| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology Volume 42 Number 8 Pages 1401-1427 2001
© Oxford University Press 2001
Feldspathic Mare Basalts at the Apollo 17 Landing Site, Taurus–Littrow
DEPARTMENT OF EARTH SCIENCES, CARDIFF UNIVERSITY, PO BOX 914, CARDIFF CF10 3YE, UK
Received June 20, 2000; Revised typescript accepted January 19, 2001
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONTHE APOLLO 17 REGOLITHSANALYSIS OF CHEMICAL VARIATION...WHY HAVE BASALT LITHIC...SUMMARY AND PREDICTIONSREFERENCES |
|---|
The basalt target rocks that have been converted to regolith across the lunar maria are everywhere more feldspathic and less mafic than the basalt hand specimens recovered from four Apollo landing sites, an effect not due to either horizontal or vertical mixing with adjacent highland materials. These crushed target rocks need to be characterized by direct chemical and petrographic analysis of the lithic fragments of basalt in the regoliths and by determination of the phase equilibria in and adjacent to these compositions at low pressure. Such data are available for the basalts of Mare Crisium and Mare Nubium (Luna 16, 24) and for Very Low Titanium basalt, first defined by three lithic fragments from the Apollo 17 core. These are all feldspathic basalts, as are those from the Mare Tranquillitatis and Oceanus Procellarum soils (Apollo 11, 12). Such data are lacking for the principal basalt components at Mare Imbrium and Mare Serenitatis (Apollo 15, 17). The thoroughly investigated Apollo 17 landing site at Taurus–Littrow, SE Mare Serenitatis, provides an example where other published information may be used to arrive at estimates of the composition of the feldspathic mare basalt that was the principal target material for regolith formation. This crushed basalt composition is that of a liquid close to being in simultaneous equilibrium with all of olivine, plagioclase, calcium-rich pyroxene, spinel, armalcolite and ilmenite at low pressure. The simplest explanation would be that the basalt that dominated the formation of the regolith comes from a different flow unit than the hand specimens, but it strains credulity that not a single hand specimen can be positively assigned to that upper unit, and not a single soil sample can be positively identified as having formed principally from the unit that provides the hand specimens.
KEY WORDS: cotectic; lithic fragment; lunar; target rock; regolith
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONTHE APOLLO 17 REGOLITHSANALYSIS OF CHEMICAL VARIATION...WHY HAVE BASALT LITHIC...SUMMARY AND PREDICTIONSREFERENCES |
|---|
The conventional interpretation of lunar petrogenesis viewed the mare basalt hand specimens as representing primary magma compositions. These were postulated to have originated by small mass fractions of renewed partial melting of (plagioclase-saturated) mantle cumulate sequences produced during consolidation of a global magma ocean, which was itself generated by impact heating during initial accretion of the Moon. This interpretation requires revision (O’Hara, 2000
), because the Al/Si ratios of 
3·5 x 106 tonnes of mare basalt remotely sensed from orbit cannot be explained if the erupted magma composition was similar to that of the hand specimens. Spectral reflectance studies would appear to extend this conclusion beyond the limited tracks of the orbital X-ray spectrometer to the greater part of the nearside maria surfaces (Pieters, 1978; and see Basaltic Volcanism Study Project, 1981, section 2.2.1, and plates 2.8 and 2.9). Liquids of hand-specimen composition have high-pressure phase equilibria totally inconsistent with their derivation as low mass fraction partial melts from originally plagioclase-saturated, even if now completely plagioclase-free, mantle assemblages (Longhi, 1982; O’Hara, 2000, fig. 4, and notes 93 and 96), or some more general mantle bulk composition close to saturation with an alumina-rich phase.
The pronounced negative europium anomalies in the basalts have been attributed to europium depletion of their source region due to flotation of plagioclase crystals from that magma ocean to form the lunar highland crust. There is, however, no positive europium anomaly in the average lunar highlands (Korotev & Haskin, 1988; O’Hara, 2000, fig. 3, and notes 61 and 62). In the absence of that anomaly, there is no compelling evidence for plagioclase flotation, for the former existence of a global magma ocean, or for the postulated cumulate mantle precipitated from it. Without a positive europium anomaly in the highlands there can be no expectation of a negative europium anomaly in the source regions of the mare basalts. Their negative europium anomalies cannot, therefore, be inherited and must be imposed by later events such as plagioclase (gabbro) fractionation within the crust. One consequence of such a model is that the average basalt erupted at the lunar surface should be close to being in cotectic equilibrium at low pressure with plagioclase, olivine and pyroxene plus sundry oxide phases. Although this is in accord with the observed Al/Si ratios it implies that the average basalt fragmented to form the regolith should be much more feldspathic in composition than the hand specimens.
Apollo 17 visited an embayment of mare basalts into the lunar highlands at the SE edge of Mare Serenitatis, about 2 km from the nearest point of the highland outcrops at the North massif. Site geology has been summarized by Spudis & Pieters (1991) with a map, and another readily accessible map with site photograph and sketch will be found in the paper by Korotev & Kresmer (1992). The Lunar Excursion Module landed on high-titanium basalts that form the surface of Mare Serenitatis in an annulus of darker basalt surrounding the lighter-coloured, lower-titanium basalts of the basin centre. Boundaries between the three rock types are sharp, as may be verified with binoculars or a small telescope (the colour contrast in the basalts requires favourable viewing conditions) and testifies to the limited extent of both vertical and horizontal mixing in the formation of the post-mare regolith. This site is particularly important in the study of lunar petrogenesis because of the wealth of information gathered from the mare and closely adjacent highlands, allowing characterization of both rocks and regoliths in greater detail and with a better control from field relationships than at any other site. Taurus–Littrow is also important as a site where evident pyroclastic material (the orange and black glasses) is abundant. It is convenient for study in that the local basalt is very strongly contrasted in titanium content from the adjacent highland materials, allowing TiO2 content to be used as an indicator of basalt contribution to the regoliths. It is also convenient because, unlike at the Apollo 15 site, there are no features suggesting late drainage of a substantial quantity of partially consolidated basalt magma from the region, nor possible effects arising from the winnowing of soils at the edge of steep slopes such as may affect soils collected at the rim of Hadley Rille on the Apollo 15 mission (Howard et al., 1972).
