Journal of Petrology

This Article Abstract FREE Full Text (PDF) Supplementary Data Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Request Permissions Google Scholar Articles by Neumann, E.-R. Articles by Wulff-Pedersen, E. Search for Related Content Social Bookmarking          What's this?

Journal of Petrology | Volume 38 | Number 11 | Pages 1513-1539 | 1997
© Oxford University Press 1997

The Origin of Highly Silicic Glass in Mantle Xenoliths from the Canary Islands

E.-R. Neumann^* and E. Wulff-Pedersen

Mineralogisk-Geologisk Museum, University of Oslo, Sarsgt. 1, N-0562 Oslo, Norway

Received April 5, 1997; Revised typescript accepted July 1, 1997

	ABSTRACT

TOP ABSTRACT Introduction The Host Xenoliths Microstructures of Glass Analytical Procedure Chemical Compositions of the... Minerals Coexisting with Glass... Chemical Compositions of the... Discussion Conclusions References

 
Spinel harzburgite, lherzolite, dunite and wehrlite mantle xenoliths from the Canary Islands (La Palma, Hierro, Tenerife and Lanzarote) contain a spectrum of silicate glasses as inclusions in minerals, along grain boundaries, and in interstitial glass pockets. These glasses show a range in composition form basaltic (44 wt % SiO₂), to highly silicic, TiO₂–FeO–MgO–CaO–P₂O₅-poor types (up to 71 wt % SiO₂). Glasses in spinel harzburgites and lherzolites are generally silica oversaturated, whereas those in spinel dunites and wehrlites have somewhat lower SiO₂ contents and are generally silica undersaturated. Glasses in xenoliths from La Palma and Tenerife are rich in K₂O compared with those from Hierrro and Lanzarote. Daughter minerals coexisting with highly silicic glass in polyphase inclusions are similar in composition to the main phases in the host xenoliths (Fo>90, Cr-diopside, chromite), whereas those in less silicic glasses are richer in Al₂O₃, TiO₂ and FeO, and poorer in MgO. The systematic relations found to exist between glass composition, mineralogy of the host xenolith and locality (island) cannot reflect random variations in the geochemistry of ‘exotic’ melts infiltrating the mantle lithosphere, but instead suggest a cogenetic relationship between the melts and their mantle host xenoliths. The silicic glasses are interpreted as the products of reactions at 8–12 kbar between infiltrating alkali basaltic magmas and peridotitic wall-rocks which, in orthopyroxene-bearing rock-types, involves formation of silicic melt+olivine at the expense of orthopyroxene. In xenoliths from La Palma and Tenerife, where interstitial phlogopite is commonly present, phlogopite has been partly or totally consumed by the reactions between relatively mafic melts and peridotite, giving rise to silicic glasses with high K₂O contents and K₂O/Na₂O ratios. The low K₂O concentrations and K₂O/Na₂O ratios in glasses in anhydrous xenoliths suites from Hierro and Lanzarote are believed to result for reactions between infiltrating melts and anhydrous and/or amphibole-bearing mantle wall-rocks. The silicic melts appear to have been mobile over distances exceeding the diameter of a xenolith, that is, at least 20–30 cm.

KEY WORDS: Canary Islands; silicic glass inclusions; mantle xenoliths; melt–wall-rock reactions

	Introduction

TOP ABSTRACT Introduction The Host Xenoliths Microstructures of Glass Analytical Procedure Chemical Compositions of the... Minerals Coexisting with Glass... Chemical Compositions of the... Discussion Conclusions References

 
Highly silicic glass (up to 72 wt % SiO₂) is commonly found as inclusions and as interstitial glass pockets in upper-mantle peridotites (e.g. Frey & Green, 1974; Francis, 1976, 1987; Jones et al., 1983; Siena et al., 1991; Schiano et al., 1992, 1994, 1995; Ionov et al., 1994; Schiano & Clocchiatti, 1994; Neumann et al., 1995; Zinngrebe & Foley, 1995; Wulff-Pedersen et al., 1996a). Reported host xenoliths include both anhydrous and hydrous spinel-bearing harzburgites, lherzolites and dunites from continental, oceanic and island-arc tectonic settings (e.g. Schiano et al., 1992, 1994, 1995; Schiano & Clocchiatti, 1994; Ionov et al., 1994; Neumann & Wulff-Pedersen, 1995).

The origin of silicic glass in mantle xenoliths, and its role in mantle processes, has been the subject of vigorous debate. A number of workers (e.g. Frey & Green, 1974; Francis, 1976) have attributed the formation of silicic melts to the breakdown of amphibole in response to decompression during transport of the xenoliths to the surface, and heating by the host lava. It has also been proposed that silicic melts may form by partial melting of mantle xenoliths during short residence times (up to a few years) in crustal magma chambers during ascent to the surface in the host magma (Klügel et al., 1996). Such melts would have no bearing on mantle processes. Other models imply formation of silicic melts at mantle depths. These include immiscible separation of a single melt into coexisting silicic and carbonate melts (Schiano et al., 1994); small degrees of partial melting of subducted crust followed by percolation of the melts into the overlying depleted mantle wedge in volcanic arcs (Schiano et al., 1995); infiltration by migrating metasomatic melt phases genetically unrelated to the mantle rock in which they are found (Edgar et al., 1989; Schiano et al., 1992, 1994, 1995; Schiano & Clocchiatti, 1994); reactions between infiltrating basaltic melts and peridotite wall-rock (Zinngrebe & Foley, 1995; Wulff-Pedersen et al., 1996a); in situ melting involving breakdown of amphibole±phlogopite (Amundsen, 1987); in situ melting of clinopyroxene + spinel±amphibole (Francis, 1987; Chazot et al., 1996); disequilibrium in situ melting involving largely clinopyroxene and spinel owing to reaction with migrating fluids (Ionov et al., 1994); and partial melting of peridotite which has previously been metasomatized by carbonatitic melts (Hauri et al., 1993).

Recently, experimental investigations have indicated that highly silicic melts may form by small degrees of in situ partial melting in the upper mantle. Baker et al., (1995) showed that at 10 kbar, near-solidus melts (melt fraction F=0.02–0.05) are enriched in SiO₂ (57 wt % SiO₂ at F=0.02), Al₂O₃ and Na₂O, and depleted in FeO_total, MgO and CaO relative to melts formed by higher degrees of melting, and exhibit a strong increase in SiO₂ and alkalis and decrease in FeO, MgO and CaO with decreasing melt fraction at near-solidus conditions. Similar results were obtained by Drury & FitzGerald, (1996) with a melt fraction of 0.0002. Draper & Green, (1997) observed that under upper-mantle pressures and temperatures, silicic (56–62 wt % SiO₂), aluminous, alkaline melts, typical of silicic glasses found in mantle xenoliths, have near-liquidus mineral assemblages and mineral compositions which indicate equilibrium with a harzburgite residue, both in the presence of a CO₂–H₂O fluid and under anhydrous conditions. Draper & Green, (1997) proposed that silicic, aluminous, alkaline melts may form by low-degree partial melting of peridotite enriched in alkalis, volatiles, and other low-melting-temperature components.

The aim of this study is to establish the origin of silicic glasses in unveined upper-mantle xenoliths from the Canary Islands. Wulff-Pedersen et al., (1996a) suggested that highly silicic glass in veined xenoliths from La Palma have formed as the result of reactions between infiltrating basaltic melts and peridotite wall-rock. An important question is whether this model has general application to silicic melts in peridotites in the Canary Islands (and other localities). Our approach in the present study is to test if any relationship exists between glass composition and type of host xenolith, mode of occurrence (or relative age) of the glass, and/or locality. As a basis for this study we have chosen ultramafic xenolith suites from four islands: essentially anhydrous xenoliths from Hierro and Lanzarote, and hydrous xenoliths from La Palma and Tenerife. The main rock types are spinel harzburgites and dunites; lherzolites and wehrlites are relatively rare, but have been included where available. This study has revealed systematic relationships between glass composition and type of host xenolith, mode of occurrence of the glass, and locality (island), which strongly support the ‘infiltrating melt–wall-rock reaction model’ and indicate that the compositions of highly silicic melts to a large extent are controlled by the modal and chemical composition of the mantle wall-rock. We found no relationship between the compositions of the glasses and the host lavas.

	The Host Xenoliths

TOP ABSTRACT Introduction The Host Xenoliths Microstructures of Glass Analytical Procedure Chemical Compositions of the... Minerals Coexisting with Glass... Chemical Compositions of the... Discussion Conclusions References

 
The silicic glasses discussed in this paper occur in ultramafic xenoliths from the Canary Islands belonging to Group I of Frey & Prinz, (1978). Group II xenoliths (wehrlites, clinopyroxenites, dunites with Fo<87 in olivine and relatively Ti–Fe-rich clinopyroxene and spinel) from the same localities contain basaltic glasses (<50 wt % SiO₂; Neumann & Wulff-Pedersen, 1995), but these glasses will not be discussed here. The Group I xenoliths comprise refractory spinel harzburgites, rare spinel lherzolites, spinel dunites, and rare spinel wehrlites. All these rock-types contain Mg-rich olivine (Fo>89), Cr-diopside and chromite; orthopyroxene is a major phase in harzburgites and lherzolites, but is generally absent in dunites and wehrlites. Phlogopite is a common accessory phase in all types of xenoliths from La Palma and Tenerife, but has not been observed in xenoliths from Lanzarote, and only in one Group I xenolith from Hierro. The Group I xenoliths are interpreted as fragments of the oceanic lithospheric mantle that have been subjected to alternating episodes of partial melting and metasomatic enrichment in highly incompatible trace elements (Neumann et al., 1995; Wulff-Pedersen et al., 1996a). Detailed discussion of the petrography and mineral chemistry, chemical composition and origin of the xenoliths may be found in Johnsen, (1990), Neumann, (1991), Hansteen et al., (1991), Neumann et al., (1995) and Wulff-Pedersen et al., (1996a). A summary of these data is given below. P–T estimates, based on conventional mineral geothermometry and densities of CO₂ inclusions, give minimum temperatures of about 900°C, and minimum pressures of origin of 12 kbar for Hierro (Hansteen et al., 1991; Neumann, 1991), and 6–8 kbar for Lanzarote (Johnsen, 1990; Neumann et al., 1995).

