32. TRACE AND MAJOR ELEMENT GEOCHEMISTRY OF BASALTS FROM LEG 811

Colin Richardson, P. J. Oakley, and J. R. Cann, The University, Newcastle-upon-Tyne2

ABSTRACT

Fifty-two samples of basalt from the four holes drilled on the Leg 81 transect across the Rockall margin were ana-lyzed by X-ray fluorescence for Rb, Sr, Y, Zr, and Nb. On the basis of these results 13 samples were chosen for majorand supplementary trace-element analysis. The results show no progressive change in the character of the volcanism,from Hole 555 in the continental domain through Holes 552 and 553A in the dipping reflector sequence to Hole 554Aon the outer high. Two distinct magma types are present, apparently reflecting heterogeneity of the underlying mantle,but both types are present in both Holes 553A and 555, while Hole 552 and Hole 554 are each composed of a singletype. Both magma types have a clear ocean-floor basalt signature when examined by discrimination diagrams, as doesthe basalt from Deep Sea Drilling Project Site 112, which formed at the same time as the Leg 81 basalts slightly farthersouth along the spreading center. In contrast, the basalts of East Greenland, formed at the same time, are more en-riched in incompatible elements and have a within-plate geochemical signature, as is found in some basalts of Icelandtoday. Clearly the present distinction in geochemistry between the basalts of Iceland and those erupting well south onthe Reykjanes Ridge was already established when continental splitting took place.

INTRODUCTION

One of the major aims of Leg 81 of the Deep SeaDrilling Project (DSDP) was to sample both the seriesof dipping seismic reflectors that occurs in the westernmargin of the Rockall microcontinent (Roberts et al.,1979) and the adjacent outer high, which is thought tobe related to the initial formation of the oceanic crust inthe area. Oceanward-dipping seismic reflectors are nowknown to be a feature characteristic of many passive con-tinental margins (Talwani, Udintsev, et al., 1976; Ger-ard, 1981; Hinz, 1981) and are well seen on the EastGreenland margin (Featherstone et al., 1977), which isthe conjugate margin of the Rockall Plateau. Both thedipping reflectors and the outer high were successfullysampled during Leg 81 and proved to consist for themost part of both submarine and subaerial basaltic lavaflows.

These basalts can potentially provide crucial infor-mation about the evolution of volcanic activity duringcontinental breakup. Previous work shows two distinct(and contradictory) models. One, pioneered by Gass(1970), considers breakup to be exemplified by the south-ern Red Sea and African Rift Valley. Here breakup waspreceded by copious volcanism evolving from alkalineto tholeiitic as the split grew, eventually turning to oce-anic volcanism. The early volcanism would be classifiedas having within-plate geochemistry by discriminationdiagrams such as those of Pearce and Cann (1973) andwas attributed by Gass (1970) and later workers to ther-mal plumes rising from the mantle along the line of thesplit. Such copious precursor volcanism is concentratedat points along the rift, and there are long stretches of

Roberts, D. G., Schnitker, D., et al., Init. Repts. DSDP, 81: Washington (U.S. Govt.Printing Office).

^ Department of Geology, The University, Newcastle-upon-Tyne, NE1 7RU, UnitedKingdom.

young margin (such as that of the central Red Sea) wheresigns of early volcanism of this type are lacking. Theother model, not yet set out so formally, would place thegeneration of oceanic basalts of mid-ocean ridge basalt(MORB) type early in the splitting process. Thus Essonet al. (1975) showed that magmas of MORB type wereavailable in Skye in the British Tertiary Province, eventhough the rifting there never reached continental split-ting, and the eventual split occurred on the Rockall mar-gin, much farther to the west. Drilling in the mouth ofthe Gulf of California on Leg 64 of the DSDP showedthat submarine MORB basalts can be found within afew kilometers of subaerially weathered granitic base-ment (Saunders et al., 1982), although the general appli-cability of this observation might be challenged becauseof the predominantly transform nature of opening inthe Gulf of California. Similarly, of course, the modelof Gass might be criticized for being based on subaerialevidence, while most margins are rapidly submerged andremain so.

In an attempt to shed more light on the problem, andespecially to understand as far as possible the processesoperating in the Rockall area, 52 basalts samples werechosen for analysis from Holes 552, 553A, 554, and 555of Leg 81. The position of the holes is shown in Figure 1.This chapter presents the results of those analyses anddiscusses their relevance to this important problem.

RESULTSThe 52 samples were chosen from the core recovered

in Holes 552, 553A, 554, and 555 using the shipboardcore descriptions. The samples were chosen from thecenters of flow units to avoid as far as possible the ef-fects of secondary alteration. All 52 samples were ana-lyzed by X-ray fluorescence spectroscopy for the ele-ments Pb, Th, U, Rb, Sr, Y, Zr, Mo, and Nb, usingour Philips PW1410 instrument, calibrated with a ba-salt sample spiked with standard additions of the ele-

795

C. RICHARDSON, P. J. OAKLEY, J. R. CANN

Table 1. Rb, Sr, Y, Zr, and Nb analyses for 52samples from Holes 552, 553A, 554A, and555, DSDP Leg 81.

Figure 1. Map showing the locations of holes drilled during DSDPLeg 81.

merits in question and checked with a selection of stan-dard rocks. The detection limit for U, Th, and Pb usingthis method is 4 ppm, whereas for Rb, Sr, Y, Zr, Nb,and Mo it is 0.7 ppm. Precision at the 0- to 5-ppm levelis ±0.3 pm, whereas at the 100-ppm level it is ±2 ppm.On the basis of these analyses, 13 representative sampleswere selected for further analysis by atomic-absorptionspectroscopy, adding major element analyses and a fur-ther range of trace elements (Li, Ni, Co, Cr, Zn, Cu,and V). Details of the analytical methods may be ob-tained from Peter Oakley on request.

The results obtained are given in Table 1 (trace ele-ments) and Table 2 (major elements). Pb, U, Th, andMo were below the limit of sensitivity for the methodand have therefore been omitted from Table 1. Some in-terelement relationships are shown in Table 3.

PETROGRAPHYAlthough the petrography of rocks from Leg 81 is the

subject of detailed study elsewhere (Harrison et al., thisvolume), we undertook a thin-section examination ofour samples to check for evidence of cumulus phasesand secondary alteration.

The rocks form a homogeneous group of fine-grained,aphyric to microphyric, usually vesicular basalts. Mi-crophenocrysts make up no more than 10% of the rocksand are commonly of labradorite and more rarely of au-gite set in a microlitic matrix with iron oxide as a com-mon accessory, particularly in samples from Hole 552.Pseudomorphs after olivine microphenocrysts are scarcebut are present in basalts from every hole. One sample,from the center of a massive flow unit at the bottom ofHole 555 (Sample 555-97-4, 75-77 cm), is a medium-grained subophitic dolerite, containing labradorite, au-gite, and rare pseudomorphed olivines, with accessorymagnetite. The lack of large phenocrysts and the lowabundances of microphenocrysts suggest that the lavasare close to liquid compositions.

