Hafnium Isotope Ratio in Mid-Ocean Ridge Basalts : a Global Survey and a Focus on the Southern Mid-Atlantic Ridge ( 40 ° S-55 ° S )

Using high-precision MC-ICP-MS, 176Hf/ 177Hfwas measured in 28 MORB as part of a global survey. The goal of the survey is to establish the full range EHf in EMO RB and NM ORB from sections · of the mid-ocean ridges which are within or immediately adjacent to areas of mantle plume influence. In addition, 176Hf/177Hfwas measured in 64 MORB from the Mid-Atlantic Ridge between 40°S and 5?0 S and in l basalt from the Discovery Tablemount in order to study mantle mixing processes beneath the southern South Atlantic Ocean. EHf ranges from -2.8 to + 11.5 in EMORB included in the global survey and from + 12.6 to 24.2 in NM ORB from the global survey. This range of EHf in MORB completely overlaps the range that is observed in OIB. Most of the basalts from the global survey fall on the well-established EHf-ENd mantle array. Only the depleted basalt from the Mohns Ridge falls above this array to higher EHf (+24.2) for a given ENd (9.1). EHf versus 206Pb!204Pb and EHf versus 87Sr/86Sr of basalts from the MidAtlantic Ridge (MAR) confirm the existence of the "IDMU province" from 24°S 34°N which is likely due to the pollution of the upper mantle by plume-head restites from a family ofIDMU-type plumes. In addition to this broad upper mantle pollution roughly centered beneath the equator, there is a longwavelength gradient in EHf along the MAR from + 24.2 in the Arctic down to + 14. 7 in the southern South Atlantic. This EHf gradient, which correlates positively with ENd and negatively with 207Pb/204Pb and ~207Pb!204Pb, may be due to a low~r average, time-weighted mean degree of melting, <F>, of the Atlantic upper mantle towards the south. It is also possible that the observed isotope gradients are due to a gradient in the onset of melting, T, in the Atlantic upper mantle. This model requires that T decreases to the south. Alternatively, the gradients may be due to pollution of the upper mantle by a Dupal-type component. In this case, the concentration of pollutant must increase to the south. EHfmeasured in basalts from the southern South Atlantic (40°S to 55°S) range from -2.8 to + 16.6. Large-scale pollution of the mantle beneath the southern South Atlantic Ocean can be modeled by a three component mixture of "normal" upper mantle (DM) with recycled ancient oceanic crust (OC) and recycled ancient pelagic sediment (SED). This pollution model requires that less than 5% recycled


UST OF TABLES
In this thesis, lutetium-hafnium isotope systematics of mid-ocean ridge basalts (MORB) are used to study mantle mixing and melting processes. The study is split into two parts. In the first part, which is presented in chapter 2, the global variation of 176 H£1 177 Hf in normal and hotspot-influenced mid-ocean ridge basalts is documented. The second part of the study, presented in chapter 3, focuses in more detail on the mantle beneath the southern South Atlantic Ocean. 176 Hf/ 177 Hf is measured in basalts from the Mid-Atlantic Ridge between 40°S-55°S.

Lu-Hf isotope system
176 Lu decays by beta decay to 176 Hf with a half-life of 36.9 Ga. (Dalmasso et al. 1992, Nir-EI andLavi 1998). Since Lu and Hf are refractory lithophile elements, their concentrations in the Bulk Silicate Earth can be estimated fairly accurately from their concentrations in carbonaceous chondrites.
Their concentrations in the Bulk Silicate Earth are about 2.7 times their concentrations in Cl chondrites (McDonough and Sun 1995). The modern day Bulk Silicate Earth values for isotope ratio and parentdaughter ratio for the Lu-Hf isotope system also have been determined from chondrites (C, 0 and E classes). These values are 176 H.£1 177 Hf = 0.282772 ± 29 and 176 Lu/ 177 Hf = 0.0332 ± 2 (Blichert-Toft and Albarede 1997).

Geochemistry of Lu
Lu is the heaviest of the rare earth elements (REE). REE are generally in the 3+ valence state.
This decrease in size is due to the "lanthanide contraction" (Goldschmidt 1958) which results from the successive addition of electrons to the internal f orbital. Because of this regular decrease in size, the heavier REE are accommodated more readily in the crystal structure of many mantle minerals (Wedepohl 1978). This is reflected in the REE crystal/melt partition coefficients for silicate melts which increase with increasing REE atomic number. Therefore, REE concentrations normalized to chondritic abundances are smooth functions of atomic number in basic to intermediate igneous rocks derived from the mantle (Wedepohl 1978, Bau 1995. In aqueous fluids and evolved (highly siliceous) magmas, however, the behavior of REE is no longer controlled only by charge and ionic radius. Chemical complexation, which depends on electron configuration and the nature of the chemical bond (covalent verses electrostatic), becomes important (Bau I 995). Therefore REE patterns in rocks crystallized from siliceous magmas or affected by aqueous fluids are often not smooth functions of atomic number (e.g. Bau 1996).

Geochemistry of Hf
Hf is a high field strength element (HFSE). It is in the 4+ valence state and has an ionic radius of 0. 79 A. HFSE are less soluble than REE in aqueous fluids. In those magmatic processes where charge and ionic radius control the partitioning behavior of elements, Hf behaves similarly to REE. For mantle-mineral/ silicate-melt systems, the compatibility of Hf generally falls between that of Nd (Z = 60) and Sm (Z = 62) which are REE (Sun and McDonough 1989).
In magmatic processes, Hf also behaves similarly to Zr (also a HFSE) because the valence state of these elements is the same ( 4+) and their ionic radii are very similar, again due to the "lanthanide contraction". Zr (Z = 40) has an ionic radius of0.80 A (Wedepohl 1978).
An interesting exception to this close behavior of Hf and Zr is in clinopyroxene (CPX) (Hart and Dunn 1993). According to experimental data collected on CPX crystallized from an alkali basalt, the CPX-melt partition coefficient for Zr is -0.12 while that for Hf is -0.26. Thus, MORB produced during melting in the presence of CPX should have a negative Zr anomaly (Zr normal ized < Hfnonnalized). Hart and Dunn (1993) note that this negative anomaly is not commonly observed in MORB, possibly because the presence of garnet in the bulk rock will reduce the effect of CPX on the bulk partition coefficient.

Lu-Hf in the mantle
During mantle melting, the parent element (Lu) is more compatible than the daughter element f) Therefore, solid residues have high 176 Lu/ 177 Hf in comparison to melts. Over time the upper (H . mantle, which is a residue of melt extraction to form the continents, has developed relatively high 176Hf!I77Hf by radiogenic decay of 176 Lu. The continents on the other hand have developed relatively The melting behavior of the Lu-Hf system is in many ways analogous to that of the Sm-Nd isotope system. As in the Lu-Hf system, the parent (Sm) is more compatible than the daughter (Nd). In general, terrestrial samples fall on a well-defined correlation line in Hf-Nd isotope space (Patchett et al. 1984, Vervoort et al. 1999

Zircon effect
Several aspects of the Lu-Hf isotope system make it particularly useful in studyjng mantle processes. One application of the Lu-Hf isotope system is to help distinguish different reservoirs of recycled sediments in the mantle. Pelagic sediments have relatively radiogenic Hf (high 176 Hf/ 177 Hf) while the isotope ratio in terrigenous sediments is unradiogenic (low 176 Hf/ 177 Hf) (Patchett et al. 1984).
The reason for tills fractionation in the Hf isotope system is the zircon effect. Hafnium is a minor element, as opposed to a trace element, in the mineral zircon since it can replace zirconium. Zircons are resistant to chemical and mechanical weathering. Therefore, zircon grains tend to be large and stay close to continental margins. Since the hafnium is sequestered in zircons at continental margins, pelagic clays in the deep sea have low Hf concentrations resulting in high Lu/Hf. Over time, these  (Patchett et al. 1984). Because of this characteristic hafuium behavior, it may be possible to detect recycled pelagic sediments in the source ofMORB.

Garnet effect
The Lu-Hf system is a powerful tool for studying mantle melting processes when used in conjunction with the Sm-Nd isotope system. Combined Lu-Hf and Sm-Nd data can be used to constrain the degree of melting and the amount of melting which occurs in the presence of garnet (Salters andHart 1989, Salters 1996). In the presence of most mantle minerals, Sm, Nd, Lu and Hf are all incompatible during partial melting. In the presence of garnet, however, Lu is compatible since Lu 3 + can replace the major element AJ3+. When garnet is present in the mantle lithology, the difference in compatibility is much greater for Lu verses Hf than for Sm versus Nd. Thus, melting in the presence of garnet fractionates Lu more from Hf than it fractionates Sm from Nd.
The major element composition of garnet can vary greatly. Peridotitic garnet is Ca poor in comparison to eclogitic garnet. van Westrenen et al. (2001) report that the partition coefficient of Hf is sensitive to garnet major element composition. Hf is more incompatible in peridotitic garnet than in eclogitic garnet. Lu, on the other hand, is equally compatible in both. Thus, Lu/Hf is fractionated more by melting peridotite than eclogite. This results in a more pronounced garnet effect from melting garnet peridotite than from melting eclogite.

Methods Used
In order to measure 176 Hf/ 177 Hf in MORB samples, Hf must be separated from the bulk rock.
For this thesis, the chemical separation used to isolate Hf was carried out at the University of Rhode Island. The isotope measurements were carried out on these separated samples by Janne Blichert-Toft at Ecole Normale Superieure de Lyon in France. The measurements were made on a magnetic sectormultiple collector Inductively Coupled Plasma Mass Spectrometer MC-ICP-MS (also called the Plasma 54 or P54).

