Role of subducted sediments in island-arc magmatism: constraints from REE patterns

Earth and Planetary Science Letters, 54 ( 1981 ) 423-430 Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands

423

[2]

Role of subducted sediments in island-arc magmatism: constraints from REE patterns Scott M. McLennan and S.R. Taylor Research School of Earth Sciences, A tzs'tralian National UnicersiO,, Canberra, A.C.T. (Australia) Received February 18, 1981 Revised version received May 7, 1981

Continental sedimentary rocks of post-Archean age are characterized by europium depletion relative to the other REE. Typical values for E u / E u * are 0.65+0.05 (where Eu* is the theoretical value for no chondrite-normalized Eu anomaly). Basalts and andesites from island-arc suites rarely display significant europium anomalies. Calculations indicate that the m a x i m u m amount of sediment that can be admixed without producing a signature of Eu depletion is: (a) 10% for a MORB source: (b) 1% for primitive mantle or a single-stage depleted mantle, and (c) 0. I-0.3% for a two-stage depleted mantle.

1. Sediment subduction

Few subjects in the earth sciences have received more attention than that of island-arc magmatism. It is generally held that continental accretion occurs as a result of such igneous processes [1,2] although it is not clear if they have been responsible for crystal growth throughout geological time [3,4]. In any context, island-arc magmatism is fundamental to the understanding of the differentiation history, of the earth. A long-standing problem has revolved around the role of sedimentary rocks, of continental origin, in subduction and the subsequent development of island-arc igneous rocks. A brief survey of the literature indicates that estimates of the amount of sediment incorporated into the subducted slab range from 0 to 20%. The exact value is of crucial importance. For example, if no continental material is recycled into the mantle, then it is quite clear that the continents have grown throughout earth history in a continuous or quasi-continuous fashion [5,6, also see reference 4]. On the other hand, if significant amounts of continental material are recycled into the mantle, then models involving very early for-

mation of the continental crust, with subsequent large-scale recycling are valid [7,8]. In the basalt-andesite-dacite-rhyolite volcanic association of island-arcs, the basaltic rocks are likely to provide the least equivocal information concerning the nature of the source. Island-arc basalts are enriched in large-ion lithophile elements (LIL) such as K, Ba, U, Sr, Pb and light rare earths (LREE), and depleted in the high field strength ions (eg. Zr, Nb,Hf) when compared to normal mid-ocean ridge basalts (N-type MORB). This indicates that island-arc basalts cannot be explained easily by simple partial melting of the mantle and that a more complex origin involving a LIL-enriched component is called for [9-11 ]. Sedimentary rocks are commonly appealed to as the LIL-enriched component. Historically, the task of quantitatively constraining the role of sediment in subduction has fallen to the isotopic geochemists. The Pb isotopes appear particularly sensitive and it has been noted that island-arc volcanics have Pb-isotopic characteristics intermediate between oceanic sediments and MORB, thus indicating a fairly small component ( ~ 2%) of sediment may have been incorpo-

0012-821X/81/0000-0000/$02.50 ~ 1981 Elsevier Scientific Publishing Company

424 rated or mixed with their source (e.g. [12,13]). On the other hand, it has been noted [10,11] that LIL-enriched types of ocean basalts (E-type MORB) have Pb isotopic characteristics which encompass island-arc volcanics and in some instances are even more radiogenic than pelagic sediments, thus giving an alternative explanation for the Pb isotopic characteristics of island-arc volcanics. One unambiguous example of crustal (perhaps sedimentary) contamination has been noted in the Banda Arc where 87Sr/86Sr143Nd/144Nd of volcanic rocks form a linear trend between the Sunda Arc to the west (apparently derived from uncontaminated mantle) and continental crust [14]. It has been noted, however, that calc-alkaline volcanics erupted in continental settings display similar behaviour in terms of Sr and Nd isotopes [10]. After reviewing the isotopic evidence, Perfit et al. [10] concluded, " . . . the Pb-, Srand Nd-isotopic data do n o t rigorously constrain ... the existence or amount of sedimentary contamination..." [see also reference 9]. Less work has been done to apply geochemical constraints to this problem [15,16]. A feature of sedimentary rocks which has not been previously emphasized in constraining sediment subduction is that the average Post-Archean sedimentary REE pattern (indicative of upper crustal abundances) is characterized by a significant negative Eu anomaly [2-4,17-21] with an average E u / E u * of about 0.65 ~ 0.05 (where Eu* is the theoretical value for Eu where there is no chondrite-normalized Eu anomaly). On the other hand, basaltic magmas derived from the mantle, including island-arc basalts, have no systematic Eu anomaly [22,23]. Thus, if sediments are added to some mantle source (or a partial melt of the mantle), they would impose a negative Eu anomaly, the magnitude of which would be dependent on the nature of the source and the amount of sediment added. Partial melts from this material would retain their negative Eu anomaly signature during subsequent melting events [24]. The importance of the negative Eu anomaly in sedimentary rocks in addition to other characteristic sedimentary signatures such as T h / U ratios and boron contents was briefly commented on by Taylor and McLennan [4] and Taylor et al. [25].