The degree of alteration is variable, with samples fromHole 552 showing the most pervasive alteration. Themargins of flows are usually most highly altered, andthe variation we observe in degree of alteration may berelated to flow thickness. The principal secondary phas-

Sample(interval in cm)

Hole 552

21,CC (6-9)22-1, 13-1722-2, 66-6823-1, 133-135

Hole 553A

38-1, 39-4138-2, 75-7639-1, 81-8340-2, 79-8142-2, 67-6943-1, 99-10145-1, 51-5345-3, 79-8146-2, 115-11747-3, 30-3249-3, 75-7750-2, 107-10951-2, 80-8252-2, 68-7053-1, 123-12553-3, 116-11854-1, 94-9654-4, 60-6255-1, 99-10155-2, 75-7755-4, 130-13256-2, 64-6657-1, 106-10858-1, 79-8159-1, 57-5959-4, 64-66

Hole 554A

7-1, 40-427-4, 27-298-1, 85-879-1, 79-8110-1, 69-7111-1, 9-1112-1, 7-913-1, 2-414-1, 2-414-1, 97-99

Hole 555

69-3, 37-3970-1, 109-11176-1, 85-8776-4, 82-8477-2, 65-6782-1, 19-2182-2, 2-482-2, 140-14286-1, 126-12890-3, 125-12793-2, 48-5097-4, 75-77

Rb

1.30.60.92.1

1.22.91.60.60.4

11.920.7

5.21.90.20.80.70.11.51.92.91.81.5

10.43.30.80.71.13.91.62.0

8.97.01.98.19.47.8

117.5

102.4

1.71.31.91.04.45.4

24188.51.13.82.9

Element

Sr

123124116117

10681917777626382697773646769728782796765697681787964

91809493819693889793

106103908874928088

11210010978

Y

59517064

2725283823223224263134242424232732301651272615211421

22222427242730263224

362426262831292822502925

Zr

211206213195

936468

12055776757537576484949496966674545544945484446

52525758616259616163

1015850527168686464

1276750

Nb

6.16.56.36.1

3.23.43.33.13.33.13.93.23.23.93.32.32.53.72.53.72.43.52.32.33.12.72.22.52.73.0

3.93.23.23.53.53.93.93.23.43.6

3.53.32.93.43.02.92.83.63.23.73.13.7

es are brown-yellow smectites and a distinctive green cel-adonite. These occur in the groundmass and as liningsor fillings to vesicles, where the celadonite appears to bethe later phase. The presence of these clay minerals ischaracteristic of low-temperature alteration produced byocean-floor weathering. The dolerite from Hole 555 con-

796

TRACE AND MAJOR ELEMENT GEOCHEMISTRY

Table 2. Major and supplementary trace element analyses of 13 selected samples from Holes 552, 553A, 554A, and 555, DSDP Leg 81.

Sample(interval in cm)

552-22-2,66-68

553A-40-2,79-81

553A-45-3,79-81

553A-51-2,80-82

553A-53-3,116-118

553A-54-4,60-62

553A-59-1,57-59

553A-59-4,64-66

554A-7-4,27-29

554A-14-1,2-4

555-69-3,37-39

555-76-1,85-87

555-82-2,2-4

Major element oxide (wt. %)

Siθ2A12O3

Fe 2O 3

FeOMnOMgOCaONa2OK2OTiO 2

P2O5H 2 O +

Total

Trace elements (ppm)

LiVCrCoNiCuZnF e 2 O 3 T

Mg/Mg + Fe

46.911.87.58

10.190.137.895.532.640.093.300.302.90

99.25

12700

954744

260163

18.90.45

52.514.37.376.330.255.507.612.900.071.420.161.77

100.18

7408

45753792

12614.40.43

49.414.9

5.967.240.167.128.762.460.191.210.021.93

99.35

7435

885265

167157

14.00.50

49.314.32.749.230.277.53

11.52.070.031.070.061.21

99.31

64061365576

161103

13.00.53

49.115.25.436.810.316.758.982.550.031.450.092.54

99.44

5523

525547

110143

13.00.51

49.614.44.668.320.316.730.412.470.041.460.092.21

99.70

6503

495951

144136

13.90.49

47.514.27.675.520.139.967.982.160.040.950.053.28

99.44

6339335

5891

3108613.80.59

49.214.3

3.567.960.268.29

11.41.890.030.940.061.35

99.24

7370337

5295

1529212.40.57

50.114.12.909.270.216.99

11.42.280.151.120.070.49

99.08

8406

605848

146103

13.20.51

48.315.16.107.290.255.05

11.72.380.361.160.131.52

99.34

11423239

6462

154120

14.20.41

49.214.4

3.648.690.236.40

10.92.520.041.710.121.53

99.38

7478212

5578

236111

13.30.49

48.313.93.766.880.237.96

11.92.610.091.060.062.44

99.19

8386349

5275

1808611.40.54

50.012.8

3.076.780.227.75

10.32.222.201.270.082.32

99.01

63841105451

176104

10.60.59

Table 3. Interelement ratios for 13 selected Leg 81 basalts.

Sample(interval in cm) Mg/Mg + Fe Ti/Zr Ti/Y Zr/Y Zr/Nb Ti/V Y/Nb

Hole 552

22-2, 66-68

Hole 553A

40-2, 79-8145-3, 79-8151-2, 80-8253-3, 116-11854-4, 60-6259-1, 57-5959-4, 64-66

Hole 554A

7-4, 27-2914-1, 2-4

Hole 555

69-3, 37-3976-1, 85-8782-2, 2-4

0.45

0.430.500.530.510.490.590.57

0.510.41

0.490.540.59

93

71128131126131130122

129115

101128112

283

224304267322293407267

305219

283246262

3.0

3.22.42.02.62.23.12.2

2.41.9

2.81.92.3

34

39182019191615

1618

291724

28

21171617181715

1717

211720

11

127.59.67.38.65.27.0

6.99.4

109.0

10

tains, in addition to the clays, minor amounts of chlo-rite, which was probably produced during an earlier phaseof high-temperature alteration.

Neither the grade nor degree of alteration are veryhigh, and many trace elements can be expected to re-main nearly immobile under these conditions (Cann,1970; Humphris and Thompson, 1978). However K andRb will certainly have been mobile (Hart, 1971), as to alesser degree may have been Mg and Sr (Matthews, 1971;Staudigel et al., 1980). Elements mobile during ocean-floor weathering must be used with care in any petroge-netic arguments.

GEOCHEMICAL VARIATION WITHIN ANDBETWEEN HOLES

This section examines the geochemical variation with-in holes, to see whether there is any systematic change

during volcanism, and between holes, to investigate wheth-er there is any systematic progression in volcanism alongthe transect that might be related to the tectonic evolu-tion of the Rockall margin. We concentrate on elementslikely to be less mobile during the low-grade alterationthat has affected the rocks.