Chemical Separation of Hf
The separation method used to isolate Hf from basalt samples is described in detail in . A brief summary is given here. A flow diagram of the procedure is presented in figure 1.1. In order to avoid isobaric interference in the measurement of 176 Hf (HFSE), Yb and Lu (both REE) must be removed from the sample. In addition, Ti (also a HFSE) must be removed from the sample because its presence reduces the transmission of Hf during mass spectrometry. It is not necessary to remove Zr (HFSE) from the sample.
In order to remove the bulk matrix, the basalt samples are leached in hydrochloric acid and then dissolved for 2 days in a 3: 1 mixture of concentrated hydrofluoric and nitric acid. This solution is then dried. The remaining residue is dissolved for I day in concentrated hydrofluoric acid creating a milky solution. The precipitate (containing insoluble Ca, Mg, Al and REE fluorides) is separated from the liquid (containing Hf, Zr, Cr, Ti and U) by centrifuging. The precipitate is then treated with hydrofluoric acid and centrifuged two more times to ensure complete dissolution of the hafnium. The combined supemantants are evaporated leaving a residue containing the HFSE (and other elements) and no REE. This is re-dissolved in dilute HCl-HF. This solution is loaded onto an anion exchange chromatographic column (AGI x 8 100-200 mesh). Washing with dilute HCl-HF pushes the other constituents through the column while the HFSE (Zr, Hf and Ti) remain on the resin as fluoride complexes. These are eluted from the column with 6N HCI. This is then evaporated with perchloric acid to drive off any fluorides. Next, the Ti is removed from the sample using a cation exchange column (AG50W x 8 200-400 mesh). First H20 2 is added to reduce the HFSE to negative complexes.
The Ti complex moves through the column more quickly than the Hf and Zr complexes. Once the Ti has been eluted from the column, the Hf(and Zr) is eluted with 2.5N HC1-0.3N HF. This is evaporated.
Since it is not necessary to separate the Zr from the Hf the sample is ready for Hf isotope measurement.

Hf Isotope ratio measurement with MC-IC-PMS
176 Hf/ 177 Hf is measured on the Plasma 54 MC-ICP-MS in Lyon, France. The dried sample containing the 176 Hf/ 177 Hf is dissolved in dilute nitric acid and aspirated into a chamber containing argon plasma. The plasm a, which is at very high temperatures, ionizes the Hf. The ions are accelerated and extracted from the plasma chamber through sample and skimmer cones with 6000V applied to them. The sample passes through openings in the cones and is accelerated by ion optics through an electrostatic sector. Here the ion beam is focused, reducing the energy dispersion so the beam is monoenergetic. This beam then passes into the mass spectrometer where the ions are separated through a magnetic field and collected in one of9 Faraday cups. Because of the mass difference of isotopes, each isotope travels along a different path and is collected in a different cup. The ratio of the signals from 2 cups gives the ratio of the isotopes collected in those cups. A bias factor must be applied to the signals in order to correct for different cup efficiencies. In order to determine the bias factor . for each cup, JMC-475 standard solution is measured. Since isotope ratios in this standard are known, the bias factor for each cup can be determined. The standard solution is run at the start of machine operation, as well as throughout the day to correct for possible machine drift (changes in cup bias factor). The suggested value for the JMC-475 standard is 176 Hf/ 177 Hf = 0.282160 ± 0.000010 . The internal precision of the P54 is better than 20 ppm and the external precision is better than 40 ppm . 176 Hf/ 177 Hf measurement by MC-ICP-MS replaces the thermal ionization (TIMS) method of Patchett and Tatsumoto (l980a). In TIMS the sample is loaded onto a metal filament and heated by passing a current through the filament. Since the first ionization potential of Hf is very high, the sample must be heated to very high temperatures. The yield of Hf ions in the TIMS method is poor . Therefore, this method requires large sample sizes and long measurement times for each sample. MC-ICP-MS on the other hand, requires only 50 ng of Hf and run times are on the order of 15 minutes .

Previous data collection on thesis samples
The samples included in this study have previously been analyzed for major and trace element concentrations and Pb, Sr and Nd isotope ratios. Trace element concentrations, including Lu, Hf, Sm and Nd were measured on dissolved samples at the University of Rhode Island using JCP-MS (Douglass 2000 and unpublished data). Trace element concentrations on glasses from 40°S -55°S were CHAPTER2 GLOBAL MORB SURVEY

Introduction
The earth's mantle is heterogeneous. Reservoirs of geochemically distinct material are continually formed and remixed in the mantle as a result of plate tectonics and mantle convection.
These reservoirs form primarily along mid-ocean ridges and subduction zones through differentiation processes such as partial melting, fractional crystallization and metasomatism. Through these processes, reservoirs develop different parent/daughter trace element ratios and thus, over time also different radiogenic isotope ratios. A goal of geochemistry is to identify these reservoirs and to determine how they form as well as how they remix in the mantle.
Mantle reservoirs contain one or more end-member components. These end-member components are defined based on their Pb, Sr and Nd isotope characteristics. While reservoirs are real and exist as distinct entities in the earth's mantle (e.g. anchored on the bottom of continents, piled up at the core-mantle boundary, floating as chunks or stretched schlieren in the mantle), end-members are hypothetical isotope compositions which are used to classify reservoir material. End-member isotope compositions may coincide with the actual composition of a real reservoir as in the case of DMM (endmember) and the depleted upper mantle (reservoir). Since heavy isotope ratios (e.g. Pb, Nd, Sr and Hf isotope ratios) are not fractionated by partial melting, the isotope ratios measured in oceanic basalts are representative of the isotope ratios in their mantle source. Thus, basalts can be used to help determine which end-members are involved in mantle mixing and the characteristics of the mantle reservoirs which contain these end-members.
As the technology for measuring 176 Hfi' 177 Hfhas improved (Salters andZindler 1995, Blichert-Toft et al. 1997), the body of data for the range of 176 Hfi' 177 Hf in the earth's reservoirs has increased. To date most EHf reported for oceanic samples are for ocean island basalts (OIB) and normal mid-ocean ridge basalts (NMORB) (e.g. Patchett and Tasumoto 1980b, Patchett 1983, Stille et al. 1983, Stille et al. 86 Salters and White 1998, Salters and Hart 1991, Blichert-Toft and 19 ' Albarede 1999. In order to fully document the range of eHf in MORB, the present study includes NMORB as well as enriched mid-ocean ridge basalts (EMORB) from sections of the mid-ocean ridges which are affected by mantle plumes. (Here MORB with La/Smn < 0.75 are classified as NMORB.)

End-member mixing components
The number of mantle end-members and their isotopic compositions is controversial since basalts typically sample mixtures of more than one reservoir and these reservoirs may themselves contain more than one end-member. There is disagreement on how many end-members are necessary to produce the observed variation in basalt isotope ratios. In addition, there is disagreement about the evolution of the reservoirs in which these end-members reside.
Those mantle end-members, which enclose most of the observed basalt compositions in isotope space (Hart et al. 1992), were identified by White (1985) and named by Zindler and Hart (1986). These are "depleted MORB mantle" (DMM), "high time integrated U/Pb" (HIMU, µ = U/Pb), "enriched mantle 1" (EMl), and "enriched mantle 2" (EM2). A fifth end-member, "low time integrated U/Pb" (LOMU), was introduced by Douglass et al. (1999). In addition to these five end-members, which each have extreme isotope characteristics, isotopically intermediate end-members have been defined as well. It is not clear whether reservoirs containing intermediate end-members are mixtures of reservoirs that contain the extreme end-members or whether they have evolved independently, isolated from mantle mixing. Names given to the intermediate end-members are "bulk silicate earth" (BSE), "common plume component" (C) (Hanan and Graham 1996), "focus zone" (FOZO) (Hart et al. 1992), "primitive helium mantle" (PHEM) (Farley et al. 1992) and "primitive mantle" (PRIMA) (Zindler and Hart 1986).