They suggested that contributions of more than 5 10% sediment in the formation of island-arc magmas would be detectable. In this paper we present detailed calculations which more rigorously constrain the role of sediments, of continental origin, in the formation of island-arc volcanics.

2. REE in island-arc volcanics

The common basalts, basaltic andesites and andesites of island-arc suites rarely display europium anomalies, and averages of many analyses in this laboratory have E u / E u * values of 1.00 ~ 0.05 which is within our analytical uncertainty. Where negative or positive Eu anomalies are encountered, as in the South Sandwich Islands basalts, these are readily attributed to removal or addition of plagioclase. The parental magmas do not exhibit europium anomalies (e.g. [9]). Negative Eu anomalies commonly occur in the acidic end members of the calc-alkaline suite (e.g. rhyolite). These are either due to fractionation processes [24,26-28], or to a separate origin. It is commonly held that the voluminous acidic ignimbrites and rhyolites, restricted to continental regions, result from intra-crustal melting. In this event, their characteristic Eu depletion results from retention of plagioclase (and Eu) in the lower crust. We exclude these rocks from the subsequent discussion and concentrate on the basalts, basaltic andesites and andesites for which mantle origins are unequivocal. To minimise the effects of secondary fractionation, we further concentrate upon island-arc basalts, although the conclusions extend to andesites as well. Table 1 clearly shows that island-arc basalts are not significantly anomalous with respect to Eu. In calculations of the type presented here, it is important to determine the magnitude of the Eu anomaly which would be identifiable. From purely statistical considerations, one could confidently identify anomalous Eu behaviour of about 10% by analytical techniques such as neutron activation and spark source mass spectrometry, while anomalous behaviour of about 5% could be identified confidently by the more precise techniques of isotope dilution mass spectrometry. When fairly large

425 TABLE 1 R a r e earth elements in basalts from island-arc settings (concentration in p p m )

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Yb LaN/Yb N Eu/Eu*

1

2

3

4

5

6

5.2 14.6 2.0 9.2 2.4 0.86 2.7 0.48 3.0 0.69 2. l 2.0

14.6 32.1 4.0 17.6 4.0 1.3 4.0 0.68 4.0 0.89 2.5 2.4

23.1 53.6 5.5 21.3 4.1 1.3 3.5 0.54 3.0 0.67 1177 1.5

4.0 I 1.0 1.7 8.6 2.5 0.91 3.0 0.55 3.6 0.79 2.3 2.2

4.4 12 1.8 8.5 2.4 0.85 2.8 0.47 3.0 0.65 1.9 1.8

1.8 1.03

4.1 0.98

10.4 1.02

1.2 1.01

1.7 1.00

7

3.84 2.33 2.13

4.78 12.2 1.69 8.23 2.21 0.79 2.52 0.43 2.82 0.63 1.85 1.78

2.7 ~ 1.04

1.8 1.02

21.1 13.5 3.42 1.23 3.81

a Estimated. 1 = average of 4 tholeiitic basalts, Sunda Arc [28]: 2 - a v e r a g e of 4 calc-alkaline basalts, Sunda Arc [28]; 3 - 1 high-K t a l c - a l k a l i n e basalt, Sunda Arc [28]; 4 - average of 3 M U W - g r o u p basahs, Witu Islands [26]; 5 = average of 2 island-arc basahs, N e w Britain [29]: 6 = average of 6 basalts, South Shetland Islands [27]: 7 - a v e r a g e of 12 island-arc basahs, New Britain island-arc [25].

numbers of samples are involved, however, these values would probably diminish significantly. We would suggest that if Eu/Eu* was less than 0.95 in island-arc basalts, it would not have escaped detection and the following calculations are made on the basis of this assumption.

