to be 72 km thick, and the potential temperature of the plume to be 1560°C. Melting in the Hawaiian plume is suggested to commence at a depth of ∼125 km, producing an average melt fraction of just under 7%. One significant difference between the Hawaiian and eastern NAIP melting regimes is that beneath Hawaii, the plume is believed to have attained steady state after some 80 my of activity. In contrast, the ancestral Iceland plume was young at 61 Ma and may have cycled mantle through the melting region at an unusually rapid rate. A second difference in melting regimes may be the thickness of the mechanical boundary layer: parts of the European lithosphere, notably the Lewisian craton of NW Scotland, are likely to have been substantially thicker than the Pacific oceanic plate, and will have restricted the geographical area beneath which melting was possible.

Comparisons may also be made between the British Isles and other plume-generated igneous provinces. For example, White & McKenzie (1995) estimated that basalts from the Deccan and Karoo provinces and the Mackenzie dyke swarm were formed by melting at elevated mantle temperatures (∼1500–1600°C) beneath mechanical lithosphere ∼70–75 km thick. Similar inferences were made by Kent & McKenzie (1994) for flood basalts on the Kerguelen Plateau [Ocean Drilling Program (ODP) Site 747] and Ninetyeast Ridge (ODP Site 756), erupted along the track of the Kerguelen plume. In each case, melting is inferred to have commenced at depths of >100 km, giving rise to modest melt fractions (8–12%, corrected for fractionation).

CONCLUDING SUMMARY

The results of this study may be summarized as follows:

  1. Palaeogene basaltic dykes corresponding in composition to magma types M1 and M2 (Table 1) occur throughout the onshore British Palaeogene igneous province. Magma types M3 and M4 (Table 1) appear to be absent from the southern part of this province, corresponding to the onshore English and Welsh portions and the island of Lundy.

  2. During the Palaeogene, the NW European continental shelf, including the British Isles, was underlain by two mantle sources: ‘Icelandic’ (plume) mantle and an N-MORB source containing blebs of ‘Icelandic’ material. Both sources were tapped during Palaeocene magmatism.

  3. Within the British Isles, the proportion of Palaeocene dykes with ‘Icelandic’ Nb–Y–Zr values (ΔNb > 0) in a given area is extremely variable, ranging from 58% in parts of the Outer Hebrides to 3% on Lundy. Mean values of ΔNb in the British Palaeocene basalts increase with time.

  4. The precise distribution of the ‘Icelandic’ and N-MORB sources beneath the NW European shelf during the Palaeogene is unclear, but was certainly complex. At 61 Ma, the thermal and compositional influence of the ancestral Iceland plume extended at least 1300 km away from the plume centre, i.e. twice the extent of the present-day compositional anomaly.

  5. During the Palaeocene, melting to form magma types M1 and M2 occurred below intact lithosphere (the British Isles). Melting began at depths well in excess of 100 km and extended into the thermal boundary layer of the lithosphere (to ∼75 km depth). The mechanical lithosphere was not eroded. Melting beneath intact lithosphere was made possible by high temperatures (∼1550°C) in the ancestral Iceland plume.

  6. Integration of polybaric melt fractions before ascent through the lithosphere was efficient at first (M1 compositions), but efficiency declined with time (M3). The cause of such variations is not well understood at present.

  7. The parent magmas of British Palaeogene basalts were produced as second-stage melts from residues ascending along streamlines away from the centre of the ancestral Iceland plume. M2 and M3 basalts have incompatible element abundances (notably Ti) indicating that they could not have been produced by remelting of M1-like residues at shallower depths.

  8. The melting regime beneath the British Isles in the Palaeogene shows many similarities to that below Hawaii at the present day: a vigorous plume, intact lithosphere, polybaric melting and average melt fractions of ∼7%.

*Corresponding author. Present address: European and External Funding Office, Coventry University, Priory Street, Coventry CV1 5FB, UK. Telephone: +44-2476-88-8157. Fax: +44-2476-88-8004. e-mail: r.kent@coventry.ac.uk

Extended data set can be found at: http://www.petrology.oupjournals.org

This work was inspired by Mike Norry. James Brodie, Lynne Chambers, Bjorn Hardarson, Andrew Kerr, Mike Norry and Andy Saunders provided helpful discussions during the course of the work. Assistance with XRF and ICP–MS analyses was provided by Dodie James, and Kate Sampson and Tracy Shimmield, respectively. Peter Dagley kindly provided splits of samples from Canna, Mull and Lundy. We thank Lotte Larsen, Ray Macdonald and Joel Baker for constructive reviews. The study was made possible via an NERC Fellowship (GT5/F/92/GS/4) awarded to R.W.K.

