Abundances of major and trace elements in whole rocks and minerals in lherzolites and harzburgites from the northern Oman ophiolite are used to understand the mantle processes creating compositional variation in oceanic lithospheric mantle. Detailed mapping shows that lherzolites occur near the base of a mantle section in the northern Fizh block. Geochemical analyses identify two types of basal lherzolite. The first type (Type I lherzolite) displays porphyroclastic microstructure and occurs sporadically in the basal mylonite zone. Whole rock and clinopyroxene are highly depleted in incompatible elements such as Na, Ti, Zr, and rare earth elements (REE). The chondrite-normalized patterns of Type I lherzolites show steep slopes from heavy REE (HREE) to light REE (LREE) that are ascribed to melt extraction, up to 12–18%, from a source containing a small amount of garnet. The chondrite-normalized patterns have slight enrichment in LREE relative to the patterns expected for residues of partial melting thereby indicating reaction with a LREE-enriched melt or fluid at a low melt/rock ratio. The second type (Type II lherzolite) shows mylonitic microstructure and only occurs at the contact between the mantle section and the metamorphic sole. Abundances of incompatible elements in whole rocks and clinopyroxenes are greater than those of Type I lherzolites, and clinopyroxenes in Type II lherzolites have high Na2O contents (>1 wt.%). To a first approximation, the high Na content of clinopyroxenes and whole rocks and the LREE-depleted, chondrite-normalized whole rock REE patterns are consistent with Type II lherzolite being in equilibrium with a mid-ocean ridge basalt (MORB)-type melt at relatively high pressure (>2 GPa). However, the flatness of chondrite-normalized patterns for middle and heavy REE are inconsistent with residual garnet peridotite. The characteristics of Type II lherzolites are better explained by a mixing process in which residual peridotite was refertilized by addition of a LREE-depleted melt. The large compositional gradient near the basal thrust in the northern Fizh block may have recorded a transient state in which the degree of partial melting was progressively decreased as a result of reducing mantle temperature and upwelling rate. This scenario is consistent with the inferred failing ridge associated with a transform zone in the western side of the northern Fizh block proposed by Nicolas et al. . In the detachment stage of the Oman ophiolite, a small amount of ascending melt may have crystallized near the basal part of mantle section thereby forming Type II lherzolites. Basal lherzolites and their spatial chemical variations in the northern Fizh block may provide a key for understanding the processes of ridge segmentation and detachment at fast spreading ridges.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The Oman ophiolite is a cross section of oceanic lithosphere exposed over 300 kilometers along the Arabian coast. It is one of the best field areas to study dynamics of the oceanic crust-mantle system and ophiolite genesis [e.g., Nicolas, 1989]. In particular the Oman ophiolite provides information for understanding mantle processes that create spatial compositional variability in oceanic lithosphere. This information cannot be obtained from abyssal peridotites which are typically highly serpentinized and limited in occurrence. The ophiolite is a typical harzburgite-type formed by high extents of partial melting at a fast spreading ridge [Boudier and Nicolas, 1985]. The mantle sections mainly consist of harzburgite and dunite accompanied by crosscutting dikes of gabbro and pyroxenite. However, basal lherzolites are sparsely distributed at some places such as the base of the northern Fizh block [Lippard et al., 1986] and the Wadi Tayin massif [Godard et al., 2000].
 Basal lherzolite has also been reported from the Table Mountain in the Bay of Island ophiolite [e.g., Suen et al., 1979; Girardeau and Nicolas, 1981; Suhr and Ronbinson, 1994; Suhr and Batanova, 1998]. The problem with basal lherzolite in harzburgite-type ophiolites is the large compositional gradient near the base of mantle section [Suhr and Batanova, 1998]. The vertical compositional variation in the mantle section of the northern Fizh block in the Oman ophiolite is similar to that in the Table Mountain in the Bay of Island ophiolite [Suhr and Batanova, 1998]. This vertical gradient is too steep to be formed by partial melting of upwelling mantle in a steady state [Suhr and Batanova, 1998]. Suhr and Batanova  proposed that basal lherzolite represents accreted, telescoped material that records a horizontal melting gradient; i.e., along the detachment path and toward a paleo-fracture zone, less-depleted mantle material accreted underneath the ophiolite sole preserving a condensed vertical melting gradient. This model presumes that mantle composition progressively changed from depleted to less depleted toward the paleo-fracture zone.
 Basal lherzolite was also reported from the Wadi Tayin massif in the southern Oman ophiolite [Godard et al., 2000]. Structural studies of this area showed a NW-SE propagating ridge within slightly older lithosphere [Nicolas and Boudier, 1995; Boudier et al., 1997]. Godard et al.  proposed that the basal lherzolite (cpx-harzburgite in their definition) formed by thermal erosion of the base of preexisting oceanic lithosphere by asthenospheric melt when the ridge axis jumped and opened a new oceanic rift. Because the northern Fizh block is considered to be a site of propagating and failing ridge segments that were separated by a 10–20 km-wide transform zone [Nicolas et al., 2000], the basal lherzolite may have formed by a similar process. In this study, we report the field occurrence and geochemical characteristics of basal lherzolites in the northern Fizh block and discuss their origin with regards to the melting and refertilization processes beneath this ridge segment.