The discussion is based on 674 analyses of glasses in 50 mantle xenoliths (29 spinel harzburgites, 5 spinel lherzolites, 14 spinel dunites and 2 spinel wehrlites). The xenoliths were collected in one locality in each of the islands La Palma, Hierro and Tenerife, and four localities in Lanzarote (Fig. 1); each locality contains different types of xenoliths. All the xenoliths were collected in cinder cones. In Hierro and Tenerife the xenoliths appear to be concentrated in a single layer. Sample identifications are of the type XXii-jj, where the letters XX indicate island (PAT, La Palma; H, Hierro; TF, Tenerife; LA, Lanzarote), the number ii gives locality in that island, and the number jj is sample number at that locality (sample TF14–52 thus means sample 52 from locality 14 in Tenerife).

View larger version (25K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 1. Map of the Canary Islands showing xenolith localities as black crosses.

 
Spinel harzburgite and lherzolite xenoliths
Harzburgite and lherzolite xenoliths are protogranular or porphyroclastic [following the nomenclature of Mercier & Nicolas, (1975)]; three generations of crystal growth may be distinguished. The oldest generation consists of highly strained porphyroclasts of olivine (up to about 25 mm long) and orthopyroxene (containing exsolution lamellae of spinel±clinopyroxene) with highly irregular grain boundaries. In La Palma xenoliths large rounded spinel grains are also interpreted as porphyroclasts. Porphyroclasts are commonly very rich in glass and fluid inclusions (see below). The second generation of crystal growth is represented by mildly strained to unstrained neoblasts of olivine+Cr-diopside+chromitephlogopite±orthopyroxene. Neoblasts occur as granular, equidimensional grains with interlocking grain boundaries (<1.0 mm in diameter), as irregular, interstitial grains, or as symplectitic intergrowths of clinopyroxene + spinel±olivine. Equidimensional neoblasts with interlocking grain boundaries are often found along the rims of phenocrysts with very irregular grain boundaries, and in narrow zones crosscutting porphyroclasts, and appear to grow at the expense of these. Interstitial grains of clinopyroxene frequently enclose corroded grains of olivine and orthopyroxene, and/or vermicular spinel. Clinopyroxene neoblasts often exhibit spinel exsolution lamellae, and orthopyroxene neoblasts occasionally have exsolution lamellae of clinopyroxene. The neoblast generation has been related to in situ heating and metasomatism during the Canary Islands magmatism (Neumann, 1991; Neumann et al., 1995; Wulff-Pedersen et al., 1996a). In addition to these textures, many spinel harzburgites from Tenerife exhibit large, irregular, poikilitic, mildly strained or unstrained orthopyroxene (up to about 6 mm long) and clinopyroxene grains (up to about 4 mm long), enclosing smaller grains of rounded to corroded olivine, pyroxene and spinel; fluid inclusions are rare. Some samples contain exsolved orthopyroxene porphyroclasts with exsolution-free domains surrounding olivine or clinopyroxene grains (up to 1 mm in diameter). Where several such domains occur close to one another, they resemble the poikilitic pyroxenes. Poikilitic orthopyroxene and Cr-diopside thus appear to have formed as the result of extensive recrystallization during the neoblast generation. Another interesting feature is closely spaced, parallel rows of minute, platy spinel inclusions that may sometimes be followed continuously from orthopyroxene porphyroclasts into olivine neoblasts (Neumann, 1991). Some samples show similar parallel rows of spinel inclusions cutting straight through clusters of olivine neoblasts of different crystallographic orientations (Fig. 2). Interstitial silicic glass is occasionally present in such areas. Olivine neoblasts with parallel rows of minute spinel inclusions are believed to have formed by incongruent melting of orthopyroxene, and to have inherited the spinel inclusions from the pre-existing orthopyroxene. Phlogopite (in xenoliths from La Palma and Tenerife) occurs as interstitial grains, and in polyphase inclusions (glass + phlogopite ± clinopyroxene ± spinel). Interstitial phlogopite often encloses spinel or small neoblasts of olivine, and is frequently associated with interstitial glass. In Hierro phlogopite has only been found in a single polyphase inclusion in olivine (glass + phlogopite + spinel) in a spinel harzburgite xenolith. Sulphide globules occur as scattered monomineralic inclusions in minerals, as parts of polyphase inclusions, and as inclusions in interstitial glass. The third, and youngest, generation of crystal growth consists of microlites of spinel, olivine, and clinopyroxene in interstitial glass pockets.

View larger version (139K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 2. (a) Closely spaced parallel rows of myriads of minute, platy spinel inclusions (black) cutting several olivine neoblasts of different crystallographic orientation in spinel dunite PAT2–113 from La Palma (crossed polars). (b) Same in ordinary light. The spinel lamellae show a sharp termination inside a larger olivine grain. This is believed to mark the boundary of the pre-existing orthopyroxene grain. (c) Clear glass (gl) (medium grey), skeletal lazurite (laz) (bright), and euhedral clinopyroxene (cpx) in interstitial glass pocket in spinel dunite PAT2–116.

 
Spinel dunite and wehrlite xenoliths
Spinel dunites and wehrlites are porphyroclastic to equigranular. Olivine porphyroclasts are strongly deformed whereas neoblasts and equigranular rocks are moderately deformed. Orthopyroxene is rarely present, and the xenoliths do not contain glass–olivine aggregates which might be interpreted as the result from incongruent melting of pre-existing orthopyroxene. Phlogopite is common in dunites from La Palma and Tenerife. Amphibole (pargasite) has only been observed in Group I dunites and wehrlites from La Palma where it is a rare accessory phase in dunites, but present in considerable amounts in wehrlites. Sulphide globules are more common in dunites and wehrlites than in harzburgites and lherzolites.

	Microstructures of Glass

TOP ABSTRACT Introduction The Host Xenoliths Microstructures of Glass Analytical Procedure Chemical Compositions of the... Minerals Coexisting with Glass... Chemical Compositions of the... Discussion Conclusions References

 
Glass shows different modes of occurrence and relative age. Glass inclusions in olivine porphyroclasts generally form trails, indicating a secondary origin. The inclusion trails sometimes stop at the boundaries of neoblasts, indicating an age younger than the porphyroclasts but older than the neoblasts. The inclusions in these trails range from negative crystal shape (Fig. 3a) to elongate irregular shapes (Fig. 3b and c), indicating different stages of healing. The glass is colourless or brownish, and both types may occur in the same thin-section (e.g. spinel harzburgites H1–4 and H1–7). Glass inclusions generally contain fluid bubbles (Fig. 3b) and in some cases also daughter minerals (clinopyroxene, spinel, phlogopite; Figs 2d, 4a and b). With the exception of some dunites from Lanzarote that contain CO₂ + N₂ (Andersen et al., 1995), fluid-bearing inclusions in xenoliths from the Canary Islands have been found to consist of pure CO₂ (Hansteen et al., 1991; Frezzotti et al., 1994; Neumann et al., 1995; Wulff-Pedersen et al., 1996a). In some samples glass inclusions or glass-bearing inclusions also contain sulphide globules (e.g. H1–4).

View larger version (63K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 3. (a) Secondary trail of glass (g) + fluid (F) inclusions with negative crystal shape in olivine porphyroclast (o) cutting fluid inclusion trails with very small inclusions (spinel dunite TF14–4). The glass is in the early stages of devitrification. (b) Secondary trail of irregular glass (g) + fluid (F) inclusions cutting fluid inclusion trails in olivine porphyroclast (o) (spinel harzburgite PAT2–59). (c) Glass (g) in poorly healed fracture through olivine porphyroclast (o) (spinel dunite PAT2–116). (d) Polyphase inclusion consisting of phlogopite (ph), glass (g) and fluid (F) in spinel dunite PAT2–49.

 

View larger version (50K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 4. (a) Inclusion consisting of clear glass (g) and fluid (F) (empty) and euhedral olivine daughter mineral or neoblast (o) inside orthopyroxene porphyroclast (p) (spinel harzburgite PAT2–61). The presence of small, scattered, primary glass (g) and spinel (s) inclusions in the olivine grain should be noted. (b) Polyphase inclusion consisting of glass (g) + cpx (c) + fluid (F) (empty) in olivine porphyroclast (o) (spinel harzburgite PAT2–59). The compositions of different phases are given in the tables to the right of each figure; analysis locations are indicated in the figures by numbers corresponding to the column numbers.

 
Olivine and clinopyroxene neoblasts frequently exhibit a central domain rich in small inclusions of colourless to pale brown glass±fluid±spinel±phlogopite; these inclusions have negative crystal shape, or they are rounded to vermicular (Fig. 3a). Their mode of occurrence suggests that they are primary, and represent melt trapped during growth of the neoblasts. Small, scattered, vermicular glass±fluid inclusions also occur in the rims of olivine porphyroclasts.

Orthopyroxene porphyroclasts commonly contain numerous, scattered inclusions of colourless glass. Small glass inclusions (a few micrometres in diameter) frequently exhibit negative crystal shape, whereas larger ones are irregular and are often found along the rims of euhedral to subhedral olivine neoblasts located inside porphyroclasts (Fig. 4a). These inclusions may contain fluid bubbles and/or daughter minerals of chromite and/or Cr-diopside, and in some cases sulphide globules. In xenoliths from La Palma and Tenerife the phase assemblage in glass-bearing inclusions in orthopyroxene porphyroclasts may include phlogopite. The observed transition from exsolved orthopyroxene with silicic glass + olivine inclusions, to large clear poikilitic orthopyroxene in Tenerife xenoliths, and parallel rows of platy spinel inclusions cutting several olivine neoblasts, indicates that the neoblast generation involved formation of silicic glass + olivine±clinopyroxene at the expense of orthopyroxene.