The highland soils from the Apollo 17 mission, and the majority of highland rock samples collected, have low Al/Si ratios relative to the lunar highlands in general. Concentrations of incompatible trace elements are relatively high and these samples mostly display small negative Eu anomalies. The basalts are rich in incompatible trace elements and also display deep negative europium anomalies. Their petrogenesis encapsulates a more general petrogenetic discussion. Do such low mg-number, incompatible element-enriched liquids owe their enrichment in incompatible elements to very small mass fractions of partial melting in the mantle, or to very advanced partial crystallization of some less well-endowed, higher mass fraction partial melt after its collection? In this specific case, is the deep negative Eu anomaly inherited from the mantle source region, or is it the product of extensive near-surface partial crystallization of gabbro from the parent magma? Resolution of the question in this instance reduces to a decision whether the average mare basalt composition erupted at the site was that of a low-pressure plagioclase-saturated cotectic liquid also saturated with olivine, pyroxene and opaque minerals, or was it that of a liquid which could be in equilibrium with a plagioclase-saturated cumulus mineral assemblage at depth within the Moon? Both questions may be answered decisively in the negative if the liquid composition is to be represented by the average hand-specimen composition (O’Hara, 2000). The most pertinent questions here are what was the average magma composition erupted at this site (assuming it to be better represented by the average compositions of the lithic fragments of basalt present in the regolith) and what are its phase equilibria at low and high pressure?
General aspects of regolith petrology have been reviewed by McKay et al. (1991). The stratigraphy of the drill core in the mare regolith collected at the Apollo 17 landing site has been studied in exceptional detail and provides data for numerous horizons with very variable proportions of basaltic lithic fragments and mineral grains relative to (a) the sparse highland-derived lithic fragments, (b) the dark matrix breccias and agglutinates which may represent matured soil components, and (c) locally abundant orange and black pyroclastic glasses. Direct determinations of the average composition of the basalt lithic fragments at different horizons in the core have not been published, but the other data available are sufficient to allow their composition to be estimated, and at the very least a good case made for further efforts to determine those average compositions.
Outline of this treatment
This study restates, in the specific context of the Apollo 17 site, the maria-wide problem that soils do not match hand-specimen compositions. It then (1) examines the discrepancy between the amount of highland debris observed in the soils and the higher amount of highland debris which has been calculated to be present in the soils on the assumption that the average basalt is represented by the feldspar-poor hand specimens. It also examines the discrepancy between the amount of basaltic debris observed in those soils and the lower amount of basaltic debris that has been calculated to be present on the same assumption regarding its composition. The underlying assumption regarding the composition of the target basalt is believed to be incorrect.
Next (2) this study examines the consequences of assuming that the proportions of mineral fragments in the most basalt-rich horizons of the core may reflect directly the mode of a coarse-grained target basalt. In an alternative approach (3) the composition of the basalt component in those basalt-rich horizons has been calculated by subtracting the contributions from the other identified components and scaling to 100%. In another approach (4) oxide concentrations in the sieved coarse and medium grain-size fractions of the core have been regressed against nine different combinations of the volume percentages of the modally identified constituents of those grain-size fractions, to obtain estimates of the average composition of the basalt component.
These results are compared with other chemical data in simple oxide–oxide plots and in two more complex data projections, which essentially display variations in the ratios of normative constituents in a fashion that allows them to be related to the known low-pressure phase equilibria in these compositions.
|
THE APOLLO 17 REGOLITHS |
|---|
|
TOPABSTRACTINTRODUCTIONTHE APOLLO 17 REGOLITHSANALYSIS OF CHEMICAL VARIATION...WHY HAVE BASALT LITHIC...SUMMARY AND PREDICTIONSREFERENCES |
|---|
An upper coarse-grained basalt-rich unit of the soil in the Apollo 17 drill core probably came as ejecta some 48–100 cm thick from the 650 m diameter Camelot crater about 920 m distant (Taylor et al., 1977
). Camelot crater is some 3 km distant from the nearest point in the highlands of the North massif. The Camelot debris would have been excavated from a maximum depth of 37 m and fossil soils from the Camelot site would have comprised about 23% of the mass of the deposited ejecta blanket, in reasonable agreement with the fused soil component in the appropriate levels of the drill core. The basaltic component in the upper layers of the drill core (unit D) is thus expected to comprise 77% by mass of basalt from depths of 8·5–37 m in the original flow(s) and pyroclastics, plus whatever proportion of the fossil soil was of basaltic origin, estimated here as not less than 85% by mass, implying that there can be no more than 3·5% by mass of highland contribution to these soils. This is consistent with the low abundance of identifiable highland lithic fragments in these horizons. The feldspathic character of these soils must be inherent in the basalt target material and it certainly owes nothing to the admixture of the orange and black glasses, which are feldpar impoverished. It has been suggested that the lower part of the drill core may contain a deposit akin to the avalanche deposits at stations 2 and 3 on the edge of the South massif (Nagle, 1982), although the chemistry of the lower drill core is much more basaltic than the soils at those stations, matches a known titaniferous basalt clast, and allies it with the soil at station 9 (see Figs 3–10, below).
|
|
|
|
|
|
|
|
Soils are persistently more feldspathic than the hand-specimen compositions
In general, the bulk regoliths from the Apollo 11, 12, 15 and 17 sites have compositions with significantly higher Al/Si ratios than the majority of hand specimens from the same sites (O’Hara, 2000, fig. 2). Explanation of these regolith compositions in terms of a basalt component with the composition of the average of the local hand specimens requires a calculated percentage of highland input which exceeds that which can be identified by petrographic examination and exceeds that anticipated at the Apollo 11, 12 and 15 sites on the basis of studies of cratering dynamics, which predict that most of the material will have come from within 5 km of the collection site (Arvidson et al., 1975; Melosh, 1989). In the case of the Apollo 17 mare regoliths, this distance criterion plus the elevation of the highland massifs would permit somewhat higher proportions of highland-derived materials in many of the soils, via both impact dispersion and gravity slides.
|
) assigned as occult highland component; and the latter plus also a proportional share of the mineral fragments (•), calculated as discussed in the text. All points are plotted as a function of the weight percentages of calculated highland-derived components in each sample. (b) As in (a), but assigning an arbitrary 60% of the fused soil component as occult highland component. Although this produces reasonable agreement between calculation and observation it is unsustainable as a model (see text).