cf{j in mantle reservoirs
The DMM end-member is contained in the convecting upper mantle. The upper mantle is tapped at mid-ocean ridges and is the source of MORB. NMORB generally arise from pure DMM 2 while EMORB arise from mixtures of DMM with one or more of the other end-members (EMl, EM2, HIMU and LOMU). It is commonly held that the upper mantle is the residue of the melting events that formed the continents from the primitive mantle. The residue (upper mantle) is depleted in incompatible elements (depleted in Hf relative to Lu) and therefore over time has developed radiogenic l76Hf/l77Hf (positive i;Hf) because of its high Lu/Hf In NM ORB, i::Hf ranges from about + 11 to + 25 (Salters and White 1998 and references therein). This is a large range, so i::Hf of DMM is not wellconstrained (Salters and Hart 1991). i;Hf in NMORB is not well correlated with i::Nd (Patchett and Tatsumoto l 980b ).
The continents are the complement to the upper mantle. Since continents are enriched in incompatible elements (enriched in Hf relative to Lu), continental material as a whole has unradiogenic 176 Hfi' 177 Hf(negative i::Ht). After formation from the mantle, the continents themselves have undergone differentiation processes (weathering, partial melting, metasomatism etc.). Thus, the range of i;Hf in continental samples is large (about -30 to + 20) (Vervoort et al. 1999 and references therein). In contrast to NMORB, i::Hf in crustal samples (sediments, continental basalts, granitoids, juvenile crustal rocks) is well correlated with i::Nd (Vervoort et al. 1999). This correlation line in i::Hf-i::Nd isotope space is called the crustal array (Vervoort et al. 1999).
The end-members EMI and EM2 have enriched isotope signatures (high 87 Sr/ 86 Sr coupled with low EHf and i::Nd) which is characteristic of continental rather than mantle material. EM2 has significantly higher 87 Sr/ 86 Sr than EMI. Which mantle reservoirs contain the EM end-members is controversial. Oceanic crust that has been subducted together with sediment is a possible EM reservoir. 2 NMORB and EMO RB are defined based on a trace element concentration ratio· (La/Sm) while endmembers like DMM are defined by isotope ratios. Thus, it is possible to have an NMORB, which does ? 0~ have isotope characteristics ofDMM. In this case, the NMORB must have been recently depleted m m~ompatible elements resulting in decoupling of trace element ratios and isotope ratios. It is also f~ssible that a basalt has isotope characteristics of DMM, but is enriched in incompatible elements (i.e. a/Sm)n > 1). In this case, the basalt source has been recently enriched in incompatible elements. th . ase the continental material in reservoirs with EMI characteristics may be small amounts of Jn IS C , ancient pelagic sediment while reservoirs with EM2 characteristics likely contain terrigeneous sediment (Weaver et al. 1986, Weaver 1991, Chauvel et al. 1992). An alternate model suggests that the reservoir containing each of the EM end-members is metasomatized subcontinental lithosphere (Tatsumoto et al. 199 2). Finally, a third model for the EMl end-member is that it resides in recycled subducted plateau basalts (Gasperini et al. 2000). The most enriched basalts from Pitcairn, which is a typical EMl ocean island, have &Hf of about -4 (Salters and White 1998). In the EM2 end-member, which is present most clearly in the Society Islands, &Hf is about -1 (Salters and White 1998). In EM l and EM2-type OIB &Hf is strongly correlated with &Nd and forms the mantle array (Patchett and Tatsumoto 1980b, Patchett et al. 1984, Salters and Hart 1991, Salters and White 1998. The mantle array is essentially collinear with the crustal array. The crustal array and mantle array together form the terrestrial array (Vervoort et al. 1999). NMORB lie in a cluster at the high (radiogenic) end of this array. Figure 2.3 shows the different end-members relative to this array.
Recently another end-member" was proposed by Douglass et al. (1999). This is the LOMU end-member. Like EMl, it is characterized by low 206 Pb!2 04 Pb. However, LOMU has higher 87 Sr/ 86 Sr and 207 Pb/2 04 Pb than EMl . There are no clear examples of OIB containing this end-member. It seems to be present (together with DMM) in MORB from a short segment of the MAR between 48.5°S and 49°S (Douglass et al. 1999) and in several basalts from the Afanasy-Nikitin Rise (Mahoney et al. 1996). In addition, an individual fresh MORB glass (S60-18/ 1) from just north of the Bouvet triple junction seems to be tapping relatively pure LOMU material (Kamenetsky et al. 2001). It is not clear where in the mantle LOMU resides though it may be in the upper mantle since it has been found in MORB but not in OIB. Possible LOMU reservoirs include delaminated subcontinental lithosphere or lower crust.

I • i
dl ess of where the reservoir resides, it is clear that this reservoir must have been isolated for a Regar t . e because of its high 207 Pbi2 04 Pb. If Kamenetsky's MORB is representative of LOMU, EHf in long un the LOMU end-member is about -25.6 (V.S. Kamenetsky personal communication) and ENd is -18 (Kamenetsky et al. 200 l ). This falls on the mantle array, but at values much less radiogen ic than any other oceanic basalts (figure 2.3, inset).
As mentioned earlier, the evolution of the intermediate reservoir, which has C, FOZO, PHEM or pRJMA-type composition, is not certain. The material may represent pure primitive mantle material that has not yet undergone any differentiation or mixing processes. Alternately, the intermediate

Global Survey Samples
Basalts included in this global survey come from sections of the mid-ocean ridge system which are near mantle plumes. In some cases, the plumes are ridge-centered. In other cases, the plumes are off-axis and the plume material flows from the plume to the ridge (e.g. Schilling 1985). From each region, those basalts with the strongest plume signal and the strongest DMM signal are included in the survey. These were chosen based on previously collected Pb-Sr-Nd isotope ratios. Generally, the basalts with the strongest plume signal lie closest to the plumes. The basalts with the lowest plume signals (with the strongest DMM signals) lie far enough away from the plumes so that they are outside of the regions of plume influence (i.e. outside of the plume anomaly). The basalts that lie between these isotopic extremes exhibit a systematic isotopic gradient indicating that as one moves along the ridge away from a plume, the proportion of plume material mixing with DMM decreases. Sometimes this gradient along the ridge is as long as 1000 km. For those regions where basalts have been dredged from both sides of the gradient, two DMM basalts are included in the survey. Basalts locations are listed in I 2 1 Figure 2.1 shows a map of these basalts. The caption lists the references for the Nd-Sr-Pb tab e · · data used to select them. The plumes influencing MORB included in this study are the Jan Mayen plume in the Arctic, the Azores and Great Meteor plumes in the North Atlantic, the Sierra Leone plume Three basalts from the Greenland Sea are included in the survey. Two are from the Mohns ridge which lies just northeast of the ridge-centered Jan Mayen mantle plume. EN026 16D-2g from the northern end of the Mohns ridge is isotopically similar to DMM. EN026 2D-l from the southern end of the Mohns ridge exhibits mildly-HIMU or C-type isotope characteristics similar to basalts from the Jan Mayen platform . The third basalt from this region is from the center of the Kolbeinsey ridge near the Spar Fracture Zone. This depleted (DMM-like) basalt is TRI 39 25D-3g.
Isotope profiles from the Azores plume anomaly along the Mid-Atlantic Ridge (MAR) show two distinct peaks (Yu et al. 1997) possibly because the Azores anomaly is caused by two plumes or by blobs detaching and rising from a single bending plume. The Azores plume is essentially ridgecentered and mildly-HIMU/C-type except for some basalts from Sao Miguel Island, which are EM2 type (Hawkesworth et al. 1979). The samples representing DMM are TR 138 6D-l Bg from north of the Azores anomaly and TR123 5D-3g from south of the Azores anomaly. The enriched sample, TR 31Dl, is from the Azores Plateau and lies at the maximum of the northern peak of the isotope profiles on the MAR One basalt from the Great Meteor plume anomaly is also included. This anomaly is superimposed on the southern end of the Azores anomaly and straddles the Oceanographer Fracture Zone. The basalt from this region, which contains an EMl or EM2 component, is TRI 19 7D-l (e.g. Shirey et al. 1987).
The Sierra Leone plume is probably HIMU-type and affects the MAR just north of the equator . The DMM-like basalt, which lies north of the anomaly, is RC2806 49Dlg. The enriched basalt included from this area is RC2806 40D-3g.
A group of basalts that have very depleted Pb, Sr and Nd isotope characteristics lies south of the Sierra Leone plume anomaly. These basalts are from the MAR between 2°S and 7°S . Their depleted nature lead Schilling et al. (1994) to suggest that the upper mantle being tapped by the basalts here is a good example of pure unpolluted DMM end-member since these basalts approach the composition of the depleted mantle calculated through inversion of global basalt data by Allegre and coworkers , Allegre et al. 1988, Allegre and Levin 1989. One basalt from this region, EN061 40-lg, is included in this survey to try to establish E:Hf ofOMM.
St. Helena, which is a HIMU plume, lies ~800 km from the Mid-Atlantic Ridge. The MORB showing the strongest St. Helena plume influence is EN06 I l 8D-lg. RC 16 3D-l is a DMM-like basalt north of this and ENO 63 2D-5g is a DMM basalt south of this.
The EMl-type Tristan plume lies about 400 km from the MAR. The depleted sample from north of the Tristan anomaly is EN063 24D-5g. The most plume-like basalt is All 107-7 14-77. The depleted basalt, which lies between the Tristan plume anomaly and the Discovery anomaly to the south, is EW9309 400-1 g.
The most enriched sample from the Discovery anomaly is EW9309 25D-5g. This basalt has a strong EMl+EM2 signal as does the ,Discovery plume, which lies about 425 km east of the MAR (Douglass et al. 1999). Basalt EW9309 7D-6g is from the short LOMU anomaly just south of the main Discovery anomaly. Whether this enriched basalt arises from the influence of a heterogeneous Discovery plume or from a mixture of DMM with LOMU material that is separate from the Discovery plume (Douglass et al. 1999) is uncertain (see chapter 3). The depleted basalt which lies between the Discovery/ LOMU anomaly and the Shona anomaly to the south is EW9309 1 OD-3g.
The enriched basalt from the Shona anomaly is EW9309 19D-lg. The Shona plume is ridgecentered and mildly-HIMU or C-type (Douglass et al. 1999).
In contrast to the Atlantic, the Gulf of Aden is a region where the continent is breaking apart to Oung ocean Whether the Afar plume, which is rising beneath this area, initiated the breakup of fonn a Y · the continent or not is controversial although Schilling et al. (1992)  The second region included from the Pacific Ocean is the Galapagos Spreading Center which is influenced by the mildly-lllMU or C-type Galapagos plume. There is evidence that the Galapagos plume is heterogeneous (White et aJ. 1993, Blichert-Toft and White submitted). Samples included from this region are TR164 22D-lg from west of the anomaly, STD 7D-l from east of the anomaly and TR164 26D-3g from the peak of the anomaly. The total range of EHf measured in basalts from this global MORB survey is -2.8 to + 24.2. As pected within a given region, the MORB influenced by the local plume always have lower EHf than ex '

Results
the regional basalts that are outside of the area of plume influence. In MORB influenced by plumes, sHf ranges from -2.8 to + 11.5. In MORB outside of the plume anomalies (i.e. in the NMORB), EHf ranges from+ 12.6 to +24.2 which is comparable to the previously reported range of EHf in Pacific and Atlantic NMORB (+9.5 to +24.9) (Salters 1996, Salters andWhite 1998).
While NMORB from this study show a limited range of EHf, NMORB and EMORB taken together cover a range of EHf which is as large as the range found in OIB. Most of the plumeinfluenced basalts from this study (EMORB) have EHf close to 9 which is typical of the intermediate-