Spinel porphyroclasts often contain numerous rounded to vermicular inclusions and ‘tunnels’ consisting of glass±clinopyroxene ± orthopyroxene ± olivine ± fluid. These inclusions are commonly concentrated along the rims, but may be found throughout large grains. Locally glass-filled ‘tunnels’ continue into interstitial glass. The host spinels have embayed, highly irregular outlines and are zoned with higher Cr₂O₃, FeO_total and TiO₂ in inclusion-rich than in inclusion-free domains (Wulff-Pedersen et al., 1996a).

In addition, many xenoliths contain brownish or colourless interstitial glass. Contacts between olivine grains often display a thin coating of very small, glass ‘droplets’. Locally these layers expand into continuous films of glass along grain boundaries, and larger domains or glass pockets, and may continue into narrow glass + fluid veinlets that cut porphyroclasts. These veinlets are clearly younger than the inclusion trails. However, we wish to emphasize that the interstitial glass and associated veinlets are local phenomena inside the xenoliths; they are not connected with the enclosing basaltic magma (host magma), or veinlets of host magma penetrating into the xenoliths. Large pockets of interstitial glass are most common around the rims of orthopyroxene and spinel porphyroclasts. In xenoliths from La Palma and Tenerife, glass is also commonly found along the rims of interstitial phlogopite, which shows corroded contacts against this glass. Interstitial glass commonly contains large, rounded, empty vesicles, sometimes microlites of olivine±clinopyroxene±spinel, and occasionally sulphide globules. Sulphide globules are more common in glass in spinel lherzolites, dunites and wehrlites than in harzburgites, and more common in hydrous than in anhydrous xenoliths, and are believed to be related to metasomatism. In two spinel dunites from La Palma (PAT2–27 and PAT2–116) lazurite {Na_5.7K_0.2Ca_2.0[Al_5.9Si_6.1O₂₄](SO₄,S)_1.6; E. Wulff-Pedersen & E.-R. Neumann, unpublished data, 1996} has been found in highly Na₂O-rich, silica-undersaturated, colourless glass (SiO₂ 51–54 wt %) present in inclusion trails in olivine and in interstitial glass pockets (Fig. 2c). In sample PAT-116 the silicic glass carries sulphide globules in addition to lazurite.

Contacts between colourless, silicic glass inclusions and host minerals are sharp, showing no evidence of reaction. This is in direct contrast to the contacts between xenolith minerals and the enclosing alkali basalt and basaltic veinlets, which are frequently marked by a reaction zone. Reaction zones may consist of (a) a vermicular intergrowth of clinopyroxene and spinel, both of which are considerably more Ti–Al-rich than the clinopyroxene and spinel inside the xenoliths; (b) an intergrowth of Ti–Al-rich clinopyroxene and amphibole, (c) formation of clinopyroxene±olivine±spinel±glass at the expense of orthopyroxene, or (d) a combination of (a)–(c). Similar reaction zones are found against basaltic glass in veined spinel harzburgite and dunite from La Palma (Wulff-Pedersen et al., 1996a).

	Analytical Procedure

TOP ABSTRACT Introduction The Host Xenoliths Microstructures of Glass Analytical Procedure Chemical Compositions of the... Minerals Coexisting with Glass... Chemical Compositions of the... Discussion Conclusions References

 
Major element analyses of glasses and minerals were obtained using an automatic wavelength-dispersive CAMECA Camebax Microbeam electron microprobe fitted with a LINK energy dispersive system at the Mineralogisk-geologisk museum in Oslo. Glass was analysed by scanning an area of 5 µm * 5 µm to 20 µm * 20 µm, counting light elements first, and minerals were analysed by point analyses, using an acceleration voltage of 15 keV, sample currents of 10 nA for glass, 20 nA for minerals, and counting times of 10–30 s per element. The composition of groundmass in host lavas was estimated as the average of 10–20 scanning analyses (20 µm * 20 µm) of adjacent areas, avoiding phenocrysts and xenocrysts. Oxides and natural and synthetic minerals were used as standards. Matrix corrections were performed by the PAP procedure in the CAMECA software. Analytical precision (2) evaluated by repeat analyses of individual grains is <1% for oxides in concentrations of 20 wt %, <2% for oxides in the range 10–20 wt %, <5% for oxides in the range 2–10 wt %, and <10% for oxides in the range 0.5–2 wt %. Representative analyses of glass in different types of mantle xenolith are listed in Table 1, together with (CIPW) normative quartz, nepheline and leucite. An Fe³⁺/Fe_total ratio of 0.5 was arbitrarily chosen for the norm calculations. As the glasses contain very little iron, the estimated amounts of quartz and feldspathoids in the norms are only marginally influenced by the choice of Fe³⁺/Fe_total ratio. A reduction in this ratio from 0.5 to 0.0 causes an increase in normative quartz (or reduction in nepheline + leucite) of <1%. The compositional ranges of glass inclusions in different types of host xenolith from each of the islands included in this study are given in Table 2.

View this table: [in this window] [in a new window]   Table 1: Representative major element analyses and normative quartz (Qz), nepheline+leucite (Ne) contents of glass inclusions and interstitial glass in mantle xenoliths from the Canary Islands, listed by island and type of host xenolith

 

View this table: [in this window] [in a new window]   Table 2: Compositional ranges in glass inclusions and interstitial glasses from diVerent types of mantle xenoliths from the Canary Islands

 

	Chemical Compositions of the Glasses

TOP ABSTRACT Introduction The Host Xenoliths Microstructures of Glass Analytical Procedure Chemical Compositions of the... Minerals Coexisting with Glass... Chemical Compositions of the... Discussion Conclusions References

 
Glasses in mantle xenoliths from the Canary Islands exhibit a considerable range in composition (e.g. 44–71 wt % SiO₂, 0.0–5.8 wt % TiO₂, <1–8 wt % MgO, <1–15 wt % CaO, <1–12 wt % K₂O, 1–2 wt % P₂O₅ and 0.0–0.8 wt % Cl), and show general trends of decreasing concentrations and ranges of TiO₂, FeO_total, MgO, CaO and P₂O₅ with increasing SiO₂ (Tables 1 and 2; Figs 5 and 6). Na₂O and K₂O exhibit considerable scatter (K₂O/Na₂O 0.02–2.0) with the highest concentrations in the most silicic glasses, whereas the Al₂O₃ contents appear to reach a maximum at 55–60 wt % SiO₂. The compositional ranges of glass inclusions in spinel harzburgites and lherzolites from Hierro and Lanzarote obtained by us overlap those published by Schiano et al., (1994), whereas Siena et al., (1991) reported a slightly higher SiO₂ range for glasses in spinel harzburgites from Lanzarote (Table 2).

View larger version (71K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 5. Compositional variations vs SiO2 content in glasses in anhydrous (left column) and hydrous spinel harzburgite and lherzolite xenoliths (right column) from the Canary Islands. Groundmass and glass in host lavas of the xenoliths are indicated by letters: P, La Palma; H, Hierro; T, Tenerife; L, Lanzarote. For comparison are also shown trends defined by glasses in the vein system of veined spinel harzburgite from La Palma, PAT2–4 (grey field; Wulff-Pedersen et al., 1996a), by aphyric lavas in Tenerife (field outlined by dotted line; E.-R. Neumann & E. Wulff-Pedersen, unpublished data, 1996), aphyric lavas in Hierro (field outlined by continuous line; data from Pellicier, 1977, 1979), and mafic MORB (star). Incl, inclusions in orthopyroxene and olivine. The horizontal dashed line in the K2O–SiO2 figure shows the lower limit of the range defined by glasses in hydrous xenoliths. (See text for discussion.)

 

View larger version (66K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 6. Compositional variations vs SiO2 content in glass inclusions in olivine and interstitial glass in anhydrous (left column) and hydrous (right column) Group I spinel dunite xenoliths in the Canary Islands. Symbols as in Fig. 5. (See text for discussion.)

 
In each island the most SiO₂-rich, and TiO₂–FeO–MgO–CaO–P₂O₅-poor glasses are found in spinel harzburgites and lherzolites (Table 2; Figs 5 and 6), that is, in orthopyroxene-bearing rock-types. In La Palma all glasses in harzburgites and lherzolites fall within the range 62–69 wt % SiO₂ (Table 3; Fig. 5). Furthermore, most glasses in harzburgites and lherzolites are moderately to strongly silica oversaturated, whereas those in dunites and wehrlites tend towards silica undersaturation (Fig. 7). We have found no systematic compositional differences between glasses in anhydrous and hydrous xenoliths within the same island, nor between harzburgites and lherzolites. However, Amundsen, (1987) reported higher K₂O contents and K₂O/Na₂O ratios in glasses in phlogopite-bearing than in phlogopite-free xenoliths from Gran Canaria (Canary Islands). Available data (Fig. 7) indicate that silicic glass in spinel harzburgite and lherzolite xenoliths from other locations and tectonic settings also tends towards silica oversaturation. Unfortunately, very few data on glass in spinel dunites and wehrlites are as yet available. The glasses also show inter-island differences with respect to Na₂O–CaO–K₂O relations. Most of the glasses in the anhydrous xenolith suites from Hierro and Lanzarote are poorer in K₂O and richer in CaO and show lower K₂O/Na₂O ratios than glasses with similar SiO₂ contents in the xenolith suites from La Palma and Tenerife, where phlogopite is a common accessory phase (Figs 5 and 6). In an Na₂O–CaO–K₂O diagram, glasses in harzburgite and lherzolite xenoliths from La Palma and Tenerife define trends towards the central part of the Na₂O–K₂O side-line, whereas those from Hierro and Lanzarote fall close to the Na₂O–CaO tie-line (Fig. 8). Glasses in dunites show less clear differences in Na₂O–CaO–K₂O relations.