3. REE in sediments The uniformity of REE abundances in PostArchean sedimentary rocks is well established [1720]. It is generally accepted that the average REE pattern in fine-grained sedimentary rocks closely

TABLE 2 R a r e earth elements in pelagic sediments (concentration in p p m )

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Yb Lay/Yb N Eu/Eu*

1

2

3

4

Average pelagic

PAAS

26 7.2 30 5.8 1.6 6.5 1.1 1.0 3.2 3.0

45.5 101 43 8.35 1.85 1.42 3.82

58 60 14 3.6 15 13 7.2 7.3

28 24 5.1 1.2 4.8 4.2 2.3 2.4

42

3.8

38 80 8.9 32 5.6 1.1 4.7 0.77 4.4 1.0 2.9 2.8

7.5 0.67

9.2 0.65

5.9 0.79

8.1 0.65

5.4 0.75

7.9 0.73

41 8.0 1.8 8.3 a

Estimated. 1 = average of 7 ocean s e d i m e n t s (one s a m p l e with very large Eu u n c e r t a i n t y was excluded) [17]; 2 average of 50 pelagic s e d i m e n t s [32]; 3 = a v e r a g e of 3 low-latitude pelagic clays [33]; 4 = a v e r a g e of 6 high-latitude pelagic clays [33]. Average p e l a g i c = w e i g h t e d average of c o l u m n s 1, 2, 3, 4. P A A S = P o s t - A r c h a e a n average A u s t r a l i a n Shale [20].

426

4. Mixing calculations ®

-o

•

PAAS

L Dy

I Ho

LO0

o= E

X Q. o.

I0

4 I Lo

I Ce

J Pr

I NB

I Srn

I Eu

I Gd

I Tb

I Er

I Yb

Fig. 1. Chondrite-normalized REE patterns of pelagic sediments (from Table 2). Also included is the REE pattern for PAAS (Table 2). All patterns are characterized by light rare earth element enrichment and significant negative Eu anomalies.

reflects the pattern in the upper continental crust [2-4,21,30]. One such average is the Post-Archean average Australian Shale (PAAS) determined by Nance and Taylor [20, see table2]. The abundances in this average is essentially the same as other composite samples such as the NorthAmerican Shale Composite (NASC) [30] and the European Shale composite (ES) [31]. Table 2 and Fig. 1 illustrate several average analyses of pelagic sediments and compare them to PAAS. All are characterized by L R E E enrichment and significant negative Eu anomalies [30]. For the calculations below, PAAS is used as the sedimentary component since the REE pattern is better constrained (i.e. data on more elements available) and probably best reflects the upper continental crust. The pelagic sediments may have marginally lower L a y / Y b N and higherZREE and E u / E u * than PAAS but it is not certain if these differences are significant. If the average pelagic sediment listed in Table 2 is used in the following calculations, the effects noted would be even more severe, because of their higher absolute abundances. It should be noted, however, that such calculations do not apply to first-cycle volcanogenic sediments which may be derived from the arc itself, since R E E patterns o f such material simply reflect the volcanic source [34]. On the other hand, detritus from more differentiated arcrelated plutonic rocks or late-stage tufts commonly have negative Eu anomalies and thus are relevant to the present discussion.

To examine the effects of sediment subduction on REE patterns, we have considered four possible primary sources for island arc volcanics: (1) the subducted oceanic crust (i.e. MORB); (2) primitive mantle; (3) single-stage depleted mantle (i.e. similar to the source of MORB); and (4) two-stage (highly) depleted mantle (i.e. MORB source following MORB extraction). The values for the various components are listed in Table 3, and illustrated on chondrite-normalized diagrams in Fig. 2. Chondrite values are from Taylor and McLennan [4], and are also listed in Table 3. The value for MORB was arbitrarily taken to be 15 times chondritic for the H R E E and 7 times chondritic for La. These values are comparable to typical N-type MORB [35-38] and if anything, have somewhat higher abundances than an average MORB. Values for primitive mantle were taken

I

I

I

I

I

I

J

I

I

I

I

L

I

I

I

I

I

I00

v

I0 L,, q~

g

PM 1.0

E o. Q.