However, not one of the 114 bulk or sieved sub-samples of horizons within the Apollo 17 drill core which have been analysed by two groups of investigators, nor any of the other valley floor mare soils from Taurus–Littrow analysed by several other groups, has TiO2 significantly exceeding that of the analysed low-pressure plagioclase-saturated cotectic liquids developed from these hand specimens (8·96–10·15% TiO2, O’Hara, 2000, table A3). Few have TiO2 as high as even the lower end of the range of the analysed hand-specimen groups (9·56–13·43% TiO2) of Rhodes et al. (1976) or the site hand-specimen average of 12·0% TiO2 (Taylor et al., 1978), in spite of there being several horizons in the core described as consisting almost entirely of mare basalt related material. None of these soil samples has Al2O3 lower than that of the analysed basaltic low-pressure plagioclase-saturated cotectic liquids (9·61–10·17% Al2O3) and none as low as in most of the analysed basalt hand specimens (7·34–10·23% Al2O3) of Rhodes et al. (1976) or the site basalt hand-specimen average of 9·0% Al2O3 (Taylor et al., 1978). These generalizations alone point to an average mare basalt composition at this site which is less mafic, less titaniferous, more feldspathic and more aluminous than the typical hand specimen.
Recalculating soil compositions assuming the basalt component is represented by the average hand specimen predicts more highland component and more orange glass than is present
The analysed compositions of sieved fractions of the soils have been fitted to a mixture of orange glass, highland materials and the average hand-specimen composition by Vaniman et al. (1979, fig. 5), who showed that unit D in the Apollo 17 core was calculated to contain almost twice as much highland-derived material, and substantially less basalt, than is actually observed in the samples. Vaniman & Papike (1977b) provided modal data for coarse and fine sieved fractions of drill core samples 70006–70009. The compositions of all these soils with respect to SiO2, TiO2, Al2O3, Cr2O3, FeO, MgO and CaO fall within ranges which allow them to be described chemically as mixtures of observed local highland components, the orange glass and a basalt component represented by the hand specimens (see Figs 3–8, below). But these soils can be described equally well by substituting a more feldpathic basalt composition than that of the hand specimens, resulting in lower percentages of calculated highland material and orange soil in the results. Which of the two yields the better description of the modal compositions?
Laul et al. (1978) provided the compositions and calculated mass fractions of highland derived materials, pyroclastic glass (74220) and basalt (assumed to be of hand-specimen average composition) from the sieved soil samples modally analysed by Vaniman et al. (1979). The observed volume fractions of particular components may, in principle, be compared with the mass fractions calculated with the assumption that the basalt has the hand-specimen compositions.
However, the soils consist mainly of dark matrix breccias and agglutinates (here grouped as the fused soil component), orange and black pyroclastic glasses, basalt lithic fragments, highland lithic fragments, and mineral fragments of olivine, pyroxene, plagioclase and opaques (here grouped as the mineral fragment component) that derive from both basalt and highland sources. The comparison has to be viewed in light of (1) the different densities of the components (addressed later), (2) the consideration that appropriate proportions of the mineral fragments should be assigned to the mare and highland sources, respectively, and (3) that the fused soil component should also be assigned between glass, highland and basalt sources in some appropriate manner. The procedure adopted in this section has been to assign the total mineral fragment component to highland or basalt, in proportion to the volume fraction of the lithic fragments that are identified as highland- or basalt-derived, respectively. The modified amounts of highland and mare fragmental materials obtained by adding this contribution to the directly observed component were then summed with the observed orange and black glass component and their volume fractions in this total then used to apportion the volume of the fused soil component between glass, modified highland and modified basalt and summed with the respective components.
This procedure is vulnerable to the possibility that the fused soil component is fossil from an earlier soil which contained different proportions of glass, highland lithologies and basalt (see below) and to the overestimation of the basalt ‘share’ of the mineral fragments, because some of the basalt contribution to the basalt:highland ratio in the lithic fragments will be fine grained or vitrophyric and less likely to have contributed to the mineral fragment total.
Figure 1 plots the individual data points for the basalt (large filled circles), glass (small open circles), and highland (filled squares) components on the above basis. Strictly, the ‘observed’ volume percentages ought to be corrected for the differences in density between the various components, but this would be both tedious and potentially contentious with respect to the density to be adopted for the basalt component. All workers can probably agree that the basalts are on average denser than the highland materials and these in turn denser than the glasses. Consequent corrections to observed volume percentages of the basalt will tend to increase the ratio of observed/calculated in Fig. 1 whereas similar corrections to the observed volume percentages of highland material and glass will tend to decrease the ratio of observed/calculated in Figs 1 and 2, these adjustments enhancing the discrepancies between calculated and observed in both cases. The calculated basalt component is consistently low (by a factor 0·75) relative to the ‘observed’ volume fraction of basalt in Fig. 1. The calculated highland component is consistently high (by a factor 2) relative to the ‘observed’ volume fraction of highland materials. The calculated glass component is also somewhat higher (by a factor 1·3) than that ‘observed’. The discrepancies in the data for the highland and basalt components would obviously be reduced if a more feldspathic average basalt composition were utilized in the calculated proportions. The discrepancies will also be reduced because the average agglutinate present may be more feldspathic than most of the bulk soils in the upper part of the core (Taylor et al., 1978) but this does not appear to be the whole explanation.
|
), orange and black glasses (small symbols) and mare basalt (•) point counted in sieved fractions of horizons of the Apollo 17 drill core are plotted in each case as a function of the percentage of that component which was calculated to be present assuming that the basalt component has the composition of the hand specimens. Basalt points mostly fall above the 1:1 line (more basalt is actually present than is ‘required’, a bias which will be enhanced when relative densities are taken into account); highland and to a lesser extent pyroclastic glass points mostly fall below the 1:1 line (less of both can be seen than is ‘required’, a bias which will also be enhanced when relative densities are taken into account).
Figure 2a plots the volume fraction of observed highland lithic fragments (small filled circles); the latter plus the proportional share of the fused soil component arbitrarily assuming that highland material comprises 20% by volume of that component (open circles); and the latter plus also the proportional share of the mineral fragments (large filled circles) as functions of the mass fractions calculated on the above assumptions. Even the most favourable estimate suggests that there is significantly less highland material actually present in the soils than is indicated by calculations employing an average composition for the basalts which is that of the (plagioclase-poor) hand specimens. Vaniman et al. (1979) argued that highland-derived plagioclase and pyroxene among the mineral fragments could be identified from their probe data and that the plagioclase:pyroxene ratio is much higher in these materials. If correct, this would lead to the anomalous situation that the putative highland-derived component in some soil samples would exceed that anticipated from lateral scattering even before the contributions from fused soil and lithic fragments were taken into account (the possibility of substantial vertical mixing is dismissed below). Furthermore, there is a high probability that petrographically identified ‘highland’ materials may include a substantial proportion of feldspar and pyroxene derived from ancient coarse-grained mare gabbros (O’Hara, 2000, note 61), undermining the criteria by which mineral fragments at the Apollo 17 site were assigned to highland or mare provenance.