Mohns Ridge
While most of the EHf-ENd data for basalts from this global MORB survey do fall along the &Hf-&Nd mantle array (figure 2.3), there is one region that deviates from this correlation. This is the northern Mohns Ridge in the Arctic. The NMORB measured from this ridge, EN026 16D-2g, has a very high EHf for its ENd. The reason for this requires a more in depth study of the area, but there are several possible mechanisms that can lead to high Elli for a given ENd in an NMORB. The first possibility is that the source of this basalt contains a small amount of ancient pelagic sediment. Pelagic sediment has high Ellf for a given ENd due to the zircon effect (Patchett et al. 1984) as described in chapter 1. However, the amount of pelagic sediment mixing with DMM to produce EN026 16D-2g must be very small for two reasons. First pelagic sediment is enriched in incompatible elements and this basalt is depleted in incompatible elements. Second pelagic sediment has low ENd and the Mohns Ridge basalt does not have particularly low ENd in comparison to other NMORB.
Another possibility is that the source of this Mohns Ridge NM ORB is the residue of an ancient melting event, during which the upper mantle melted in the garnet stability field rather than in the spine! stability field (Johnson and Beard 1993). The residue of garnet peridotite melting has very high Lu/Hf compared to the residue of spine! peridotite melting because Lu is compatible in garnet. Over time, this high Lu/Hf leads to radiogenic 176 Hf7 177 Hf. The presence of garnet will not have a big effect on Sm/Nd of this ancient residue. Therefore, 143 Nd/ 144 Nd in the residue will be similar to that in upper mantle, which has undergone previous melting events in the spinet stability field. Thus, the result of melting ancient garnet peridotite residue is a basalt with "normal" ENd and high EHf. This is an example of the garnet effect described in chapter 1. Since the Mohns Ridge is close to the continents, it is quite possible that fragments of the sub-continental mantle which have experienced an ancient garnet effect are present in the upper mantle beneath the ridge and contribute to the source ofMORB here.

dffofDMM
As noted by Salters and White ( 1998), the range of EHf documented for NM ORB is + 11 to + 25. This large range in EHf is accompanied by a smaller range in ENd ( + 7 to + 13) (Salters and White 1998, Atlantic and Pacific MORB field in their figure 1). Because of the large range of EHf in NMORB, defining EHf of the DMM end-member is difficult. The basalt from the depleted 2°S -7°S region along the equatorial MAR (EN06 l 4D-lg) has EHf = 17.6 which a little higher than the average of the fifteen NMORB from this study ( 17. l ). If 2°S -7°S really is representative of pure unpolluted DMM, as suggested by Schilling et al. (1994), then 17.6 may be a reasonable value for EHf of DMM.
An alternative method to determine EHf ofDMM is to calculate it from ENd of DMM (from Hart et al. 1992 ) and the equation for the mantle array. This results in i:;Hf of DMM = 20.8 (see caption of figure   23 for equation of mantle array).
There is evidence from the St. Helena plume anomaly that 20.8 is a better estimate of i:;Hf in DMM than 17.6. Since St. Helena is a HIMU plume, OIB from this island fall noticeably below the mantle array (figure 2.3) (e.g. Salters and White 1998). Thus, the most enriched basalt from the MAR affected by this plume (EN061 I 8D-1 g) should fall below the mantle array as well. Figure 2.3 shows that EN061 18D-lg does fall slightly below the mantle array, but not nearly as much as typical HIMUtype OIB do. Presumably, this is because EN061 18D-lg is not pure HIMU material, but a mixture of (This argument assumes that mixtures between HIMU and DMM fall along straight lines. In order for this to be true, Hf/Nd in HIMU must be the same as Hf/Nd in the upper mantle. This is a reasonable assumption since Hf behaves similarly to Nd and Sm in most mantle melting processes.)

MORBfrom the MAR between 34°N and 24°S -the "HIMU province"
Based on i:;Nd versus 87 Sr/ 86 Sr for a large set of Atlantic MORB, Schilling et al. (1994) divided Atlantic MORB from 6°S to 79°N into two isotopic groups. The northern group includes basalts from 79°N to 34 °N and the southern group includes basalts from 34 °N to 6°S. The southern group can be extended to 24°S if the subsequent data of Fontignie and Schilling (1996) from the MAR near Ascension and St. Helena are included. The isotopic distinction between these two groups is evident both in EMORB and in NMORB. The southern group has a lower i:;Nd for a given 87 Sr/ 86 Sr than MORB from the northern group (figure 2.6a). Since low ENd for a given 87 Sr/ 86 Sr is a characteristic of the HIMU end-member, Schilling et al. (1994) call this 24°S -34°N group the "HIMU province". (This tn . ce includes basalts from the 14°N MAR anomaly documented by Dosso et al. (1991).) prov &Hf versus 87 Sr/ 86 Sr for the basalts from the current global survey confirms the existence of these two distinct MORB populations between 24°S and 79°N. Figure 2.6b shows that for a given s1sr/s6sr, the southern population has lower &Hf than the northern population, both for MORB influenced by plumes (solid symbols) and for normal MORB (open symbols). In contrast to the population originally considered by Schilling et al. (1994), figure 2.6 includes only a small basalt population. However, work in progress at URI indicates that &Hf of the other basalts from these regions are consistent with the trends reported here.
The isotopic differences between these two regions are likely caused by the ditforent plumetypes which are upwelling in the two regions. Schilling et al. (1994) noted that the "HIMU province" (24°S -34°N) is influenced by strongly HIMU-type plumes (l4°N, Sierra Leone, Circe (or Ascension) and St. Helena). The northern region, on the other hand, is influenced by mildly-HIMU/ C-type and EM2-type plumes (Jan Mayen, Iceland, Azores and Great Meteor). Since the plume-types in these two regions are distinct, EMORB, which are mixtures of upper mantle and plume material, fall into distinct groups. Since NMORB also fall into these groups, NMORB from the "HfMU province" likely are affected by plume material as well. However, NMORB are incompatible element depleted by definition. Thus, this plume material influencing the upper mantle in the "HIMU province" must have been depleted in incompatible elements relatively recently. This can be achieved by melting HIMU plume-heads (which depletes the plume-head residue in incompatible elements without changing isotope ratios) followed by dispersal of the plume-head residue into the upper mantle (e.g. Hanan et al. 1986).

Long-and medium-wavelength signals in Atlantic NMORB
To further investigate the cause for the two isotopic provinces discussed in the previous section, and to investigate the nature of the entire upper mantle beneath the Atlantic, it is helpful to consider isotope profiles for the Atlantic NMORB. &Hf collected in this global survey show that Atlantic NMORB, rather than falling into two groups, actually exhibit a remarkable 15,000 km long 19 gradient along the entire MAR from 79°N to 55°S. For this MAR basalt population there is a clear in i::Hf from north to south from +24.2 down to +14.7 (figure 2.7a, red line). A similar decrease gradient is also visible in i::Nd versus latitude for NMORB (figure 2.7b, red line), although this is not as un ced The opposite trend is visible for these NMORB in 207 Pbi2 04 Pb and ~2 07 Pb/ 204 Pb 3 versus prono · latitude which both increase from north to south (figure 2.7c and 2.7d, red lines). These four trends contrast with 206 Pbi2 04 Pb and 87 Sr/ 86 Sr versus latitude. The 206 Pbi2 04 Pb profile has a broad maximum roughly centered on the equator (figure 2.7e, red line) and 87 Sr/ 86 Sr has a broad minimum over the equator (figure 2. 7f, red line).
From the NMORB profiles in figure 2. 7, it is apparent that the isotopic trends can be divided into two categories. The first category is a long-wavelength trend represented by the 15,000 km gradient along the entire MAR (apparent in i::Hf, i::Nd, 207 Pbi2 04 Pb and ~2 07 Pb!2 04 Pb). The second category is a medium-wavelength trend which is manifested in a broad maximum or minimum at the equator (apparent in 206 Pb/ 204 Pb and 87 Sr/ 86 Sr). This indicates that there are at least two processes affecting the Atlantic NMORB population and by inference, the Atlantic upper mantle. One process produces long-wavelength signals and the other produces medium-wavelength signals.
The isotope signals from these two processes are superimposed on each other. The isotope profile observed along the MAR is a function of the magnitudes of the isotope signals from the two processes. For example, i::Hf is relatively unaffected by the medium-wavelength process and therefore increases linearly to the south without any increase or decrease near the equator due to the mediumwavelength process. 206 Pb!2 04 Pb on the other hand is much more affected by the medium-wavelength process and therefore exhibits a maximum over the equator. There is however, a very slight effect on 206Pb/204 . Pb from the long-wavelength process as well smce NMORB from the south have slightly higher 206 Pb!2 04 Pb than NMORB from the north (figure 2.7e, red line, 206 Pbi2 04 Pb = 17.6 the Arctic NMORB and 17.9 in the southern South Atlantic NMORB). 4 ~0f 07 Pbi2 04 Pb represents the deviation from the Northern Hemisphere Reference Line in 206 Pbi2 04 Pb -Pb/ 204 Pb space (Hart 1984 The medium-wavelength process affects the 206 Pbi2°4Pb and 87 Sr/ 86 Sr profiles along the MAR which have a broad maximum and a broad minimum respectively near the equator. This process likely is the dispersal of IDMU plume-head residue, discussed in the previous section, which results in the "HlMU province" from 34°S -24°N. 206 Pb/ 204 Pb of filMU is significantly higher than that of DMM (figure 2.2a or 2.4a). In comparison, i;Hf and i;Nd of IDMU and DMM, while certainly not equal, are much closer in value (fig 2.2a and 2.4b). In addition, since HIMU material is old but not ancient (perhaps -2Ga rather than -3Ga), it does not have particularly high 207 Pbi2 04 Pb for its 206 Pbi2 04 Pb. Thus, the dispersal of HIMU plume-head restites into the upper mantle does not affect sHf, sNd, or 201pb/ 204 pb nearly as much as 206 Pbi2 04 Pb. sHf, sNd and 207 Pbi2 04 Pb are essentially "immune" to the dispersal offllMU plume-head restite. Therefore, these three isotope signals do not exhibit a maximum (or minimum) near the equator. (1) wavelength plume-ridge interactions affect EMORB much more strongly than NMORB. Thus, by consi~ering just the NM ORB, it is possible to analyze just the large-scale and medium-scale processes affectmg the upper mantle.