Of the elements determined in all 52 samples, Zr, Y,and Nb are relatively immobile under these conditions,whereas Sr shows significant mobility only under condi-tions of extreme ocean-floor weathering (Staudigel etal., 1980), or during greenschist facies metamorphism(Pearce and Cann, 1973). Figures 2, 3, and 4 show Y,Nb, and Sr plotted in turn against Zr. On Figure 2 (Yagainst Zr), most of the samples from Holes 553A, 554A,and 555 plot in a single cluster around a Zr value of60 ppm, which is rather low for ocean-floor basalts andat Y about 25 ppm, which is normal. Those from Hole552 are consistently much more enriched in both ele-ments, containing levels of Y, in particular, which, atabout 60 ppm, are abnormally high for ocean-floor ba-salts. Some of the points for Holes 553A and 555 extendfrom the unevolved cluster toward the cluster for Hole552 and suggest that all of the samples might be relatedby fractional crystallization of a single magmatic par-ent. Assuming Rayleigh fractionation and a bulk solid-liquid distribution coefficient (D) of 0.15 for Zr (Pearceand Norry, 1979), such fractional crystallization wouldinvolve 77°/o solidification of the less evolved liquids toproduce the more evolved liquids of Hole 552 and wouldimply a bulk solid-liquid distribution coefficient for Yof 0.39.

That this is probably too simple a conclusion can beseen from a plot of Nb against Zr (Fig. 3). The main,unevolved cluster plots around Zr = 60 ppm, Nb =3.5 ppm, while the evolved cluster is near Zr = 210 ppm,Nb = 6.5 ppm. If the two clusters are related by Ray-leigh fractionation, with DZT = 0.15, this implies the

797

C. RICHARDSON, P. J. OAKLEY, J. R. CANN

70 r

50 100 150

Zr (ppm)

200 250

Figure 2. Plot of Y against Zr for 52 basalt samples from Holes 552, 553A, 554A, and 555, DSDP Leg 81.

7 r

250

Figure 3. Plot of Nb against Zr for 52 basalts samples from Holes 552, 553A, 554A, and 555, DSDP Leg 81.

entirely unrealistic bulk solid-liquid distribution coeffi-cient of 0.6 for Nb, and a change in Zr/Nb from about17 to about 32. In addition, the more evolved samplesfrom Holes 553A and 555 contain the same level of Nbas the less evolved samples. If these too were to be re-lated in a simple Rayleigh fractionation scheme, thiswould imply the even more unrealistic bulk solid-liquiddistribution coefficient for Nb of about 1. Where simi-lar variations in Zr/Nb have been found elsewhere, theyhave been attributed, with supporting arguments fromother elements, to the presence of heterogeneities in themantle (Erlank and Kable, 1976; Macdonald, 1980).

Because of the need for further evidence, the Sr-Zrdiagram must be used (Fig. 4) despite the potential prob-lem of the mobility of Sr. The mobility of Sr can be as-sessed in two ways, first by examining points that areclustered closely in the two preceding diagrams and sec-ond by comparing the relative positions of points in allthree diagrams. Hole 552 and 554A basalts produce tightclusters in both Figures 2 and 3, and the clustering onFigure 4 is as good if not better. The relative position ofpoints is also very similar. Thus the group of Zr-poorpoints from Hole 553A (Zr = 45-50 ppm) lies in thesame relative position on all diagrams. For these rea-

798

TRACE AND MAJOR ELEMENT GEOCHEMISTRY

120 r

100 -

80 -

α 60

40

20 -

/ U / ' '

/ / / ^/ / ^

/

/

Δ Δ

Δ Hole 552• Hole553Ao Hole554A+ Hole 555

i

0 50 100 150 200 250

Zr (ppm)

Figure 4. Plot of Sr against Zr for 52 basalt samples from Holes 552, 553A, 554A, and 555, DSDP Leg 81.

sons, it seems that mobility of Sr has not been impor-tant in this set of samples, except perhaps for one or twopoints that may be aberrant.

If this is accepted, then the diagram can be used tosupplement Figure 3. In particular, the relative placingof the unevolved group, the evolved samples of Holes553A and 555, and the highly evolved group from Hole552 are very similar in both diagrams. This suggestsstrongly that the pattern on the Nb-Zr plot has a defi-nite petrogenetic significance. It appears that the evolvedmagmas of Holes 553A and 555 are the most suitableprecursors for the highly evolved magmas of Hole 552,as they have very similar Zr/Nb ratios and lie in the oth-er diagrams in a suitable relation to each other. If a rela-tionship by Rayleigh fractionation is assumed for thesetwo groups of magmas (specifically, a mean of Samples553A-40-2, 79-81 cm and 555-90-3, 125-127 cm on theone hand, and of all 552 samples on the other), and thebulk solid-liquid distribution coefficient for Nb is takento be zero, then the transition from one group to theother would require a fraction solidified of 0.46, andbulk solid-liquid distribution coefficients of 0.18 for Zr,0.47 for Y, and 0.52 for Sr. These values seem very rea-sonable, although that for Y is somewhat high and thatfor Sr low, suggesting that clinopyroxene would have toform rather a large part of any precipitated solid.

On this basis, the unevolved group of magmas wouldnot be parental to either of the two groups consideredabove but would have a separate origin, probably frommantle of a different composition than that supplyingmagma to the more evolved groups. Two broad groupsof magmas emerge from this analysis, one with Zr/Nb15-20, sampled in Holes 553A, 554A, and 555 and rep-resented by the large unevolved group on Figures 2, 3 and4, and the other with Zr/Nb 30-40, sampled in Holes552, 553A, and 555, represented by the more evolvedgroups. Both types of magma are represented at Sites

553A and 555, where some intermediate basalts are alsofound, possibly the result of mixing of the two types ofmagma in a magma chamber.

This conclusion is reinforced by the results from the13 samples chosen for further analysis (Table 2). Select-ed elements are plotted against Zr in Figure 5. Many ofthe elements analyzed in this further group appear tohave been mobile during alteration. This not only ap-plies to K2O, Na2O, and CaO, which are mobile duringthe formation of clays, and the alteration of plagioclasebut also apparently to both MgO and total iron. MgO isadded to the rock from seawater to make smectites andother sheet silicates during this style of alteration, andFe may be mobilized either in this connection or duringthe formation of pyrite in the basalt. Thus MgO ap-pears to have been added to the basalt from Hole 552,which contains 18.9% of Fe2O3T, as is appropriate forits high content of incompatible traces, but also 7.9%of MgO, rather than 4.5-5%, which would be expectedin lavas with such high Fe2O3T. Iron mobility seems tohave affected particularly the basalts in the unevolvedgroup. The result is that Mg/Mg + Fe cannot be reliedon as an index of evolution, and certainly the plot ofMg/Mg + Fe against Zr shows a scatter of points thatmust in part be caused by this mobility.