0.1

L

I

I

LoCePr Nd

L

L

SmEuGdTbDyHoEr

I

Yb

Fig. 2. Chondrite-normalized REE patterns of possible sources for island-arc magmas. Included are mid-ocean ridge basalt (MORB), primitive mantle (PM), single-stage depleted mantle (DM) and two-stage (highly) depleted mangle (HDM). Also included is the REE pattern for PAAS. Data from Tabled (see text for discussion).

6.1 l 15.3 2.12 10.2 3.26 1.16 4.32 0.82 5.48 1.24 3.65 3.63 1.14 0.945

0.47 1.00

LaN/YbN9.2 E u / E u * 0.65

0.9 MORB 0.1 PAAS

2.57 8.13 1.37 7.82 3.00 1.17 4.28 0.83 5.60 1.27 3.73 3.72

MORB

38 80 8.9 32 5.6 1.1 4.7 0.77 4.4 1.0 2.9 2.8

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Yb

PAAS

1.00 1.00

0.551 1.436 0.206 1.067 0.347 0.130 0.459 0.087 0.572 0.128 0.374 0.372

Primitive mantle (PM)

1.58 0.955

0.925 2.222 0.293 1.376 0.400 0.140 0.501 0.094 0.610 0.137 0.399 0.396

0.99 PM 0.01 PAAS

0.33 1.00

0.184 0.622 0.110 0.676 0.277 0.113 0.428 0.082 0.556 0.127 0.373 0.372

Depleted mantle (DM)

0.96 0.953

0.562 1.42 0.198 0.989 0.330 0.123 0.471 0.089 0.594 0.136 0.398 0.396

0.99 D M 0.01 PAAS

0.15 1.00

0.0110 0.0479 0.0096 0.0604 0.0277 0.0122 0.0490 0.0104 0.0724 0.0166 0.0498 0.0496

Highly depleted mantle (HDM)

Mixing calculations involving Post-Archean average Australian shale (PAAS) and various island-arc sources (concentration in ppm)

TABLE 3

1.07 0.923

0.0870 0.208 0.0274 0.124 0.0388 0.0144 0.0583 0.0119 0.0811 0.0186 0.0555 0.0551

0.998 H D M 0.002 PAAS

-

0.367 0.957 0.137 0.711 0.231 0.087 0.306 0.058 0.381 0.0851 0.249 0.248

Chondrite

428 i

i

i

i

i

i

i

i

i

i

i

i

0 . 9 MORB 't- 0 . 1 P A A S

models of mixing phenomenon in subduction zones are not constrained and extremely difficult to test.

I0

5. Discussion

"1o J= ¢,J

__

0,99

PM "t- 0 . 0 1 P A A S

0.99

DM -t" 0.01 P A A S

E E.~1.0 o. ¢x 0 , 9 9 8 H D M + O.O02PAAS

I

I

I

I

LO

Ce

Pr

Nd

I

I

Sm Eu

I

I

I

I

I

I

Gd

Tb

Dy

Ho

Eu

Yb

Fig. 3. Chondrite-normalized REE patterns showing the effects of mixing PAAS to the possible sources of island-arc magmas (see Table 3). Mixing proportions shown result in E u / E u * ~ 0.95. It is suggested that such Eu anomalies would probably be detected in rocks derived by partial melting of such sources.

at 1.5 times chondritic. The values for depleted mantle (DM) was taken to be a factor of ten lower than MORB for H R E E with slightly greater LREE depletion. Such values are consistent with the nodule data [39]. Values for strongly depleted mantle ( H D M ) was taken to be 0.2 X chondrite for the L H R E E and L a N / Y b N =0.15. Such values are consistent with those given by Kay [40]. The effects of mixing continentally-derived sediment to the various sources are shown in Table 3 and Fig. 3. If it is assumed that E u / E u * < 0.95 would be detected in island-arc basalts, then for the values adopted, the maximum sediment that could be accommodated in the source of island-arc volcanics is as follows: (1) 10% for a MORB source; (2) 1% for primitive mantle; (3) 1% for a single stage depleted mantle [DM] source; and (4) 0.2% for two-stage depleted mantle [HDM] source. Such calculations assume thorough mixing of the various components prior to melting since such conditions are called for by adherents of crustal recycling models [7,8,40]. Whether such mixing takes place is questionable. It is well known that mantle inhomogeneities persist at very small scales, arguing against efficient mixing in the mantle [41]. If sediments were subducted, they might well remain as discrete bodies. At this stage, conceptual