2. Geological Background
 To understand the lithological distribution of peridotite in the mantle section of the northern Fizh block we conducted detailed mapping of this area (Figure 1). The planar and linear structures summarized in Figures 2 and 3are mostly consistent with previous studies [Boudier et al., 1988; Nicolas et al., 1988, 2000; Reuber, 1988]. The distributions of coarse granular and porphyroclastic microstructures are also comparable to those for the High-T and Medium-T harzburgites shown in the Oman & UAE ophiolite structural maps by Nicolas and Boudier . The region with mylonitic microstructure in Figures 2 and 3 corresponds to the area for the Low-T and a part of the Medium-T harzburgites of Nicolas and Boudier . Figure 4 shows cross sections along the three section lines indicated in Figure 2. Similar to other parts of the Oman ophiolite, the mantle section of the northern Oman ophiolite mainly consists of harzburgite and dunite. Pyroxenite and gabbroic dikes are abundant and obliquely cut the foliation plane of host peridotite. Our study confirmed that clinopyroxene-rich harzburgite and lherzolite are largely confined to the basal part of the northern Fizh block [Lippard et al., 1986]. Occasionally cpx-rich peridotites also occur in the upper part of mantle section. However, in such cases, clinopyroxenes are present only in narrow bands several centimeters thick that have diffuse boundaries with the host harzburgite.
 In this study, we report two types of basal lherzolite from the northern Fizh block. The clinopyroxene in the first type (Type I lherzolite, hereafter) is very poor in Na (Na2O < 0.7 wt.%), as typical of abyssal peridotites [e.g., Rivalenti et al., 1996] while clinopyroxene in the second type (Type II lherzolite, hereafter) is characterized by very high Na contents (Na2O > 1 wt.%). Type I lherzolites sporadically appear in the basal mylonite zone and grade into clinopyroxene-rich harzburgite over several meters while Type II lherzolites only occur at the basal thrust as massive pyroxene-rich blocks or as bands alternating with dunite and harzburgite on a scale of one meter to a few meters.
 The northern Fizh block is structurally the most complicated area in the Oman ophiolite, and many studies have been conducted in this area [Smewing, 1980; Lippard et al., 1986; Boudier et al., 1988; Nicolas et al., 1988, 2000; Reuber, 1988]. Here, we summarize the structural features in this area based on our new observations and previous studies. The peridotites with high-temperature coarse granular microstructure are distributed within 5 km of the preserved Moho (Figure 2). The lineations gently plunge southward throughout the region. These foliations and lineations represent paleo-flow lines subparallel to an isothermal plane in the uppermost mantle [Nicolas, 1986]. Therefore, preservation of high-T deformation features near the preserved Moho could have resulted from conductive cooling from the top of the oceanic lithosphere without further influence from deformation during detachment. At the eastern border of the mantle section, the dip of the N-S directed Moho becomes steeper toward the north from Wadi Zabin to Wadi Rajmi [Boudier et al., 1988; Reuber, 1988; Nicolas et al., 1988, 2000]. The Moho dips southeastward at 10–40° at Wadi Zabin and in the southern portion of Wadi Rajmi while it dips eastward at 30–80° in the northern portion of Wadi Rajmi where a dextral and steep NNW-SSE shear zone has been developed [Boudier et al., 1988]. This shear sense is consistent with southward thrusting throughout the entire northern Fizh block (Figure 3) [Reuber, 1988; Boudier et al., 1988; Nicolas et al., 2000] whereas it is unlike the shear sense in the southern Fizh and Hilti blocks where flow is directed to the east and is normal to the strike of the Moho [Ceuleneer et al., 1988; Boudier et al., 1988; Michibayashi et al., 2000].
 The central part of the mantle section is dominated by harzburgite with porphyroclastic microstructure. The strike of foliation in this domain ranges from NW-SE to E-W and the lineations dominantly plunge to the north although others are EW trending (Figure 2). The shear sense in the porphyroclastic microstructure domain is top to the south, similar to those in the high T deformation zone near the Moho [Boudier et al., 1988; Nicolas et al., 2000]. The basal shear zone which is more than 2–3 km above the basal thrust, consists of low-T deformation features including mylonitic microstructure. These latter microstructures are related to the detachment and early thrusting of the ophiolite [Boudier et al., 1988; Nicolas et al., 2000]. The lineation is consistently oriented N-S [Boudier et al., 1985; Ceuleneer et al., 1988; this study]. It is noteworthy that, within the mylonitic domain, porphyroclastic microstructure sporadically appears in cpx-rich harzburgite and lherzolite.