However, the other elements plotted are believed tobe more resistant to alteration, and the patterns of plot-ted points are consistent with such a conclusion. Theplots of TiO2, V, and P2O5 all show a very similar distri-bution of points to those of Y and Nb and are clearly ofelements behaving similarly incompatibly. That of P2O5

has little character to it, presumably because P and Zrhave similar distribution coefficients in all of the pro-cesses that have taken place. The plots of TiO2 and Vclosely parallel the pattern of the same points on theNb-Zr plot, and both clearly identify (even if in skeletalform, because of the smaller number of points) the dif-

799

C. RICHARDSON, P. J. OAKLEY, J. R. CANN

100

50

0

•*

f +*o

o t

1

Δ Hole 552• Hole553Ao Hole554A+ Hole 555

0.6 -

tuu

300

Q.

3 200

ò

100

n

M

0

-

° .

Δ Hole 552• Hole553A° Hole554A+ Hole 555

Δ

i

100Zr (ppm)

200

Zr(ppm)

Figure 5. Cr, Ni, V, P 2 O 5 , TiO2, Fe2O31> and Mg/Mg + Fe plotted against Zr for 13 selected samples of basalt from Leg 81.

ferent magma types. Thus the variation in Zr/V ratios isclosely parallel to that of Zr/Nb ratios, both pointingtoward similar conclusions to those drawn above.

The plots for Cr and No also emphasize this variety,although for compatible rather than incompatible ele-ments. Both plots are very alike, and their correspon-dence suggests little mobility of either element. Bothshow that, at levels of Cr and Ni corresponding to mag-nesian basalts, a range of Zr contents must have existed,unless there were, in the early stages of evolution ofthese magmas, large and synchronized changes in thebulk solid-liquid distribution coefficients for both ele-ments. Since this is unlikely, the evidence here, too, sug-gests that the two groups of magmas are not related bycrystal fractionation.

The chemical evidence thus provides two kinds of in-sight into the magmas of Leg 81. First, it emphasizesthat two essentially different types of primary magmawere available in this area, one enriched relative to theother in Zr, Y, Ti, P and Sr, depleted in V, and with sim-ilar Nb content. Intermediate magmas also occur. Sec-ond, it emphasizes the overall unity of the magmatism,since, although Hole 552 shows only the Zr-enrichedtype (with very small basement penetration) and Hole554 only the Zr-poor type, the other two holes (with thegreatest basement penetration), show both magma types

in the same section. The diversity of magma types canprobably, as has been said above, be related to mantleheterogeneity, such as has already been documented onLeg 49 of the DSDP (Cann et al., 1979) and south ofthe Azores (Bougault and Treuil, 1980).

The within-hole variation can be seen from the plotsof Figures 2-5 but is also shown in Figure 6, where twovariables have been chosen to plot against depth in eachhole. The Zr/Nb ratio relates to the magma type present:values less than about 22 represent Zr-poor magmas,and values greater than about 30 Zr-rich magmas. Yabundance relates to the degree of evolution of a givenmagma type. In Hole 552, the magmas are both Zr richand highly evolved, although the depth of penetrationhere was small and the interval from which basalt wasrecovered even smaller. In Hole 553A, the basalts in thelower part of the hole are generally of the Zr-poor typeand become on average gradually more evolved from thebottom of the hole upward for about 100 m. The upperpart of the hole is more varied, with Zr-rich and Zr-poor types of different degrees of evolution present to-gether. Hole 554A is made up entirely of Zr-poor mag-mas, and here the degree of evolution decreases upwardin the hole, in contrast to the lower part of Hole 553A.Hole 555 was sampled at less closely spaced intervals. Inthe upper part, thick volcaniclastics separate units of la-

800

TRACE AND MAJOR ELEMENT GEOCHEMISTRY

Hole552

Zr/Nb Y

40 20 0 20 40 60

Hole 553A

Zr/Nb Y

40 20 0 20 40

Hole 554A

Zr/Nb Y

40 20 0 20 40

Hole 555

Zr/Nb Y

40 20 0 20 40

-g 100

200

300 L

100

200

J300

Figure 6. Zr/Nb and Y plotted against depth in hole for basalt samples from DSDP Leg 81.

va, and no description of the lower part was available tous when we chose the samples for analysis. However, itis clear that here both magma type and degree of evolu-tion vary with depth in the hole, and no regular patternis visible. It is thus not possible to see any consistency ofeither degree of evolution or magma type with depth inthe section. However, these holes have sampled only thevery upper part of the dipping reflector sequence, and amore systematic pattern might be evident if longer sec-tions were available.

The variation of magmas between holes shows noregularity either. This is surprising because of the greatrange in tectonic position of the four holes. Hole 555 isapparently well onto continental crust (Site 555 chapter,this volume), while Hole 544A is on the outer high be-yond the zone of dipping reflectors and clearly in theoceanic domain, if not clearly on oceanic crust (Site 554chapter, this volume). Holes 552 and 553A occupy anintermediate position in the zone of dipping reflectors.Yet, as can be seen from Figure 6 as well as from the pre-vious figures, the characteristics of the magmas sampledfrom Holes 555 and 553A are very similar, as is the ir-regularity with which these magmas occur within the sec-tion. There are only two possible indications that somechanges may be taking place along the transect. One isthe uniformity of the chemistry of Hole 554A, on theouter high. However, the penetration at the hole was notreally deep enough to be sure that this uniformity ischaracteristic of the volcanic section here as a whole,and the magma type found here is also found abundant-ly in Holes 553A and 555. The other is the higher Sr lev-el at Hole 555 than in other Zr-poor magmas in thetransect. This may be related to continental crustal con-tamination and may be a sign of the position of Hole555 in a continental setting. However, the high Sr mightalso be an effect of alteration by seawater.

To summarize, the main result from this assessmentof within-hole and between-hole variation is not thatthere is a variety of magma types available in the tran-sect (although this is the case), but that magmatism alongthe transect was remarkably consistent in overall charac-ter both in time and space. No evolution from one kindof magma to another can be seen in the splitting pro-cess. It appears that as the magmatism began, so it con-tinued from the continental into the oceanic domain.

TECTONIC AFFINITIES OF THE MAGMAS

Given the consistent character of the magmatismthroughout the transect, the next question is whetherthe magmas show any compositionally clear tectonic af-finity. In particular whether they resemble chemicallythe within-plate/hot-spot type of basalt (WPB), or theocean-floor basalts (OFB) of normal mid-ocean ridges.This can be examined using the methods of Pearce andCann (1971, 1973) based on the elements, Cr, Ti, Zr,Nb, Y, and Sr, which are relatively immobile during low-grade alteration.

