Journal of Petrology | Pages |
© 1998 Oxford University Press |
We have always carefully weighed the criticisms of our colleagues, notably Dr Dorrit Jacob and Professor Emil Jagoutz, and welcome the opportunity to debate in print the issues they have raised. Our program of study for Yakutian eclogites has encompassed many years and involved comprehensive mineralogical and chemical studies of many eclogite xenoliths from several kimberlite pipes, including the Udachnaya pipe
(Jerde et al., 1993;
Snyder
et al., 1993,
, 1995,
, 1997;
Sobolev et al., 1994), the Obnazhennaya pipe
(Qi et al., 1994,
, 1997), and the Mir pipe
(Beard et al., 1996), as well as on silicate inclusions in diamonds taken from within eclogites from Mir and Udachnaya (
Taylor et al., 1996b). Such a program, involving a nearly overwhelming body of data, has allowed our group to look deeply into the issue of eclogite genesis, and, necessarily, has led to some complex interpretations. These interpretations are at odds with the earlier work of
Jacob et al., (1994), which comprised eight samples from one kimberlite pipe. Contrary to the statements of
Jacob et al., (1998), we will show that:
One might ask why the genesis of eclogites is so important as to warrant such a comment and reply presented here. Eclogites may represent some of the earliest evidence of subduction and plate tectonics in the early Earth (e.g.
Helmstaedt & Doig, 1975;
Jagoutz
et al., 1984;
MacGregor & Manton, 1986). In contrast, eclogites could also represent an important early mantle component of the Earth (e.g.
Shervais et al., 1988;
Smyth
et al., 1989). In fact, some researchers have speculated that eclogite xenoliths may even represent fragments of accretion of the early Earth (e.g.
Anderson, 1989;
McCulloch, 1989). That eclogitic material may make up a large part of the present-day mantle
(Anderson, 1989) also makes understanding its genesis worth while. Furthermore, the purported abundance of eclogitic material residing in the mantle sets important constraints on the role of subduction and recycling of oceanic crust during the evolution of the Earth's mantle. The crux of a large proportion of the discussion and criticisms of
Jacob et al. (1998) hinges on the oxygen isotope data for Yakutian eclogites. However, to state that our oxygen isotope dataset [as presented by
Snyder et al., (1995)] is `only fragmentary' and involves samples that have been `only sparingly analysed' is either an unfair misrepresentation of our body of work or a misunderstanding of the known, high-temperature, oxygen isotopic fractionation between minerals, or both.
Snyder et al., (1995) presented 19 whole-rock, 19 clinopyroxene, five garnet, and one kyanite oxygen isotopic analyses of Udachnaya eclogites (a total of 44 separate analyses) determined in the laboratories of R. N. Clayton at the University of Chicago. Furthermore, our group has published 13 whole-rock, 13 clinopyroxene, and 15 garnet oxygen isotopic analyses (a total of 41 separate analyses) from eclogites of the Mir kimberlite
(Beard et al., 1996), and four garnet and six clinopyroxene analyses from Obnazhennaya eclogites
(Qi et al., 1997), let alone several other analyses from Obnazhennaya and Zagadochnaya kimberlites that remain unpublished. In all, nearly 100 analyses have been published by our group, compared with 14 analyses by
Jacob et al., (1994).