View this table: [in this window] [in a new window]   Table 3: Representative compositions of euhedral to subhedral crystals coexisting with highly silicic glass in polyphase inclusions (incl) and interstitial glass (interstitial), compared with wall-rock phases (host rock)

 

View larger version (39K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 7. Normative quartz and nepheline + leucite (CIPW) plotted against SiO2 for glass inclusions in olivine and orthopyroxene and interstitial glass in different types of host xenoliths in the Canary Islands (a–d, f). The figure shows that glasses in spinel harzburgites and lherzolites are generally silica oversaturated, whereas those in spinel dunites and wehrlites are silica undersaturated. The ‘open-star’ symbol represents average N–MORB. Groundmass and glass in host lavas of the xenoliths are indicated by letters: P, La Palma; H, Hierro; T, Tenerife; L, Lanzarote. Trends defined by glasses in the vein system of veined spinel harzburgite PAT2–4 and veined spinel dunite PAT2–62, are shown as grey fields in (a) and (b), and in (c) and (d), respectively (grey fields; Wulff-Pedersen et al., 1996a). Trends are defined by aphyric lavas from Tenerife (field outlined by dotted line; E.-R. Neumann & E. Wulff-Pedersen, unpublished data, 1996), and Hierro (field outlined by continuous line; data from Pellicier, 1977, 1979). (f) Published data on glass in spinel harzburgite and lherzolite xenoliths from other locations: Yemen—interstitial glass, glass in glass pockets and veinlets in spinel lherzolites (Chazot et al., 1996); W Eifel—interstitial glass in spinel harzburgites (Zinngrebe & Foley, 1995); Mongolia—glass in glass pockets in spinel lherzolites from Mongolia (Ionov et al., 1994); Oceanic and Continental—glass inclusions in spinel harzburgites and lherzolites from various oceanic and continental localities, respectively (Schiano et al., 1992, 1994;Schiano & Clocchiatti, 1994); Philippines—glass inclusions in spinel harzburgites from Philippine arc lavas (Schiano et al., 1995). Incl, inclusions in orthopyroxene and olivine.

 

View larger version (23K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 8. Na2O–CaO–K2O relations among glasses in (a) spinel dunite and (b) harzburgite and lherzolite xenoliths from the Canary Islands. Glass in veined spinel harzburgite (PAT2–4) and dunite (PAT2–62; Wulff-Pedersen et al., 1996a) are shown as grey fields marked ‘vein glass’. Mafic basaltic lavas (MgO>7 wt %) from the islands of La Palma, Hierro, Tenerife and Lanzarote (data from Fuster et al., 1968a; Pellicier, 1977, 1979; Staudigel, 1981; Hernandez-Pacheco & Valls, 1982; Staudigel et al., 1986; E.-R. Neumann & E. Wulff-Pedersen, unpublished data, 1996) are shown as grey fields marked ‘basalts’. (c) Published data on series of glass inclusions and glass pockets in spinel harzburgites and lherzolites from other localities: GC hydr and GC anhydr—glasses in phlogopite-bearing and phlogopite-free spinel harzburgite xenoliths from Gran Canaria (Amundsen, 1987); Mongolia (Ionov et al., 1994); Gees, West Eifel (Zinngrebe & Foley, 1995), Mt Lessini, Southern Alps, and Cape Verde (Siena & Coltorti, 1993). (See text for discussion.)

 
We have also tested compositional differences against mode of occurrence: (a) glass in secondary inclusion trails in olivine porphyroclasts (assumed to be the oldest generation of glasses), (b) irregular glass-bearing inclusions in orthopyroxene porphyroclasts, and (c) glass in interstitial glass pockets. Clusters of (primary) inclusions in olivine and clinopyroxene neoblasts (intermediate in age between type a and type c) are generally too small to allow analysis; we have therefore not been able to test those as a separate group. Glasses in spinel harzburgites and lherzolites from La Palma show no compositional relation to mode of occurrence. In the other islands there is a weak tendency for glass found in secondary inclusion trails in olivine and as inclusions in orthopyroxene to show a more restricted range in high-SiO₂ compositions and higher average SiO₂ than interstitial glasses (Fig. 5). Dunites show no chemical differences between glass inclusions in olivine and interstitial glass (Fig. 6).

	Minerals Coexisting with Glass in Inclusions and Glass Pockets

TOP ABSTRACT Introduction The Host Xenoliths Microstructures of Glass Analytical Procedure Chemical Compositions of the... Minerals Coexisting with Glass... Chemical Compositions of the... Discussion Conclusions References

 
The compositions of daughter minerals and microlites are clearly correlated with the coexisting glass. Euhedral to subhedral daughter minerals in polyphase inclusions associated with highly silicic glass have compositions typical of the main phases in the host xenoliths, that is, olivine with Fo>90, and Mg–Cr-rich, Ti–Al-poor pyroxenes (Table 3, Fig. 4). The same is true for microlites in glass pockets with highly silicic glass. Less silicic glass, in contrast, contains minerals which are markedly richer in TiO₂, Al₂O₃ and FeO, and poorer in MgO than those in the mantle wall-rock. Minerals of similar compositions are typically found as phenocrysts in alkali basalts, and in Group II wehrlites and clinopyroxenites. The latter are interpreted as mantle cumulates formed from alkali basaltic Canarian magmas (Hansteen et al., 1991; Neumann, 1991). Similarly, veined xenoliths from La Palma show a continuous shift in mineral compositions from relatively Fe-rich olivine and Ti–Al–Fe-rich clinopyroxene, amphibole and phlogopite in the least silicic glass, to Mg-rich olivine and Ti–Al-poor, Mg-Cr-rich clinopyroxene and phlogopite coexisting with the most silicic glass (Wulff-Pedersen et al., 1996a). Corresponding relations between the compositions of glass and daughter minerals-microlites were observed by Zinngrebe & Foley, (1995) in mantle xenoliths from Gees, Germany.

	Chemical Compositions of the Host Lavas

TOP ABSTRACT Introduction The Host Xenoliths Microstructures of Glass Analytical Procedure Chemical Compositions of the... Minerals Coexisting with Glass... Chemical Compositions of the... Discussion Conclusions References

 
As the xenoliths were collected in cinder cones, their host lavas were often difficult to analyse. However, we obtained data on host lava (bulk rock, groundmass, or glass) for a number of samples. Representative analyses are presented in Table 4 and plotted in Figs 5-7. Like Canary Islands basalts in general, the host lavas of the xenoliths are TiO₂ rich, and silica saturated to undersaturated. K₂O/Na₂O values range from 0.31 to 0.71. Within each locality we found only minor compositional differences between host lava attached to different xenoliths. This means that different types of xenoliths are hosted by lava of the same composition. Furthermore, host lavas from different localitites and islands fall on, or close to, the trends defined by aphyric lavas in Tenerife (E.-R. Neumann & E. Wulff-Pedersen, unpublished data, 1996) and Hierro (data from Pellicier, 1977, 1979; Figs 5-7).

View this table: [in this window] [in a new window]   Table 4: Representative analyses of host lavas of xenoliths from diVerent localities

 

	Discussion

TOP ABSTRACT Introduction The Host Xenoliths Microstructures of Glass Analytical Procedure Chemical Compositions of the... Minerals Coexisting with Glass... Chemical Compositions of the... Discussion Conclusions References

 
The origin of the silicic glasses
We noted above that silicate glasses in peridotite xenoliths from the Canary Islands cover a wide range in compositions from basaltic, to SiO₂–Al₂O₃–Na₂O–K₂O-rich, TiO₂–FeO–MgO–CaO–P₂O₅-poor types (Tables 1 and 2; Figs 5 and 6). There is considerable evidence that the highly silicic melts are in, or close to, equilibrium with the spinel peridotites in which they are found, whereas the basaltic melts are not. This is indicated by the sharp contacts between silicic glass and peridotite minerals, in contrast to the reaction-type contacts between xenoliths and basaltic glass seen in both unveined (this study) and veined xenoliths. Furthermore, the compositions of daughter minerals and microlites in highly silicic glass are similar to those of the main phases in the host xenolith (Mg-rich olivine Cr-diopside, chromite; Fig. 4, Table 3), whereas those in basaltic glass are not (e.g. Ti–Fe-rich augite and titanomagnetite; Figs 5 and 6). Another important result of this study is that it reveals systematic relations between glass composition and type of host xenolith, mode of occurrence (or relative age) of the glass, and locality (island). The mineral-melt relations, and contact relations between glass and peridotite minerals found for unveined Canary Islands xenoliths closely resemble those found in the vein systems of veined xenoliths from La Palma (Wulff-Pedersen et al., 1996a). The small compositional differences found to exist among the host lavas cannot account for the chemical contrasts exhibited by glasses in different types of peridotites and different localitites (islands). Most striking is the fact that the host lavas fall on a common K₂O–SiO₂ trend, whereas the silicic glasses in hydrous xenoliths are markedly richer in K₂O than glasses with similar SiO₂ contents in anhydrous xenoliths (Figs 5 and 6). Furthermore, whereas all the host lavas fall close to the TiO₂–SiO₂ trend defined by aphyric lavas in Tenerife and Hierro (Figs 5 and 6), glasses in Hierro xenoliths appear to define both a high-TiO₂ and a low-TiO₂ trend.

The above relations have important implications for our understanding of the origin of silicic glasses in mantle xenoliths. They indicate that the compositional diversity found to exist among glasses in Canary Islands xenoliths does not reflect random chemical variations among infiltrating melts migrating undisturbed through a column of different mantle rock types (Model 1: Edgar et al., 1989; Schiano et al., 1992, 1994, 1995; Schiano & Clocchiatti, 1994). Instead, the observed relations imply a direct association between melts and host xenolith which may arise through reactions between infiltrating basaltic melts and peridotite wall-rock (Model 2: Zinngrebe & Foley, 1995; Wulff-Pedersen et al., 1996a), or through in situ partial melting (Model 3: Amundsen, 1987; Francis, 1987; Hauri et al., 1993; Ionov et al., 1994; Baker et al., 1995; Chazot et al., 1996; Draper & Green, 1997).