The observations are, however, also open to the interpretation that the fused soil component is not representative of the bulk soil composition and contains a much higher proportion of highland-derived materials. Taylor et al. (1978) did find that the average agglutinate composition in some of the soils, particularly those richest in basalt fragments, was significantly more feldspathic than that of the average soil at the same depths, an effect that they attributed to the agglutinates having formed in mature soils to which the basaltic debris had been added later. This does not, however, establish the provenance of the extra feldspar, which might be derived either from a feldspathic basalt forming the upper parts of a single flow or of a sequence of flows, or be derived from highland sources.
Figure 2b plots the volume fraction of observed highland lithic fragments (small filled symbols); the latter plus the proportional share of the fused soil component now arbitrarily assuming that highland material comprises 60% by volume of that component (large open symbols), the value necessary to produce acceptable agreement between calculated and ‘observed’ highland components; and the latter plus also the proportional share of the mineral fragments (large filled symbols). Such a high value for highland contribution in the ‘fossil’ soil represented by the fused soil component, taken with the indicated 20% contribution from orange glass, leaves only a minor contribution (20%) from basalt and implies a fossil soil less basaltic than any other analysed valley soil, less basaltic even than the titanium-poor sieved fines from the drill core samples, and with a TiO2 content of barely 5 wt %. This is not a satisfactory solution to the problem and the simplest explanation is that the average basalt component in the regolith is more feldspathic than the average of the hand specimens.
Horizontal and vertical mixing
The Apollo 17 drill core was collected at a site 3 km from the nearest point on the North massif and 5·5 km from the nearest point on the South massif. The results of Arvidson et al. (1975) suggest that 2·5% of the material at the site will be derived by lateral scattering from distances >3 km, and of that, no more than half (1·25% of the total) would be expected to derive from the North massif. About 1% of the material at the site is expected to have arrived from distances >5·5 km to the south and most of that might be highland material. Accordingly, 2·25% of the total debris in the drill core might be expected to have arrived by lateral scattering from highland sources. Both calculated and recorded amounts of highland material significantly exceed these expectations (Figs 1 and 2).
Korotev & Kresmer (1992) conducted a related exercise to that of Laul et al. (1978), using the surface and trench soils recovered from 25 sampling stations, and extending the fitting to a broad selection of trace elements as well as to major elements. They found that the most satisfactory fit was obtained when the basalt component was assumed to consist of 92% average hand-specimen high-titanium basalt mixed with 8% Very Low Titanium (VLT) basalt (thus producing an ‘average’ basalt which was more feldspathic and less titaniferous than the hand specimens). They also noted that the least satisfactory fits using this basalt composition were obtained in the most basalt-rich samples, an indication that the composition of the basalt component had been incorrectly specified. Furthermore, the presence of 8% VLT component cannot be sustained in the light of the extreme scarcity of VLT lithic fragments in the core. Korotev & Kresmer recognized the difficulty of accounting for their calculated highland component in the soils by lateral scattering and opted in favour of efficient vertical mixing from some underlying highland floor, quoting Rhodes (1977), who explored a similar mechanism to explain a similar problem at the Apollo 15 site.
It is worth reciting briefly the reasons why vertical mixing is an improbable explanation of the ‘excess’ highland component in many mare soils. Solid layered rocks are visible at shallow depth below the mare surface in the walls of Hadley Rille; apparent bedrock basalt is exposed in the floors of relatively small craters at the Apollo 11 and 12 landing sites; basalt, not highland-rich material has been excavated by the large Camelot crater at Taurus–Littrow, and is a probable source of many of the large hand specimens which are anything but biased towards highland compositions; preservation of flow fronts on some maria; preservation of ‘draped basalt crust’ over low hills at the Apollo 15 site without exposure of the presumably shallow underlying highland floor; preservation of thin layers of ancient pyroclastic deposits (the orange soil) on the mare surface at Apollo 17 and other localities; preservation of sharp rims to partially drowned craters in the maria; failure to mix underlying basalt upwards into the avalanche debris from the South massif at stations 2 and 3; and the lack of gradation of composition in the mare surfaces where the mare thickness is decreasing towards the edge against highland rocks. The very limited extent of vertical mixing is further underlined by the excavation of a few conspicuous dark halo craters through thin debris sheets, some probably from the relatively ancient Orientale basin event, overlying earlier mare surfaces (Hawke & Bell, 1982).
Proportions of mineral fragments in basalt-rich soils indicate a feldspathic basalt source
Some soil fractions in the core are very rich in basalt debris. Let it be assumed that the mineral fragments in the coarse- and medium-grained fractions of those soils are wholly derived from the mare basalt component (exclusive of the orange and black glasses). The proportions of these mineral fragments may then indicate the mode of the average coarsely crystalline basalt that was the target for the projectile which formed Camelot crater. Using the modal data of Laul et al. (1979) for the coarse and medium fractions at 28, 38, 47, 57, 71 and 81 cm depth, weighted according to the abundance of these fractions, and then taking a simple average of the six results, one obtains a putative mode for the average target basalt of olivine 2·3%, pyroxene 51·4%, plagioclase 30·7% and opaques 15·4%. This is significantly richer in feldspar and less rich in opaques than the modes of many Apollo 17 hand specimens (Papike et al., 1998, pp. 5–205), but is somewhat analogous to the mode of 75055, which has silica 3·0%, pyroxene 51·4%, plagioclase 29·1% and opaques 13·4% with a chemical composition which contains TiO2 10·79%, Al2O3 9·67%. Sample 75055 is much closer to the low-pressure plagioclase-saturated cotectic liquid compositions than are the majority of the hand specimens (see Figs 3–10, below). This result, although a strong pointer, is open to the obvious alternative interpretation that a substantial proportion of the plagioclase mineral fragments might be derived from highland rocks, even in those soils that contain very low percentages of identifiable highland lithic fragments.