5
The effect of this long-wavelength process on 206 Pbi2 04 Pb and 87 Sr/ 86 Sr is difficult to discern because ~ese two isotope ratios are affected so strongly by the medium-wavelength HIMU plume-head ~spersal. In order to include 206 Pbi2 04 Pb and 87 Sr/ 86 Sr in an analysis oflong-wavelength processes, it fi~ul~ be n~cessary to filter out the effect of the medium-wavelength process. In the absence of such termg, neither of these isotope ratios is a good tool for determining what process causes the longwavelength isotope trends. Such a gradient in <F> is consistent with the observation that the EHf gradient is much more pronounced than the i;Nd gradient. Since Lu/Hf is fractionated by melting much more than Sm/Nd is, a given degree of ancient melting will result in a more pronounced signal in EHf than in i;Nd. In addition, since the half-life of 176 Lu is shorter than that of 147 Nd, i;Hf increases more quickly than i;Nd. Since the gradient is apparent in 207 Pb/2°4Pb, it is also necessary that the melting event is ancient because 207 Pb/ 204 Pb is produced by the almost extinct 235 U.
Model lb: long wavelength gradient in the age of melting, T (constant < 176 Lu/ 177 Hf>) The observed decrease of EHf and i;Nd from north to south is also consistent with a model where < 176 Lu/ 177 Hf> is constant with latitude, but the age of upper mantle melting, T, decreases to the south (figure 2.10). This model of earlier melting in the north than in the south will cause an increase in 207 Pb/ 204 Pb from north to south. This is because the parent element (U) is more incompatible than the daughter element (Pb) for melt extraction at mid-ocean ridges.
~increased U/Pb also should lead to increased 206 Pb!2 04 Pb towards the south. However, this is not 0 ved. It is likely that this is because the observed 206 Pb!2 04 Pb profile is obscured by the overwhelming effect on 206 Pb!2 04 Pb of the medium-wavelength dispersal of HIMU plume-heads. Models la and 1 b indicate that ancient melt processes may indeed be the cause of the long-I ngth isotope gradients observed along the MAR. This is either via a gradient in the average, wavee time-weighted mean degree of melting, <F> or via a gradient in the age of melting, T (or both). Either <F> decreases to the south or T decreases to the south (or both).
Model 2: large-scale pollution ~ Another possible model for the process controlling the long-wavelength isotope gradients in the NMORB population is that the Atlantic upper mantle has been polluted on a large-scale by a Dupal component (Hart 1984, Dupre and Allegre 1983, Hamelin and Allegre 1985, Hamelin et al. 1985.

Conclusions
The Hf isotope system in MORB was largely ignored for many years because of its similarity to the Nd isotope system and difficulty in 176 Hfl 177 Hf measurement. This global MORB survey shows that the two systems are in fact not redundant. There are important differences between them that make i:Hf a valuable additional source of information about mantle processes.
In the Arctic, high i::Hf for a given i::Nd indicates that the two isotope systems have become decoupled in the local upper mantle. This mantle may have been polluted locally by ancient pelagic sediment. Alternatively, this upper mantle may contain material which has undergone ancient melt extraction in the garnet stability field rather than in the spine! stability field. This ancient garnet restite e from the sub-continental mantle which is fairly close to the Mohns Ridge. Either of these may com 5 lead to high dff for a given ENd. processe EHf of DMM is difficult to define because of the relatively large range of EHf in NMORB. t:Hf from this global survey provides a reasonable estimate for EHf of DMM. Based on the EHf-ENd composition of the EM ORB from the St. Helena plume anomaly and the composition of HIMU-type ocean island basalts, EHf of DMM is estimated as 20.8.
EHf for MORB from the Mid-Atlantic Ridge confirms the existence of a "HIMU province" from 24°N to 34°S. This province is formed by a medium-wavelength process, namely, the pollution of the upper mantle around the equator by HIMU plume-head restites. Low 87 Sr/ 86 Sr for a given EHf and t:Nd distinguishes this "HIMU province" from the rest of the Atlantic. The isotopic signal of this process is most clear in profiles of 206 Pbi2 04 Pb and 87 Sr/ 86 Sr versus latitude, which exhibit a broad maximum and minimum respectively over the "HIMU province".
The 15,000 km long decrease of EHf in NMORB southwards along the Mid-Atlantic Ridge clearly shows that there is a long-wavelength process, which affects the entire Atlantic upper mantle.
This process also produces the decrease in ENd and increase in ~ 207 Pbi2 04 Pb and 207 Pbi2 04 Pb southwards.
It is possible that these gradients are due to an ancient gradient in the average, time-weighted mean degree of melting, <F>, experienced by the upper mantle. This model is supported by the stronger gradient in EHf compared to ENd. In this model, <F> decreases to the south. The gradients can also be explained by a gradient in the age, T, when melting began in the Atlantic upper mantle. In this case, melting in the north must have started earlier than in the south. A final possibility is that these isotope gradients are due to pollution (either ancient or recent) and that the proportion of the Dupal-type pollutant present in the upper mantle increases to the south. In addition, localized plume-upper mantle mixing relationships may be complex because plumes themselves may be heterogeneous. The Hf isotope data presented here build on previous major element, trace element, rare gas and isotope data for the same sample set (Douglass et al. 1995, Douglass et al. 1999, Douglass 2000, le Roux et al. a in press, le Roux et al. b submitted, le Roux et al. c in preparation, Sarda et al. 2000, Moreira et al. 1995, Schiano et al. 1997, and Eiler et al. 2000.

South Atlantic plume-ridge interactions
Numerous mantle plumes are upwelling beneath the South Atlantic Ocean. According to the plume-source/ridge-sink model (Schilling 1985), material can travel from an upwelling mantle plume along the base of the lithosphere to a ridge where it mixes with upper mantle material and is incorporated in mid ocean ridge basalts (MORB) (Morgan 1978, Schilling 1985, Kincaid et al. 1996.
Plume-ridge interactions have been documented between the Mid-Atlantic Ridge (MAR) and the Circe (or Ascension), St. Helena, Tristan da Cunha, Gough, Discovery, and Shona mantle plumes (Hanan et al. 1986, Graham et al. 1992, Fontignie and Schilling 1996, Douglass et al. 1995, Douglass et al. 1999. Previously collected Pb, Sr and Nd isotope data indicate that compositions of some MORB from the present study area (40°S-55°S) are influenced by the Discovery and Shona plumes (figures 3.1 and 3.2) (Douglass et al. 1995, Douglass et al. 1999. The Discovery plume is located about 425 km east of the ridge at 44.45°S, 6.45°W (Douglass et al. 1995). MORB influenced by the Discovery plume lie along the MAR between 40°S and 48.2°S. Since the Discovery plume is EM-type, the Discovery anomaly is most apparent in a 87 Sr/ 86 Sr profile (figure 3.3c). The Shona plume is ridge centered at 51.5°s (Small 1995, Douglass et al. 1995. The Shona plume is mildly-HIMU or C-type and influences MORB compositions from 49.2°S -52.65°S. This HIMU influence is most clearly seen on a 206pb/ 204 Pb profile (figure 3.3d).
The location of the mildly-HIMU or C-type Bouvet plume, which may also influence the compositions of the southernmost basalts in this study area (le Roex et al. 1987, Douglass et al. 1999, is uncertain. It may be located beneath Bouvet Island on the Antarctic Plate (Kurz et al. 1998, le Roex et al. 1983). Another suggestion by H.J.B. Dick (personal communication) is that the plume is currently located at the Bouvet triple junction (intersection of the MAR, South West Indian Ridge and American Antarctic Ridge) and that it oscillates back and forth between this location and Bouvet Island.
large-scale pollution of South Atlantic MORB source (Dupal, plume-heads, LOMU) In addition to the localized influences of plumes on the MORB source, the southern South Atlantic may represent a transition region from a MORB source that is normal Pacific/North Atlantictype upper mantle to a source that is Dupal-type upper mantle. The Dupal anomaly was originally defined as a longitudinal band of ocean island basalts (OIB) and MORB stretching from about 0° -60°S (centered around 30°S -40°S) (Dupre and Allegre 1983, Hart 1984, Hamelin and Allegre 1985, Hamelin et al. 1986). In comparison to basalts which lie outside of this band, these Dupal anomaly basalts have high 208 Pbi2 04 Pb and 207 Pb/ 204 Pb for a given 206 Pb/ 204 Pb as well as high 87 Sr/ 86 Sr. There is disagreement about the area influenced by the Dupal anomaly. Mahoney et al. (1992) find evidence of a transition region from Dupal to North Atlantic/ Pacific mantle along the South West Indian Ridge l ?oE _ 26°E. However, Dupal material may extend further to the west into the South Atlantic between . fr ro 24°S to 50°S along the MAR (Hanan et al. 1986, Castillo 1988, Fontignie and Schilling region o 1996, Douglass and Schilling 2000. The cause of the anomalous Dupal basalts must be a reservoir with high time integrated Rb/Sr, 'Jh/Pb, and Th/U, which has been isolated for a long time since it also has high 207 Pbf2<»Pb. {2°7Pb is produced by the now almost extinct 235 U.) The source of this anomalous reservoir may be mantle contaminated by recycled continental sediments or delaminated subcontinental lithosphere (Hart 1984).
In the model supported by Hart (1984) and Dupre and Allegre (1983), the Dupal source is deep in the mantle and is brought up by plumes thereby producing the Dupal OIB. Then these plumes contaminate the upper mantle and this imparts a Dupal signature on MORB as well. In contrast, the Dupal source may reside in the subcontinental lithosphere (Hawkesworth et al.1986). In this case, the Dupal signature in basalts may be due to erosion and dispersal of this shallow Dupal reservoir into the upper mantle by upwelling plumes (e.g. Storey et al. 1989) or by remobilization of the subcontinental lithosphere during continental breakup (Fontignie and Schilling 1996).
A synthesis of these models has been suggested for large-scale contamination of the South Atlantic upper mantle. The region has been polluted by dispersal of the St. Helena plume-head (which is not Dupal-type) from -2°S -47°S and the Tristan plume-head (which is Dupal-type) from -24°S -47"S (Hanan et al. 1986). In addition to this upper mantle pollution by plume-heads, delaminated subcontinental lithosphere may be present as passive heterogeneities in the upper mantle (Douglass et al. 1999, Mahoney et al. 1992. In the view of these authors, these heterogeneities, which Douglass et al. (1999) call LOMU material, were delaminated during the breakup ofGondwana. The heterogeneities may be widely dispersed throughout the region, but since they are refractory, large heterogeneities are only melted and incorporated in MORB when plumes are available to provide excess heat (Douglass et al. 1999, Mahoney et al. 1992). According to Douglass et al.