Because of the difficulties inherent in analyzing garnets by conventional oxygen isotopic methods, our group analyzed a clinopyroxene separate and whole-rock split for the Udachnaya eclogites, analyzing the occasional garnet separate to insure that equilibrium was maintained between garnet and clinopyroxene. As we have conclusively shown for the Mir eclogites
(Beard et al., 1996), expected high-temperature, equilibrium, oxygen isotopic fractionations [as per
Chiba et al., (1989) and
Rosenbaum et al., (1994)] do persist between clinopyroxene and garnet, such that clinopyroxene is always higher than garnet in [delta]18O by 0·1-0·4%o. Thus, if hand-picked, clean, unaltered, mineral separates are used (and they were), only one or the other analysis, garnet or clinopyroxene, is required to effectively calculate the oxygen isotopic composition of the other mineral. The four eclogites from Udachnaya in which we analyzed both garnet and clinopyroxene are confirmation of this high-temperature equilibrium fractionation between these two minerals. Clinopyroxene is heavier by 0·15, 0·20, 0·31, and 0·11 over garnet in samples U-79, U-86, U237, and U-281, respectively
(Snyder et al., 1995). The data of
Jacob et al. (1994) further confirm this oxygen isotopic fractionation, as clinopyroxene is heavier by 0·42, 0·32, 0·36, 0·27, 0·31, and 0·50%o over garnet in their samples from Udachnaya. One need not analyze both clinopyroxene and garnet to obtain an accurate estimate of the precursor eclogite oxygen isotopic composition. Furthermore, we have reanalyzed many of the samples determined by conventional methods, and published by
Snyder et al., (1995), by the laser-fluorination method of
Valley et al., (1995). The data determined by the laser-fluorination method (
Taylor et al., 1996b) agree with those previously published
(Snyder et al., 1995). Thus, all eight of their samples and all 19 of our samples from Udachnaya must be included in any discussion of oxygen isotopes in eclogites from this pipe. With this in mind, we have replotted fig. 1 of
Jacob et al. (1998) with all Udachnaya eclogite data (
Fig. 1), a discussion of which is presented below.
Correlations of [delta]18O with chemical parameters (Fe, Ca, Sm/Nd, and 143Nd/144Nd) that might suggest `seawater alteration before subduction' of an oceanic crustal fragment are not found in the complete (i.e. including our data) Udachnaya dataset
(Snyder et al., 1995, fig. 9;
Snyder et al., 1997). Thus, the interpretation of the Udachnaya eclogites as fragments of subducted oceanic crust cannot be predicated on these grounds alone. Again, we state that the trends of
Jacob et al., (1994) do not exist when all of the data are considered [also see fig. 9 of
Snyder et al., (1995)].
We have stated that we do see a `broad trend of increasing [delta]18O with increasing 87Sr/86Sr values', when all of the data are considered
(Snyder
et al., 1995). However, such a trend can be produced by a variety of methods, including kimberlite contamination of the eclogite in transit to the surface and continental crustal addition (see below). It was the supporting correlations of [delta]18O with Fe, Ca, Sm/Nd, and 143Nd/144Nd that gave the arguments of
Jacob et al. (1994) their weight. These correlations are no longer visible with the complete dataset [again, see fig. 9 of
Snyder et al., (1995)]. We have neither `fail[ed] to recognize these correlations' nor `co-opt[ed]' their interpretations. These correlations do not exist and, thus, interpretations based on these correlations are not supported. Metasomatism is an important and pervasive mantle process which can potentially compromise the integrity of our mantle sample. Nearly all whole-rock eclogites show evidence of metasomatism in the form of secondary minerals (such as phlogopite) in the groundmass and in alteration of clinopyroxene. However, careful hand-picking and acid leaching of ultrapure mineral separates can often `see through' such patent metasomatic effects (e.g.
Zindler & Jagoutz, 1988;
Snyder
et al., 1993,
, 1997). In a trace- and major-element comparative study of pairs of garnet-clinopyroxene inclusions from several Yakutian diamonds, we have shown that cryptic metasomatism has not affected many of the primary minerals in the eclogite host (
Taylor et al., 1996a). However, this study is in direct opposition to that of
Ireland et al., (1994), who, on the basis of analyses of a single garnet in one diamond and a single clinopyroxene in another, stated that all eclogites have experienced metasomatism. Such postulated pervasive metasomatism would lead to changes in the contents of Fe, the REE, and Sm/Nd ratios (and, consequently, also 143Nd/144Nd). Thus, even if
Jacob et al. (1998) wish to lend credence to the conclusions of
Ireland et al., (1994), many of their own arguments for a subducted oceanic crustal origin, such as correlations of [delta]18O with these chemical parameters, can be ruled out.