Model 2 (reactions between infiltrating melts and mantle wall-rocks) is also supported by a strong decrease in TiO₂/Al₂O₃ ratios from the least silicic to the most silicic glasses (exemplified by data on Hierro in Fig. 9). Experiments at pressures of 10–20 kbar (e.g. Mysen & Kushiro, 1977; Jaques & Green, 1980; Falloon & Green, 1987) indicate only moderate fractionation of Ti relative to Al over a wide range of partial melting of spinel peridotite. Similar results were obtained in near-solidus melting experiments (2–5% melting of spinel lherzolite) at 10 kbar (Baker et al., 1995). This implies that a silicate melt and spinel peridotite in equilibrium should have similar TiO₂/Al₂O₃ ratios. Spinel harzburgites and lherzolites from La Palma, Hierro and Lanzarote are characterized by TiO₂/Al₂O₃ <0.07. This is within the range of ‘normal’ oceanic peridotites and primitive tholeiitic basalts (highest MgO/FeO ratios) collected along ‘normal’ segments of the Mid-Atlantic Ridge (Bryan et al., 1981; Sigurdsson, 1981; Schilling et al., 1983; Weaver et al., 1985; Menzies, 1991). Mafic Canarian lavas (MgO >7 wt %), in contrast, have TiO₂/Al₂O₃ >0.15. Glass in veined spinel peridotite xenoliths from La Palma shows a gradual decrease in TiO₂/Al₂O₃ from about 0.2 in alkali basaltic glass in broad veins, to TiO₂/Al₂O₃ <0.08 in the most SiO₂-rich glass in very narrow veinlets penetrating peridotite fragments (Wulff-Pedersen et al., 1996a; Fig. 9). The variation in glass chemistry in the veined xenoliths was interpreted by Wulff-Pedersen et al., (1996a) as the consequence of melt–wall-rock reactions. These reactions start with infiltration by alkali basaltic melts (with high TiO₂/Al₂O₃ ratios) out of equilibrium with the peridotite wall-rock, and their end-products are silicic melts (with low TiO₂/Al₂O₃ ratios) in, or close to, equilibrium with the wall-rock. The postulate that highly silicic, FeO–MgO–CaO-poor melts may be in equilibrium with spinel peridotite is supported by the experiments of Baker et al., (1995) and Draper & Green, (1997).

View larger version (23K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 9. TiO2/Al2O3 vs SiO2 for glass inclusions and interstitial glasses in spinel harzburgite and lherzolite xenoliths from Hierro. Xen, peridotite xenoliths from Hierro; basalt, basaltic lavas from Hierro with MgO>7 wt % (data from Pellicier, 1977, 1979); vein, field of vein glasses in veined spinel harzburgite PAT2–4 (data from Wulff-Pedersen et al., 1996a). (See text for discussion.)

 
Model 3 (different degrees of in situ partial melting) cannot account for the large differences in TiO₂/Al₂O₃ ratios (Fig. 9) observed among glasses in Canary Islands xenoliths. It is also difficult to reconcile the fact that the least silicic glasses appear not to be in equilibrium with the host peridotite with partial melting sensu stricto. However, mixing between melts formed by very small degrees of in situ partial melting (low TiO₂/Al₂O₃ ratios) and infiltrating mafic melts with high TiO₂/Al₂O₃ ratios cannot be excluded.

Decompression melting (Model 4: e.g. Frey & Green, 1974; Francis, 1976) has been dismissed as a general mechanism for the formation of highly silicic melts by Edgar et al., (1989) and Zinngrebe & Foley, (1995). Textural relations indicate that in the Canary Islands xenoliths SiO₂-rich melts were present in the peridotites before and during the neoblast generation of crystal growth (see description of glass inclusions). This was followed by a period of cooling and exsolution of pyroxenes before entrainment of the xenoliths in the magma which transported them to the surface. Decompression melting has therefore been discarded as a general model for formation of silicic melts in the Canary Islands xenoliths, although we cannot exclude the possibility that some silicic melt has also formed during ascent.

Partial melting during short residence times in shallow magma chambers during ascent (Model 5: Klügel et al., 1996) has also been discarded as a relevant model for the xenoliths of this study. Studies on dissolution rates indicate that ultramafic xenoliths can survive in a host lava for only a few days before they are dissolved and consumed [e.g. Scarfe & Brearley, (1987), and references therein]. Although the Canary Islands xenoliths frequently show narrow reaction rims against the enclosing basaltic magma, their surfaces are well defined. In contrast to glasses in the veined xenoliths from La Palma (Wulff-Pedersen et al., 1996a), occasional veinlets of host magma penetrating xenoliths have maintained basaltic compositions. Formation of the silicic glass by partial melting of mantle xenoliths in crustal magma chambers during ascent to the surface therefore seems highly unlikely.

Also, immiscible separation of mantle melts into a silicate + carbonate melt pair (Model 6: Schiano et al., 1994) seems an unlikely mode of formation for the silicic glasses discussed here. Kogarko et al., (1995) reported interstitial silicate glass (64 wt % SiO₂, mildly silica oversaturated) with carbonate and sulphide globules in wehrlitic alteration zones of Group II composition in spinel harzburgite xenoliths from the Montana Clara Island, Canary Islands, and interpreted this as the result of immiscible separation. The composition of the silicate glass is within the range of glasses analysed by us. Mixed silicate glass + carbonate inclusions have also been observed in composite Group II clinopyroxene–spinel dunite xenoliths cut by clinopyroxenite veinlets from Gomera (Fo_83–85; Frezzotti et al., 1994). However, there the silicate glass associated with carbonate is ultramafic (33–46 wt % SiO₂, 24–38 wt % MgO, 5–18 wt % FeO) and has very low concentrations of Al₂O₃ (0.1–2.1 wt %) and alkalis (<0.1 wt % Na₂O and K₂O), and has no apparent relation to the highly silicic, alkali-rich glass which is the topic of this paper. We have observed carbonate aggregates in interstitial glass in xenoliths from Tenerife, but these aggregates appear to be secondary, filling rounded vesicles. Occurrences which may represent carbonate melt inclusions have not been observed in Group I xenoliths. Without visible evidence that carbonate melts were present, we find it hard to believe that liquid immiscibility has been an important process in the formation of silicic melts in the Group I xenoliths.

We conclude that silicic melts in the upper mantle under the Canary Islands mainly result from reactions between infiltrating, alkali basaltic melts ascending from deeper parts of the mantle, and spinel peridotite wall-rocks (Model 2). The observed tendency for interstitial glass pockets to contain a higher proportion of relatively SiO₂-poor glasses than the inclusions in olivine and orthopyroxene porphyroclasts (Figs 5 and 6) suggests that many interstitial melts have not had time to go through the extensive reactions which may eventually lead to equilibrium with the peridotite wall-rock. Glass in secondary inclusion trails in olivine represents interstitial melts that were trapped in fractures at some earlier stage. The more restricted range of high SiO₂ contents found among these glasses may be the result of reaction and crystallization during healing of the fractures in which the melts were trapped. However, although we regard reaction processes as most important, this does not preclude the possibility that formation of silicic melts by in situ melting, and mixing between infiltrating melts and in situ melts (Model 3, modified) may also have taken place at some stage of the evolution.

Reaction processes
Silicic glasses in unveined, hydrous peridotite xenoliths from the Canary Islands define compositional trends which resemble those exhibited by melts in veined xenoliths from La Palma (Wulff-Pedersen et al., 1996a). It therefore seems likely that the reactions which caused the evolution from infiltrated, basaltic melt to highly silicic melts in the veined xenoliths may be used to gain insight into the evolution from basaltic to silicic melts in the unveined xenoliths. Textural relations in the veined xenoliths combined with petrographic mixing calculations (Wulff-Pedersen et al., 1996a) imply that the highly silicic melts are the products of a series of reactions between infiltrating basaltic melts (and melts formed through the reactions) and peridotite. Reactions between basaltic melts and peridotite include formation of the daughter phases cpx±amph±ol + melt at the expense of primary opx±ol±phlog in harzburgite and lherzolite, and the daughter phases cpx + sp±amph±phlog±melt at the expense of primary ol±phlog in spinel dunite and wehrlite. Daughter minerals formed through these reactions are TiO₂–Al₂O₃–FeO rich (e.g. augite, kaersutite, Fe-rich olivine). As the reactions progress, the melts produced become progressively enriched in SiO₂ and alkalis, and depleted in TiO₂, FeO, MgO and CaO, and phlogopite takes the place of amphibole in the daughter mineral assemblages. The daughter minerals become progressively depleted in TiO₂, Al₂O₃ and FeO, and enriched in MgO, and in highly silicic glass they have compositions close to the peridotite wall-rock phases.

Also, the unveined harzburgites and lherzolites provide considerable petrographic evidence (see description above) for the formation of olivine and silicic melt at the expense of orthopyroxene (e.g. Figs 2 and 4a). This is clearly a major source of SiO₂ enrichment in the melt. Under dry conditions, incongruent melting of orthopyroxene is restricted to pressures below about 5 kbar, but this pressure limit increases to more than 20 kbar in the presence of an H₂O-rich fluid (e.g. Eggler, 1972; Mysen & Boettcher, 1975a, 1975b). Incongruent melting of orthopyroxene within the pressure range of origin of the xenoliths of this study, 8–12 kbar (Johnsen, 1990; Hansteen et al., 1991; Neumann et al., 1995), requires the presence of an H₂O–CO₂ fluid. Silicic glass and phlogopite in harzburgite and dunite xenoliths from La Palma contain 0.21–0.36 and 2.7–3.1 wt % H₂O, respectively (Wulff-Pedersen et al., 1996b). The large vesicles frequently found in glass pockets (Fig. 2) indicate that a fluid phase unmixed from the silicic melts during ascent of the xenoliths, implying that the original fluid content of the melts was considerably higher than that indicated by analyses of the degassed glass. The H₂O + CO₂ content in the upper mantle may thus have been high enough to allow incongruent melting of orthopyroxene to take place at pressures of 8–12 kbar. The frequent presence of phlogopite in mantle xenoliths from Tenerife suggests the presence of H₂O in addition to CO₂ (seen as fluid inclusions) also in the upper mantle under Tenerife. The near absence of phlogopite and amphibole in xenoliths from Hierro and Lanzarote suggests higher CO₂/(CO₂ + H₂O) ratios, or lower fluid partial pressure in the upper mantle below these islands. However, also, mantle xenoliths from these islands bear evidence of incongruent melting of orthopyroxene.