 The structural complexities in the northern Fizh block have been interpreted in terms of segmentation in a fast spreading ridge system. Smewing  proposed a leaky transform model in which spreading axes were aligned N-S and offset by a 30 km WNW-ESE zone that represents the transform section of an oceanic fracture zone. Alternatively, Reuber  suggested large-offset overlapping spreading centers while MacLeod and Rothery  proposed a propagating rift structure. Recently, Nicolas et al.  synthesized all observations and interpretations into a model of propagating and failing ridges that were oriented parallel to NNW-SSE and separated by a 10–20 km-wide transform zone located in the northern Fizh block (Figure 1).
3. Sample Description
 In this study we selected five Type I lherzolites, four Type II lherzolites and eight harzburgites for chemical analyses of minerals and whole rocks. An additional eighteen harzburgites, from a traverse from the preserved Moho to the basal thrust in the northern Fizh block, were selected for determination of mineral compositions. Mineral modes of primary phases were calculated using a least squares mass balance of the whole rock and mineral compositions (Table 1a). For comparison, modal analyses by point-counting were also determined for six representative samples (Table 1b). The lherzolites do not contain plagioclase. Because of replacement by serpentine, the point-counted modes for olivine and orthopyroxene are significantly lower than the calculated mode. Also clinopyroxene is heterogeneously distributed in hand specimen. In addition, errors arise from difficulty in distinguishing clinopyroxene from olivine in the fine-grained matrix. We use the calculated modes for defining rock type. For example, lherzolite is peridotite with modal clinopyroxene greater than 5%.
Table 1a. Calculated Mineral Modes for Peridotites From the Northern Oman Ophiolite
Unit in wt.%.
Extremely fresh boulders from the southern Fizh block.
Sample from the Wadi Farfar in the Hilti block that neighbors southward on the Fizh block.
Table 1b. Modes of Oman Peridotites Estimated From 2000 Points on a 2.5 × 1.5 cm Section
Unit in vol.%. Others include magnetite, anthophyllite, tremolite, brucite, and talc; tr. = trace.
Type I Lherzolite
Type II Lherzolite
 Type I lherzolites consist of olivine, orthopyroxene, clinopyroxene and spinel as primary phases. Orthopyroxene and clinopyroxene occur as porphyroclasts or neoblasts. The size of orthopyroxene porphyroclasts ranges from 1 to 3 mm while clinopyroxenes are smaller than 2 mm. Spinel occurs as fine interstitial grains ranging from 0.05 to 0.3 mm. They are commonly spatially associated with orthopyroxene. Alteration phases include serpentine and small amounts of anthophyllite, brucite, magnetite and calcite veins. In Type II lherzolites, orthopyroxene porphyroclasts occur in either spherical or stretched shapes in a fine-grained matrix. Clinopyroxene occurs as spherical porphyroclasts or neoblasts that commonly form aggregates. The maximum amount of clinopyroxene is 13% in the calculated mode. Most spinel occurs as fine grains associated with orthopyroxene. Alteration phases include tremolite in addition to those in Type I lherzolite.
 Most harzburgites are heavily serpentinized and have a small amount of clinopyroxene. However, two boulders (981201-6 and 97075) from southern Fizh block are extremely fresh and serpentine is absent. A harzburgite (99121803) from the crust/mantle transition zone at southern Wadi Rajmi contains clusters of orthopyroxene, clinopyroxene and saussurite probably replacing plagioclase. These minerals may have been precipitated from trapped melt [Nicolas, 1989, and references therein].
4. Analytical Methods
 Major element compositions of primary minerals (olivine, orthopyroxene, clinopyroxene and spinel) were determined using an electron probe microanalyzer with wavelength dispersive X-ray spectrometry (JEOL JXA-8600SX) at Niigata University. Operating conditions were 15 kV accelerating voltage, 13 nA beam current, and ∼1 μm beam diameter, using oxide ZAF matrix correction. Whole rock major element compositions were analyzed using X-ray fluorescence spectrometry (Rigaku RIX3000) at Niigata University following the analytical method of Takahashi and Shuto  with an optimization for ultramafic rocks. The calibration curve for Na was tested with 13 whole rock powders from the Horoman peridotite, whose Na content was determined by instrumental neutron activation analyses (INAA) [Takazawa et al., 2000]. Analytical results for Na by XRF agree very well with the INAA data to concentrations as low as 50 ppm.