Our study of coexisting garnet and clinopyroxene inclusions from within diamonds from four separate eclogites, when compared with the host eclogites, indicates a unique and complex history for each eclogite (
Taylor et al., 1996a). Some eclogites do indicate post-diamond-formation metasomatism, but others indicate little or no metasomatism, and still others indicate only light rare earth element (LREE) depletion (not enrichment) of the eclogite after diamond formation, indicating the possibility of partial melting. Although care must be taken to select samples that have not experienced metasomatism, such samples can be and have been found (
Taylor et al., 1996a).
For the interested reader, we reiterate a portion of the reasoning given in two of our earlier papers
(Snyder et al., 1993,
, 1997) leading us to the conclusion that metasomatism was minor in most (not all!) of the primary minerals in these samples. Metasomatism, either from the kimberlite itself or within the mantle, often includes the interaction of fluids containing high abundances of the large ion lithophile elements (LILE). The LILE include not only the LREE, but also Rb and K. The LREE are relatively highly charged cations that diffuse much more slowly and are more likely to be retained in most silicate crystal structures, especially garnet and clinopyroxene, than the larger, singly charged Rb and K. The whole-rock eclogites analyzed from Udachnaya and other Yakutian localities, indeed, do have elevated Rb and K, but this is more probably due to the infiltration of kimberlitic components during entrainment and transport of the xenoliths to the surface and precipitation of secondary minerals along grain boundaries. However, carefully hand-picked mineral separates do not contain elevated abundances of Rb, but, instead, relatively small quantities of this element [see table 8 of
Snyder et al. (1997) ] and very low Rb/Sr ratios. In fact, the Rb abundances (0·0207-0·330 ppm) in all but one of the reconstructed whole-rock eclogites [table 9 of
Snyder et al., (1997)] are approximately an order of magnitude or more lower than the Rb abundances in present-day n-MORB (1·52-1·97 ppm;
Viereck et al., 1989), a proposed protolith for some of these eclogites
(Rudnick, 1995). Potassium is somewhat elevated in these clinopyroxenes in general, but this is probably a primary feature and is due to the nuances of the high-pressure crystal structure of pyroxenes
(Harlow & Veblen, 1991;
Harlow, 1997). Thus, if
Jacob et al. (1998) are correct and metasomatism has caused the higher La values seen in Group A eclogitic clinopyroxenes U-25, U-236 and U-281, then these samples should also exhibit elevated K and Rb relative to other samples. This is definitely not the case [see table 8 of
Snyder et al., (1997)].
Kyanite eclogites (grospydites) are a minor component of the eclogite population world-wide, but they place important constraints on models of eclogite origins. Our group has supported a subducted oceanic crust, anorthosite precursor origin for such aluminous eclogites world-wide (e.g.
Taylor & Neal, 1989,
, 1993). However, these observations are not related to our evidence for ancient carbonatitic metasomatism in a few of the eclogites. It is
Jacob et al. (1998) who confuse `apples with pears' in this instance, as it is not the kyanite eclogites which are suspected to contain evidence of ancient carbonatitic metasomatism. The eclogite which best defines this speculative carbonatitic component, U-5, is not a grospydite. In point of fact, only two of the four kyanite eclogites, U-1 and U-112, have relatively high Sr/Nd ratios and plot along the array with seven other kyanite-free eclogites that point to ancient metasomatism [see fig. 12 of
Snyder et al., (1997)]. Furthermore, our speculation on ancient carbonatitic metasomatism in the mantle beneath Yakutia is supported by other lines of evidence, including fluid inclusions in Yakutian diamonds that point to carbonatitic fluids
(Schrauder et al., 1994), the occurrence of calcite in most of the Udachnaya eclogites, and the presence of other minerals such as celestite, barite, anhydrite, various sulfides, apatite, and schorlomite, which are consistent with a carbonatite origin
[Sobolev et al., (1994) and petrographic observations]. We have stated that `there are no definitive mantle signatures in igneous rocks. It is only the absence of crustal signatures that allows the interpretation of a mantle origin'
(Snyder et al., 1995).
Jacob
et al. (1998) find this statement to be troubling. We find this statement equally bothersome, but, to the best of our knowledge, it is true.