Present-day thoughts on the origin of island-arc volcanics appear to be in a state of flux and recent reviews [10,11] have emphasized the uncertainties in all of the presently available models. An origin of island-arc volcanics by direct melting of subducted ocean crust [42] received early support but more recent geochemical and isotopic constraints indicate such a source probably can be discounted [10,11,40] even if an LIL-enriched component, such as sediments, is added. Generation of island-arc volcanics from primitive mantle appears inconsistent with the Nd isotope data [9,29,43]. Similarly, generation of island-arc volcanics from highly depleted mantle (i.e. H D M of this study) such as that suggested by Green [44] can also be discounted on the basis of geochemical balance arguments [40]; Thus the extreme models diSCussed in this paper are probably least relevant (however, cf. [40]). Although no concensus is available, a popular model for the generation of island-arc volcanics involves melting of depleted mantle, probably above the subduction zone, which has some added LIL-enriched component [9-11]. Clearly, from the calculations discussed above, models of this nature, which appeal to sediment as the LIL-enriched component, are restricted to about 1% sediment. Recently, Kay [40] has proposed a complex model for the origin of island-arc volcanics whereby small amounts of sediment (0.1-0.125%), variable amounts of ocean crust (MORB) and seawater-derived material such as K and Rb are added to highly depleted peridotite and melted to variable degrees to form the different rock types found in island-arc regimes. Although this model appears to encounter considerable difficulty in explaining a number of trace element and isotopic characteristics in island-arc rocks [10,11,16] it is not clear if the constraint from Eu anomalies is exceeded. The limit on sediment subduction indicated for this model from Table 3 and Fig. 3 is 0.2%. This value depends crucially on the average

429

value of highly depleted mantle chosen, a value not well known. More realistically, it can probably only be said that the upper limit on sediment subduction for such models lies in the range 0.10.3%. The value for sediment subduction adopted by Kay [40] for the origin of an island-arc tholeiite was 0.1-0.125% but he also adopted very low REE abundances. Thus, the Eu/Eu* for the melt in his model island-arc tholeiite would be expected to have Eu/Eu* ~0.93-0.95 and might be expected to be detected. His model compositions which also involve components of MORB, cannot easily be tested from these constraints.

6. Conclusions The amount of sedimentary rock which can be included in the source of island-arc magmas is reasonably constrained by the Eu anomaly that such mixing would impart on the source compositions. For direct melting of MORB, the least likely candidate, the fraction of sediment can only be constrained to about 10%. Models which appeal to melting of primitive mantle or depleted mantle (similar to the source of MORB) are constrained to about 1% sediment, models which involve depleted mantle [44] are constrained to about 0.10.3% sediment, but variations on this model, involving complex mixing-melting events [40] cannot be stringently tested from these constraints.

Acknowledgements We are grateful to R.J. Arculus and M.R. Perfit for several helpful discussions and for reading an earlier draft of this paper. Gilbert Hanson, Chris Hawkesworth and Keith O'Nions provided perceptive reviews. We also thank Carmel Neagle and Gail Stewart for assistance in the preparation of the manuscript.