Results
The MORB included in this part of the study come from three cruises. Fifty-two samples were collected during cruise 93-09 aboard the RIV Maurice Ewing (EW9309) (Douglass et al. 1995). Four samples between 40°S and 46°S are from cruise 107-07 aboard the RIV Atlantis II (Aii 107-7) . Eight of the southernmost samples are from cruise 32 aboard the RIV Agulhas (AG32) (le Roex et al. 1987). The MORB are all glass rims, except for six of the AG32 basalts, which are pillow interiors. In addition to the 64 MORB, one basalt from the Discovery Tablemount is included in the study (Kempe and Schilling 1974, le Roux et al. in preparation). This sample is a crushed whole rock powder. As expected, the 176 Hf/ 177 Hf profile along the MAR between 40°S and 55°S resembles that of 143 Nd/ 144 Nd ( figure 3.3 a and b). In regions of plume-ridge interaction, both 176 Hf/ 177 Hf and 143 Nd/ 144 Nd are lower (less radiogenic) than in the surrounding NMORB population. This indicates that these plume-influenced basalts arise from a source that has time integrated enrichment in incompatible elements (i.e. low Sm/Nd and low Lu/Ht) which is typical of mantle plumes.
Basalts from 40°S to 45.2°S are largely NMORB with relatively high (non plume-like) l76fID177 Hf ( figure 3.3a). For the most part, the ridge here is probably not currently being fed by plume ·a1 Hf ranges from + l 0.9 to + 16.6 which is a little lower than the estimates for normal North matert . E . /Pacific type NMORB (i:Hf = 20.8 or i:Hf = 17.6 depending on how i:Hf of DMM is defined). Atlantic (See chapter 2 for a discussion of the methods used to determine i:Hf of DMM.) However, one of the two EMORB from this region (EW9309 42D-lg) does have moderately low (moderately plume-like) &Hfof7.5. This is probably due to mixing of the upper mantle with a small amount of Gough plume material as has been suggested by Douglass et al. (1999) to account for elevated Sr and Pb isotope ratios. Gough Island lies approximately at 40°S, l 0°W and a small chain of seamounts extends from the island to the east just south of the Walvis Ridge-Tristan hotspot track (Humphries et al. 1985).
The Hf data collected in this study support the model of Discovery and Shona plume-ridge interactions along the MAR between 45°S and 53°S deduced from Pb-Sr-Nd isotope space (Douglass et al. 1995, Douglass et al. 1999. The 176 Hf/ 177 Hfprofile (figure 3.3a) shows two broad regions of low t76Hf/ 177 Hf due to the influences of the Discovery plume from 45.2°S to 48.2°S (i:Hf= -2.8 to 14.4) and the Shona plume from 49.2°S to 52.6°S (i:Hf= 6.6 to 14.7). In addition to these two broad regions of low 176 Hf/ 177 Hf, there is one narrow region of low 176 Hf7 177 Hf just south of the main Discovery anomaly from 48.5°S to 49 .I 0 S ( i:Hf = -1.6 to 7. 7). This region is called the LOMU segment by Douglass et al. (1999), although if the Discovery plume is heterogeneous, it may be part of the region of Discovery plume influence.
Establishing the southern end of the Shona anomaly with confidence is difficult because there is a gap in dredging between 53°S and · 54°S (Douglass et al. 1995(Douglass et al. , 1999. Gravity data shows that an axial valley is present in the short region between 52.2°S and 53°S while most of the surrounding regions to the north and south have smooth topography with no axial valley (Douglass et al. 1999). If the Shona anomaly ends at 52.6°S, the low Hf ratios just north of the triple junction are likely due to material from the Bouvet plume. Sorting out these mixing relationships is not unambiguous since both the Bouvet and Shona plumes are mildly-HIMU or C-type and therefore do not have distinctive isotope signals.
It is interesting to note that 176 Hf/ 177 Hf of the two basalts from dredge station EW9309 25D along the MAR are even lower (more plume-like) than 176 Hf7 177 Hf in the basalt from the Discovery Tablemount. The Discovery Tablemount (42°S, 0°) is northeast of the present location of the Discovery plume (Douglass et al. 1995). The Discovery Tablemount basalt is 25 Ma and was probably formed by intraplate Discovery plume volcanism (Kempe andSchilling 1974, Douglass et al. 1999, le Roux et al. in preparation).

Mixing relationships
Upper mantle -plume mixing between 40°S -55°S Jn order to help establish whether mixing is occurring and which mantle reservoirs are involved in the mixing process, the data can be plotted on isotope-isotope plots. The three mixing The Discovery, LOMU and Shona mixing vectors are not distinguishable in i::Hf-i::Nd space (figure 3.5). All of these basalts fall on the Hf-Nd mantle array (Patchett andTatsumoto 1984, Blichert-Toft and. However, basalts from the Discovery and LOMU anomalies extend to much lower i::Hf and i::Nd values than the Shona anomaly basalts. This indicates there is a continental component in the sources of the Discovery and LOMU basalts. This is supported by the i::Hf -87 Sr/ 86 Sr (fi ure 3 6) which shows that Discovery and LOMU basalts extend to much higher 87 Sr/ 86 Sr than plot g · ' ano maly basalts. It is also noteworthy that for a given 87 Sr/ 86 Sr value, eHf is lower for Shona Shona anomaly basalts than for Discovery or LOMU basalts.

Ambient depleted mantle verses "normal" depleted mantle
Each of the three mixing trends identified in the previous section is a mixture between an enriched source (i.e. the Shona plume, Discovery plume or LOMU rafters) and a depleted source (i.e. the ambient upper mantle). Douglass et al. (1999) found that the upper mantle with which the Discovery and Shona plumes are mixing is not the normal North Atlantic/Pacific -type upper mantle.
They defined the likely isotopic composition of this ambient upper mantle from the intersection of the Discovery and Shona mixing trends on isotope-isotope plots. They found that the ambient upper mantle is radiogenic relative to the North Atlantic and Pacific in terms of Pb and Sr isotopes and unradiogenic in tenns of Nd isotopes. This indicates that in comparison to normal upper mantle, the South Atlantic ambient upper mantle may have experienced a slight long term enrichment in incompatible elements (i.e. model 2 of chapter 2) or less long term depletion in incompatible elements (i.e. model la of chapter 2) leading to high Rb/Sr, U/Pb , Th/Pb and low Sm/Nd parent-daughter ratios. Alternately, the mantle may have been polluted relatively recently with a component with radiogenic Pb and Sr and unradiogenic Nd. The isotopic composition of this ambient upper mantle is compared to normal upper mantle in table 3.2. (There is much controversy surrounding the "real" composition of the normal upper mantle. Some workers advocate a composition like that of Hart et al. (1992) which is based on the mean isotopic composition of NMORB rather than the most depleted NMORB. Other workers support a more extreme end-member type composition for the normal upper mantle which is either based on the most depleted NMORB documented from 2°S-7°S in the Equatorial Atlantic  or on inversion of global NMORB isotope data , Allegre et al. 1988, Allegre and Lewin 1989.) The data collected in this study now allow characterization of the 176 lID 177 Hf of the South Atl · · antic ambient upper mantle (figure 3.7). Following the method of Douglass et al. (1999), the . ection of the Shona and Discovery trends is roughly 176 Hf/ 177 Hf = 0.28324 (&Hf= 16.6). This mters ambient mantle value compares to &Hf= 20.8 for normal upper mantle (see caption of table 3.2 for calculation). Thus, the South Atlantic ambient upper mantle is relatively unradiogenic in comparison to the North Atlantic and Pacific in terms of Hf as one would expect based on Nd isotopes. This indicates that the ambient upper mantle (or a pollutant present in the ambient upper mantle) has experienced long-term depletion of Lu relative to Hf. This is consistent with a slightly incompatible element enriched (or slightly less depleted) ambient upper mantle (compared to normal upper mantle) since Hf is more incompatible than Lu.
In order to investigate the source of the South Atlantic upper mantle "pollution" which causes Helena, Tubaii, Mangaia (e.g. White 1985)). It is much more common to find mildly-HIMU or C-type plumes (Shona, Bouvet, Iceland, Azores (e.g. Hanan and Graham 1996)). This is probably because subduction of oceanic crust that is completely free of sediments is unlikely and even a small amount of sediment quickly erases the high 206 Pb!2 04 Pb associated with pure HIMU plumes.
Many sources of upper mantle pollution have been proposed (i.e. subcontinental lithosphere, subduction zone fluids, garnet granulites from the lower crust). The calculations described above do not rule out these other sources of pollution, but the calculations do suggest that if pelagic sediments are involved in the pollution of the upper mantle, they must be present together with recycled oceanic crust.
'Jbe mixing calculations also do not constrain how the pollutants were dispersed in the upper mantle.