Jacob et al. (1998) go on to state that `it would be more constructive to compile positive geochemical and petrological lines of evidence' for mantle-derived eclogites. They cite as an example previous work from our group
(Shervais et al., 1988) where high Cr contents were used to designate so-called Group A eclogites as mantle derived. However, their last sentence in this section negates their whole objection and is further support of our statement; they write that `local high-Cr regions may also result from metamorphism during subduction of chromite layers within ophiolitic or anorthositic rocks of low-pressure [oceanic crustal] igneous origin'. We agree, and therein lies the rub. Characteristics of rocks that workers have previously considered as `evidence' of mantle derivation can also be found in rocks taken from oceanic crust. Thus, the development of an `updated catalogue of the geochemical and petrological characteristics which are to be expected from cumulates from mantle melts at high pressures' is a false panacea and will not likely be unique to mantle rocks alone.
It is always problematic to attempt to extend knowledge on element partititioning at 1 atmosphere to that at much higher pressures. The presence of diamond in these eclogites indicates that low-pressure equilibrium partitioning will not give one an adequate understanding of melting and crystallization at the depths of eclogite genesis and diamond formation (in excess of 40 kbar pressure). Thus, the element partititioning arguments of
Jacob et al. (1998) are suspect from the outset.
Jacob et al. (1998) say that `the variation of D values as a function of mineral composition is too small to explain the observed variation' in our dataset and that our interpretation is `demonstrably erroneous'. However,
Jacob et al. (1998) do not, in fact, demonstrate any error in this interpretation and suggest only that disequilibrium caused by metasomatic exchange, an explanation that has been effectively ruled out [see above and
Snyder et al., (1997)], could possibly be the culprit. In support of our interpretations, the relative partitioning of REE and other trace elements between garnet and clinopyroxene in eclogite xenoliths (especially Groups B and C) from Yakutia are similar to that seen in other eclogites from the Roberts Victor pipe, South Africa
(Harte & Kirkley, 1997). Furthermore,
Harte & Kirkley, (1997) have shown that clinopyroxene-garnet partition coefficients for the LREE can vary by `approximately three orders of magnitude', and that this variation is satisfactorily explained by changes in the relative Ca contents of the minerals. Thus, these mineral partition coefficients for the REE are not unusual in mantle xenoliths and not due to disequilibrium. It is of paramount importance that one prove chemical equilibrium between minerals if internal, mineral-mineral Sm-Nd and Rb-Sr isochrons are to have any meaning. Positive slopes alone on garnet-clinopyroxene Sm-Nd isochron diagrams do not prove that equilibrium has been achieved between the two minerals; however, this is a necessary requirement for proof of chemical equilibrium. The Udachnaya eclogites pass this important test.
We were very careful in both our Earth and Planetary Sciences Letters paper
(Snyder et al., 1993) and our Journal of Petrology paper
(Snyder et al., 1997) not to overstate the significance of the positive slopes on Sm-Nd isochron diagrams, as well as the derivative age information. In fact, we stated openly that these positive slopes allow only `speculation about age information' and that with some exceptions the Sm-Nd systematics in the Udachnaya eclogites `approximate the time of emplacement of the host kimberlite' (p. 103,
Snyder et al., 1997). We then went on to indicate specifically those samples that yield ages which differ from the weighted average age. We find it compelling that the weighted average age of all the garnet-clinopyroxene isochrons for all of our Udachnaya eclogites is 389 ± 4 Ma, similar to an 40Ar/39Ar age of 384 Ma
(Burgess et al., 1992), U-Pb zircon ages from two other kimberlites (344 and 358 Ma) in the Daldyn-Alakit field (which contains the Udachnaya kimberlite)
(Davis, 1978;
Davis et al., 1980), and U-Pb perovskite ages for the Udachnaya East (367 ± 3 and 367 ± 5 Ma) and Udachnaya West (353 ± 5 and 361 ± 4 Ma) kimberlites
(Kinny et al., 1997). Of the 11 samples where internal Sm-Nd garnet-clinopyroxene ages were determined by our research group, eight of them yielded ages that were within 60 Ma of the range of ages (353 ± 5 to 367 ± 5 Ma) determined in perovskites from the Udachnaya kimberlite. This is important confirming evidence of rough, albeit not perfect, mineral-mineral equilibration in eight out of 11 samples at the time of kimberlite eruption. Only three of the six samples of
Jacob et al. (1994) fulfill this criterion; i.e. only three yield Sm-Nd ages (353 ± 9, 379 ± 10, and 394 ± 10 Ma) similar to the accepted age range for the Udachnaya kimberlite
(Kinny et al., 1997). Their other three eclogites yield demonstrably different ages (194 ± 16, 808 ± 77, and 813 ± 25 Ma) [see table 9 of
Snyder et al., (1997)], suggesting gross mineral-mineral disequilibria in these samples.