Target basalt composition calculated by subtraction of other debris components from the most basalt-rich soils
Calculation of the average composition of the basaltic component in the soils has been restricted to the two most basalt-rich samples from depths of 38 and 47 cm in the core, to minimize the uncertainties introduced when assumptions are made about the average compositions of the extraneous components. Only data for the coarse-grained sieved fractions of these two soils have been used because these fractions can be expected to contain the lowest proportion of far-travelled debris, the highest proportion of polymineralic fragments of undoubted provenance, and will display the least effect due to surface contamination of the grains by ultrafine particles. The analysed coarse fractions comprise respectively 47 and 44% by weight of their respective soils. Major element compositions of these fractions have been provided by Laul et al. (1978, table 3). Of all the examined portions of the core, these two samples are the richest in basalt, the poorest in orange or black glass and agglutinate, and among the poorest in identified highland component. They comprise (by volume), respectively, 41·58% and 28·06% of basalt lithic fragments; 35·91% and 46·00% mineral fragments dominated by pyroxene, plagioclase and opaques; only 3·78% and 3·68% highland lithic fragments; 9·45% and 10·48% of orange or black pyroclastic glass; and 9·45% and 11·04% fused soil component, predominantly agglutinate (Laul et al., 1978, table 2). ‘Other’ components, including colourless glasses, comprise nil and 0·92% in these samples and are ignored in what follows. The mineral fragment component has been partitioned between highland and basalt sources in proportion to the relative abundance of the respective lithic fragments in the core. Thus a volume of monomineralic fragments of 35·91 x 41·58/(41·58 + 3·78)% has been assigned to the basalt component in the 38 cm depth sample. The revised total volume percentages of basalt- and highland-derived materials contributing to the regolith samples thus become 74·50% and 68·73% basalt-derived, and 6·77% and 9·01% highland-derived. These values were then corrected for their differing densities to obtain approximate weight percentages of the components in these soils. These density corrections assumed mean densities for the basalt, highland-derived, glass and agglutinate components of 3·3, 3·0, 3·0 and 2·7, respectively,
|
|
The mean composition of the agglutinate component has been assumed to be that reported from this depth zone of the core by Taylor et al. (1978). The composition of the orange and black glasses has been assumed to be that given by Papike et al. (1998) and the composition of the highland component has been assumed to be that of the titanium-poor soils collected from stations 2 and 3 on the South massif, an assumption which will maximize the titanium content of the calculated basalt. The compositions of the basalt components in the two samples have then been calculated and are presented in Table 1.
One might argue indefinitely about the details of this calculation without changing the broad conclusion. The average basalt target rock converted to regolith must be more feldspathic and less rich in titanium-bearing oxide phases than the average of the hand specimens.
Regression of oxide concentrations against modal proportions of non-basaltic material in sieved sub-samples of the core
Vaniman & Papike (1977a) provided modal analyses of sieved coarse and medium grain-size fractions for many levels of the Apollo 17 drill core, samples 70006–70009. Some of the horizons were described as ‘almost entirely composed of basaltic material’. Laul et al. (1978) provided chemical analyses of the same samples. It is possible to regress the weight percentage of the oxides against the percentages of various combinations of the modal constituents with the aim of estimating the composition of the basaltic component in the soil. Results are given in Tables A1–A6 of Appendix A in the electronic supplement, regressing each oxide percentage (i) against the modal percentages of identified highland lithic materials and orange and black glass; (ii) against this total plus materials loosely classified as ‘other’; (iii) against identified highland lithic fragments only; (iv) against these plus ‘other’; (v) against the sum of highland lithic fragments plus glasses and a pro rata share of the mineral fragments and fused soil components in each sample as discussed below; (vi) against the sum of highland, dark matrix breccias and agglutinates; (vii) against this total plus the orange and black glasses; (viii) against the sum of everything present which is not either basalt or mineral fragments; and finally (ix) against everything present which is not basaltic lithic fragments. The ranges of values for the modal percentage of non-basaltic materials in each regression are a measure of how far the best-fit line has to be extended to obtain the intercept which yields the estimate of the oxide percentage in the putative basaltic component. These ranges are respectively, 7–30, 9–30, 0·5–10, 1·5–16·5, 10–70, 12–70, 22–78, 22–82, 57–96 modal %. The published data used in the preparation of the Tables A1–A6, and used in the preparation of Figs 3–10, are presented in Appendix D of the electronic supplement.
For each oxide, the solutions for the intercepts obtained tend to decline (SiO2, Al2O3, MgO, CaO) or increase (TiO2, FeO) in the sequence listed, passing from solutions more favourable to the interpretation favoured in this paper towards solutions more favourable to the interpretation that the lithic basalt component in the soils is represented by the composition of the hand specimens.
The quality of the fit is particularly poor in cases (i) and (ii), suggesting that these sieved soil fractions cannot be satisfactorily explained as mixtures of a fairly uniform basaltic component with variable amounts of the orange or black glass and highland-derived materials alone. However, the data points fall close to the oxide axis and the intercepts may have some meaning in indicating the average oxide contents of the basalt + mineral fragment + fused soil component. Two further correlations, with (v) the highland and glass components plus a share of the mineral fragments and fused soil, (vi) and with the highland component plus the fused soil component may have greater significance. These four compositions, together with values of minor oxides obtained from the regressions, have been chosen for representation in later plots (Figs 3–10). They are shown in Table 2, columns 2–5, respectively, where they are compared with the average low-pressure plagioclase-saturated cotectic liquid from experimental charges, the average agglutinate present in this core, the composition of orange glass 74220, and the average composition of hand-specimen basalts from the Apollo 17 site.
The compositions of the putative basic components in the soils derived from the regressions are those of an aluminous basalt (Al/Si 0·31–0·34) comparable with other randomly sampled basalt fragments from a variety of sites (O’Hara, 2000, fig. 2). The compositions might be interpreted variously as a mixture of a basalt with the composition of the cotectic liquids (Table 2, column 1), agglutinate (Table 2, column 6) and orange glass (Table 2, column 7). They can also be interpreted as a mixture of average hand-specimen material (Table 2, column 8) with agglutinate, and at this point the interpretation of the agglutinate composition becomes crucial.