The tectonic discrimination of the chemical methodsused is not exactly parallel to the tectonic distinctionsdrawn by geophysicists. Thus the WPB class of basaltsis found not only in within-plate environments, such asHawaii, but also in rifting/slow-spreading environments,such as the East African Rift Valley. Even in Iceland,which clearly forms a segment of the mid-ocean ridgesystem, some of the lavas may be classified chemicallyas WPB, apparently because of the anomalous charac-ter of the mantle beneath Iceland. Other lavas in Ice-land, especially the more magnesian class of basalts (Si-gurdsson et al., 1978; Wood, 1976, 1978), plot as OFBand along the Reykjanes Ridge south of Iceland there isa clear transition to a consistently OFB character, paral-leling the transition described by Schilling (1973) in, for

801

C. RICHARDSON, P. J. OAKLEY, J. R. CANN

example, the degree of light rare earth element (LREE)enrichment. It should be stressed, however, that tectonicdiscrimination by the Pearce and Cann methods doesnot necessarily relate to the degree of LREE enrichmentin basalts.

The OFB class of basalts is represented mostly alongspreading centers as MORB in one or another of its va-rieties. However, it can also be erupted in zones of rift-ing as a precursor to spreading which may never takeplace (Esson et al. 1975).

Figure 7 is a modification into cartesian coordinatesof the Ti-Zr-Y triangular plot of Pearce and Cann (1973),to which it is topologically identical (Cann and Heath,1976). It is essentially used to distinguish WPB basaltsfrom other basalts (OFB and arc basalts). All 13 sam-ples for which analyses of the three elements are avail-able fall outside of the WPB field. The nearest point tothe field boundary is Sample 553A-59-1, 57-59 cm, anunevolved basalt from near the bottom of Hole 553A.All of the other points fall well away from the bound-ary.

Once the non-WPB basalts are separated, they can beplotted on a Zr/Sr versus Ti/Sr diagram (Fig. 8), againa cartesian transformation of the triangular Ti-Zr-Srdiagram of Pearce and Cann (1973). As expected, thepoints all fall into the OFB field rather than the arcfield. If the basalts of Hole 555 have picked up somecontinental crustal Sr, it is not enough to push themover the field boundary.

Thus all of the basalts erupted on the transect, whetherof the Zr-rich or Zr-poor groups, have close affinitieswith the basalts of mid-ocean ridges, showing that thiskind of basalt was available early in the history of split-ting within the continental domain, through the time ofdevelopment of the dipping reflectors, to the building ofthe outer high.

REGIONAL COMPARISONS OF LEG 81 BASALTS

To assess the significance of the geochemistry of Leg81 basalts, it is necessary to compare them with other

e553A

G Greenland

2 i-

5

4

3

2

1

0

\-\

\\\\

\

\ /\ /\ /

\ /\ /\ /V

\\\

\\ Q.

-

i i

A

/

/

A

+

Λ\ 8

\ 7

\\\

B \i \

•

9 /

c ^

I

G

αG

\\

A\\

1

G

•

D

\

\

i \

7 Ho8 Ho

e407e408

9 Hole 409

G Lower series

• Middle series

• Upper seriesFaeroes

200 300

Ti/Y

B

o

••

1

100

Ti/Sr

Δ

•

O

+•

Hole 552

Hole 553A

Hole 554A

Hole 555

Upper series-

Faeroes

Figure 7. Zr/Y versus Ti/Y for selected basalts from the North Atlan-tic region. (A,C = fields of volcanic arc basalts; B = field ofocean floor basalts; D = field of within-plate basalts.)

200

Figure 8. Zr/Sr versus Ti/Sr for selected basalts from the North At-lantic region. (A = field of volcanic arc basalts; B = field ofocean floor basalts.)

basalts from the North Atlantic region. Particularly im-portant in this respect are comparisons with basalts fromthe Faeroe Islands and Eastern Greenland, with whichthe Rockall basalts are broadly coeval. The other com-parison must be with basalts currently being eruptedalong the Mid-Atlantic Ridge.

The Faeroe Islands are mostly composed of a 3-kmthick sequence of plateau basalts, with a minimum ageof 50-60 m.y. (Tarling and Gale, 1968). The stratigra-phy, petrology, and geochemistry of the lavas have beendescribed by a number of workers, all of whom accept athreefold division into lower, middle, and upper series,separated by minor unconformities (Noe-Nygaard andRasmussen, 1968; Schilling and Noe-Nygaard, 1974; Bol-lingberg et al., 1976). The lower and middle series aredominated by Fe- and Ti-rich quartz tholeiites, enrichedin incompatible elements and LREE. The upper series iscomposed predominantly of olivine tholeiites depletedin incompatible elements and LREE, although enrichedlavas do occur (Gariepy et al., 1983).

The early Tertiary basalts of East Greenland are bestdeveloped on the Blosseville coast between Scoresby Sundand Kangerdlugssuaq (Brooks et al., 1976). The truethickness of the pile is difficult to determine because offaulting, and estimates vary from over 7 km (Soper etal., 1976) to an average thickness of 2 km (Nielson andBrooks, 1981). The pile certainly thins to a feather edgeinland, where it overlies continental crust. Where the con-tinental crust ends to the seaward is not entirely clear. Itis certain, however, that the volcanic episode lasted lessthan 3 m.y. (Soper et al., 1976) and that it occurredimmediately prior to magnetic Anomaly 24 (54 m.y.),which is found just offshore (Larsen and Jacobsen, 1982).The Rockall margin basalts have the same relationship

802

TRACE AND MAJOR ELEMENT GEOCHEMISTRY

to Anomaly 24 and thus appear to have erupted at thesame time as the East Greenland basalts. The Greenlandpile is divisible into a lower series of compound flowsand interbedded pyroclastics, which contains some pi-crites, overlain by a generally thicker series of simpleflows, the Plateau basalts, dominated by saturated Fe-and Ti-rich tholeiites (Brooks et al., 1976; Brooks andNeilson, 1982).

Representative analyses of basalts from the Faeroesand East Greenland are given in Table 4 together withmeans of the analyses from Hole 552 (highly evolved Zr-rich magma type) and Hole 554 (unevolved, Zr-poormagma type), representing the extremes of variation withinthe Leg-81 transect. There is a good correspondence be-tween the major elements of all of the basalts, taking in-to account the probable mobility of Mg in the Leg 81basalts, except for the upper Faeroese basalts and theEast Greenland picrite. Incompatible elements give asomewhat different story, however. At an equivalent stageof evolution of the basalts, (measured by Y, Cr, and Nicontents, Mg/Mg + Fe or total iron) both the Faeroesand the East Greenland basalts are more enriched in in-compatibles such as Zr, Sr, Ti, and Nb. Again mantleheterogeneity seems to be the cause of this difference,with the mantle supplying the Leg 81 basalts being sig-nificantly depleted in incompatibles relative to that sup-plying the basalts further north. It is interesting, though,that Zr/Nb is approximately the same in the Zr-poor la-vas of Leg 81 as it is in the basalts of East Greenland.

The presence of picrites at the base of the East Green-land lava pile is unique in the North Atlantic region, al-though they do occur in other flood basalt sequences[e.g., Deccan (Krishnamurthy and Cox, 1977), WestGreenland (Clarke and Pederson, 1976)], where they

have been interpreted as representing parental magmaderived more or less directly from the mantle. Althoughpicrites were not recovered during Leg 81, the drillingonly sampled at most the top 300 m of the lava sequencewhich may be as much as 7 km thick.