The postulated effects of reactions between infiltrating basaltic melts and peridotite wall-rock are also supported by Kelemen's, (1990) modelling. He showed that liquids saturated in olivine and undersaturated in low-Ca pyroxene, reacting with lherzolite or harzburgite, will dissolve low-Ca pyroxene and crystallize a smaller mass of olivine. Kelemen, (1990) predicted that this kind of reaction will lead to silica enrichment in the derivative liquids. We believe that addition of SiO₂ to the melt through reaction between harzburgite or lherzolite and alkali basaltic melts and derivative liquids, all undersaturated in low-Ca pyroxene, is responsible for the high SiO₂ contents and silica oversaturation in the majority of glasses in spinel harzburgites and lherzolites, as compared with glasses in spinel dunites and wehrlites (Figs 5-7), which, in the Canary Islands, generally do not contain orthopyroxene. Addition of SiO₂ to the melts as the result of breakdown of orthopyroxene accounts for the fact that the most silicic glass in spinel harzburgites and lherzolites is significantly richer in SiO₂ than any of the mineral phases in the host xenolith. Furthermore, Kelemen, (1990) predicted that melts saturated in olivine and calcic pyroxene will stay that way during melt–wall-rock reactions. Formation of clinopyroxene and olivine through melt–wall-rock reactions is in agreement with the decreasing concentrations of CaO, FeO and MgO with increasing SiO₂ in the glass (Figs 5 and 6).

In unveined Canary Islands xenoliths high K₂O contents and high K₂O/Na₂O ratios are found in glasses of xenolith series where phlogopite is a common accessory interstitial phase (La Palma and Tenerife), whereas low K₂O contents and low K₂O/Na₂O ratios are typical of glasses in xenolith series where interstitial phlogopite is very rare (Hierro and Lanzarote; Tables 1 and 2; Figs 5, 6 and 8). It seems likely therefore that the K₂O content in glass is directly related to the presence or absence of phlogopite in the mantle wall-rock with which the melt has reacted. This is in agreement with petrographic mixing calculations which indicated that the strong increase in K₂O with increasing SiO₂ defined by the veined xenoliths required consumption of a K₂O-rich phase, such as phlogopite (Wulff-Pedersen et al., 1996a). Consumption of phlogopite is supported by the corroded contacts between interstitial phlogopite and glass in unveined xenoliths, which indicate that interstitial phlogopite was already present in the upper mantle under La Palma and Tenerife before formation of the silicic melt and was partially consumed during its generation. The following reactions involving phlogopite are compatible with petrographic observations:

Where phlogopite is present, it will play an important role in melt–wall-rock reactions, resulting in melts with high K₂O contents, and high K₂O/Na₂O ratio compared with the initial infiltrating melt. In anhydrous mantle domains alkalis will not be added to the melt through the breakdown of hydrous phases, although some Na₂O may be partitioned into the melt from clinopyroxene. The main change in Na₂O–CaO–K₂O relations will result from removal of CaO from the infiltrating melt (to form new clinopyroxene), causing the Na₂O/CaO ratio in the melt to increase as reactions progress. The K₂O/Na₂O ratio in glasses formed by reactions with anhydrous peridotite is expected to be similar to, or slightly lower than, that of the infiltrating melt, and (Na₂O + K₂O)/CaO and K₂O/Na₂O ratios are expected to be significantly lower than in melts with similar SiO₂ contents in hydrous peridotite. If amphibole is present in the peridotite wall-rock, breakdown of amphibole will add sodium and minor amounts of potassium to the melt, resulting in marked Na₂O enrichment and low K₂O/Na₂O ratios compared with the infiltrating melt.

The high K₂O contents and K₂O/Na₂O ratios in silicic glasses in xenoliths from La Palma and Tenerife as compared with those in xenoliths from Hierro and Lanzarote (Fig. 8) are compatible with reactions involving phlogopite in the peridotite wall-rocks. It should also be noted that the K₂O/Na₂O ratios of these glasses are higher than those of the host basalts and Canary Islands basalts in general (Fig. 8). Within the xenolith suites from La Palma and Tenerife, we see no difference in K₂O content between glasses in xenoliths with, and those without, interstitial phlogopite. This suggests that phlogopite may originally have been more common in the upper mantle under La Palma and Tenerife than is indicated by the modal composition of the xenoliths, but that it locally has been totally consumed through reactions and/or partial melting. Glasses in the anhydrous xenolith suites from Hierro and Lanzarote show K₂O/Na₂O ratios similar to, or below, those in Canary Islands basalts. These ratios are compatible with reactions between anhydrous mantle wall-rock and infiltrating melts covering a range in K₂O/Na₂O ratios, including very low values. However, if the K₂O/Na₂O ratios of the Canary Islands basalts are representative of the infiltrating melts, reactions involving amphibole-bearing peridotite are likely. Amphibole has not been reported in mantle xenoliths from Hierro or Lanzarote (Fuster et al., 1968b; Sagredo Ruiz, 1969; Johnsen, 1990; Neumann, 1991; Neumann et al., 1995), but pre-existing amphibole may have been consumed by reaction processes.

Our data imply that the phase assemblage of the mantle wall-rocks is not only a factor controlling the compositions of melts formed by in situ partial melting, but is also a factor that strongly affects the compositions of silicic melts which are modified through melt–wall-rock reactions. The composition of glass inclusions in a mantle rock series may thus give important information about the evolutionary history of the mantle domain.

Silicic glasses in mantle xenoliths from other localities and tectonic settings also show wide ranges in Na₂O–CaO–K₂O relations (Fig. 8c). Some of these peridotite suites show ‘expected’ relationships between the Na₂O–CaO–K₂O relations in silicic glasses and the modal composition of their host xenoliths, whereas in other suites such connections appear to be lacking. In Gran Canaria glasses in phlogopite-bearing xenoliths are relatively enriched in K₂O and define a trend parallel to the K₂O–Na₂O tie-line, whereas glasses in anhydrous xenoliths are significantly poorer in K₂O and define a trend parallel to the CaO–Na₂O tie-line (Amundsen, 1987). Amundsen, (1987) interpreted these glasses as the results of in situ partial melting involving breakdown of phlogopite in the phlogopite-bearing xenoliths, and breakdown of pre-existing amphibole in the anhydrous xenoliths. The highest degrees of K₂O enrichment are observed in glasses in xenoliths from Cape Verde [data from Siena & Coltorti, (1993)] and Gees, West Eifel [data from Zinngrebe & Foley, (1995)]. The high K₂O contents of these glasses strongly suggest partial melting of phlogopite (or another K-rich phase) in the mantle wall-rock. Phlogopite is common in West Eifel xenoliths, both as interstitial grains and in veins where it is associated with amphibole (C. Shaw, personal communication, 1997). Descriptions of xenolith suites from Cape Verde (Siena & Coltorti, 1989, 1993) indicate no K-bearing minerals which may account for the high K₂O contents in the silicic glasses. Glasses in xenoliths from Mt Lessini, Southern Alps, show moderate enrichment in K₂O, and no hydrous minerals are reported in their host xenoliths (Siena & Coltorti, 1989, 1993). Glasses in xenoliths from Yemen define a trend of strong Na₂O enrichment, interpreted by Chazot et al., (1996) as the result of in situ partial melting of clinopyroxene and/or amphibole, which are observed as residual phases. Glasses in anhydrous spinel lherzolite xenoliths from Mongolia [data from Ionov et al., (1994)] define a kinked trend of Na₂O enrichment with decreasing CaO among the more CaO-rich glasses, and marked K₂O enrichment with decreasing CaO among the most CaO-poor ones. These rocks show a clear negative correlation between modal clinopyroxene and glass, and Ionov et al., (1994) interpreted the glasses ‘as the result of disequilibrium in situ melting, involving largely clinopyroxene and spinel, resulting from reaction with migrating fluids’. Resorption of clinopyroxene explains the increase in Na₂O relative to CaO, whereas the mechanism behind the trend of K₂O enrichment remains an open question.

Origin of the infiltrating melts
Based on the discussion above, we assume that the least silicic glasses are closest to the compositions of the infiltrating melts. The mildly to strongly silica-undersaturated nature of these glasses (Table 1, Fig. 7) suggests that the reaction processes started with infiltration by alkali basaltic melts. Low-SiO₂ glasses are most common in xenoliths from Hierro and Tenerife. These glasses form two groups: a low-TiO₂ group, and a group with relatively high TiO₂, CaO and P₂O₅ contents (Figs 5-7). Low-SiO₂ glasses in harzburgites and lherzolites from Tenerife belong to the high-TiO₂ group (Fig. 5), whereas low-SiO₂ glasses in dunites from La Palma appear to belong to the low-TiO₂ group (Fig. 6). Neither group shows affinity to mid-ocean ridge basalt (MORB). The observed compositional diversity among the low-SiO₂ glasses strongly suggests that the melt–wall-rock reactions involved infiltrating melts of different compositions. Furthermore, although the reaction processes which gave rise to the silicic glasses must have taken place during the formation of the Canary Islands, the glasses include more extreme compositions (TiO₂–P₂O₅ rich and TiO₂–P₂O₅ poor) than seen among the aphyric basalts from Hierro and Tenerife, and lavas from the Canary Islands in general. This suggests that the total compositional range of melts produced during the Canary Islands magmatism is greater than that reflected in the exposed lavas. It is unclear if the compositional diversity seen among the low-SiO₂ glasses is the result of the melting processes in the source region (different degrees of partial melting, different pressures, etc.), or if they are imposed through interaction with different types of wall-rock during ascent through the upper mantle.