 Abundances of trace elements in whole rocks and mineral separates (clinopyroxene and orthopyroxene) from representative samples were determined using an inductively coupled plasma-mass spectrometer (ICP-MS) (Yokogawa HP4500) at Niigata University. Whole rock powders (0.1g) and mineral separates (11–65 mg) were dissolved in HF-HNO3 mixture on a hotplate and diluted by a factor of 1,000 with 2% HNO3. Spinels in whole rock powder did not completely dissolve. Because of low abundances of spinel as well as incompatible trace elements in spinel their incomplete dissolution should not influence the results. For orthopyroxene a few drops of HClO4 were added to HF-HNO3 mixture to enhance digestion. The analytical procedure employed for ICP-MS in this study is similar to the method of Eggins et al. . A single solution of USGS reference material, BHVO-1, was used as an external calibration standard with reference values from Eggins et al. . Sensitivity variation during the analytical runs was corrected using four internal standards (In, Tm, Re, Bi). External standardization was performed for individual elements in unknown samples by interpolating results for replicate analyses of BHVO-1 after 5–6 unknown samples. Geological reference materials for peridotite (JP-1 and DTS-1) were also analyzed to monitor analytical quality.
 Abundances of most elements in peridotite standards are consistent with reference values except for Nb and Ta. Abundances of Nb and Ta are systematically higher in this study than those reported by Makishima and Nakamura . Because our results (Nb = 0.052 ppm; Ta = 0.006 ppm) are similar to our long-term average of Nb and Ta in JP-1 (Nb = 0.050 ppm ± 0.008 ppm (1σ); Ta = 0.0056 ppm ± 0.0013 ppm (1σ); n = 5 over two years) these anomalies do not simply result from insufficient removal of Nb and Ta scavenged by tubing during ICP-MS analysis. Further analyses are necessary for evaluating the true abundances of Nb and Ta in the peridotite standards JP-1 and DTS-1.
5.1. Major Elements
 Forsterite molar percent of olivine in lherzolite and harzburgite varies from 90 to 92 while Cr/(Cr + Al) molar ratio of spinel varies from 0.1 to 0.6. These ranges overlap with the fields for oceanic harzburgite and lherzolite defined by Dick and Bullen  and Arai [1987, 1994]. The cores of orthopyroxenes in Type II lherzolites have greater Al contents than orthopyroxene cores in Type I lherzolites. Figure 5a shows Na2O versus Al2O3 contents of clinopyroxenes from the Oman peridotites. Abundances of these oxides in clinopyroxenes increase from harzburgites to Type I lherzolites to Type II lherzolites. Figure 5b compares the Na2O and Al2O3 contents of clinopyroxenes in the Oman peridotites with those in abyssal peridotites [Hamlyn and Bonatti, 1980; Dick, 1989; Johnson et al., 1990; Johnson and Dick, 1992; Bonatti et al., 1992; Dick and Natland, 1996; Bonatti et al., 1986]. The fields for clinopyroxenes in Type I lherzolites and harzburgites are similar to the range for clinopyroxenes in abyssal peridotites from the Mid-Atlantic Ridge [Bonatti et al., 1992]. However, the Na2O contents of clinopyroxene from Type II lherzolites exceed the range for most abyssal peridotites, and are comparable to clinopyroxenes in Zabargad peridotites [Bonatti et al., 1986].
Figure 6 shows compositional variations of minerals from harzburgites along a traverse (A–B in Figures 2 and 4) in the mantle section from the Moho to the basal thrust. Forsterite molar percent (Fo) of olivines and Mg/(Mg+Fe*) ratio (Mg#) of orthopyroxenes vary slightly from 91 to 92 and from 0.91 to 0.93, respectively. In contrast, both Mg# and Cr# (Cr/[Cr + Al]) of spinels vary widely and are inversely correlated in the mantle section. The Cr# of spinels has a maximum value of 0.6 in the middle of mantle section and gradually decreases to 0.2 near the basal thrust over a lateral distance of 10 km. The upper part of the mantle section also contains spinels with lower Cr# relative to those from the middle part. Pyroxene compositions are also variable. The wollastonite molar percent (Wo) of orthopyroxenes mostly ranges between 1 to 2. The heterogeneity in the Wo content of orthopyroxenes in the upper part of the section reflects the presence of clinopyroxene lamellae. As in the case of spinel Cr#, the Al2O3 content of orthopyroxenes and clinopyroxenes and the Na2O content of clinopyroxenes increase downward from the middle part of the section. The Na2O content of clinopyroxenes also increases in the upper part of mantle section, especially near the shear zone which may have been affected by hydrothermal alteration or focused melt/fluid flow [Kelemen and Dick, 1995]. The variation of TiO2 content of clinopyroxenes is limited and slightly increases from the middle part of the section toward the basal thrust and toward the Moho. The variations of Al2O3 content of orthopyroxenes and of Cr# and Mg# of spinels along the traverse in this study reproduce the results for the Rayy–Ragmi traverse reported by Lippard et al. . In contrast the variations of Fo content and of Mg# of orthopyroxene in our study differ from this previous study which showed increases of both values toward the basal thrust.