REFERENCES

Ade-Hall, J. M., Dagley, P., Wilson, R. L., Evans, A., Riding, A., Smith, P. J., Skelhorn, R. R. & Sloan, T. (
1972
). A palaeomagnetic study of the Mull regional dyke swarm.
Geophysical Journal of the Royal Astronomical Society
 
27
,
517
–545.
Barrat, J. A. & Nesbitt, R. W. (
1996
). Geochemistry of the Tertiary volcanism of Northern Ireland.
Chemical Geology
 
129
,
15
–38.
Berggren, W. A., Kent, D. V., Swisher, C. C. & Aubry, M.-P. (
1995
). A revised Cenozoic geochronology and chronostratigraphy. In: Berggren, W. A., Kent, D. V., Aubry, M.-P. & Hardenbol, J. (eds)
Geochronology, Timescales and Global Stratigraphic Correlation
 . Tulsa, OK: Society for Sedimentary Geology, pp.
129
–212.
Brodie, J. A. (
1995
). Early Tertiary magmatism in the North Atlantic. D.Phil. Thesis, University of Cambridge.
Brodie, J. A. & Fitton, J. G. (
1998
). Composition of basaltic lavas from the seaward-dipping reflector sequence recovered during DSDP Leg 81 (Hatton Bank). In: Saunders, A. D., Larsen, H. C., Wise S. & Allen, J. R. (eds)
Proceedings of the Ocean Drilling Program, Scientific Results
 ,
152
 . College Station, TX: Ocean Drilling Program, pp.
431
–435.
Brodie, J. A. & White, N. (
1995
). The link between basin inversion and igneous underplating. In: Buchanan, J. G. & Buchanan, P. G. (eds)
Basin Inversion. Geological Society, London, Special Publication
 
88
,
21
–38.
Brodie, J. A., Latin, D. & White, N. (
1994
). Rare-earth element inversion for melt distribution: sensitivity and application.
Journal of Petrology
 
35
,
1155
–1174.
Campbell, I. H. & Griffiths, R. W. (
1990
). Implications of mantle plume structure for the evolution of flood basalts.
Earth and Planetary Science Letters
 
99
,
79
–93.
Carter, S. R., Evensen, N. M., Hamilton, P. J. & O’Nions, R. K. (
1979
). Basalt magma sources during the opening of the North Atlantic.
Nature
 
281
,
28
–30.
Chambers, L. M. & Fitton, J. G. (
2000
). Geochemical transitions in the ancestral Iceland plume: evidence from the Isle of Mull Tertiary volcano, Scotland.
Journal of the Geological Society, London
 
157
,
261
–263.
Chambers, L. M., Fitton, J. G. & Pringle, M. S. (
1999
). Spatial and temporal variations in ΔNb: implications for the developing structure of the ancestral Iceland plume.
EOS Transactions, American Geophysical Union
 
80
,
F647
.
Clift, P. D., Carter, A. & Hurford, A. J. (
1998
). The erosional and uplift history of NE Atlantic passive margins: constraints on a passing plume.
Journal of the Geological Society, London
 
155
,
787
–800.
Dagley, P. & Mussett, A. E. (
1981
). Palaeomagnetism of the British Tertiary igneous province: Rhum and Canna.
Geophysical Journal of the Royal Astronomical Society
 
65
,
475
–491.
Dam, G., Larsen, M. & Sonderholm, M. (
1998
). Sedimentary response to mantle plumes: implications from Paleocene onshore successions, West and East Greenland.
Geology
 
26
,
207
–210.
Dickin, A.P. (
1981
). Isotope geochemistry of Tertiary igneous rocks from the Isle of Skye, Scotland.
Journal of Petrology
 
22
,
155
–189.
Eldholm, O. & Grue, K. (
1994
). North Atlantic volcanic margins: dimensions and production rates.
Journal of Geophysical Research
 
99
,
2955
–2968.
Ellam, R. M. (
1992
). Lithosphere thickness as a control on basalt geochemistry.
Geology
 
20
,
153
–156.
England, R. W. (
1992
). The role of Palaeocene magmatism in the tectonic evolution of the Sea of the Hebrides basin: implications for basin evolution on the NW seaboard. In: Parnell, J. (ed.)
Basins on the Atlantic Seaboard: Petroleum Geology, Sedimentology and Basin Evolution. Geological Society, London, Special Publication
 
