The origin of compositional variation in basalts recovered by submersible drill from Mount Glooscap, Mid-Atlantic Ridge at 36'25'N JAMESA. WALKER,' PATRICK J. C. RYALL,AND MARCOS ZENTILLI Department of Geology, Dalhousie University, Halifax, N.S., Canada B3H 35.5
IANL. GIBSON Department of Geology, University of Waterloo, Waterloo, Ont., Canada N2L 3GI AND
JARDADOSTAL Department of Geology, Saint Mary's University, Hal*, N.S., Canada B3H 3C3 Received December 21, 1983 Revision accepted March 23, 1984 A large peak in the crestal mountains of the Mid-Atlantic Ridge, about 16 km west of the AMAR rift valley at 36'25'N, was sampled for basalt with a submersible electric rock core drill on a comparable surficial scale as the FAMOUS area. Twenty-eight basalt samples from seven drilling stations have been analyzed for major and trace elements. Many of the samples come from flows lying under a cover of carbonate rocks and therefore could not have been sampled by a submersible or a dredge. Through comparisons with published compositional data, it appears that, unlike "FAMOUS-generated" basalts, "AMARgenerated" basalts are, on average, more evolved and are always LREE enriched. Most of the in- and between-hole compositional variation can be accounted for by low-temperature alteration, accumulation of phenocrysts, and low-pressure, relatively low-temperature fractional crystallization. A source heterogeneous in trace elements or undergoing variable degrees of partial melting is necessary to explain the remaining compositional variation. If the large peak can be interpreted as a single volcano, it may be that lavas become progressively more differentiated with time at mid-ocean ridge volcanoes as they commonly do at subduction zone volcanoes. Un pic important des montagnes sommitales de la CrEte mCdio-atlantique, situC a environ 16 km B I'ouest du fossC central de la crste AMAR, i.e., a 36"25'N, et oh le basalte a CtC kchantillonnk par forage B I'aide d'un carottier Clectrique d'un submersible, a une Cchelle de surface comparable a celle de la rkgion FAMOUS. Vingt-huit Cchantillons de basalte provenant de sept stations de forage ont CtC analysts pour les ClCments majeurs et en traces. Plusieurs des Cchantillons proviennent de coultes recouvertes de roches carbonatCes et donc n'auraient pu stre prklevks par un simple submersible ou une drague. Les comparaisons des donntes chimiques obtenues ici et celles publikes antkrieurement indiquent que contrairement aux basaltes produits ii FAMOUS, ceux d'AMAR sont, en moyenne, plus Cvoluks et toujours enrichis en klkments ICgers de terres rares. En gCnCral, la variation entre la composition des roches prklevkes dans les trous et entre les trous peut s'expliquer par une alteration a basse [email protected]
, une accumulation de phknocristaux et une faible pression et une tempkrature relativement basse de la cristallisation fractionnke. Une source-mtre de composition httkrogtne en ClCments traces, ou assujettie ii des degrks variables de fusion partielle est nCcessaire pour comprendre la variation de la composition observke. Si on considtre que ce pic important soit un volcan unique, il est possible que les laves aient CtC progressivement plus diffCrenciCes avec le temps pour les volcans de la Crste mCdio-atlantique tout comme on ]'observe souvent pour les volcans des zones de subduction. [Traduit par le journal] Can. 1. Earth Sci. 21, 934-948 (1984)
Introduction Many important hypotheses concerning the petrogenesis of mid-ocean ridge basalts have grown out of detailed studies in the French -American Mid-Ocean Undersea Study (FAMOUS) area of the Mid-Atlantic Ridge (e.g., Bryan and Moore 1977; Langmuir et al. 1977). One of the considerable attractions of the FAMOUS work was that collection by acoustically positioned submersibles permitted sampling on a very fine scale. Geochemical and petrological data, therefore, had accompanying field constraints. An underwater electric rock core drill developed at the Bedford InBtitute of Oceanography, designed originally for continental shelf work (Fowler and Kingston 1975) and briefly described below, furnishes precisely located surface sampling on roughly the same scale as a submersible. In addition, up to 6 m of core can be recovered from each site, thus establishing a small vertical as well as horizontal rock distribution. In June of 1980, the drill was successfully operated in the 'Present address: Department of Geology, Northern Illinois University, DeKalb, IL 601 15, U.S.A.