5.1.2. Whole Rocks
 Whole rock compositions of the Oman peridotites show a linear array in SiO2, FeO*, Al2O3, CaO versus MgO diagrams, similar to those defined by upper mantle xenoliths and other massive peridotites (Figure 7) [Takazawa et al., 2000, and references therein]. The low MgO end of these arrays extends toward estimates of primitive mantle compositions [McDonough and Frey, 1989; McDonough and Sun, 1995, and references therein]. The melt-impregnated harzburgite (99121803) is characterized by higher FeO* content (9.5 wt.%) and lower Mg# (0.893) relative to the other harzburgites. Corresponding to the clinopyroxene compositions, whole rock compositions of Type I lherzolites are characterized by lower Na and Ti contents than those of Type II lherzolites. In SiO2, FeO*, Al2O3, CaO versus MgO diagrams the compositional trends are, to the first order, consistent with formation of the Oman peridotites as residues of partial melting of fertile source mantle. Moreover, the low Na2O contents of Type I lherzolites are consistent with melting at pressure lower than 2.5 GPa. On the other hand, the high Na2O contents of Type II lherzolites are difficult to fit to a single melting curve. Later we will evaluate a hypothesis in which Type II lherzolites were formed by mixing of residual peridotites with a component with low MgO and high Na2O contents, such as basaltic melt.
5.2. Incompatible Elements
5.2.1. Whole Rocks
Figure 8 shows chondrite-normalized rare earth element (REE) patterns for the Oman lherzolites and harzburgites. The patterns for Type I lherzolites show depletion in light rare earth elements (LREE) relative to heavy rare earth elements (HREE) and have steep slopes from Tb to Nd. However, they are slightly enriched in the lightest REE, i.e., (La/Ce)N > 1. A Type I lherzolite (99120513) is enriched in La, Ce and Pr relative to Nd and thereby shows a spoon-shaped REE pattern. Abundances of REE in Type II lherzolites are greater than those in Type I lherzolites. The chondrite-normalized patterns for Type II lherzolites are depleted in LREE without any relative enrichment in the lightest REE.
Figure 9 shows primitive mantle-normalized incompatible element patterns for the Oman lherzolites and harzburgites. These peridotites show enrichments in highly incompatible elements such as Rb, Ba, Th, U, Nb, Ta, Pb and Sr relative to their adjacent REE. The pattern for the melt-impregnated harzburgite (99121803) is nearly straight from La to Lu except for Pb. The pattern does not have a positive anomaly in Sr. These characteristics are probably inherited from the impregnated melt. The abundances of incompatible elements in a harzburgite (99120610) that was collected from the basal shear zone resemble those of Type I lherzolites.
5.2.2. Pyroxene Separates
 The primitive mantle-normalized incompatible element patterns for selected clinopyroxenes and orthopyroxenes show strong depletion in LREE relative to HREE (Figure 10) and are similar to the whole rock patterns. The patterns for clinopyroxene are slightly convex upward at the MREE and have a steep slope from MREE to LREE which is associated with negative anomalies in Sr, Zr and Hf. The patterns for the Type I lherzolite clinopyroxenes show strong positive anomalies in Pb, similar to those of whole rocks. In contrast, the clinopyroxenes from Type II lherzolites do not have anomalous Pb abundances. Abundances of highly incompatible elements such as Rb, Ba, Th, U, Nb and Ta are similar in clinopyroxenes from Type I and II lherzolites and unlike the whole rocks, U and Nb are not strongly fractionated from Th and Ta. Primitive mantle-normalized patterns for orthopyroxenes decrease from HREE to LREE. There are positive anomalies in Pb, Zr and Hf. The opposite sense of anomalies in Zr and Hf for clinopyroxenes and orthopyroxenes indicate that the intermineral partition coefficient (Dopx/cpx) differs significantly from unity for Zr and Hf relative to adjacent REE [McDonough et al., 1992; Rampone et al., 1993].
5.3. Evaluation of Alteration Effect
 Strong positive anomalies in Sr are observed in whole rocks, however, pyroxene separates from the same samples have no anomaly in Sr. This observation indicates that most of the Sr does not reside in pyroxene but is probably present along grain boundaries or in secondary phases formed by alteration. On the other hand, both whole rock powders and mineral separates display positive anomalies in Pb. Although hydrothermal alteration may have increased the abundance of Pb in peridotites, these Pb anomalies may reflect the relatively high instrumental blank for Pb and, as a consequence, a high detection limit for Pb in ICP-MS analyses.
Figure 11 shows measured abundances for whole rocks compared to whole rock abundances estimated from pyroxene separates. If the abundance ratio exceeds unity the element may reside in alteration phases or at grain boundaries. Most ratios are near unity, varying from 0.5 to 2. This range is within the uncertainties associated with estimated mineral modes. However, the measured abundances for Cs, Rb, Ba, Th, Sr and Pb (for Type I lherzolites) exceed 200% of the calculated abundances. Apparently, these elements were affected by alteration processes. We assume in the following that all the elements except Cs, Rb, Ba, Th, Pb and Sr retain their igneous concentrations.
