| ||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology Volume 42 Number 6 Pages 1221-1224 2001
© Oxford University Press 2001
Flood Basalts, Basalt Floods or Topless Bushvelds? Lunar Petrogenesis Revisited: a Reply
DEPARTMENT OF EARTH SCIENCES, CARDIFF UNIVERSITY, PO BOX 914, CARDIFF CF10 3YE, UK
 |
INTRODUCTION |
|---|
 TOP INTRODUCTION REFERENCES |
|---|
I thank Taylor (2001)
for drawing my attention to some additional references which I had overlooked and I welcome his tacit concession that there is as yet no evidence for a positive europium anomaly in the lunar highland crust. I surmise from the nature of Taylor’s comments that he also has changed his views little in the past 30 years, but he has undeniably enjoyed a more acquiescent audience. Numbers in parentheses in the following text and otherwise unattributed figure mentions are references to the corresponding notes and figures in O’Hara (2000). Only references additional to those in O’Hara (2000) or Taylor (2001) are listed here.
The Lunar Sample Preliminary Examination Team (LSPET) (1969) team included a notable primary magmatist in P. W. Gast, and also included S. R. Taylor, who came from a school which 2 years earlier had committed itself to that ultimate in primary magma hypotheses, the pyrolite model of terrestrial upper-mantle composition. The LSPET report on the Apollo 11 lunar samples from Mare Tranquillitatis asserted that soils and hand specimens had ‘similar but unique compositions’ and further asserted that ‘It is particularly significant that the unique composition is that of a silicate liquid’, thus perhaps laying the foundations for subsequent events. By the first lunar science conference the hand specimens were predetermined by many workers to be primary magma compositions, despite the manifest and unexplained discrepancy which had emerged on further study (Fig. 9) between the hand-specimen compositions and that of the average target rock which had formed the regolith (Fig. 2) at the Apollo 11 site. Samples from a second purely mare site in Oceanus Procellarum, also remote from highland outcrops, received similar treatment early the following year, although LSPET (1970) this time had commented specifically on the discrepancy between the hand specimens and the average target rock forming the soils. These high- and low-titanium basalt compositions, however, possessed deep negative europium anomalies, probably resulting from a major plagioclase separation event somewhere in their petrogenesis, yet plagioclase is never saturated at or anywhere near the liquidus of the hand-specimen compositions at any pressure. The compositions of the average target rocks which had formed the regoliths, on the other hand, implied an erupted magma composition which could be plagioclase saturated at the low-pressure liquidus, but this would have implied that they were derivative liquid compositions from gabbro fractionation at low pressure, and not primary magmas. The Apollo scientific community which had embraced the primary magma hypothesis for the hand specimens were thus impaled on a sword of their own forging. Rather than re-examining the concept of primality or addressing the composition discrepancy, the hypothetical primary status of the hand specimens was conserved by postulating an in-built negative Eu anomaly in the lunar mantle. But how to implant this anomaly? How better than by removal of plagioclase from that mantle, carrying its excess of Eu over the other REE? Enter the hypothesis of formation of an impact-generated global magma ocean, during the subsequent fractional crystallization of which plagioclase-saturated assemblages were eventually precipitated. Flotation of the plagioclase crystals was hypothesized to form the lunar highland crust, and deposition of the mafic minerals was postulated to form an olivine pyroxenite mantle of relatively low mg-number. Later incipient remelting of this plagioclase-depleted cumulate was then postulated as the source of the primary mare magmas—failing to appreciate the force of the italicized statement above.
The tail of this elaborate snake of hypotheses was a non-negotiable requirement for a substantial positive Eu anomaly in the average pre-3·9 Ga lunar highland crust, which was duly reported. Belief in this positive anomaly has succoured the lunar geochemical community for 30 years. The burden of Taylor’s (2001) argument is that the required anomaly may yet emerge. It is, however, the head of this snake which I have attacked; the fate of its body and the terminal rattle are simply collateral damage.
There are six threads to my objection to the original promotion of lunar mare basalt hand specimens as primary magmas: (A) the chemical engineering problems posed by extracting hot, highly corrosive silicate melts, unmodified and uncontaminated, through 150–400 km of vulnerable silicate wall rock with which the magma would not be in equilibrium, and then through a density filter of 50 km of feldspathic crust which is supposed to have floated on such magmas in the first place; (B) the probability of substantial sodium and sulphur loss during eruption; (C) the probability of assimilation of KREEPy materials by the basalts during their passage through or over the lunar crust; (D) the major element mismatch between the compositions of the hand specimens of basalt and the composition of target rock forming the mare regoliths; (E) the phase equilibria disqualification of the hand-specimen compositions as small mass fraction partial melts of an originally plagioclase-saturated mantle source; (F) the difficulty of accounting for the requisite negative Eu anomalies in the source region of the alleged primary magmas in the absence of a demonstrable substantial positive Eu anomaly in the lunar crust. Taylor’s comments do not fully address (A)–(E). The above criticisms of the primary magma hypothesis are independent of each other and that hypothesis falls if any one of them can be sustained. If the primary magma hypothesis falls, so does the raison d’être for the rest of conventional lunar petrogenesis.