62
,
163
–174.
Fitton, J. G., Saunders, A. D., Norry, M. J., Hardarson, B. S. & Taylor, R. N. (
1997
). Thermal and chemical structure of the Iceland plume.
Earth and Planetary Science Letters
 
153
,
197
–208.
Fitton, J. G., Saunders, A. D., Larsen, H. C., Hardarson, B. S. & Norry, M. J. (
1998
). Volcanic rocks from the East Greenland margin at 63°N: composition, petrogenesis and mantle sources. In: Saunders, A. D., Larsen, H. C., Wise, S. & Allen, J. R. (eds)
Proceedings of the Ocean Drilling Program, Scientific Results
 ,
152
 . College Station, TX: Ocean Drilling Program, pp.
331
–350.
Ford, T. D. (
1993
). The Isle of Man.
Geologists’ Association Guide
 
46
,
94
pp.
Govindaraju, K. (
1994
). 1994 compilation of working values and sample description for 383 geostandards.
Geostandards Newsletter
 , Special Issue.
158
pp.
Hald, N. & Waagstein, R. (
1991
). The dykes and sills of the early Tertiary Faeroe Island basalt plateau.
Transactions of the Royal Society of Edinburgh, Earth Sciences
 
82
,
373
–388.
Hall, B. D. & White, N. J. (
1994
). Origin of anomalous Tertiary subsidence adjacent to North Atlantic continental margins.
Marine and Petroleum Geology
 
11
,
702
–714.
Hamilton, M. A., Pearson, D. G., Thompson, R. N., Kelley, S. P. & Emeleus, C. H. (
1998
). Rapid eruption of Skye lavas inferred from precise U–Pb and Ar–Ar dating of the Rum and Cuillin plutonic complexes.
Nature
 
394
,
260
–263.
Hardarson, B. S., Fitton, J. G., Ellam, R. M. & Pringle, M. S. (
1997
). Rift relocation—a geochemical and geochronological investigation of a palaeo-rift in northwest Iceland.
Earth and Planetary Science Letters
 
153
,
181
–196.
Hémond, C., Arndt, N., Lichtenstein, U., Hofmann, A. W. & Oskarsson, N. (
1993
). The heterogeneous Iceland plume: Nd–Sr–O isotopes and trace element constraints.
Journal of Geophysical Research
 
98
,
15833
–15850.
Hofmann, A. W. (
1988
). Chemical differentiation of the Earth: the relationship between mantle, continental crust and oceanic crust.
Earth and Planetary Science Letters
 
90
,
297
–314.
Jarvis, G. T. & McKenzie, D. P. (
1980
). Sedimentary basin formation with finite extension rates.
Earth and Planetary Science Letters
 
48
,
42
–52.
Jarvis, K. E. (
1997
). Inductively coupled plasma–mass spectrometry (ICP-MS). In: Gill, R. C. O. (ed.)
Modern Analytical Geochemistry
 . Harlow, UK: Longman, pp.
171
–186.
Kempton, P. D., Fitton, J. G., Saunders, A. D., Nowell, G. M., Taylor, R. N., Hardarson, B. S. & Pearson, G. (
2000
). The Iceland plume in space and time: a Sr–Nd–Pb–Hf study of the North Atlantic rifted margins.
Earth and Planetary Science Letters
  (in press).
Kent, R. W. (
1995
). Magnesian basalts from the Hebrides, Scotland: chemical composition and relationship to the Iceland plume.
Journal of the Geological Society, London
 
152
,
979
–983.
Kent, R. W. & McKenzie, D. (
1994
). Rare-earth element inversion models for basalts associated with the Kerguelen mantle plume.
Mineralogical Magazine
 
58A
,
471
–472.
Kerr, A. C. (
1995
). The geochemistry of the Mull–Morvern Tertiary lava succession, NW Scotland: an assessment of mantle sources during plume related volcanism.
Chemical Geology
 
122
,
43
–58.
Kerr, A. C. (
1995
). The geochemical stratigraphy, field relations and temporal variation of the Mull–Morvern lava succession.
Transactions of the Royal Society of Edinburgh, Earth Sciences
 
86
,
35
–47.
Kerr, A. C. (
1998
). Mineral chemistry of the Mull–Morvern Tertiary lava succession, western Scotland.
Mineralogical Magazine
 
62
,
295
–312.
Kerr, A. C., Kent, R. W., Thomson, B. A., Seedhouse, J. K. & Donaldson, C. H. (
1999
). Geochemical evolution of the Tertiary Mull volcano, western Scotland.
Journal of Petrology
 