6.1. Inferences From Na Content of Clinopyroxene
 The partition coefficient for Na2O between clinopyroxene and basaltic melt is strongly dependent on pressure and increases with increasing pressure [e.g., Blundy et al., 1995] (Figure 12). When clinopyroxene is in equilibrium with a melt, the Na2O content of clinopyroxene is controlled both by Na2O content of the melt and by equilibrium pressure. If Na2O content of the melt is fixed, equilibrium pressure can be determined from the Na2O content of the clinopyroxene. When clinopyroxene is reequilibrated at subsolidus conditions, the Na2O content of clinopyroxene does not significantly change in a four phase peridotite, except for a slight increase due to formation of exsolved orthopyroxene. Thus the Na2O content of clinopyroxene has the potential to indicate the pressure of final equilibration with melt prior to emplacement.
 The liquid in equilibrium with clinopyroxenes in Type II lherzolites are similar to mid-ocean ridge basalt (MORB) melt except for slight depletion in La and Ce (Figure 13). To calculate equilibrium pressure, we assume that the clinopyroxenes were equilibrated with a MORB in which the Na2O content ranges from 0.5 to 2.8 wt.% [Elthon, 1992]. Because the average Na2O content of clinopyroxenes in Type II lherzolites is about 1.2 wt.%, the inferred range of partition coefficient for Na2O between clinopyroxene and melt is between 0.43 and 2.4. On the basis of experimental data compiled in Figure 12 a minimum pressure of 2 GPa can be inferred for Type II lherzolites. Use of melt compositions with lower Na2O content would result in a higher minimum pressure estimate. The only way that these clinopyroxenes could have equilibrated with a melt at pressures less than 2 GPa would be for the melt to have had more than 2.5 wt.% Na2O. In view of the depleted REE contents of the lherzolites and calculated liquids, it seems implausible that the melt that equilibrated with these samples would have had a Na2O content higher than 2.5 wt.%. On the other hand, clinopyroxenes in Type I lherzolites have Na2O contents lower than 0.7 wt.%. Such low Na2O contents are consistent with equilibration between clinopyroxene and melt at lower pressures. However, in this case, the inferred range of partition coefficients for Na2O between clinopyroxene and melt, from 0 to 1.4, cannot be used to quantitatively constrain the pressure (Figure 12).
6.2. Inference From REE Abundances in Basal Lherzolites
 The lherzolites in the northern Oman ophiolite are strongly depleted in LREE relative to HREE. To a first approximation, the LREE depleted patterns are consistent with residues after partial melting. Type I lherzolites show strong depletion in LREE with slight enrichment in La and Ce. These spoon-shaped patterns can be explained by melt extraction followed by a reaction between residual peridotite and LREE-enriched melt if the peridotite has low porosity [Navon and Stolper, 1987; Bodinier et al., 1990; Takazawa et al., 1992]. Alternatively, both melt extraction and reaction operated simultaneously during percolation of partial melt through peridotite [Godard et al., 2000]. In this study, we calculated chondrite-normalized REE patterns for residues after incremental melting (1% batch increments) of primitive mantle using the nonmodal melting equation of Shaw  (Figure 14a). Note that the degree of partial melting in this study has not been corrected for the depletion of MORB source mantle relative to primitive mantle composition. The chondrite-normalized patterns for Type I lherzolites are consistent with residues after 12 to 18% melting of a garnet peridotite source containing 3% garnet. The steep slopes from MREE to HREE in Type I lherzolites require this small amount of garnet in the source [Matsukage et al., 2001]. Moreover, on the basis of the trend for Type I lherzolites in the Na2O versus MgO diagram the melting may have started around 2.5 GPa (Figure 7) which is close to the boundary between garnet- and spinel-stability fields near the solidus [Takahashi and Kushiro, 1983; Takahashi, 1986].
 On the other hand, whole rock REE patterns for Type II lherzolites are relatively flat between the MREE and HREE (Figure 8). Based on higher Na content in clinopyroxene from the Type II lherzolites, we inferred that these lherzolites experienced partial melting at pressures greater than those for Type I lherzolites (Figure 7). However, the flat patterns at MREE-HREE in the chondrite-normalized patterns for Type II lherzolites are not consistent with residual garnet peridotite. What is the explanation for this apparent paradox? One possibility is that the high Na content in the clinopyroxenes from these lherzolites is an alteration effect. However, petrographical observations of Type II lherzolites preclude this possibility. A second possibility is that a large proportion of the clinopyroxenes in Type II lherzolites (along with smaller proportions of the spinel and olivine, and perhaps orthopyroxene) formed via crystallization of a trapped melt component. In this case, it is not appropriate to discuss partitioning between residual clinopyroxene and a migrating melt that was later removed from the rock. We found that mixtures of 4 to 6% LREE-depleted melt (chondrite-normalized [La/Yb]CN ∼0.01) with residual peridotite can reproduce the chondrite-normalized REE pattern for Type II lherzolites (Figure 14c). In this model the melt is the incremental melt produced after 10% melting of the garnet peridotite source (Figure 14b). The depleted peridotite end-member is residual lherzolite equivalent to Type I lherzolites that was formed after 18% melt extraction (Figure 14a). The mixing ratio of 4–6% melt is close to the 6 to 8% melt inferred from TiO2 and Na2O (Figure 7).