Mobility of the silicic melts
Wulff-Pedersen et al., (1995, 1996a, 1996b) showed that silicic melts in mantle rocks from the Canary Islands are highly enriched in strongly incompatible as compared with mildly incompatible elements [e.g. heavy rare earth elements (HREE), Ti and Zr], and proposed that silicic melts may be important agents of cryptic metasomatism in the mantle. The potential of a melt to act as a metasomatic agent is highly dependent on its mobility. Silicic glass in Canary Islands xenoliths generally makes up a very small proportion of the total rock volume (0.1–4.3% in La Palma; Wulff-Pedersen et al., 1996a). Although glass in secondary inclusion trails represents melts which at one time moved into cracks and fractures, it is clear that the ability of a fluid to move through the mantle is greatly reduced if the fluid is present in such small amounts that fluid–mineral cohesive forces are large relative to the buoyancy of the fluid. Furthermore, the most silicic glasses consist almost exclusively of SiO₂, Al₂O₃, Na₂O and K₂O (Tables 1 and 2). Unless they are relatively rich in H₂O, such melts are expected to be highly polymerized and viscous. As indicated above, it is likely that the silicic glass and phlogopite of this study contained H₂O + CO₂. There is also evidence in Canary Islands xenoliths that the silicic melts have moved over moderate distances. The presence of glass that is clearly not in equilibrium with the host peridotite in some unveined xenoliths implies that silicic melts have been mobile over distances at least corresponding to the dimensions of the xenoliths, that is, up to about 30 cm. The existence of an open system is supported by the films of glass and glass inclusions commonly present along grain boundaries.

	Conclusions

TOP ABSTRACT Introduction The Host Xenoliths Microstructures of Glass Analytical Procedure Chemical Compositions of the... Minerals Coexisting with Glass... Chemical Compositions of the... Discussion Conclusions References

 
Detailed studies of glasses and associated minerals in Group I xenoliths (spinel harzburgite, lherzolite, dunite and wehrlite) from different Canary Islands lead to the following observations and conclusions:

1. The glasses cover a considerable compositional range from TiO₂–FeO–MgO–CaO–P₂O₅-rich basaltic glasses with 44 wt % SiO₂, to highly silicic, TiO₂–FeO–MgO–CaO–P₂O₅-poor glass with up to 71 wt % SiO₂.
   
2. Several features suggest that highly silicic melts are in, or close to, equilibrium with the host spinel peridotites, whereas the basaltic melts are not: daughter minerals associated with highly silicic glass in polyphase inclusions are similar in composition to the main phases in the host xenoliths (Fo>90, Mg-Cr-rich, Ti–Al-poor pyroxenes, chromite); minerals in less silicic glasses are richer in Al₂O₃, TiO₂ and FeO, and poorer in MgO; reaction rims are common between basaltic glass and peridotite minerals, whereas contacts between highly silicic glass and peridotite are smooth.
   
3. Each island shows a systematic relationship between glass composition and type of host xenolith. Glasses in spinel harzburgites and lherzolites are typically silica oversaturated and extend to higher SiO₂ contents than glasses in dunites and wehrlites (no orthopyroxene), which are generally silica undersaturated.
   
4. The Canary Islands show differences with respect to Na₂O–CaO–K₂O relations which are related to the phase assemblages of their host peridotites. Glasses in xenoliths from La Palma and Tenerife where interstitial phlogopite is common have higher K₂O contents and K₂O/Na₂O ratios than glasses in xenoliths from Hierro and Lanzarote, where phlogopite is very rare.
   
5. The silicic melts are interpreted as the results of complex series of reactions between infiltrating alkali basaltic melts and peridotite wall-rocks. The higher SiO₂ contents in melts in orthopyroxene-bearing than in orthopyroxene-free rock types result from formation of SiO₂-rich melt + olivine at the expense of orthopyroxene in the presence of an H₂O–CO₂ fluid.
   
6. The high K₂O contents and K₂O/Na₂O ratios in glasses from La Palma and Tenerife reflect reactions involving partial or total consumption of phlogopite produced by earlier metasomatic events. The low K₂O concentration and K₂O/Na₂O ratios of glasses in the anhydrous xenolith suites from Hierro and Lanzarote reflect reactions between infiltrating melts and anhydrous and/or amphibole-bearing mantle wall-rocks.
   
7. In the upper mantle under the Canary Islands highly silicic melts have been mobile over a distance at least equal to the xenolith dimension, that is, 20–30 cm.
   

	Acknowledgements

 
This project was supported by the Commission of the European Communities, DGXII, Environment Programme, Climatology and Natural Hazards Unit, under Contract EV5V-CT-9283, and grants from the Norwegian Research Council (NFR) and Nansenfondet and associated funds. We also gratefully acknowledge the permission from the Ayuntamiento de Fuencaliente de La Palma (given to E.-R.N. in 1988) to collect xenolith samples from the volcanoes of San Antonio and Teneguía. P. Bottazzi, T. H. Green, W. L. Griffin, B. B. Jensen, L. Ottolini, C. Shaw, R. Vannuzzi and M. Wilson are gratefully acknowledged for enlightening discussions and constructive criticism of earlier versions of this paper. Fieldwork in Tenerife was made pleasant because of the very comfortable accommodation at the Parador de Cañadas del Teide, and the efforts of the staff there.

---

^* Corresponding author

	References

TOP ABSTRACT Introduction The Host Xenoliths Microstructures of Glass Analytical Procedure Chemical Compositions of the... Minerals Coexisting with Glass... Chemical Compositions of the... Discussion Conclusions References

 
Amundsen H. E. F. Peridotite xenoliths from Gran Canaria, Canary Islands; evidence for metasomatic processes and partial melting in the lower oceanic crust. Neues Jahrbuch für Mineralogie (1987) 156:121–140.

Andersen T., Burke E. A. J., Neumann E.-R. Nitrogen-rich fluid in the upper mantle: fluid inclusions in spinel dunite from Lanzarote, Canary Islands. Contributions to Mineralogy and Petrology (1995) 120:20–28.[Web of Science]

Baker M. B., Hirschmann M. M., Ghiorso M. S., Stolper E. M. Compositions of near-solidus peridotite melts from experiments and thermodynamic calculations. Nature (1995) 375:308–311.

Bryan W. B., Thompson G., Ludden J. N. Compositional variations in normal MORB from 22°-25°N: Mid-Atlantic Ridge and Kane Fracture Zone. Journal of Geophysical Research (1981) 86:11815–11836.

Chazot G., Menzies M. A., Harte B. Silicate glasses in spinel lherzolites from Yemen: origin and chemical composition. Chemical Geology (1996) 134:159–179.[Web of Science]

Draper D. S., Green T. H. Anhydrous and C–O–H fluid-saturated PT phase relations of silicic, alkaline, aluminous mantle-xenolith glasses. Journal of Petrology (1997) 38:1187–1224.[Abstract/Free Full Text]

Drury M. R., FitzGerald J. D. Grain boundary melt films in an experimentally deformed olivine-orthopyroxene rock: implications for melt distribution in upper mantle rocks. Geophysical Research Letters (1996) 23:701–704.[Web of Science]

Edgar A. D., Lloyd F. E., Forsyth D. M., Barnett R. L. Origin of glass in upper-mantle xenoliths from the Quaternary volcanics SE of Gees, West Eifel, Germany. Contributions to Mineralogy and Petrology (1989) 103:277–286.[Web of Science]

Eggler D. H. Role of CO₂ in melting processes in the mantle. Carnegie Institution of Washington, Yearbook (1972) 72:457–467.

Falloon T. J., Green D. H. Anhydrous partial melting of MORB pyrolite and other peridotite compositions at 10 kbar: implications for the origin of primitive MORB glasses. Mineralogy and Petrology (1987) 37:181–219.[Web of Science]

Francis D. Mantle–melt interaction recorded in spinel lherzolite xenoliths from the Alligator Lake Volcanic Complex, Yukon, Canada. Journal of Petrology (1987) 28:569–597.[Abstract/Free Full Text]

Francis D. M. The origin of amphibole in lherzolite xenoliths from Nunivak Island, Alaska. Journal of Petrology (1976) 17:357–378.[Abstract/Free Full Text]

Frey A. F., Green D. H. The mineralogy, geochemistry and origin of lherzolite inclusions in Victorian basanites. Geochimica et Cosmochimica Acta (1974) 38:1023–1059.[Web of Science]

Frey F. A., Prinz M. Ultramafic inclusions from San Carlos, Arizona: petrologic and geochemical data bearing on their petrogenesis. Earth and Planetary Science Letters (1978) 38:129–176.[Web of Science]

Frezzotti M. L., Touret J. L. R., Lustenhouwer W., Neumann E.-R. Melt and fluid inclusions in dunite xenoliths from La Gomera, Canary Islands: tracking the mantle metasomatic fluids. European Journal of Mineralogy (1994) 6:805–817.[Abstract/Free Full Text]

Fuster J. M., Fernandez S., Sagredo J. Geologia y Volcanologia de las Islas Canarias. Lanzarote (1968a) Madrid: Inst. Lucas Mallada. 177.

Fuster J. M., Páez A., Sagredo J. Significance of basic and ultramafic rock inclusions in the basalts of Canary Islands. Bulletin of Volcanology (1968b) 33:665–693.

Hansteen T. H., Andersen T., Neumann E.-R., Jelsma H. Fluid and silicate melt inclusions in spinel lherzolite xenoliths from Hierro, Canary Islands: implications for mantle metasomatism. Contributions to Mineralogy and Petrology (1991) 107:242–254.[Web of Science]

Hauri E. H., Shimizu N., Dieu J. J., Hart S. R. Evidence for hotspot-related carbonatite metasomatism in the oceanic upper mantle. Nature (1993) 365:221–227.

Hernandez-Pacheco A., Valls M.C. Las erupciones volcanicas historicas de la isla de La Palma (Canarias). Arquipelago Revy Universidade dos Acores (1982) 3:83–94.

Ionov D. A., Hofmann A. W., Shimizu N. Metasomatism-induced melting in mantle xenoliths from Mongolia. Journal of Petrology (1994) 35:753–785.[Abstract/Free Full Text]

Jaques A. L., Green D. H. Anhydrous melting of peridotite at 0–15 kb pressure and genesis of tholeiitic basalts. Contributions to Mineralogy and Petrology (1980) 73:287–310.[Web of Science]

Johnsen K. The nature and evolution of the lithosphere beneath Lanzarote, Canary Islands: evidence from upper mantle xenoliths. (1990) 106. Cand. Scient. Thesis, University of Oslo.