The interpretation of the
Kinny et al., (1997) U-Pb perovskite data, as presented by
Jacob et al., (1998), raises some troubling questions. Russian mine geologists, who have extensively studied the Udachnaya pipe, have reported that `most investigators believe that the Eastern Body [Udachnaya east] has formed before the Western Body [Udachnaya west]'
(Sobolev et al., 1995), in opposition to the statements of
Jacob et al., (1998). At the present level of the mine, the two bodies are separate and cross-cutting relationships are not seen (our observations). Therefore, the relative ages of the two bodies are still in question.
Kinny et al., (1997) noted this inconsistency, but claimed that other unspecified field observations are supportive of the `East pipe' (which contained the eclogite xenoliths) being younger than the `West pipe'. Regardless of the superposition of these two sub-pipes, we still do not understand how an older event could be manifest in these samples, as
Jacob et al. (1998) suggest. We have stated that the eclogite xenoliths were equilibrated at the time of kimberlite eruption, probably as a result of the heating caused by that eruption. Previous effects would necessarily have been reset by this later event. Contrary to their statements,
Jacob
et al. (1994) did not `demonstrate' that the Udachnaya eclogites are Archean in age. Although
Jacob et al. (1994) mentioned unpublished Pb-Pb analyses which yield an age of 2·76 ± 0·1 Ga [stated by
Jacob et al. (1998) as 2·7 Ga], an old age for the Udachnaya eclogites was not confirmed until the next year, when Re-Os isotopic dating by our group and collaborators (
Pearson et al., 1995b) gave an age of 2·90 ± 0·38 Ga. The Pb-Pb analyses mentioned by
Jacob et al., (1994), as of this writing, remain unpublished; i.e. the data have not been included in any abstract, article, report, or other form for public use and evaluation.
We are glad that
Jacob et al. (1998) have raised the spectre of reconstructed whole-rock compositions. They mention that we have spent a considerable amount of time and effort concerning ourselves with whole-rock reconstructions and their interpretations. They also correctly point out that, because of the nature of the whole-rock samples, this is about the only way to derive information about the evolution and origin of mantle xenoliths. However, errors in determining the modal proportions of minerals can lead to large errors in model ages and initial radiogenic isotopic compositions. In fact, we purposely showed this in our earliest isotopic work on Yakutian eclogites [see fig. 3 of
Snyder et al., (1993)].
Jacob et al. (1998) question the validity of fig. 14 of
Snyder et al., (1997). This figure attempts to show all of the reconstructed Yakutian eclogite data on a so-called Sm-Nd isochron plot (147Sm/144Nd vs 143Nd/144Nd). To construct such a plot, one must know the proportions of garnet and clinopyroxene in each sample. Because
Jacob et al. (1994) did not provide modal information on the eclogites that they studied, we had to estimate these proportions. Albeit flawed, this was the only way that we could compare reconstructed radiogenic isotopic data from our study with those of
Jacob et al., (1994). Our experience has shown that the garnet:clinopyroxene modal proportions for nearly all Yakutian eclogites vary from 30:70 to 70:30
(Sobolev et al., 1994), thus 50:50 seemed to be a reasonable estimate. We did not `arbitrarily' choose the 50:50 proportions, and, in hindsight, we might have been better served by excluding their data altogether, or at least having used mineral proportions only when they are determined. We have done just this in
Fig. 2, where reconstructed whole-rock data for Yakutian eclogites, where mineral proportions are given in our work and in the literature, are presented on a plot of 147Sm/144Nd vs 143Nd/144Nd. The observations of
Snyder et al., (1997) still hold. Samples U-41, U-79, and U-464 [analyzed by
Pearson et al., (1995a)] indicate an old, ultra-depleted source; the bulk of samples indicate a nearly chondritic or only mildly depleted source; and two samples, U-5 and UJ-43, suggest an enriched component.