6.3. Origin of Basal Lherzolites in the Northern Fizh Block
Godard et al.  described “cpx-harzburgites” from the base of Wadi Tayin massif in the southeastern Oman ophiolite. The cpx-harzburgites contain 5.3 to 6.7% clinopyroxene. The field occurrence and characteristics of Type I lherzolites from the northern Fizh block are very similar to their cpx-harzburgites. The primitive mantle-normalized REE patterns for the cpx-harzburgites also show a relatively steep slope from HREE to LREE. Godard et al.  proposed a cpx-forming melt-rock reaction at decreasing melt mass to create the spoon-shaped REE patterns of cpx-harzburgites. They envisioned a near-solidus reaction in which percolating N-MORB melt reacted with peridotite plus a small amount of trapped melt and precipitated clinopyroxene at the expense of orthopyroxene at the base of oceanic lithosphere. In the Wadi Tayin massif the strikes of dike swarms indicate the presence of a two-ridge system that developed during the opening of a NW-SE propagating ridge in slightly older lithosphere [Nicolas and Boudier, 1995]. When the ridge axis jumped and a new oceanic rift opened, the asthenospheric mantle may have thermally eroded the base of preexisting oceanic lithosphere [Godard et al., 2000]. Refertilization of preexisting lithosphere occurred during this process via percolation of an N-MORB melt as described above.
 As mentioned earlier, Type I lherzolites from the northern Fizh block are similar to the cpx-harzburgites from the Wadi Tayin massif [Godard et al., 2000]. For example, (a) both lithologies are located near the base of the mantle sections, (b) the modal proportion of clinopyroxene exceeds 5% in both peridotites, and (c) abundances of major and trace elements in both peridotites are nearly identical. However, there are also significant differences between these peridotites: (a) the cpx-harzburgites of Godard et al.  have low Al2O3/CaO ratios ranging from 0.53 to 0.69 while the ratio for Type I lherzolites in this study ranges from 0.84 to 1.1. This latter range is much closer to the primitive mantle value of 1.25 [McDonough and Sun, 1995]. Moreover, the Mg# is the same in cpx-harzburgites and in more “depleted” harzburgites in the Wadi Tayin massif [Godard et al., 2000], whereas this clearly is not the case for Type I lherzolites compared to harzburgites in the northern Fizh block. Godard et al.  explained their observations by postulating a process in which orthopyroxene was dissolved while clinopyroxene was precipitated. This model does not seem warranted for Type I lherzolites from the northern Fizh block, since these lherzolites have more orthopyroxene, as well as more clinopyroxene than the cpx-harzburgites.
 As mentioned earlier, Nicolas et al.  synthesized observations and interpretations in the northern Fizh block into a model of propagating and failing ridge segments that were oriented parallel to NNW-SSE and separated by a 10–20 km-wide transform zone located in the north of Fizh block [see Nicolas et al., 2000, Figure 11]. The NW-SE oriented shear zones in the mantle section of this area represent deformation associated with ridge propagation. According to the model of Nicolas et al. , a failing ridge was located near the base of northern Fizh block (Figure 1). We consider that this tectonic situation may have been responsible for the local presence of lherzolites in this area (Figure 15). Development of a transform zone and the progressive movement of the failing ridge may have decreased the temperature of upwelling mantle and, as a consequence, reduced the spreading rate of oceanic lithosphere. Thus, in a transient stage, the degree of melting in the upwelling mantle may have been progressively decreased over a relatively short interval so that the less depleted peridotite (i.e., Type I lherzolites) reached relatively shallow levels. We speculate that the chemical variation at the base of northern Fizh block was a relic of a transient profile in which refractory harzburgite was replaced by less depleted peridotite.
 In the model, Type I lherzolites formed by 12 to 18% of melt extraction from a fertile source mantle. The slight enrichment in LREE for Type I lherzolites requires reaction with a melt migrating within a matrix with reduced porosity. Such melt flow may have occurred at the end of melt extraction after corner flow beneath the mid-ocean ridge. Later, Type II lherzolites were formed by freezing of a trapped melt in a peridotite matrix. Because Type II lherzolites occur immediately below Type I lherzolites, we suggest that Type II lherzolites formed by refertilization of Type I lherzolites. The refertilizing melt was generated in the same melting column that produced a series of residual peridotites including Type I lherzolites. Because Type II lherzolites only occur at the base of the ophiolite, the freezing of trapped melt may have been linked to the detachment of the Oman ophiolite. Further investigation of the contact between the base of mantle section and metamorphic sole is required to identify the actual distribution of Type II lherzolites in the Oman ophiolite.