The basalts may be further compared on the tecton-ic discrimination diagrams. The Greenland and Faeroesanalyses are plotted on Figure 7, where they clearly fallin the WPB field. The exception is the upper series lavasof the Faeroes which lie in the OFB field, confirmingtheir similarity to MORB (Gariepy et al., 1983). It seemsfrom this plot that the differences between the Leg 81basalts and those from East Greenland and from thelower and middle series in the Faeroes are not simply amatter of enrichment or depletion in incompatible ele-ments, but that there are more fundamental differences,similar to those between ocean-floor basalts and within-plate basalts. These differences are confirmed by theother element ratios of Table 5, where data for compara-tive basalts are listed.

Table 5 also gives comparative data for the basalt re-covered from the bottom of DSDP Hole 112 (ShipboardScientific Party, 1972). This site is on crust that musthave been very close to the Leg 81 crust in Anomaly-24time and formed at nearly the same time, although itnow lies on the opposite side of the Rekyjanes Ridgespreading center. This basalt has many characteristics incommon with the Leg 81 basalts, although a major-ele-ment analysis is not available. On Figure 7 it plots onthe OFB side of the dividing line, close to several of theLeg 81 samples.

To extend this comparison further, it is useful to ex-amine basalts that have been erupted along the ridgecrest from Iceland to the Charlie-Gibbs Fracture Zone

Table 4. Chemical comparison of Leg 81 basalts with basalts from East Greenlandand the Faeroe Islands.

1

Hole 552Mean

Major element oxides (%)

S1O2A12O3

Fe2O3

FeOMnOMgOCaONa2OK2OTiO2

P2O5

46.911.807.58

10.190.137.895.532.640.093.300.30

Trace elements (ppm)

SrYZrNbCuCoNiVCrFe 2O 3 T

Zr/Nb

12061

2066.3

2604744

7009518.933

Mg/Mg + Fe 0.45

2

Hole 554Mean

49.214.614.508.270.236.02

11.552.450.251.140.08

902558

3.51506155

414200

13.7170.47

3 4 5

Faeroese basalts

Lower

48.812.907.608.300.205.90

10.402.400.402.90—

30037

22015

30070

12043012016.8_0.41

Middle

49.514.504.308.400.207.20

10.902.300.402.30—

25025

1708

20050

120280230

13.6

0.51

Upper

47.414.804.108.600.209.60

11.801.800.301.40—

15028

1102

19060

260290430

13.6_0.58

6 7 8 9

East Greenland basalts

47.313.404.908.620.186.31

10.792.350.262.640.30

27234

16912

2114595

33017514.5140.46

47.313.194.828.540.216.55

11.402.480.262.480.25

25333

13710

2074793

342160

14.4140.47

47.713.335.588.180.206.31

10.742.030.262.380.26

33530

1489

1554848

34310914.6160.46

47.110.7213.8

0.1611.809.442.00.382.120.31

192—

138—

13062

480270730

13.8—0.63

Notes: 1,2 = Means of analyses from Holes 552 and 554 (Tables 1 and 2, this paper); 3, 4, 5 = Lower, middle,and upper series Faeroese basalts (Bollingberg et al., 1976; Gariépy et al., 1983); 6, 7, 8 = East Greenlandbasalts from Scoresby Sund (6), Widemann's Fjord (7), and Nansen's Fjord (8) (Brooks et al. 1976); 9 =East Greenland picrite from Mikis Fjord (Brooks et al., 1976).

803

C. RICHARDSON, P. J. OAKLEY, J. R. CANN

Table 5. Trace elements in Leg 81 basalts compared with other NorthAtlantic examples.

Sample Zr Nb Zr/Nb Zr/Y Ti/Y Ti/Zr

Leg81,Hole552 a

Leg 81, Hole 553Aa

Leg 81, Hole554A a

Leg81,Hole555 a

Faeroes lower series"Faeroes middle series"Faeroes upper seriesb

E. Greenland l c

E. Greenland 2C

E. Greenland 3C

E. Greenland picrite1*Leg 12, Hole 112βLeg 49, Hole 407'Leg 49, Hole 408 f

Leg 49, Hole 409 f

Iceland iron-rich*5

Iceland magnesian

19,800

7,3006,800

8,100

17,400

13,800

8,400

15,800

14,900

14,300

7,300

6,000

13,800

10,100

8,100

18,800

8,800

120

75

90

93

300

250

150

272

253

335

192

94

193

201

103

299

194

61

26

25

29

37

25

28

34

33

30_

19

34

25

29

48

23

206

60

58

69

220

170

110

169

137

148

138

45

152

115

83

206

77

6.3

3.0

3.5

3.3

15

8

2

12

10

9_

1.0

18

15

10

27

10

33

20

16

21

10.8

14

33

14

14

16_

45

8.4

7.6

8.3

7.6

7.7

3.4

2.3

2.3

2.4

4.3

4.4

2.3

5.0

4.2

4.9_

2.4

4.5

4.6

2.9

4.3

3.3

325

281

272

279

460

549

291

465

450

476—

316

407

403

279

392

383

96

121

117

116

79

81

76

93

108

96

52

133

151

146

98

91

114

a Leg 81 basalts (this chapter).b Faeroe basalts (see Table 4, nos. 3-5).c East Greenland basalts (see Table 4, nos. 6-8).d East Greenland picrite (see Table 4, no. 9).e Basalt from DSDP Hole 112, Leg 12 (54o01'N, 46°36.2'W); (Laughton, Berggren et al.,

1972).Means of basalts from Reykjanes Ridge sites of Leg 49, all at about 63°N (Tarney et al.,1979).

8 Representative basalts from Iceland from Wood (1976): B30, typical iron-rich basalt.Representative basalts from Iceland from Wood (1976); U2, typical magnesian basalt.

since Anomaly-24 time. Extensive studies have been con-ducted on the basalts of Iceland. These can be consid-ered to fall into two broad classes (Sigurdsson et al.,1978)., a group rich in iron and titanium (FETI basalts)and a magnesian group, which sometimes shows primi-tive enough characteristics to be considered to have beenderived directly from the mantle. A member of each classhas been selected from the analyses of Wood (1976) andthese are included in Table 5 and plotted on Figure 7.Both show WPB-OFB transitional characteristics, withthe FETI basalt plotting on the dividing line and themore magnesian type within the OFB field. Their Zr/Nb ratios are considerably lower than for Leg 81.