40
,
873
–908.
Kirton, S. R. & Donato, J. A. (
1985
). Some buried Tertiary dykes of Britain and surrounding waters deduced by magnetic modelling and seismic reflection methods.
Journal of the Geological Society, London
 
142
,
1047
–1057.
Lewis, C. L. E., Carter, A. & Hurford, A. J. (
1992
). Low temperature effects of the Skye Tertiary intrusions on Mesozoic sediments in the Sea of the Hebrides Basin. In: Parnell, J. (ed.)
Basins on the Atlantic Seaboard. Geological Society, London, Special Publication
 
62
,
175
–190.
Lyle, P. (
1980
). A petrological and geochemical study of the Tertiary basaltic lavas of north-east Ireland.
Royal Society of Dublin, Journal of Earth Sciences
 
2
,
137
–152.
Mattey, D. P., Gibson, I. L., Marriner, G. F. & Thompson, R. N. (
1977
). The diagnostic geochemistry, relative abundance and spatial distribution of high-calcium, low-alkali olivine tholeiite dykes in the Lower Tertiary regional swarm of the Isle of Skye, NW Scotland.
Mineralogical Magazine
 
41
,
273
–285.
McDonough, W. F. & Sun, S.-s. (
1995
). The composition of the Earth.
Chemical Geology
 
120
,
223
–253.
McKenzie, D. & O’Nions, R. K. (
1995
). The source regions of ocean island basalts.
Journal of Petrology
 
36
,
133
–159.
Merriman, R. J., Taylor, P. N. & Morton, A. C. (
1988
). Petrochemistry and isotope geochemistry of early Palaeogene basalts forming the dipping reflector sequence SW of Rockall Plateau, NE Atlantic. In: Morton, A. C. & Parson, L. M. (eds)
Early Tertiary Volcanism and the Opening of the Northeast Atlantic. Geological Society, London, Special Publication
 
39
,
123
–134.
Moorbath, S. & Thompson, R. N. (
1980
). Strontium isotope geochemistry and petrogenesis of the Early Tertiary lava pile of the Isle of Skye, Scotland, and other basic rocks of the British Tertiary Province: an example of magma–crust interaction.
Journal of Petrology
 
21
,
295
–321.
Morton, A. C., Dixon, J. E., Fitton, J. G., Macintyre, R. M., Smythe, D. K. & Taylor, P. N. (
1988
). Early Tertiary volcanic rocks in Well 163/6-1A, Rockall Trough. In: Morton, A. C. & Parson, L. M. (eds)
Early Tertiary Volcanism and the Opening of the Northeast Atlantic. Geological Society, London, Special Publication
 
39
,
293
–308.
Morton, N. (
1987
). Jurassic subsidence history in the Hebrides, NW Scotland.
Marine and Petroleum Geology
 
4
,
226
–242.
Mussett, A. E., Dagley, P. & Eckford, M. (
1976
). The British Tertiary igneous province: palaeomagnetism and age of dykes, Lundy Island, Bristol Channel.
Geophysical Journal of the Royal Astronomical Society
 
46
,
595
–603.
Mussett, A. E., Dagley, P. & Skelhorn, R. R. (
1988
). Timing and duration of activity in the British Tertiary igneous province. In: Morton, A. C. & Parson, L. M. (eds)
Early Tertiary Volcanism and the Opening of the Northeast Atlantic. Geological Society, London, Special Publication
 
39
,
337
–348.
Nadin, P., Kusznir, N. J. & Toth, J. (
1995
). Transient regional uplift in the early Tertiary of the northern North Sea and the development of the Iceland plume.
Journal of the Geological Society, London
 
152
,
953
–959.
Nakamura, N. (
1974
). Determination of REE, Ba, Fe, Mg, Na, and K in carbonaceous and ordinary chondrites.
Geochimica et Cosmochimica Acta
 
44
,
1917
–1930.
Nicholson, H. & Latin, D. (
1992
). Olivine tholeiites from Krafla, Iceland: evidence for variations in melt fraction within a plume.
Journal of Petrology
 
33
,
1105
–1124.
Pearce, J. A. (
1983
). Role of the sub-continental lithosphere in magma genesis at active continental margins. In: Hawkesworth, C. J. & Norry, M. J. (eds)
Continental Basalts and Mantle Xenoliths
 . Nantwich, UK: Shiva, pp.
230
–249.
Pedersen, T. & Ro, H. E. (
1992
). Finite duration extension and decompression melting.
Earth and Planetary Science Letters
 