What is the reason for the relatively feldspathic, mafic-poor composition of the average agglutinate? Is it the product of homogenization of a fossil mare soil into which a substantial proportion of feldspar had been introduced (Nagle, 1982) from highland sources? If this was the case, the basalt lithic fragments present might have the composition of the hand specimens. Or is it the product of preferential conversion to agglutinate of material from the upper parts of the flows (which would be relatively more feldspathic in the interpretation favoured here) and of finer-grained material (which is also more feldspathic than the bulk soil composition, partly because of the greater ease with which plagioclase becomes comminuted)? If either of the latter is true, the agglutinate represents almost entirely basalt-derived material. Further discussion of this issue is deferred until after the presentation of chemical variation diagrams for the Apollo 17 bulk soil samples from the drill core, from various sites on the valley floor, and from the adjacent highlands in the South and North massifs.
|
ANALYSIS OF CHEMICAL VARIATION IN THE TAURUS–LITTROW SOILS |
|---|
|
TOPABSTRACTINTRODUCTIONTHE APOLLO 17 REGOLITHSANALYSIS OF CHEMICAL VARIATION...WHY HAVE BASALT LITHIC...SUMMARY AND PREDICTIONSREFERENCES |
|---|
Oxide–oxide plots
Schonfeld (1974
, fig. 2) has provided an informative graphical summary of interpretation of the bulk soil compositions, based on the assumption that the basalt component matches the hand specimens. Several groups of researchers have noted the marked variations in other oxide components with the concentration in Al2O3 in these samples, but in this study the variations with TiO2 content have been preferred because this is the most distinctive parameter defining the local basalt compositions.
Introducing the TiO2–Al2O3 plot
Figure 3 is a plot of the weight per cent of Al2O3 (the prime indicator of the local highland-derived material) as a function of the weight per cent of TiO2 (the prime indicator for the local mare basalt material) in analysed bulk soils from the Apollo 17 site, separated into ‘highland’ soils (blue) and valley surface soils or horizons from the drill core (red).
Highland rocks. The compositions of highland samples from the Apollo 17 mission are plotted in each figure as follows: C, crystalline melt breccias 72395, 73215, 76015, 77135; FB, fragmental breccia 72275; GB, granulitic breccia and granulite 77017, 78155, 79215; N, norite 77215; T, troctolite 76535 (data from Duncan et al., 1974; Nava, 1974; Rhodes et al., 1974, 1976; Rose et al., 1974, 1975).
South massif soils. Soils from stations 2 and 3 (data from Rhodes et al., 1974) are derived from a landslide that may have transported material originating at some distance from the contact with the basalts and ‘refreshed’ the regolith-forming process. These soils form a tight group at the extreme low-TiO2 end of the range. They represent a mixture of granulitic breccias and crystalline melt rocks, dominated by the latter, and these rocks in turn arguably represent averages of huge volumes of highland target material mixed and homogenized in giant impacts. The soils have titanium contents consistent with almost negligible horizontal transport or vertical mixing of high-titanium basalt from the closely adjacent or underlying mare surface by post-landslide cratering events, despite the presence of fairly large nearby craters (Bowen–Apollo, Cochise) on the mare surface.
North massif soils. The soils from station 6 were collected from the very boundary between in situ highlands and mare surface which have been exposed to the full 3·7 Ga of regolith development and any associated lateral and vertical mixing. They have slightly elevated titanium contents but their compositions emphasize the small extent of even short-range lateral transport of material during regolith formation over that time.
Low highland contamination of valley soils. On the above evidence it is reasonable to expect only minor transfer of highland-derived materials in the other direction into the soils of the Apollo 17 drill core. This core was collected from a point much farther from the mare–highland contact (3 km) than were any of the highland site soils other than those from station 2. The same argument for low highland contribution should hold for the soils collected at station 9, 1·5 km from the contact with the North massif. This interpretation is fully consistent with expectations from studies of cratering dynamics (Arvidson et al., 1975; Melosh, 1989). This interpretation further validates the use of Al/Si ratios determined by remote sensing from orbit as an indicator of the average basalt compositions present across the maria at points far more remote from any possible highland input (O’Hara, 2000, fig. 2).
Other highland soils. Other aluminous soils, mostly from stations 7 and 8 closely adjacent to the North massif, display a greater spread towards the mare basalt compositions. The field between 3·5 and 6·0 wt % TiO2 reflects the collection of these soil samples from within a few hundred metres of the boundary between highland and mare surfaces where there clearly may have been significant short-range lateral mixing (vertical mixing cannot contribute mare material to the highland soils at these localities).
Station 9 and lower drill core soils. There are indications within the valley soils of a bimodal distribution with respect to TiO2 content. The station 9 soils are much more aluminous than most of the other basaltic materials collected from Taurus–Littrow, but they are generally similar in this and other plots (Figs 3–10) to the soils from the lower part of the drill core taken at the landing site. They are also similar to the average agglutinate in the upper parts of that drill core, which is interpreted as a mature fossil soil composition excavated by the Camelot crater (Taylor et al., 1978). It is conceivable that a soil of this composition covered much of the local mare surface before the late cratering events. The station 9 soil, however, is no more feldspathic relative to the local hand-specimen basalts than is the soil developed from similar high-titanium basalts at the Apollo 11 site, 40 km distant from the nearest point of highland outcrop, where there can be no question of substantial addition of highland-derived materials. It might be argued that this ancient surface regolith at Taurus–Littrow had become extensively contaminated by transport of material from the adjacent highlands. It is equally credible that this soil is representative of an upper 7·5 m thick layer of the mare fill which was a more feldpathic basalt, either because of some combination of plagioclase flotation within a plagioclase-saturated cotectic liquid and residual plagioclase phenocryst enrichment as a result of preferential sinking of olivine and oxide phases within that liquid, or simply because this was a flow unit of different composition.
Stations 0, 1, 4, 5 and upper drill core soils. Some particularly basalt-rich horizons, thought to be ejecta from Camelot crater, are found in the upper Apollo 17 drill core, 70006–70009. They plot at the extreme high-TiO2 end of the range of soil analyses (7·5–10·5%). Stations 0, 1, 4 and 5 were well away from the North massif but all were close to relatively large late craters whose ejecta may have excavated materials from beneath the early, mature regolith. The surface soils from these localities have compositions broadly similar to those of the upper basalt-rich horizons of the drill core. This cluster of analyses at the high-titanium end of the field of valley soils and drill core analyses may be taken to represent closely the mean composition of mare target materials derived by late cratering events, predominantly from depths of 7·5–37 m. These compositions are very different in composition from, and do not overlap with, the composition field of the hand specimens.