References 1 S.R. Taylor, The origin and Tectonophysics 4 (1967) 17-34.

growth

of

continents,

2 S.R. Taylor, Island-arc models and the composition of the continental crust, Am. Geophys. Union, Maurice Ewing Ser. 1 (1977) 325-335. 3 S.R. Taylor and S.M. McLennan, The rare earth element evidence in Precambrian sedimentary rocks: implications for crustal evolution, in: Precambrian Plate Tectonics, A. Kr6ner, ed., (Elsevier, Amsterdam, 1981). 4 S.R. Taylor and S.M. McLennan, The composition and evolution of the continental crust: rare earth element evidence from sedimentary rocks, Philos. Trans. R. Soc. London, Ser. A. 5 S. Moorbath, Ages, isotopes and evolution of Precambrian continental crust, Chem. Geol. 20 (1977) 151-187. 6 S. Moorbath, Age and isotope evidence for the evolution of continental crust, Philos. Trans. R. Soc. London, Ser. A, 288 (1978) 401-413. 7 R.L. Armstrong, A model for Sr and Pb isotope evolution in a dynamic earth, Rev. Geophys. 6 (1968) 175-199. 8 R.L. Armstrong, Radiogenic isotopes: the case for crustal recycling on a near-steady-state no-continental growth Earth, Philos. Trans. R. Soc. London, Ser. A (in press). 9 C.J. Hawkesworth, R.K. O'Nions, P.J. Pankhurst, P.J. Hamilton and N.M. Evensen, A geochemical study of island-arc and back-arc tholeiites from the Scotia Sea, Earth Planet. Sci. Lett. 36 (1977) 253-260. 10 M.R. Perfit, D.A. Gust, A.E. Bence, R.J. Arculus and S.R. Taylor, Chemical characteristics of island-arc basalts: implications for mantle sources, Chem. Geol. 30 (1980) 227-256. 11 R.J. Arculus, Island arc magmatism in relation to the evolution of the crust and mantle, Tectonophysics (in press). 12 "R.W. Kay, S.-S. sun and C.-N. Lee-Hu, Pb and Sr isotopes in volcanic rocks from the Aleutian Islands and Pribilof Islands, Alaska, G e o c h i m Cosmochim. Acta 42 (1978) 263 -273. 13 S.-S. Sun, Lead isotopic study of young volcanic rocks from mid-ocean ridges, oceanic islands and island arcs, Philos. Trans. R. Soc. London, Ser. A (in press). 14 D.J. Whitford, W.M. White, P.A. Jazek and I.A. Nicholls, N d isotope composition of Recent andesites from Indonesia, Carnegie Inst. Washington, Yearb. 78 (1979) 304308. 15 S.E. Church, Limits of sediment involvement in the genesis of orogenic volcanic rocks, Contrib. Mineral. Petrol. 39 (1973) 17-32. 16 R.J. Arculus and R.W. Johnson, Island-arc magma, sources: a geochemical assessment of the roles of slab-derived components and crustal contamination (submitted to Geochem.

J.). 17 T.R. Wildeman and L.A. Haskin, Rare earth elements in ocean sediments, J. Geophys. Res. 70 (1965) 2905- 2910. 18 T.R. Wildeman and L.A. Haskin, Rare earths in Precambrian sediments, Geochim. Cosmochim. Acta 37 (1973) 419-438. 19 L.A. Haskin, T.R. Wildeman, F.A. Frey, K.A. Collins, c.R. Keedy and M.A. Haskin, Rare earths in sediments, J. Geophys. Res. 71 (1966) 6091-6105. 20 W.B. Nance and S.R. Taylor, Rare earth patterns and