A complete series of samples has been dredged fromthe spreading center between Iceland and the Charlie-Gibbs fracture zone, and has been the subject of inten-sive investigation by Schilling and co-workers (Schilling,1973; Schilling and Sigurdsson, 1979). Unfortunately,only partial analyses of this collection have been pub-lished, and these do not include the critical elements ofTable 5. Schilling and Sigurdsson (1979) relate the geo-chemistry to a number of factors, including liquidustemperature of the basalts. Their Figure 3 shows somechemical parameters plotted against latitude, and veryclearly demonstrates the change in rare-earth patternsfrom Iceland southward. One of their samples, fromabout 57°N, must be very close to the point on thespreading center corresponding to the position at whichthe Leg 81 crust was formed. It has a LREE-depletedpattern, with (La/Sn‰ = 0.4, which suggests a MORBgeochemistry for the basalt, although, as was said earli-er, the correlation between REE pattern and tectonic af-finity is far from perfect.

Other evidence comes from the basalts drilled onDSDP Leg 49 at Sites 407, 408, and 409 on a transectwestward from the crest of the Rekyjanes Ridge at about63°N, relatively near Iceland. Means are given for eachsite in Table 5, taken from Tarney et al., 1979. In several

ways the analyses are similar to those from Iceland, par-ticularly in the Zr/Nb ratios which for both Sites 407-409 and Iceland lie close to 8. On Figure 7, the oldersites (Site 407 in crust of age about 40 m.y. and Site 408in crust about 20 m.y. old) plot close to the WPB/OFBboundary and close to the East Greenland basalts andthe iron-rich Icelandic basalt. The mean for Site 409,however, plots well within the OFB field, close to theLeg 81 points.

It is quite clear that the Leg 81 lavas are significantlydifferent from those of East Greenland, the Faeroes,and Iceland, which are, however, very similar to eachother. This is to be expected since all three areas areclosely associated with the Greenland-Iceland-FaeroesRidge, which has been a topographic high probably con-tinuously during the opening of the North Atlantic/Norwegian Sea (Talwani, Udintsev, et al., 1976). On anypredrift reconstruction of the North Atlantic the crustdrilled during Leg 81 would have lain well to the south-west of the Greenland-Iceland-Faeroes area on crustlacking the topographic and magmatic anomalies asso-ciated at present with Iceland. The other basalts form-ing at that latitude then and now (DSDP Site 112 andthe Trident sample from 57°N) appear to share many ofthe characteristics of the Leg 81 basalts and are differentfrom the basalts erupted further to the north. The ba-salts of the Leg 49 transect, lying closer to Iceland thanto the Leg 81 sites, share many features in common withIceland but, in Site 409 at least, have moved from theWPB/OFB transitional geochemistry of Iceland into theOFB field.

CONCLUSIONS

The basalts drilled on Leg 81 are derived from twodistinct types of magma, one relatively rich in Zr andother incompatible elements and the other relatively poorin these elements. Hole 552 contains a very evolved Zr-rich magma type throughout the very short section pen-etrated. Hole 553A contains both magma types, withthe Zr-poor type the most abundant and the Zr-richtype concentrated more in the upper part of the section.Hole 554A basalts are uniformly of the Zr-poor type,while Hole 555 again contains both types, although with-out any preferential distribution of either within the sec-tion. The two types of magma cannot be related by crys-tal fractionation and must reflect heterogeneity of themantle source. Although two magma types are present,there is no systematic variation in magma type along theLeg 81 transect, and so no discernible evolution in mag-matism during continental splitting. Both magma typeshave ocean-floor basalt affinities, and it seems as thoughthis magma type was readily available for eruption veryearly in the splitting process, even at Site 555, which liesessentially within the continental domain. There is nosign of precursor within-plate or plume-type magmatism.

Comparison with other basalts from the North At-lantic province shows that the basalts of East Greenlandand the Faeroes, although erupted at the same time andpossessing very similar major-element characteristics, havea distinct trace-element signature with a strong with-

804

TRACE AND MAJOR ELEMENT GEOCHEMISTRY

in-plate or plume character. This character is shared bysome of the present basalts of Iceland and, to a lesserdegree, by basalts drilled just to the south of Iceland onDSDP Leg 49. Thus the character of the basalts eruptedduring splitting at those latitudes was probably alreadyinfluenced by the same factors that give Iceland itstopographic and magmatic anomaly at the present time.There too the character of the magmatism continues to-day as it was established when splitting occurred.

At the latitude of the Leg 81 transect, DSDP Hole112 basalts show very similar geochemical characteris-tics to the basalts of the transect, and the indications arethat chemically similar basalt is still being erupted at theridge crest today.

ACKNOWLEDGMENTS

We would like to thank the chief scientists and the entire staff in-volved in Leg 81 for the opportunity to contribute to this volume. Inparticular the advice and encouragement of Dr. David Roberts is greatlyappreciated. In addition we gratefully acknowledge the constructivecriticism of Dr. B. Weaver, who reviewed the manuscript. Finally wethank Paula Weiss of Lamont-Doherty Geological Observatory forsupplying additional material at short notice, Christine Jeans for draw-ing the diagrams, and Elizabeth Walton for typing the manuscript.Colin Richardson gratefully acknowledges receipt of a N.E.R.C. re-search studentship (GT4/81/GS/76).

REFERENCES

Bollingberg, H., Brooks, C. K., and Noe-Nygaard, A., 1976. Trace el-ement variations in Faeroese basalts and their possible relation-ships to ocean floor spreading history. Bull. Geol. Soc. Den., 24:55-60.

Bougault, H., and Treuil, M., 1980. Mid-Atlantic Ridge: Zero-agegeochemical variations between Azores and 22°N. Nature (Lon-don), 286:209-212.

Brooks, C. K., and Nielson, T. F. D., 1982. The East Greenland conti-nental margin: A transition between oceanic and continental mag-matism. J. Geol. Soc. London, 139:265-275.

Brooks, C. K., Nielson, T. F. D., and Peterson, T. S., 1976. The Blosse-ville Coast Basalts of East Greenland: their occurrence, compo-sition, and temporal variations. Contrib. Mineral. Petrol., 58:279-292.

Cann, J. R., 1970. Rb, Sr, Y, Zr and Nb in some ocean floor basalticrocks. Earth Planet. Sci. Lett., 10:7-11.

Cann, J. R., and Heath, R. A., 1976. Some trace elements in basaltsfrom Leg 34. In Yeats, R. S., Hart, S. R., et al., 1976. Init. Repts.DSDP, 34: Washington (U.S. Govt. Printing Office), 289-292.

Cann, J. R., Tarney, J., Varet, J., and Wood, D. A., 1979. Mantle het-erogeneity in the North Atlantic: evidence from Leg 49 geochemis-try. In Luyendyk, B. P., Cann, J. R., et al., Init. Repts. DSDP, 49:Washington (U.S. Govt. Printing Office), 841-850.

Clarke, D. B., and Pederson, A. K., 1976. Tertiary volcanic provinceof West Greenland. In Esher, A., and Watt, W. S. (Eds), Geology ofGreenland: Copenhagen (Geological Survey of Greenland), 36-85.

Erlank, A. J., and Kable, E. J. D., 1976. The significance of incom-patible elements in Mid-Atlantic ridge basalts from 45 °N with par-ticular reference to Zr/Nb. Contrib. Mineral. Petrol., 54:281-291.