Mixing lines within the Shona anomaly
In isotope-isotope space, the Shona anomaly (49.25°S-52.65°S) is quite broad in comparison to the Discovery and LOMU anomalies. 176 Hf7 177 Hffor these basalts reveal systematic structure within the Shona anomaly that has not been recognized previously based only on Pb-Sr-Nd isotope space. This structure indicates that the Shona anomaly cannot be caused by simple binary mixing of a homogeneous plume with a homogeneous ambient upper mantle. At least three mixing end-members are needed to describe the observed isotopic structure of the Shona anomaly. Two different physical models can be used to describe mixing of these three end-members.

Model l: homogeneous plume
In the first model ( figure 3.9a), a homogeneous plume is mixing with a heterogeneous upper mantle. In this case, the ambient upper mantle has passive heterogeneities embedded in it. The plume, ambient mantle and passive heterogeneities mix in varying proportions to produce the MORB.

Model 2: heterogeneous plume
In a second possible model, the local upper mantle is homogeneous but the upwelling Shona plume is heterogeneous ( figure 3.9b). In this case, the mixing of the three end-members is sequential binary mixing. In stage 1, two end-members mix to produce the heterogeneous Shona plume. Then in stage 2 this heterogeneous plume mixes with the homogeneous ambient mantle. It is possible that three component DM-OC-SED mixtures, which are used in the previous section to model the ambient upper mantle, also give rise to basalts from the Shona anomaly. In order to test this hypothesis, the model of Rehkamper and Hofinann (1997) is applied to Shona anomaly basalts. In this case, however, rather than using pure upper mantle (DM) as a mixing end-member, the polluted ambient mantle (ADM) is used. The isotope composition of ADM is taken from the intersection of the Shona and Discovery anomalies in isotope-isotope space (Douglass et al. 1999   Therefore, if the sediment fraction is less than 2% and the ratio of sediments to oceanic crust is constant, SED-ADM-OC mixing trends are linear in isotope-isotope space and point from ADM to the composition of the recycled component. It is evident that segments 2, 3 and 4 from the Shona anomaly each fall along a trend of constant sediment-oceanic crust ratio (~gure 3.14). Thus, each segment can be considered a binary mixture. The depleted end-member in all three cases is the ambient upper mantle (ADM). In segment 2 the recycled end-member is a mixture of approximately 0.7% sediment with 97.3% oceanic crust. (The best-fit in Sr-Pb isotope space is 0.4% sediment while in Nd-Pb, Hf-Pb and 207 PbJ2 04 Pb-206 PbJ2 04 Pb space the best fit is for a recycled component with 0.7% sediment.) In segment 3 the recycled endmember contains about 1.1 % sediment and in segment 4 this increases to about 1.6% sediment. These results show that very small amounts of pelagic sediment have a very large effect on the composition of the resulting mixture. Mixtures of these three recycled components (0.7%, 1.1% and 1.6% sediment) with ADM also successfully reproduce the observed Ba/Nb trends of segments 2, 3 and 4 (figure 3.14/).
The above results imply that the Shona plume consists of recycled material, namely anciently subducted oceanic crust with pelagic sediment. In this model, the Shona plume is not well mixed since there seem to be three different "flavors" of recycled component (i.e. three different crust/sediment ratios) mixing with a locally homogeneous ambient upper mantle. This ambient upper mantle was likely polluted with oceanic crust and pelagic sediment in an event prior to (and likely unrelated to) the present-day mixing with the Shona plume.
It is of course possible that the Shona plume is in fact homogeneous (i.e. model 1), either with a HIMU composition (pure oceanic crust with no sediment present) or C composition (some small, but constant fraction of sediments present). In this case, the ambient upper mantle with which the plume is mixing must be heterogeneous with the sediment fraction in the ambient upper mantle increasing to the south. In such a scenario, however, it is difficult to explain how the sediment-oceanic crust fraction remains constant over a given ridge segment since the oceanic crust resides in the plume and the sediment resides in the ambient upper mantle. There is evidence for heterogeneous plumes in other areas such as Hawaii (Hauri 1996), Iceland (Chauvel and Hemond 2000) and the Galapagos (Blichert-Toft and White submitted). It is likely that the Shona plume is also an example of a heterogeneous plume rather than a homogeneous plume mixing with a locally heterogeneous ambient upper mantle.
A final point can be made regarding ADM-OC-SED mixtures and Hf-Nd isotope systematics. Patchett et al. ( 1984) suggested that Elli can be used to evaluate whether pelagic sediment is included in the source of a basalt because EHf for basalts which include pelagic sediment will fall above the Hf-Nd mantle array. This is because, for a given ENd, Elli is high in pelagic sediment due to the high time One possibility is that pelagic sediment is in fact mixing with ambient depleted mantle and oceanic crust to produce Discovery and LOMU anomaly basalts but that SED is an inaccurate representation of pelagic sediment. If this is true, 207 Pbl2 04 Pb ofSED is too low and Pb concentration is too high to accurately represent ancient pelagic sediment. The SED end-member used in the previous ADM-OC-SED model is from Rehkamper and Hofmann (1997). These authors estimate the 206 Pb/ 204 Pb, 207 Pb!2 04 Pb and 208 Pb!2 04 Pb of 1.5 Ga. pelagic sediment from a two-stage Stacey-Kramers Pb evolution model (Stacey and Kramers 1975). There is some evidence that the Stacey-Kramers model underestimates the U/Pb in the first stage of Pb evolution. If one uses stage I U/Pb ratios suggested by Tatsumoto (1978) or Allegre et al. (1988) then pelagic sediment does have the appropriate Pb isotope . (' high enough 207 Pbi2 04 Pb for a relatively low 206 Pbi2 04 Pb) to encompass all the data on Pb-Pb ratios t.e. plots.
While the Pb isotope evolution of 1.5 Ga. pelagic sediment may be modeled such that it is an ·ate mixing component for Discovery and LOMU anomaly basalts, the Pb concentration of appropn pelagic sediment is more problematic. The SEO Pb concentration of Rehkamper and Hofinann (1997) is SSppm. While this value is in the range of pelagic sediment concentrations reported in the literature, it falls 00 the high end of the range of pelagic sediment data from Plank and Langmuir (1998)  The Discovery and LOMU anomaly basalts that fall outside of the possible EHf-ENd range have EHf values that are too low for a given ENd . . However, one of the distinguishing features of ancient pelagic sediment is that it has high time integrated Lu/Hf (and therefore high EHf) for a given time integrated Sm/Nd due to the zircon effect (Patchett et al. 1984). The trends of Discovery and LOMU data indicate that they arise from such a source with a much lower EHf (and therefore lower time integrated Lu/Hf) than pelagic sediment. Thus, pelagic sediment is probably not involved in the source of Discovery and LOMU basalts.
It is possible that material from the lower continental crust rather than pelagic sediment (which arises from the upper continental crust) is involved in mixtures that produce the Discovery and LOMU anomaly basalts. Whether or not lower crust is an appropriate mixing reservoir can be assessed from . t pie composition of lower crustal material which is brought to the surface as garnet granulite the !SO 0 l 'ths The isotopic compositions of such xenoliths, which come from depths between about 25 km xeno t .
_ 45 km (Rudnick and Fountain 1990)), vary greatly from one region to another depending on their protolithic nature Goldstein 1990, Rudnick andFountain 1995). The component which is mixing with the ambient depleted mantle to produce Discovery and LOMU anomaly basalts must have 206pbf204pb significantly lower than 18. Globally, many garnet granulite xenoliths do not have such low 206pb/204Pb. However, those xenoliths from stable regions (cratons) do have low 206 Pbi2 04 Pb (e.g. Rudnick and Goldstein 1990). The Kaapvaal Craton in southern Africa is one such region where xenoliths with low 206 Pb/ 204 Pb are found (Huang et al. 1995). Of these Kaapvaal Craton xenoliths, however, only those from the northern Lesotho region have a high enough 207 Pbi2 04 Pb for a given 206pb;2 04 Pb (Huang et al. 1995) to be appropriate end-members in ADM-granulite binary or ADM-OCgranulite three component mixtures.
While the 206 Pb!2 04 Pb and 207 Pb/ 204 Pb of these Lesotho xenoliths make them appropriate mixing end-members, their 87 Sr/ 86 Sr do not. Of the 13 xenoliths from the Lesotho region reported by Rogers and Hawkesworth (1982) only one (LT2, a felsic xenolith) has high enough 87 Sr/ 86 Sr to be an appropriate mixing end-member in the source of Discovery and LOMU anomaly basalts. Xenoliths from regions in South Africa other than Lesotho do have high enough 87 Sr/ 86 Sr but not high enough 207 Pb/2 04 Pb for a given 206 Pbi2 04 Pb. Apparently, only one xenolith from the Kaapvaal Craton region has the combined isotope characteristics tha_ t make it a suitable mixing end-member in Discovery and LOMU anomaly basalts. It is possible that some combination of South African lower crustal material is involved in mixing to produce the Discovery and LOMU anomaly basalts, but because of the heterogeneity in recovered xenoliths, this is too open ended to model qualitatively. Table 3.5 lists the average isotope ratios for the Lesotho granulites, granulites from the whole region, L T2 values and the isotope composition of the most plume like samples from the Discovery and LOMU segments.
Finally, material from the subcontinental lithosphere may be considered as a possible component in mixtures that give rise to the Discovery and LOMU anomaly basalts. Walker et al.
(1 9 89) report isotope data for peridotite xenoliths from the subcontinental lithosphere (SLC) beneath southern Africa. These SLC xenoliths come from depths of about 180 km. As with the garnet granulite l ·ths from the southern African lower crust, these SLC xenoliths exhibit great variability in their xeno 1 isotopic compositions. In general, these SLC xenoliths have 207 Pb!2 04 Pb that is too low for a given 2 06pb/204pb to be a mixing end-member producing the LOMU anomaly basalts. None of the xenoliths reported by Walker et al. (1989) is an appropriate end-member for Discovery or LOMU anomaly basalts.
Neither the SLC nor the lower crust can be ruled out definitively as reservoirs involved in producing Discovery and LOMU anomaly basalts since xenoliths from both reservoirs exhibit isotopic variability and the recovered xenoliths may not be completely representative of SLC or lower crust compositions. Although there is no overwhelming evidence in favor of a lower crustal granulite source in the Discovery and LOMU anomaly basalts from xenolith compositions, there is some evidence from the trace element characteristics of the S 18-60/1 basaltic glass described by Kamenetsky et al. (2001 ).
This basalt, collected from just north of the Bouvet triple junction, represents relatively pure LOMU end-member material (see chapter 2). It has high 87 Sr/ 86 Sr, 207 Pbi2 04 Pb and low sHf and sNd, even more extreme than any Discovery or LOMU anomaly basalts. Kamenetsky et al. (2001) feel that mafic granulite from the lower crust is a reasonable source for this basalt. In their view, based on rare earth elements, CaO, Sc, and V, the source of this basalt could have a mineralogy that is consistent with the mineralogy of lower crustal mafic granulites (i.e. the source contains olivine, orthopyroxene and garnet). Thus, although a mafic xenolith with the appropriate Sr and Pb isotope characteristics has not yet been recovered, it is still possible that the lower crust is indeed the reservoir that contains the endmember involved in generating the unusual isotope characteristics of Discovery and LOMU basalts.