The chemical engineering problems faced by would-be primary magmas have defeated vast volumes of terrestrial MORB despite much more favourable circumstances, i.e. as little as one-tenth the travel distance, a pre-heated pathway, and a less effective density filter in the crust (5, 93, 94; Grove et al., 1992). Ocean-island basalts and continental flood basalts have likewise generally failed to emerge unscathed with longer, but still short travel distances relative to those postulated for lunar magmas. The rare primitive eruptive rocks in such provinces do not typify the terminal stages of eruptive activity.
The Gibson–Hubbard argument regarding lithospheric pressure and suppression of volatilization must be incomplete in at least one respect, because mare lavas from Apollo 11, 12, 15 and 17 manifestly vesiculated to depths much greater than 10-3 cm. That argument, and its companion of rapid cooling of a thin outer skin of the basalt, are, however, irrelevant if we are dealing with small droplets in a fire fountain (80–84). Volatilization loss during fire-fountaining has been the issue since Biggar et al. (1971), an issue which Taylor has failed to address in three book-length reviews over 23 years, and the notion of significant amounts of volatiles boiling off from the surface of a lava lake is a straw man of Taylor’s own propagation. It is the style of eruption on Io and its geochemical consequences which are important, and the relevance to lunar petrogenesis cannot be so cavalierly written off on the grounds that the driving mechanism may be different. Taylor (2001) seeks to set aside the evidence from Io but is happy to introduce that from the Eucrite Parent Body (EPB), which I agree is highly relevant (85–90), although it is much smaller than the Moon and the eruption mechanisms are a matter for speculation. The parent–daughter element ratio argument relating to volatilization has been partially addressed already (81, 84). Taylor (2001) has suggested no new tests of this vital volatilization issue. Rather than shelter complacently behind 30-year-old trace element ratio arguments, should we not conduct some definitive experiments on melting and extruding basalts in hard vacuum, to supplement those already available (81)? If my concerns are justified, these should be carried out by remote control at the end of a long spoon extended from the International Space Station. If the conventional model is correct, these experiments can be safely conducted inside a depressurized cabin on the Space Shuttle. Volunteers?
I am by origin a field petrologist. I became an experimental petrologist because this was necessary in order to understand major element and petrological data for high-pressure mafic rocks from Scourie, Glenelg, Norway and South Africa. This led amongst other things to the recognition that most erupted terrestrial basalts, including MORB, blatantly cannot be primary magmas (35–44). This view was challenged by trace element geochemists in 1968–1974 on the grounds that the requisite partial crystallization at low pressure would deplete compatible trace elements excessively, and could not modify incompatible trace element ratios in the manner required (0–2, 4, 7–15). Therefore I digressed into geochemical modelling for the greater part of 25 years to demonstrate (refilled magma chambers, integrated melting, integrated crystallization, combined small packet, integrated and recharged crystallization, etc.) that these criticisms were without foundation because the geochemical modelling was naive and the conclusions severely model dependent [1, 5, 6, 16–30, 46, 47; the bad news is encapsulated in Fig. 1 and O’Hara & Fry (1996b, Table 1), but has rarely been cited, still less considered, in the many hundreds of papers on basalt geochemistry published since]. This was a task which some might think should have been performed by geochemists themselves with their impressive manpower resources, before the copious replication of non-unique interpretations. Basalts are rich in geochemical information about complex magma chamber processes and crustal interactions as well as about upper-mantle composition and process. Ignoring the former merely results in attributing erroneous properties to the latter.
Rather than my chasing further trace element ratio arguments, I now invite the geochemical community to invest more effort in exploring how and why their data fit with the field observations, major element data, well-established phase equilibria and the geophysical evidence, a pursuit which I have found intellectually stimulating and professionally rewarding. The main reservoir for Rb, Sr, U, Th and Pb in the lunar crust is KREEP, an ancient magma which may have lost its volatiles early as required, but then handed on that signal to subsequent mare lavas through assimilation and contamination. In the specific case of the above-mentioned parent–daughter element argument, the hazard of contamination by KREEP either within crustal magma chambers (33, 125, 126) or during flow across a KREEP-bearing regolith has not been adequately considered. The techniques developed by McSween et al. (2001) might usefully be applied to KREEP basalts in particular.