430 crustal evolution, I. Australian post-Archean sedimentary rocks, Geochim. Cosmochim. Acta 40 (1976) 1539- 1551. 21 S.R. Taylor, Chemical composition and evolution of the continental crust, The Earth: Its Origin, Structure and Evolution, M.W. McElhinny, ed. (Academic Press, London, 1979) 353-376. 22 A.E. Bence, T.L. Grove and J.J. Papike, Basalts as probes of planetary interiors: constraints on the chemistry, and mineralogy of their source regions, Precambrian Res. 10 (1980) 249-279. 23 A.E. Bence, T.L. Grove, J.J. Papike and S.R. Taylor, Basalts as probes of planetary interiors: constraints from major and trace element chemistry, in: NASA Planet. Basaltic Volcanism Project (Lunar Planetary Institute, Houston, Texas, in press). 24 G.N. Hanson, Rare earth elements in petrogenetic studies of igneous systems, Annu. Rev. Earth Planet. Sci. 8 (1980) 371-406. 25 S.R. Taylor, R.W. Johnson, R.J. Arculus and M.R. Perfit, The island-arc basalt reference suite, in: NASA Planet. Basaltic Volcanism Project, (Lunar and Planetary Institute, Houston, Texas, in press). 26 R.W. Johnson and R.J. Arculus, Volcanic rocks of the Witu Islands, Papua New Guinea: the origin of magmas above the deepest part of the New Britain Benioff zone, Bull. Volcanol. 41 (1978) 609-655. 27 S.D. Weaver, A.D. Saunders, R.J. Pankhurst and J. Tarney, A geochemical study of magmatism associated with the initial stages of back-arc spreading, Contrib. Mineral. Petrol. 68 (1979) 151-169. 28 D.J. Whitford, I.A. Nicholls and S.R. Taylor, Spatial variations in the geochemistry of Quaternary lavas across the Sunda Arc, Java and Bali, Contrib. Mineral. Petrol. 70 (1979) 341-356. 29 D.J. DePaolo and R.W. Johnson, Magma genesis in the New Britain Island Arc: constraints from Nd and Sr isotopes and trace element patterns, Contrib. Mineral. Petrol. 70(1979) 367 379. 30 L.A. Haskin and T.P. Paster, Geochemistry and mineralogy of the rare earths, in: Handbook on the Physics and Chemistry of Rare Earths, Vol. 3. Non-Metallic Compounds, I, K.A. Gschneidner and L. Eyring, eds. (North-Holland, Amsterdam, 1979) Chapter 21,1-80. 31 M.A. Haskin and L.A. Haskin, Rare earths in European

shales: a redetermination, Science 154 (1966) 507 509. 32 D.Z. Piper, Rare earth elements in ferromanganese nodules and other marine phases, Geochim. Cosmochim. Acta 38 (1974) 1007-1022. 33 T. Shimokawa, A. Masuda and K. Izawa, Rare earth elements in the top samples of the cores from the Pacific Ocean floor, Geochem. J. 6 (1972) 75-81. 34 W.B. Nance and S.R. Taylor, Rare earth element patterns and crustal evolution, II. Archean sedimentary, rocks from Kalgoorlie, Australia, G e o c h i m Cosmochim. Acta 41 (1977) 225-231. 35 R.A. Ayuso, A.E. Bence and S.R. Taylor, Upper Jurassic tholeiitic basalts from DSDP Leg 11, J. Geophys. Res. 81 (1976) 4305-4325. 36 B.-M. Jahn, J. Bernard-Griffith, R. Charlot, J. Cornichet and F. Vidal, Nd and Sr isotopic compositions and REE abundances of Cretaceous MORB (Holes 4170 and 418A, Legs 51, 52, and 53), Earth Planet. Sci. Lett. 48 (1980) 171-184. 37 R.W. Kay and N.J. Hubbard, Trace elements in ocean ridge basalts, Earth Planet. Sci. Lett. 48 (1980) 171-184. 38 S.-S. Sun, R.W. Nesbitt and A.Y. Sharaskir, Geochemical characteristics of mid-ocean ridge basalts, Earth Planet. Sci. Lctt 44 (1979) 119- 138. 39 E. Jagoutz, J. Palme, H. Baddenhausen, K. Blum, M. Cendales, G. Dreibus, B. Spettel, V. Lorenz and H. Wanke, The abundances of major, minor and trace elements in the earth's mantle as derived from primitive ultramafic nodulus, Proc. 10th Lunar Planet. Sci. Conf. 10 (1979) 2031 - 2050. 40 R.W. Kay, Volcanic arc magmas: Implications of a meltingmixing model for element recycling in the crust-upper mantle system, J. Geol. 88 (1980) 497-522. 41 A.W. H o d m a n n and S.R. Hart, An assessment of local and regional isotopic equilibrium in the mantle, Earth Planet. Sci. Lett. 38 (1978) 44-62. 42 T.H. Green and A.E. Ringwood, Origin of the calc-alkaline igneous rock suite, Earth Planet. Sci. Lett. 1 (1966) 307-316. 43 D.J. DePaolo and G.J. Wasserburg, The source of island arcs as indicated by Nd and Sr isotopic studies, Geophys. Res. Left. 4 (1977) 465 468. 44 D.H. Green, Experimental testing of "equilibrium" partial melting of peridotite under water-saturated, high-pressure conditions, Can. Mineral. 14 (1976) 255-268.