Conclusions
The large-scale pollution of the upper mantle beneath the southern South Atlantic is apparent from the 176 Hf/ 177 Hf data collected in this study. sHf for normal upper mantle is 20.8 while that for the ambient upper mantle is 16.6. This pollution can be modeled with a three component mixture of "unpolluted" upper mantle with a very small amount of oceanic crust and pelagic sediment (DM-OC-SED). Pollution with pelagic sediment alone cannot reproduce the observed isotope and trace element characteristics of the ambient upper mantle.
116Hf7 177 Hf of basalts from the Shona anomaly indicate that the Shona plume is heterogeneous.
Mixing of this heterogeneous plwne with the homogeneous ambient upper mantle can be modeled by a three component mixture of ambient upper mantle with recycled oceanic crust and pelagic sediment (ADM-OC-SED). According to this model, the Shona plwne contains recycled components with three different oceanic crust-sediment ratios. Each of these three recycled components (oceanic crust + sediment) contains less than 2% sediment. 176 Hf7 177 Hf data indicates that pelagic sediment is not involved in mixtures that produce Discovery or LOMU basalts. The question about which reservoirs are mixing with the ambient upper mantle to produce Discovery and LOMU anomaly basalts remains unresolved. Douglass et al. (1999) provide an isotope composition for this end-member which they call the LOMU end-member. It is a source with low 206 p;2°4Pb and 143 Nd/ 144 Nd and high 207 Pbi2 04 Pb and 87 Sr/ 86 Sr. While lower crustal or lithospheric xenoliths exhibiting this isotopic composition are elusive, there is some indication that the reservoir containing this end-member is lower crustal granulites (Kamenetsky et al. 2001 ).

SUMMARY
This study demonstrates that hafuium isotope systematics in MORB can be used to study mantle mixing processes. Despite the similarity between the Lu-Hf and Sm-Nd isotope systems, i;Hf data often reveal information about mixing processes that is less apparent in i;Nd isotope space. This is because the range of i;Hf (-3 to +24) in MORB is larger than the range of i;Nd (-4to+12). There are two causes for this larger range. First, Lu-Hf is fractionated more than Sm-Nd in mantle melting processes. This results in a larger range of Lu/Hf than Sm/Nd in the earth's reservoirs. Second, the half-life of 176 Lu is about 1 / 3 that of 146 Sm. Thus, radiogenic 176 Hf is built up more rapidly than 143 Nd.
The present study of i;Hf in MORB shows that, despite the continual convective stirring experienced by the upper mantle, material in the earth's upper mantle is not well-mixed.
Heterogeneities in the mantle are preserved on many length scales and in some cases, for long time periods. The 15,000 km long gradient of i;Hf in Atlantic NMORB clearly demonstrates that the Atlantic upper mantle is heterogeneous on a massive scale. Possible causes for this heterogeneity include ancient melting events or large-scale pollution of the Atlantic upper mantle. On the other end of the length scale spectrum, i;Hf-206 Pbi2 04 Pb indicates that the Shona plume is internally heterogeneous.
EHf data from this study, together with previously collected isotope data, are incorporated into mixing ~odels, which identify possible mixing components in the mantle. The upper mantle beneath the South Atlantic can be modeled as a three component mixture of unpolluted mantle with ancient pelagic sediment and ancient oceanic crust. The Shona plume can be modeled as a mixture of ancient oceanic crust with varying concentrations of ancient pelagic sediment. This heterogeneous plume is currently mixing with the ambient (polluted) South Atlantic mantle to produce Shona anomaly basalts.
Further, EHf data indicate that pelagic sediment is not a component of the Discovery plume.
Tue mixing models developed in this study show that despite the zircon effect, when the concentration of pelagic sediment in a mixture is very low, i::Hf cannot be used to confirm its presence.
In these cases, other isotope ratios (like 206 Pbi2 04 Pb) are better indicators of the presence of pelagic sediment. However, when the concentration of pelagic sediment in a mixture is larger than a few percent, i::Hf-i::Nd systematics can be used to confirm the presence of pelagic sediment and to estimate its concentration.
Tue main focus of the present study is identification of heterogeneities in the mantle and assessment of the components, which mix to produce these heterogeneities. Thus, the conclusions are static; they indicate what is mixing, not how it is mixing. The next step in analysis of mantle mixing processes is to relate the observed isotope data to dynamic physical mixing models. In addition, it may be possible to determine the ages of heterogeneities using Pb isotope data. ···samples are glass rims unless sample ID does not end in "g". Replicate analyses are denoted as repl.  .. Errors are reported as standard error where value listed is ±the error on the last significant figure.

TABLES
•••samples are glass rims, unless the Sample ID does not end in "g". Replicates are denoted as repl.  .. Isotope values are from Douglass et al. (1999) and this study. ·xenolith data from Huang et al. (1995) and Rogers and Hawkesworth (1982) .
.. Discovery and LOMU basalts are the most "plume-like" from each region. Data from Douglass et al. (1999) and this study.  Numbers correspond to basalts listed in table 1. Data used to select basalts for the survey come from the following sources. Jan Mayen plume anomaly: Schilling et al. 1999; Azores plume anomaly: White andSchilling 1978, Yu et al. 1997; Great Meteor plume anomaly: Rideout and Schilling 1985; Sierra Leone plume anomaly and 2°S -7°S region: Schilling et al. 1994;St. Helena and Tristan plume anomalies: Schilling et al. 1985, Hanan et al. 1986, Fontignie and Schilling 1996; Discovery and Shona plume anomalies: Douglass et al. 1999; Galapagos plume anomaly: Schilling et al. 1982, Verma and Schilling 1982, Verma et al. 1983; Salay Gomez plume anomaly: Kingsley and Schilling 1998; Afar plume anomaly: Schilling et al. 1992. Hf isotope data for Salay Gomez plume is from a separate study being conducted at URI and by Blichert-Toft.
""" ::i::: ""  (Salters and White 1998, Salters 1996, Salters and Hart 1991. Mantle end-members of Zindler and Hart (1986) and Hart et al. (1992) are labeled. EMl is represented by the most enriched Walvis Ridge basalts (Salters and Hart 1991 ). EM2 is represented by the most enriched basalts from Samoa and Societies (Patchett 1980b, Salters andHart 1991 ). HIMU is represented by St. Helena basalts (Hart et al. 1986, Salters andHart 1991 Schilling et al. 1994)). Basalts EN026 l6D-2g and EN06l 18D-lg are discussed in the text. All Hf data is corrected to JMC 475 176 Hf7 177 Hf = 0.282160 ±0.000010.     Work around Iceland is still in progress at URI. In order to include MORB near all major Atlantic plumes in the Atlantic profile, an Iceland EMORB (grey square with green cross) and NMORB (green cross) from the literature are included. For these two MORB Hf data is from Patchett 1980, corrected to the current JMC 4 75 value and Pb isotope data is from Sun et al. 1975.          Ellipses are to group the data. Inset shows data relative to mantle end-members discussed in chapter 2. C composition is based on Hanan and Graham (1996)         Since identical sets of equations can be written for each isotope ratio pair, (e.g. for 206 Pbi2 04 Pb and 87 Sr/ 86 Sr) the problem is over-determined. The values of z reported in the text were chosen so as to minimize the errors in the calculated and observed ADM isotope ratios.
The following table shows the discrepancy between the observed and the calculated ambient depleted mantle (ADM) composition which results from zoM = 96%, Zoc = 3.94% and zsrn = 0.06%.

86
:1 In light of the uncertainty regarding the input parameters to the mixing model (isotope ratios and element concentrations in the 3 components and isotope ratios observed in the ambient mantle), these errors are negligible.
The same mixing equations apply to calculations of ADM-OC-SED mixtures which reproduce the Shona anomaly basalts.

Sample EW 9309 14D-4g
This basalt is excluded from the discussion regarding the Shona anomaly segmentation 87 because its isotopic composition is anomalous in comparison to other basalts in the region. This can be · seen clearly from the isotope profiles (figure A.l). In comparison to basalts from the same dredge haul and from neighboring dredge stations, this basalt has very low 206 Pbi2 04 Pb as well as slightly high EHf and ENd and slightly low 87 Sr!8 6 Sr. The source of this basalt is unclear. It may be a very small heterogeneity in the upper mantle or in the Shona plume, but since it is an individual sample, this in not explored further. Open grey squares are all other basalts from the Shona anomaly. This basalt has more plume-like &Hf, &Nd (not shown) and 87 Sr/ 86 Sr but less plume-like 206 Pb/2 04 Pb than EW9309 14D-lg from the same dredge haul.