Jones A. D., Smith J. V., Dawson J. B. Glasses in mantle xenoliths from Olmani, Tanzania. Journal of Geology (1983) 91:167–178.[Web of Science]

Kelemen P. B. Reactions between ultramafic rock and fractionating basaltic magma I. Phase relations, the origin of calc-alkaline magma series, and the formation of discordant dunite. Journal of Petrology (1990) 31:51–98.[Abstract/Free Full Text]

Klügel A., Sachs P. M., Schmincke H.-U. How long did mantle xenoliths from La Palma (Canary Islands) reside in the transporting magma? In: Chapman Conference on Shallow Level Processes in Ocean Islands Magmatism (Abstracts) (1996) Washington, DC: AUG Meetings Department. 21.

Kogarko L. N., Henderson C. M. B., Pacheco H. Primary Ca-rich carbonatite magma and carbonatite–silicate–sulphide liquid immiscibility in the upper mantle. Contributions to Mineralogy and Petrology (1995) 121:267–274.[Web of Science]

Menzies M. Oceanic peridotites. In: Oceanic Basalts—Floyd P. A., ed. (1991) Glasgow: Blackie. 363–385.

Mercier J.-C. C., Nicolas A. Textures and fabric of the upper mantle peridotites as illustrated by basalt xenoliths. Journal of Petrology (1975) 16:454–487.[Abstract/Free Full Text]

Mysen B. O., Boettcher A. L. Melting of a hydrous mantle: I. Phase relations of natural peridotite at high pressures and temperatures with controlled activities of water, hydrogen, and carbon dioxide. Journal of Petrology (1975a) 16:520–548.[Abstract/Free Full Text]

Mysen B. O., Boettcher A. L. Melting of a hydrous mantle: II. Geochemistry of crystals and liquids formed by anatexis of mantle peridotite at high pressures and high temperatures as a function of controlled activities of water, hydrogen, and carbon dioxide. Journal of Petrology (1975b) 16:549–593.[Abstract/Free Full Text]

Mysen B. O., Kushiro I. Compositional variations of coexisting phases with degree of melting of peridotite in the upper mantle. American Mineralogist (1977) 62:843–865.[Abstract]

Neumann E.-R. Ultramafic and mafic xenoliths from Hierro, Canary Islands: evidence for melt infiltration in the upper mantle. Contributions to Mineralogy and Petrology (1991) 106:236–252.[Web of Science]

Neumann E.-R., Wulff-Pedersen E. Melt inclusions in upper mantle xenoliths from the Canary Islands. EOS Transactions, American Geophysical Union (1995) 76:268. Abstract Supplement.

Neumann E.-R., Wulff-Pedersen E., Johnsen K., Andersen T., Krogh E. Petrogenesis of spinel harzburgite and dunite suite xenoliths from Lanzarote, eastern Canary Islands: implications for the upper mantle. Lithos (1995) 35:83–107.[Web of Science]

Pellicier M. J. Estudio volcanológico de la isla de El Hierro (Islas Canarias). Estudios Geologicos (1977) 33:181–197.

Pellicier M. J. Estudio geoquímico de la isla de Hierro Archipiélago Canario. Estudios Geologicos (1979) 35:15–29.

Sagredo Ruiz J. Origin de las inclusiones de dunitas y otras rocas ultramáficas en las rocas volcánicas de Lanzarote y Furteventura. Estudios Geologicos (1969) 25:189–233.

Scarfe C. M., Brearley M. Mantle xenoliths: melting and dissolution studies under volatile-free conditions. In: Mantle Xenoliths—Nixon P. H., ed. (1987) New York: John Wiley. 599–608.

Schiano P., Clocchiatti R. Worldwide occurrence of silica-rich melts in sub-continental and sub-oceanic mantle minerals. Nature (1994) 368:621–624.

Schiano P., Clocchiatti R., Joron J. L. Melt and fluid inclusions in basalts and xenoliths from Tahaa Island, Society archipelago: evidence for a metasomatized upper mantle. Earth and Planetary Science Letters (1992) 111:69–82.[Web of Science]

Schiano P., Clocchiatti R., Shimizu N., Weis D., Mattielli N. Cogenetic silica-rich and carbonate-rich melts trapped in mantle minerals in Kerguelen ultramafic xenoliths: implications for metasomatism in the oceanic upper mantle. Earth and Planetary Science Letters (1994) 23:167–178.

Schiano P., Clocchiatti R., Shimizu N., Maury R. C., Jochum K. P., Hofmann A. W. Hydrous, silica-rich melts in the sub-arc mantle and their relationship with erupted arc lavas. Nature (1995) 377:595–600.

Schilling J.-G., Zajac M., Evans R., Johnston T., White W., Devine J. D., Kingsley R. Petrologic and geochemical variations along the mid-Atlantic ridge from 29°N to 73°N. American Journal of Science (1983) 283:510–586.[Abstract/Free Full Text]

Siena F., Coltorti M. Lithospheric mantle evolution: evidences from ultramafic xenoliths in the Lessinian volcanics (northern Italy). Chemical Geology (1989) 77:347–364.[Web of Science]

Siena F., Coltorti M. Thermobarometric evolution and metasomatic processes of upper mantle in different tectonic settings: evidence from spinel peridotite xenoliths. European Journal of Mineralogy (1993) 5:1073–1090.[Abstract/Free Full Text]

Siena F., Beccaluva L., Coltorti M., Marchesi S., Morra V. Ridge to hot-spot evolution of the Atlantic lithosphere mantle: evidence from Lanzarote peridotite xenoliths (Canary Islands). In: Orogenic lherzolites and mantle processes, Journal of Petrology Special Volume—Menzies M. A., Dupuy C., Nicolas A., eds. (1991) 271–290.

Sigurdsson H. First-order major element variations in basaltic glasses from the Mid-Atlantic Ridge: 29°N to 73°N. Journal of Geophysical Research (1981) 86:9483–9502.

Staudigel H. Der Basale Complex von La Palma. Submarine vulkanische Processe, Petrologie, Geochemie und sekundäre Prozesse im herausgehobenen, submarinen Teil einer ozeanischen Insel und ein Vergleich mit ozeanischer Kruste. (1981) 358. Dissertation Dr. Naturw. Ruhr-Universität Bochum, Germany.

Staudigel H., Feraud G., Giannerini G. The history of intrusive activity on the island of La Palma (Canary Islands). Journal of Volcanology and Geothermal Research (1986) 27:299–322.[Web of Science]

Weaver B. L., Tarney J., Saunders A. D. Geochemistry of basalts recovered from the central North Atlantic. In: Initial Reports of the Deep Sea Drilling Project, 82—Bougault H., Cande S. C., et al, eds. (1985) Washington, DC: US Government Printing Office. 395–419.

Wulff-Pedersen E., Neumann E.-R., Vanucci R., Bottazzi P., Ottolini L. Metasomatic processes in the upper mantle beneath La Palma, Canary Islands. EOS Transactions, American Geophysical Union (1995) 76:271. Abstract Supplement.

Wulff-Pedersen E., Neumann E.-R., Jensen B. B. The upper mantle under La Palma, Canary Islands: formation of K-rich silicic melt and its importance as a metasomatic agent. Contributions to Mineralogy and Petrology (1996a) 125:113–139.[Web of Science]

Wulff-Pedersen E., Neumann E.-R., Vannucci R., Ottolini L., Bottazzi P. Origin of silicic glass inclusions in refractory mantle xenoliths from the Canary Islands. In: 30th International Geological Congress (1996b) Vol. 2. Beijing. 481.

Zinngrebe E., Foley S. F. Metasomatism in mantle xenoliths from Gees, West Eifel, Germany: evidence for the genesis of calc-alkaline glasses and metasomatic Ca-enrichment. Contributions to Mineralogy and Petrology (1995) 122:79–96.[Web of Science]

CiteULike    Connotea    Del.icio.us    What's this?

This article has been cited by other articles:

				N. Raffone, G. Chazot, C. Pin, R. Vannucci, and A. Zanetti Metasomatism in the Lithospheric Mantle beneath Middle Atlas (Morocco) and the Origin of Fe- and Mg-rich Wehrlites J. Petrology, February 1, 2009; 50(2): 197 - 249. [Abstract] [Full Text] [PDF]

				C. Pin, J.L. Paquette, P. Monchoux, and T. Hammouda First field-scale occurrence of Si-Al-Na-rich low-degree partial melts from the upper mantle Geology, May 1, 2001; 29(5): 451 - 454. [Abstract] [Full Text] [PDF]

				E.-R. Neumann, E. Wulff-Pedersen, S. L. Simonsen, N. J. Pearson, J. Marti, and J. Mitjavila Evidence for Fractional Crystallization of Periodically Refilled Magma Chambers in Tenerife, Canary Islands J. Petrology, July 1, 1999; 40(7): 1089 - 1123. [Abstract] [Full Text] [PDF]

				E. Widom, K. A. Hoernle, S. B. Shirey, and H.-U. Schmincke Os Isotope Systematics in the Canary Islands and Madeira: Lithospheric Contamination and Mantle Plume Signatures J. Petrology, February 1, 1999; 40(2): 279 - 296. [Abstract] [Full Text] [PDF]

This Article Abstract FREE Full Text (PDF) Supplementary Data Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Request Permissions Google Scholar Articles by Neumann, E.-R. Articles by Wulff-Pedersen, E. Search for Related Content Social Bookmarking          What's this?

Online ISSN 1460-2415 - Print ISSN 0022-3530

Copyright © 2010 Oxford University Press

Oxford Journals Oxford University Press

• Site Map
• Privacy Policy
• Frequently Asked Questions

Other Oxford University Press sites: Oxford University Press Oxford Journals China Oxford Journals Japan American National Biography Booksellers' Information Service Children's Fiction and Poetry Children's Reference Corporate & Special Sales Dictionaries Dictionary of National Biography Digital Reference English Language Teaching Higher Education Textbooks Humanities International Education Unit Law Medicine Music Online Products Oxford English Dictionary Reference Rights and Permissions Science School Books Social Sciences Very Short Introductions World's Classics