The central challenge to the pedigree of the mare hand specimens as primary magmas is the mismatch between their compositions and the composition of mare target rock required to explain the average soil and basalt lithic fragment compositions at all mare sites (Fig. 2, 100–118, 120, 121). Confirmatory data relating to the compositions of randomly sampled basaltic materials have come to hand. Several estimates of the composition of the average basaltic lithic fragments which must be present in the Apollo 17 drill core have been extracted (O’Hara, 2001a), with the average result Al2O3 10·3%, SiO2 39·9%, Al/Si 0·29. These values place the compositions within the random basalt and low-pressure cotectic liquid fields (Fig. 2). The average compositions of two different but overlapping sub-sets of the randomly sampled basalt fragments from the Apollo 15 rake samples have been analysed by INAA (Laul & Schmitt, 1973) yielding the average result Al2O3 10·9%, SiO2 45·6%, Al/Si 0·27, and by defocused beam techniques (Dowty et al., 1973) which, after removing (O’Hara, 2001a) arbitrary correction factors, yielded the average result Al2O3 11·7%, SiO2 45·4%, Al/Si 0·29. These values place both of these average compositions within the low-pressure cotectic liquid field (Fig. 2) although the Apollo 15 site regoliths appear to contain a substantial component of mare gabbros (O’Hara, 2001b). The conclusion that the hand specimens do not represent the average erupted magma composition grows more robust, and will soon be further tested by the D-CXIS experiment aboard SMART-1, which promises to yield high-resolution major element analyses, not just Al/Si ratios, for most of the mare surfaces. Obviously I am predicting the widespread establishment of bulk compositions which, after allowance for highland contamination, will still be overwhelmingly close to those of low-pressure cotectic liquids which would approach pyroxene saturation via plagioclase saturation rather than vice versa.
The disqualification of the basalt hand specimens as low mass fraction melting primary magmas derived from a plagioclase-saturated lunar mantle mineral assemblage (Fig. 4, 93–98) was first highlighted by Longhi (1981). It is based on phase equilibria studies (i.e. manifestations of major element geochemistry) which have proved robust for synthetic systems and for terrestrial, lunar and EPB basalts over a period of nearly 40 years in studies by a multitude of independent and sometimes fiercely competitive workers in at least seven different laboratories.
Presence of a substantial positive Eu anomaly in the average pre-3·9 Ga lunar highland crust is the cornerstone of the plagioclase flotation hypothesis, the global magma ocean hypothesis, the cumulate mantle hypothesis and the mare primary magma hypothesis, not one of which can survive without it. Those who would cling to the conventional lunar petrogenetic model are entitled to hope that a positive Eu anomaly in the average lunar highland crust will yet emerge (the average of the surface composition is only part of the story, given the evidence of substantial vertical variability in the crust, but the D-CXIS experiment will add considerably to the database). Suffice it to recall that confidence in the existence of this anomaly is so low that recent compendiums (Heiken et al., 1991; Papike et al., 1998) on lunar petrology and geochemistry make no reference to it. If a positive anomaly subsequently emerges, its identification on the basis of the evidence available either during the Apollo programme (Fig. 3, 61–67) or up to the end of the last millennium, must be viewed as breathtaking serendipity or astounding prescience.
Well might the impartial observer ask ‘How could so many workers, good and true, have got something so important so wrong?’ This may be a source of amazement in the world at large, but not to one from the land of Black Wednesday, the Millennium Dome and the BSE reassurances. Comprehensive corporate misjudgements happen. Sample return from Mars is a realistic possibility within the decade. To be cultivating incomplete interpretations of crustal petrogenesis on our own planet is understandable within the context of the evolution of knowledge and ideas; to misinterpret processes on our most accessible neighbour in the teeth of the major element evidence is careless; but to repeat the errors on any further planets would be inexcusable. Rather than dispute the fine details of old arguments, would it not be prudent to conduct now a programme of open-minded experimental re-examinations of the many working hypotheses (list on request) which may have been under-addressed in the tens of thousands of pages of scientific publications emerging from the lunar exploration programme?
High on the list of immediately feasible actions come extended mathematical studies of all possible tidal contributions to lunar volcanism; the above-mentioned studies of the eruption of appropriate basalt compositions into hard vacuum; determinations of the average compositions of the basaltic lithic fragments in the remaining Apollo 11 and 12 and the multitude of Apollo 15 and 17 regolith samples; high-resolution multi-spectral imaging of the composition of materials exposed in the walls and debris aprons of mare-penetrating impact craters (e.g. Archimedes, Aristillus, Autolycus, Bullialdus, Goclenius, Helicon, Leverrier, Lambert, Pytheas, Timocharis) and critical re-examination of the published database for the Rb–Sr system in particular for evidence of variable contamination of lavas by an ancient KREEP component. Accessible with present or imminently available technology are detailed studies of the compositions and temperatures of the magmas erupting on Io and of the detailed composition of the gas cloud released (e.g. Zolotov & Fegley, 1999, 2000; Fegley & Zolotov, 2000); an automated mission to examine and sample the exposures in the walls and valley blocks of Hadley Rille; and recovery of a 50–100 m drill core through the mare basalts, logically at the Apollo 15 or Apollo 17 LM sites where surrounding provenance is best established.