Esson, J., Dunham, A. C , and Thompson, R. N., 1975. Low alkali,high calcium olivine tholeiite lavas from the Isle of Skye, Scotland.J. Petrol., 16:488-497.

Featherstone, P. S., Bott, M. H. P., and Peacock, J. H., 1977. Struc-ture of the continental margin of southeastern Greenland. Geo-phys. J.R. Astron. Soc, 48:15-17.

Gariepy, C , Ludden, J., and Brooks, C , 1983. Isotopic and trace ele-ment constraints on the genesis of the Faeroes lava pile. EarthPlanet. Sci. Lett., 63:257-272.

Gass, I. G., 1970. The evolution of volcanism in the junction area ofthe Red Sea, Gulf of Aden and Ethiopian rifts. Philos. Trans. R.Soc. Lond., Ser. A., 267:369-381.

Gerard, R., 1981. Structure of the continental margin off S.W. Af-rica. Proc. Hedberg Symposium AAPG, Galveston, Texas, Jan.1981.

Hart, S., 1971. K, Rb, Cs, Sr and Ba contents and Sr isotope ratios ofocean floor basalts. Philos. Trans. R. Soc. London, Ser. A, 268:573-587.

Hinz, K., 1981. A hypothesis on terrestrial catastrophes: wedges ofvery thick oceanward dipping layers beneath passive continentalsignificance. Geol. Jahrb., Reihe E, 22:3-28.

Humphris, S. E., and Thompson, G., 1978. Trace element mobilityduring hydrothermal alteration of oceanic basalts. Geochim. Cos-mochim. Acta, 42:127-136.

Krishnamurthy, P., and Cox, K. G., 1977. Picrite basalts and relatedlavas from the Deccon Traps of western India. Contrib. Mineral.Petrol., 62:53-75.

Larsen, H. C , and Jacobsen, N., in press. Sea floor spreading anom-alies off East Greenland between 60°N and 70°N. Earth Planet.Sci. Lett.

Macdonald, R., 1980. Trace element evidence for mantle heterogene-ity beneath the Scottish Midland Valley in the Carboniferous andPermian. Philos. Trans. R. Soc. London, Ser. A, 297:245 -257.

Matthews, D. H., 1971. Altered basalts from Swallow Bank, an abyss-al hill in the N.E. Atlantic, and from a nearby seamount. Philos.Trans. R. Soc. London, Ser. A, 268:551-573.

Nielson, T. F. D., and Brooks, C. K., 1981. The East Greenland riftedcontinental margin: a reinterpretation of the coastal flexure. J.Geol. Soc. London, 138:559-568.

Noe-Nygaard, A., and Rasmussen, J., 1968. Petrology of a 3000 m se-quence of basaltic lavas in the Faroe Islands. Lithos, 1:286-304.

Pearce, J. A., and Cann, J. R., 1971. Ophiolite origin investigated bydiscriminant analysis using Ti, Zr and Y. Earth Planet Sci. Lett.,12:339-349.

, 1973. Tectonic setting of basic volcanic rocks determinedusing trace element analyses. Earth. Planet. Sci. Lett., 19:290-300.

Pearce, J. A., and Norry, M. J., 1979. Petrogenetic implications of Ti,Zr, Y and Nb variation in volcanic rocks. Contrib. Mineral. Pet-rol., 69:33-47.

Roberts, D. G., Montadert, L., and Searle, R. C , 1979. The westernRockall Plateau: stratigraphy and structural evolution. In Monta-dert, L., Roberts, D. G., et al., Init. Repts. DSDP, 48: Washington(U.S. Govt. Printing Office), 1061-1088.

Saunders, A. D., Fornari, D. J., and Morrison, M. A., 1982. Thecomposition and emplacement of basaltic magmas produced dur-ing the development of continental-margin basins: the Gulf of Cal-ifornia, Mexico. J. Geol. Soc. London, 139:335-346.

Schilling, J.-G., 1973. Iceland mantle plume: geochemical study ofReykjanes Ridge. Nature (London), 242:565, 571.

Schilling, J.-G., and Noe-Nygaard, A., 1974. Faroe-Iceland plume:rare earth evidence. Earth Planet. Sci. Lett., 24:1-14.

Schilling, J.-G., and Sigurdsson, H., 1979. Thermal minima along theaxis of the Mid-Atlantic Ridge. Nature (London), 282:370-375.

Shipboard Scientific Party, 1972. Site 112. In Laughton, A. S., Berg-gren, W. A., et al., Init. Repts. DSDP, 12: Washington (U.S. Govt.Printing Office), 161-253.

Sigurdsson, H., Schilling, J.-G., and Meyer, P. S., 1978. Skagi andLangjökull volcanic zones in Iceland: 1. Petrology and structure.J. Geophys. Res., 83:3971-3982.

Soper, N. J., Higgins, A. C , Downie, C , Matthews, D. W., andBrown, P. E., 1976. Late Cretaceous-early Tertiary stratigraphy ofthe Kangerdlugssuaq area, East Greenland, and the age of theopening of the N. Atlantic. J. Geol. Soc. London, 132:85-104.

Staudigel, H., Frey, F. A., and Hart, S. A., 1980. Incompatible traceelement geochemistry and 87/86Sr in basalts and correspondingglasses and palagonites. In Donnelly, T., Francheteau, J., Bryan,W, Robinson, P., Flower, M., Salisbury, M., et al., Init. Repts.DSDP, 51, 52, 53, Pt. 2: Washington (U.S. Govt. Printing Office),1137-1141.

Talwani, M., Udintsev, G., et al., 1976. Init. Repts. DSDP, 38: Wash-ington (U.S. Govt. Printing Office).

Tarling, D. H., and Gale, N. H., 1968. Isotopic dating and paleo-magnetic polarity in the Faeroe Islands. Nature (London), 218:1043-1044.

805

C. RICHARDSON, P. J. OAKLEY, J. R. CANN

Tarney, J., Saunders, A. D., Weaver, S. D., Donellan, N. C. B., and Wood, D. A., 1976. Spatial and temporal variation in the trace ele-Hendry, G. L., 1979. Minor-element geochemistry of basalts from ment geochemistry of the Eastern Iceland flood basalt succession.Leg 49, North Atlantic Ocean In Luyendyk, B. P., Cann, J. R., et J. Geophys. Res., 81:4353-4360.al., 1979. Init. Repts. DSDP, 49: Washington (U.S. Govt. Printing , 1978. Major and trace element variations in the Tertiary la-Office), 657-691. vas of eastern Iceland and their significance with respect to the Ice-

Tarney, J., Wood, D. A., Saunders, A. D., Cann, J. R., and Varet, J., land geochemical anomaly. J. Petrol., 19:393-436.1980. Nature of mantle heterogeneity in the North Atlantic: evi-dence from deep sea drilling. Philos. Trans. R. Soc. London, Ser.A, 297:179-202. Date of Acceptance; October 15, 1983

806