Basalt hand specimens. The regolith at this site is expected to be 8·5 m thick (Taylor et al., 1977) and the most recent craters, formed by secondary impacts from the formation of Tycho far to the south, are thought to excavate to maximum depths of 37 m, implying that the surviving hand specimens may have been derived relatively recently and preferentially from depths of 8·5–37 m beneath the original mare surface with the majority of the fresh basaltic material coming from depths of 10–20 m. None of the soils has TiO2 as high as that of most of the hand specimens. The basalt hand specimens initially distributed for analysis, with the exception of 75055, appear to have been more mafic than the average composition of hand-specimen materials from the site that emerged later (Rhodes et al., 1976; Taylor et al., 1978). Rhodes et al. (1976) identified three chemical groups (A–C) within their hand-specimen analyses, principally on the basis of trace element criteria although the distinction is also apparent in the values for potash and phosphorus pentoxide. They were unable to classify a further 11 samples (group U). These distinctions are not resolved in any of the plots in Figs 3–10. There is, however, a clear distinction between the majority of the hand specimens and a small subgroup of five more feldspathic samples. The latter includes representatives of both group A and group B basalts and their separation is clear in Figs 3, 6, 8 and 9 but completely concealed in Fig. 10; that is, they differ from the majority of hand specimens principally in terms of their content of iron–titanium oxides and olivine.
Basalt rake samples. Warner et al. (1975) presented analyses of 27 small (0·15–0·82 g) basalt chips selected from the rake samples at stations 1, 3, 7 and 8. These analyses are not plotted in the figures, but overlap almost entirely the fields defined by the large hand-specimen analyses, extending them very slightly to high TiO2 values (77536) and more significantly to values of TiO2 lower even than in the experimental cotectic liquids (71559, 78598). Warner et al. illustrated vitrophyric texture in 78587 enclosing frequent skeletal olivine and acicular ilmenite microphenocrysts, not dissimilar from textures described in Apollo 11 lithic fragments (O’Hara et al., 1974); given that small but dense quench crystals might have sunk within these melts (see below), the average groundmass composition from samples such as this might be informative. The most calcic plagioclase (An92) reported by Warner et al. (1975) is similar to that found as microphenocrysts in Apollo 11 samples, and an extended search for plagioclase microphenocrysts among the rake samples and other lithic fragments from the Apollo 17 site might be rewarding. Warner et al. (1975) commented on the bimodal distribution of TiO2 among the rake samples (two samples with <9·0%; 25 with >10·8%) but almost as conspicuous is the bimodal distribution of Al2O3 (four samples with >10·0%; 23 with <9·3%).
Correlations and other compositions plotted. A best-fit line through all the bulk soil analyses and its equation are shown. There is also some suggestion of a less steep trend defined by the compositions of the hand specimens and the mare surface soils. This latter correlation passes through the composition of the average agglutinate, A, in the upper part of the drill core (Table 2, column 6), which may represent the average composition of a mature soil from the regolith that was scattered when Camelot crater was formed. The correlation also passes through the bulk compositions of the coarse fractions, 38C and 47C, of the sieved basalt-rich material from depths of 38 and 47 cm in the drill core and through the estimates of the average crystalline basalt component actually present in the coarse- and medium-grained fractions of the soils B–E (Table 2, columns 2–5, respectively), which are clustered at the high-TiO2 end of the soil trend. Also shown (diamond, unlabelled) is the calculated composition of the average basalt component present in the coarse-grained fraction of the soils at 38 and 47 cm (from Table 1, column 6). In this plot the experimentally produced (Longhi et al., 1974; O’Hara, 2000, table A3) low-pressure plagioclase-saturated liquids also fall on this trend and very close to composition estimates C–E. This secondary trend might project back towards a composition similar to the analysed highland norite, an observation that has compositional significance in the subsequent discussion but no necessary genetic significance. Questions about the composition control in these experimental charges have been discussed elsewhere (O’Hara et al., 1975; O’Hara, 2000, note 119). On that basis it would be expected that the charges produced in experiments by Longhi et al. (1974) would have gained some iron from the capsules and lost sodium, whereas those reported by O’Hara (2000) will have gained a small amount of molybdenum from the capsules. Armalcolite was (just) not present with plagioclase in experiments in iron capsules reported by Longhi et al. (1974), but was present with plagioclase in numerous experiments in both iron and molybdenum capsules reported by O’Hara (2000). Reported cotectic liquid compositions at or close to plagioclase saturation are comparable in the two sets of experiments.
The composition of a large mare basalt clast from Apollo 16 breccia 60639 (Dowty et al., 1974) is also included as an indicator of other ancient aluminous titanium-rich basalt compositions present in this general region of the Moon. There are numerous other examples of aluminous basaltic clasts, mainly involving basalts with lower titanium contents which are less relevant to the present discussion (O’Hara, 2000, fig. 2 and notes 113–116).
Glass component. Orange and black (partially devitrified) high-TiO2 pyroclastic glasses, represented by the analysis of the orange glass, 74220, are common in the soils of the drill core, but their composition (orange circles) falls well off the trend established by the soil compositions.
Effects of crystal addition and subtraction. The vector diagram at top right of Fig. 3 illustrates the effect in this plot of adding 5 wt % of plagioclase (Pl), olivine (Ol), spinel (Sp) and ilmenite (Ilm) respectively to the composition of hand specimen 75055. Mineral compositions were based on those analysed in plagioclase-saturated cotectic experiments (O’Hara, 2000, tables A3–A9). Specimen 75055 is close to a cotectic liquid in composition. The other analysed hand specimens could relate to 75055 by addition of ilmenite, chrome spinel and some olivine in amounts of 5–10% by mass. Conversely, the cotectic liquids may be reached by precipitation of these amounts of phases from liquids of the hand-specimen compositions. The differences between the hand-specimen compositions and the experimental liquids in Figs 3–10 do define an equilibrium liquid line of descent from starting materials of hand-specimen composition at atmospheric pressure. However, the greater part of the variation shown in these figures is not suggested to represent a liquid line of descent. The specific suggestion here is that the average target basalt was a porphyritic magma with composition of 9·5% TiO2 which separated into hand-specimen compositions by accumulation of mafic minerals and into a plagioclase-enriched top, perhaps as aluminous as the station 9 and lower drill core soils, by subtraction of those mafic minerals from the average bulk composition, i.e. the trends between 6·5 and 13·5% TiO2 are the product of movement of phenocrysts within, and relative to, a plagioclase-saturated cotectic magma with 9·5% TiO2. The trend between the station 9 and lower drill core compositions through the North massif soils from stations 8, 7 and 6 towards highland soils like those of the South massif stations 2 and 3 probably reflects physical mixing of the respective end members by impact events.