113
,
15
–22.
Saunders, A. D., Fitton, J. G., Kerr, A. C., Norry, M. J. & Kent, R. W. (
1997
). The North Atlantic Igneous Province. In: Mahoney, J. J. & Coffin, M. F. (eds)
Large Igneous Provinces: Continental, Oceanic and Planetary Volcanism. Geophysical Monograph, American Geophysical Union
 
100
,
45
–93.
Scarrow, J. S. & Cox, K. G. (
1995
). Basalts generated by decompressive adiabatic melting of a mantle plume: a case study from the Isle of Skye, NW Scotland.
Journal of Petrology
 
36
,
3
–22.
Sobolev, A. V. & Shimizu, N. (
1993
). Ultra-depleted primary melt included in an olivine from the Mid-Atlantic Ridge.
Nature
 
363
,
151
–154.
Sobolev, A. V. Gurenko, A. A. & Shimizu, N. (
1994
). Ultra-depleted melts from Iceland: data from melt inclusion studies.
Mineralogical Magazine
 
58A
,
860
–861.
Speight, J. M., Skelhorn, R. R., Sloan, T. & Knaap, R. J. (
1982
). The dyke swarms of Scotland. In: Sutherland, D. S. (ed.)
Igneous Rocks of the British Isles
 . Chichester: John Wiley, pp.
449
–459.
Stein, C. A. & Stein, S. (
1992
). A model for the global variation in oceanic depth and heat-flow with lithospheric age.
Nature
 
359
,
123
–129.
Tegner, C., Lesher, C. E., Larsen, L. M. & Watts, W. S. (
1998
). Evidence from the rare-earth element record of mantle melting for cooling of the Tertiary Iceland plume.
Nature
 
395
,
591
–594.
Thompson, D. B. & Winchester, J. A. (
1995
). Field relationships, geochemistry, and tectonic context of the Tertiary dyke suites in Staffordshire and Shropshire, central England.
Proceedings of the Yorkshire Geological Society
 
50
,
191
–208.
Thompson, R. N. (
1974
). Primary basalts and magma genesis, I: Skye, NW Scotland.
Contributions to Mineralogy and Petrology
 
45
,
317
–341.
Thompson, R. N. & Morrison, M. A. (
1988
). Asthenospheric and lower-lithospheric mantle contributions to continental extensional magmatism: an example from the British Tertiary Province.
Chemical Geology
 
68
,
1
–15.
Thorpe, R. S. & Tindle, A. G. (
1992
). Petrology and petrogenesis of a Tertiary bimodal dolerite–peralkaline/subalkaline trachyte/rhyolite dyke association from Lundy, Bristol Channel, UK.
Geological Journal
 
27
,
101
–117.
Vink, G. E. (
1984
). A hotspot model for Iceland and the Voring Plateau.
Journal of Geophysical Research
 
89
,
9949
–9959.
Vogt, P. R. & Avery, O. E. (
1974
). Detailed magnetic surveys in the north-east Atlantic and Labrador Sea.
Journal of Geophysical Research
 
79
,
363
–389.
Wallace, J. M., Ellam, R. M., Meighan, I. G., Lyle, P. & Rogers, N. W. (
1994
). Sr isotope data for the Tertiary lavas of Northern Ireland: evidence for open-system petrogenesis.
Journal of the Geological Society, London
 
151
,
869
–877.
Watson, S. (
1993
). Rare-earth element inversions and percolation models for Hawaii.
Journal of Petrology
 
34
,
763
–783.
Watson, S. & McKenzie, D. (
1991
). Melt generation by plumes: a study of Hawaiian volcanism.
Journal of Petrology
 
32
,
501
–537.
White, R. S. (
1992
). Crustal structure and magmatism of North Atlantic continental margins.
Journal of the Geological Society, London
 
149
,
841
–854.
White, R. S. & McKenzie, D. (
1989
). Magmatism at rift zones: the generation of volcanic continental margins and flood basalts.
Journal of Geophysical Research
 
94
,
7685
–7729.
White, R. S. & McKenzie, D. (
1995
). Mantle plumes and flood basalts.
Journal of Geophysical Research
 
100
,
17543
–17585.
Wood, D. A. (
1979
). Dynamic partial melting: its applications to the petrogenesis of basalts erupted in Iceland, the Faeroe Islands, the Isle of Skye (Scotland) and the Troodos Massif (Cyprus).
Geochimica et Cosmochimica Acta
 
43
,
1031
–1046.