7. Summary and Conclusions
 Structural characteristics of the mantle section of the northern Fizh block are summarized as follows. Similar to other mantle sections in the Oman ophiolite, the harzburgites with coarse granular microstructure are distributed over a wide region beneath the Moho within the ophiolite. Near Wadi Zabin, a shear zone penetrates the coarse granular peridotites and extends NW over 5 km from the Moho. The shear zone is structurally continuous with porphyroclastic microstructure in peridotites at its NW termination. Peridotites with porphyroclastic microstructure occupy a wide area in the central part of the mantle section where planar features exhibit a half-dome structure opening to the SW and discordant to the basal shear zone. The lineation in the whole area is dominantly N-S, but E-W lineation is also observed.
 Basal lherzolites occur at the base of the northern Fizh block. Two types of lherzolite (Type I and Type II) were identified on the basis of their field occurrence, petrographic and geochemical characteristics. Type I lherzolites, with porphyroclastic microstructure, are sporadically distributed in the basal mylonite domain. The whole rock compositions of Type I lherzolites are highly depleted in Na, Ti, Zr and REE. Their chondrite-normalized REE patterns are steeply inclined from HREE to MREE with slight enrichment in the LREE La and Ce. Although Type I lherzolites are somewhat similar to the cpx-harzburgites from the base of Wadi Tayin massif in the southeastern Oman ophiolite [Godard et al., 2000], they differ in Al/Ca and in Mg#. On the basis of ophiolite kinematics and the development of shear zones, a propagating and failing ridge system has been proposed in the northern Fizh block [Nicolas et al., 2000]. As the failing ridge retreated toward the south the mantle temperature dropped and, as a consequence, the degree of melting was progressively reduced. The compositional variation from harzburgite to lherzolite near the base of the northern Fizh block may indicate a transient profile of decreasing degree of melting under the failing ridge. The chondrite-normalized REE patterns (except for La and Ce) of Type I lherzolites are consistent with 12–18% incremental melting of garnet-bearing fertile peridotite. Slight enrichment in LREE may have resulted from a late stage chromatographic reaction with migrating melt.
 On the other hand, Type II lherzolites have higher concentrations of Na, Ti, Zr and REE relative to Type I lherzolites. The chondrite-normalized patterns for Type II lherzolites are flat at the middle to heavy REE with strong depletion in LREE. We conclude that Type II lherzolites formed through refertilization of refractory peridotite by freezing of 4 to 8% trapped melt. We infer that the melt was a LREE-depleted melt generated by 10–12% incremental melting of fertile source mantle. The melt ascended in the upwelling mantle and refertilized depleted peridotite to form Type II lherzolites. The field occurrence of Type II lherzolites suggests that the refertilization may have coincided with the formation of the basal thrust in the Oman ophiolite.
 Critical and constructive reviews by Peter Kelemen (as Guest Editor), Alberto Saal, Günter Suhr and an anonymous reviewer greatly improved the manuscript. Fred Frey, Peter Kelemen and Mike Roden kindly provided many suggestions for improving English. We thank Bill White for his editorial assistance and pertinent advice. We have benefited from discussions with Shoji Arai, Katsuyoshi Michibayashi, Margot Godard, Fred Frey, Tsuyoshi Ishikawa, Sumio Miyashita and Yoshiko Adachi. We would like to thank Sumio Miyashita, Shoji Arai, Susumu Umino and Hodaka Kawahata for their leaderships in our field study to the Oman ophiolite from 1997 to 2001. We greatly thank to Hilal Al Azri, the Deputy Director General of Minerals, Ministry of Commerce and Industry of Oman as a sponsor of our field study. We also thank to Yoshiaki Shibata, Kazuta Kawamura, Ryoichi Nobumoto, Hiroshi Kusaka and Tomohiro Obara for their support during our field study. We would like to send special thanks to Zenji Kaminaga, Japanese Ambassador and to Japanese embassy in Sultanate of Oman for their encouragement and support during our stay in Oman. Sample dissolution technique for ICP-MS analysis was improved by advice from Dmitri Ionov. Microprobe analysis was assisted by Toshiaki Shimura. XRF analysis was supported by Toshiro Takahashi and Kenji Shuto. Financial support for the field survey was provided by Ministry of Education, Culture, Sports, Science and Technology of Japan, and Japan Society for the Promotion of Science. Financial support for chemical analyses were partially provided by SANEYOSHI SCHOLARSHIP FOUNDATION.