Other oxide variations with TiO2 content. Figures 4–8 are plots, similar to Fig. 3, of the weight per cent of FeO, SiO2, CaO, MgO and Cr2O3, respectively, as a function of the weight per cent of TiO2. The same materials as in Fig. 3 are shown, amplified, symbolized and annotated as in that figure with minor changes to reflect the monochrome nature of Figs 4, 5, 7 and 8 (colour versions of these figures are available in Appendix B of the electronic supplement). The spinel vector in the TiO2–Cr2O3 plot (Fig. 8), however, reflects just 1 wt % addition of chrome spinel. The commentary on the TiO2–FeO correlation (Fig. 4) and TiO2–SiO2 correlation (Fig. 5) would differ little from that given for TiO2–Al2O3 in Fig. 3. The situation is different for the remaining three oxides.
Although much of the early part of the commentary still applies to Figs 6–8, the compositions of the low-pressure plagioclase-saturated cotectic liquids in these plots fall far off the correlations through the highland and valley soils, or that from the hand specimens through the valley soils towards norite. Liquids produced in experiments at higher temperatures (not plotted in Figs 3–8) fall between the plagioclase-saturated cotectic field and that of the hand specimens. Possible genetic relationships between the material represented by the hand specimens and the basaltic component that is required to explain the soil compositions from the drill core and valley soils are explored below.
The correlations between Al2O3, SiO2 and TiO2 lead to the expectation that the Al/Si ratios in these soils will range from 0·55 (somewhat low for average highland materials) at 0 wt % TiO2, through 0·45 at 4% TiO2 to 0·36 at 7% TiO2 (values of the ratio typical of large areas of mare basalt) and to 0·30 at 9% TiO2 (a value of the ratio typical of randomly sampled high-titanium basalt fragments and of the low-pressure plagioclase-saturated cotectic liquids produced in experiments). At 12–14% TiO2 the Al/Si ratios are 0·20–0·14, well below the range of values measured from orbit (Adler et al., 1972) anywhere in the maria. Al/Si ratios as a function of the TiO2 content in these samples have been obtained by combining the correlations in Figs 3 and 5. These values are shown as a secondary scale in Figs 3 and 5.
Titanium has been independently determined in soils of this region from orbit by the gamma-ray spectrometer experiment (Metzger & Parker, 1979; Davis, 1980). The values reported for the border regions of Mare Serenitatis (MS) and Taurus–Littrow (TL) are also shown against the TiO2 scale in Figs 3, 4 and 5. Total iron was also determined for the same regions by the gamma-ray spectrometer experiment and these values, as FeO, are shown against the vertical scale in Fig. 4. (NB: there are substantial uncertainties attached to these observations from orbit).
Remotely sensed Ti and Fe composition parameters point to soils which are similar in composition to those of station 9 and the upper drill core being present over substantial areas of the local mare surface. Many of these regions are much further from the nearest highland exposure and the feldspathic character of these soils cannot owe much to admixture of highland component. The remotely sensed Al/Si ratios are more consistent with widespread occurrence of soils comparable with those at stations 0, 1, 4 and 5 or the upper drill core. Nothing suggests development of soils from target rock with an average composition close to that of the hand specimens.
Specific questions addressed
Is the regolith a mixture of cotectic liquid, orange glass and agglutinate?
In the correlations between TiO2 and MgO, CaO and Cr2O3 (Figs 6–8) it appears that the required basalt composition falling on the correlations might be made up from some combination of the cotectic liquid and the orange glass, perhaps with a small admixture of agglutinate. This is not, however, an admissible solution because lithic fragments with basalt mineralogy and texture and average compositions falling close to the TiO2-rich end of the field of soils appear to be present in some abundance in the debris from Camelot crater.
Is the regolith a mixture of hand-specimen composition basalt and misattributed highland norite?
More plausible, perhaps, is the possibility that a substantial proportion of highland-derived noritic material present as lithic fragments in the soils has been misattributed to a basaltic origin. This would account simultaneously for the high MgO and low CaO of the ‘basaltic’ component in the soils (Figs 6 and 7) and would permit retention of the hand-specimen compositions as representative of the average composition of the mare basalt target rock at the site. This solution, however, would require that the lithic fragments in all soil samples consist of 30% highland norite misattributed as mare-derived, amounting to at least 12% by weight of the coarse fraction at 38 cm depth in the drill core. It would further require that all valley floor soil samples have suffered at least this extent of highland contamination. This solution seems particularly improbable in the upper layers of the drill core, which are interpreted as a debris sheet thrown out by the excavation of nearby Camelot crater largely from in situ basaltic material. This solution is also in conflict with the evidence for very low complementary transfer of high-titanium material into the closely adjacent highland soils and the expectations from studies of cratering dynamics. Furthermore, it may not fit well with the existence of some hand specimens which are close to the experimental cotectic liquids in composition.
Are the cotectic liquid, orange glass and average basalt related by liquid immiscibility?
Given the near-linear relationship between the orange glass composition, the most basalt-rich end of the valley soil field and the cotectic liquid composition field in Figs 3–8, solutions might be entertained which would involve a liquid immiscibility relationship. This, however, is unsupported by any experimental evidence and is denied by the age evidence. It would require that basalt liquid of a composition close to D either consolidated to basalt entirely without separating into its two immiscible liquids; or separated completely into an orange glass fraction that erupted entirely by fire-fountaining and a low-pressure plagioclase-saturated cotectic liquid fraction that cooled slowly without fire-fountaining and subsequently developed partial cumulates (the hand specimens) but yielded no recovered complementary soils or hand specimens. This seems improbable.
Was basalt of cotectic liquid composition contaminated by orange glass during consolidation?
Slightly less implausible is the possibility that a basalt flow of the low-pressure plagioclase-saturated cotectic liquid composition erupted at Taurus–Littrow, and was then severely contaminated during consolidation by admixture of orange glass liquid fire-fountaining close to the site. Subsequent regolith would be preferentially formed from the products of the hybridized liquid