In retrospect, the development of terrestrial and lunar petrogenesis between 1967 and 1975 may be seen as a triumph of an approach based upon oversimplified modelling of trace element geochemistry supported by opportunistic ‘cherry-picking’ experimental petrology on isolated natural samples of poor provenance. It was powered by the avalanche of materials from the ocean floor and the Moon, where provenance was necessarily less secure. This approach replaced the traditional approach centred on field and major element petrology and the systematic experimental study of phase equilibria in systems of interrelated compositions. If the resulting lunar paradigm has crumbled, just how secure are our assumptions and attitudes towards terrestrial basalts, many of which views are rooted in the same period of scientific development?
Finally, Taylor (2001) has drawn my attention to the paper by Ruzicka et al. (1998) which displays a healthy and welcome scepticism towards the giant impactor hypothesis for lunar origin, on the basis of the geochemistry of the transition elements in the terrestrial planets. This is a group of elements which tend to be compatible in oxide, sulphide, metal and silicate phases which may separate early in partial crystallization of basalts. Their individual and group behaviour will be highly sensitive to sulphide activity, oxygen fugacity and the specific partial crystallization models entertained. The planetary bulk compositions utilized in the discussion are based on implicit assumptions that surface basalt compositions from the Earth, Moon, Eucrite Parent Body and Mars are representative of the primary melts which have separated from their respective mantles. This proposition is certainly untrue for the Earth and probably untrue for the EPB. I have summarized here the reasons for thinking it is untrue for the Moon and it is my sincere hope to obstruct its facile assumption for Mars. The differences commented on by Ruzicka et al. (1998) may owe as much to crustal nurture as to planetary nature.
|
FOOTNOTES |
|---|
*E-mail: sglmjo{at}cardiff.ac.uk
|
REFERENCES |
|---|
|
TOPINTRODUCTIONREFERENCES |
|---|
Dowty, E., Prinz, M. & Keil, K. (1973). Composition, mineralogy, and petrology of 28 mare basalts from Apollo 15 rake samples. Proceedings of the Lunar Science Conference 4, 423–444.
Fegley, B. & Zolotov, M. Yu. (2000). Carbon chemistry of volcanic gases on Io (abstract). Meteoritics and Space Science 35(Supplement), A52.
Grove, T. L., Kinzler, R. J. & Bryan, W. B. (1992). Fractionation of mid-ocean ridge basalt (MORB). In: Mantle Flow and Melt Generation at Mid-Ocean Ridges. Geophysical Monograph, American Geophysical Union 71, 281–310.
Laul, J. C. & Schmitt, R. A. (1973). Chemical composition of Apollo 15, 16 and 17 samples. Proceedings of the Lunar Science Conference 4, 1349–1367.
Longhi, J. (1981). Preliminary modeling of high pressure partial melting: implications for early lunar differentiation. Proceedings of the Lunar Science Conference 12B, 1001–1018.
LSPET (Lunar Sample Preliminary Examination Team) (1969). Preliminary examination of lunar samples from Apollo 11. Science 165, 1211–1227.
LSPET (Lunar Sample Preliminary Examination Team) (1970). Preliminary examination of lunar samples from Apollo 12. Science 167, 1325–1339.
McSween, H. Y., Jr, Grove, T. L., Lentz, R. C. F., Dann, J. C., Holzheid, A. H., Riciputi, L. R. & Ryan, J. G. (2001). Geochemical evidence for magmatic water within Mars from pyroxenes in the Shergotty meteorite. Nature 409, 487–490.[Medline]
O’Hara, M. J. (2000). Flood basalts, basalt floods or topless Bushvelds? Lunar petrogenesis revisited. Journal of Petrology 41, 1545–1651.
O’Hara, M. J. (2001a). Feldspathic mare basalts at the Apollo 17 landing site, Taurus–Littrow. Journal of Petrology (in press).
O’Hara, M. J. (2001b). Lava ponds and crusts, norite sludges and quench crystal cumulates: petrogenesis of low titanium basalts from Hadley Rille, S.E. Mare Imbrium. (Submitted.)
Taylor, S. R. (2001). Flood basalts, basalt floods or topless Bushvelds? Lunar petrogenesis revisited: a critical comment. Journal of Petrology 42, 000–000.
Zolotov, M. Yu. & Fegley, B. (1999). Oxidation state of volcanic gases and the interior of Io. Icarus 141, 40–52.[ISI]
Zolotov, M. Yu. & Fegley, B. (2000). Eruption conditions of Pele volcano on Io, inferred from chemistry of its volcanic plume. Geophysical Research Letters 27, 2789–2792.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||