The criticism of our interpretations presented in the Earth and Planetary Science Letters paper
(Snyder et al., 1993) are somewhat warranted. Because of the paucity of crustal signatures that we have observed in the Udachnaya eclogites, we initially thought that many of the samples were derived directly from the mantle. However, we were careful not to assume that this model was inclusive of all samples, but stated that the model presented there applied to `at least some of the eclogites'
(Snyder et al., 1993).
We also hypothesized in this paper
(Snyder et al., 1993) that several of the Udachnaya eclogites were derived from very old, ultra-depleted mantle and enriched reservoirs. The presence of ultra-depleted mantle beneath Yakutia is consistent with previous work on peridotites from Yakutia by
McCulloch, (1989) and
Zhuravlev et al., (1991). These workers found that mantle samples with ages of 1·54, 1·7, and 2·6 Ga gave initial [epsilon]Nd values of + 24·5, + 23, and + 20, respectively, indicating an old ultra-depleted mantle source beneath Yakutia. One sample, U-5, if indeed of mantle derivation, suggested an old (possibly >4 Ga) enriched component. This interpretation was speculative, but we have since discovered an eclogite from the Mir pipe (~400 km south ofUdachnaya) which also suggests a very enriched source (
Snyder et al., 1996b). Thus, although such an interpretation may be debated, the process which led to such an isotopic signature operated in other than the Udachanaya pipe, and must be included in any overall model of eclogite genesis.
Our interpretations of both old depleted- and enriched-mantle sources were initially determined from Nd model ages. However, we were not satisfied with this or any other model calculations, so we determined Re-Os on these samples and found model ages of 2·8-3·4 Ga and a whole-rock `isochron' of 2·90 ± 0·38 Ga (
Pearson et al., 1995b).
Jacob et al. (1998) are correct to point out that these Re-Os data negate the model of 4 Ga differentiation in the early Earth, first proposed by
McCulloch, (1989), which was based on Nd model ages. Science marches on, and we no longer adhere to such a model of mantle differentiation. Previous studies of eclogite xenoliths from southern Africa consistently yielded [delta]18O isotopic values far outside the accepted mantle range, as well as extremely radiogenic Nd and Sr isotopic compositions. These chemical signatures were consistent and compelling evidence for oceanic crustal derivation of most eclogites. Our earliest work on Yakutian eclogites included only major- and trace-element analyses of garnet and clinopyroxene in 14 eclogites from Udachnaya. Thus, it was through the lens of earlier studies that we found the genesis of these xenoliths to be consistent with an oceanic crustal protolith
(Jerde et al., 1993). Surprisingly, Nd and Sr isotopic studies of the Udachnaya eclogites
(Snyder et al., 1993) indicated the distinctly nonradiogenic character of these rocks. Oxygen isotopic analyses of these eclogites also showed little variation, with only a few samples lying outside of the accepted mantle range
(Snyder et al., 1995). Thus, as we have stated in print and in countless talks at national and international meetings, the Udachnaya eclogites indicate a dearth of oceanic crustal signatures, leading us initially to the conclusion that most were igneous fractionates from the mantle, as per
Smyth et al., (1989). Our work on diamond-free eclogites from Obnazhennaya also indicated a dearth of crustal signatures
(Qi et al., 1994,
, 1997).
However, our continuing work on eclogites from the Mir pipe
(Beard et al., 1996) showed conclusive evidence of an oceanic crustal origin. In point of fact, we developed a model involving subduction of an oceanic crustal (proto-ophiolite) sequence with samples coming from different levels within this sequence
(Beard et al., 1996). We have also been careful in papers subsequent to
Snyder et al., (1993) to state the possibility that even those eclogites with no crustal signatures could have come from the mid-sections of a proto-ophiolite or oceanic crustal fragment, where oxygen isotopic compositions are within the mantle range [see fig. 8 of
Snyder et al., (1995)].
Our thoughts and ideas on eclogite genesis have been evolving as we have collected more data and studied eclogites from various kimberlites world-wide. We firmly believe that this is how science should operate. Our continuing work on Yakutian eclogites has led us to some intriguing interpretations. The combined data for Yakutian eclogites show two distinct groups (
Snyder et al., 1996b,
1997). One group, which includes the low-Ca Mir eclogites, follows a well-defined trend of increasing [delta]18O (+4·7 to +7·4%o) with slightly increasing Sm/Nd and [epsilon]Nd, and large increases in 87Sr/87Sr (
Snyder et al., 1996b). A second dominant trend, which includes the high-Ca Mir eclogites, is more diffuse, exhibits large increases in Sm/Nd and [epsilon]Nd, but is independent of [delta]18O. Although this dominant trend does indicate variation in 87Sr/86Sr, the variation is small and is not correlated with any other parameters. These two trends probably represent two distinct environments of formation for Yakutian eclogites: a shallow oceanic crustal environment with higher [delta]18O values, and exhibiting a significant continental, terrigenous sediment input (as evidenced by large variations in 87Sr/86Sr and lower [epsilon]Nd values), and a deeper oceanic crustal environment with lower [delta]18O and shielded from continental crustal sediment input.
Finally,
Jacob et al. (1998) object to our conclusion that a `multiplicity' of origins for eclogites are still warranted by the data. As evidence against this conclusion, they cite their fig. 1a and b, which excludes most of the oxygen isotope data from the Udachnaya eclogites and fully 90% of the Yakutian eclogite data. Furthermore, they assert that if we are to prove that some of the eclogites, indeed, are mantle derived, then we need to produce `irrefutable evidence in favour of a mantle origin for these samples'. We agree, but this is the crux of the problem with mantle samples, which we have already discussed above. Although the interpretations given by
Jacob et al., (1994) have merit, the bases for these interpretations have been largely discounted by
Snyder et al., (1995,
, 1997). Furthermore, the rather straightforward interpretation of many of these eclogites as subducted oceanic crust has been complicated by further data, as well as those exceptional samples which indicate direct mantle derivation. Our group is continuing to analyze eclogites from various pipes from throughout Yakutia and, as always, we will consider the useful comments of
Jacob et al. (1998) in further interpetations of eclogite genesis. As an example, our discussions with Professor Emil Jagoutz and Dr Dorrit Jacob have led us to initiate a study of Group A eclogites world-wide to determine if they are indeed of direct mantle derivation, and to continue our work on kyanite eclogites to set further phase equilibrium constraints on eclogite genesis.
We would like to thank Randy Keller, John Valley and Bob Clayton for reviewing this paper `in house', and Roberta Rudnick for a formal review. We also appreciate the careful oxygen isotopic work of Bob Clayton and Tosh Mayeda published in our earlier paper
(Snyder et al., 1995), and the subsequent work of John Valley and Michael Spicuzza that has confirmed our earlier findings (
Taylor et al., 1996b).
INTRODUCTION
OXYGEN ISOTOPIC DATA
High-temperature isotopic fractionation between garnet and clinopyroxene
Correlations of oxygen isotopes with other chemical parameters
METASOMATISM OF YAKUTIAN ECLOGITES
MINERAL CHEMISTRY
Mantle signatures?
Element partitioning
RADIOGENIC ISOTOPE DATA
Mineral-mineral isotopic equilibrium
Age of the Udachnaya kimberlite and further evidence of isotopic equilibration
The age of the Udachnaya eclogite protolith
Reconstructed whole-rock chemical compositions
Model calculations and possible mantle sources
THE SCIENTIFIC METHOD AND MODELS OF ECLOGITE ORIGINS
CLOSING STATEMENT
ACKNOWLEDGEMENTS
REFERENCES