crestal mountains of the Mid-Atlantic Ridge at 36"25'N, near the FAMOUS area, but south of fracture zone B and about 16 km west of rift valley 3 and the Alvin Mid-Atlantic Ridge (AMAR) study area (Fig. 1). All of the drill stations were located on a large volcanic or volcano-tectonic peak, about 5 km X 10 km, whose long axis roughly parallels the strike of rift valley 3 (Figs. 1, 2). Although probably modified by tectonism, this peak is possibly a single volcano or volcanic complex with a presumed age of about 2 Ma. Consequently, we have referred to this volcanic feature by a single name, Mount Glooscap. By contrast, Mont de Vtnus, the largest active volcano in rift valley 2 of the FAMOUS area, is about 4 km long and 1 km wide (Bellaiche et al. 1974). The purposes of this investigation are to (1) compare -2 Ma Mount Glooscap (MG) basalts with "0-age" basalts from both rift valley 2 (FAMOUS) and rift valley 3 (AMAR) and with more contemporaneous basalts from Deep-Sea Drilling Project (DSDP) sites 332, 333, 41 1, and 412 (legs 37 and 49), presumably formed at the next ridge segment, rift valley 2; and (2) to assess the petrogenesis of MG basalts in light of petrogenetic models developed for FAMOUS and DSDP basalts.
WALKER ET AL.
CONTOURS ( m l
FIG. 1. Regional tectonic setting of Mid-Atlantic Ridge around 36"N, modified from MacDonald (1977). Rectangular box shows approximate coverage of Fig. 2.
Sampling and analytical methods The underwater electric rotary rock core drill is powered and controlled from an unanchored surface vessel and is capable of driving a 6 m long diamond drill barrel into the bottom to take a 2.5 cm diameter core. Bottom-mounted acoustic beacons and an acoustic transponder attached to the drill rig permit it to obtain a set of cores from precisely determined locations. Without the cost and limitations of alternative methods, the drill can recover cores in situations where no other common sampling tool can. More information on the design, evolution, and operation of the drill can be found in Fowler and Kingston (1975), Ryall et al. (1981), and Ryall (1982). A total of 10 cores were successfully recovered from Mount Glooscap during the summer of 1980 in water depths ranging from about 200 to 420 m. Four of the cores consist of basalt overlain by carbonate. Two cores consist of basalt alone, and one is mostly basalt except for two thin layers of carbonate (Fig. 3). Three of the cores consist of carbonate alone. The locations of the drill stations where basalt was recovered are shown in Fig. 2. From visual inspection and thin section study, four samples were selected from each core, avoiding carbonate veining and obvious alteration. Selected samples were then pulverized in a tungsten carbide shatterbox to -200 mesh. Major-element analyses were performed on fused beads with the electron microprobe and energy dispersive unit at
Dalhousie University (McKay 1981). Zr, Y, Sr, and Rb were determined by radioactive source X-ray fluorescence spectroscopy (XRF) at St. Mary's University, and the remaining trace elements were determined by instrumental neutron activation analysis (INAA) at the University of Waterloo. Precision and accuracy of the trace-element data can be found in Dupuy et al. (1981) and Gibson et al. (1982).
Petrography As can be seen from Table 1, five of the seven drill holes contain moderately phyric to aphyric plagioclase-olivinepyroxene basalts according to the classification of Hekinian et al. (1976). Within this group, basalts from station 6 have few mafic phenocrysts and are thus plagioclase enriched. The remaining two holes, 5 and 10, contain only plagioclase-rich basalts (>9% plagioclase by volume). In comparison with plagioclase-rich basalts described from the FAMOUS area (Hekinian et al. 1976; le Roex et al. 198I), those from holes 5 and 10 have significantly larger quantities of accompanying mafic phenocrysts. All of the basalts. appear to be part of massive flows and there is no lithologic variability within a given drill core. Plagioclase phenocrysts in all basalts are generally tabular and euhedral to subhedral. Many of the larger plagioclases (-3 mm) have strongly resorbed margins and contain many inclusions and are presumably xenocrysts. A few smaller grains
CAN. J. EARTH SCI.
FIG. 2. Bathymetry of Mount Glooscap and rift valley 3 at 36'25'N. Also shown are drill stations where basalt was recovered. The first number in sample identification corresponds with station number, except for basalts from stations 10 and 11. Basalts labeled 9 and 10 are from stations 10 and I I , respectively, to conform with shipboard identification. S iOp
1. Modal analyses (volume percent) TABLE
I . . . -
FIG. 3. MgO (wt.%), SiO, (wt.%), and Sr (ppm) variations of basalts within two drill stations, one showing negligible variation, the other showing variation greater than the estimated analytical error.
share the same characteristics. The majority of olivine phenocrysts are equant and euhedral, although hopper morphologies (Donaldson 1976) are also quite abundant. Clinopyroxene occurs occasionally as isolated, subhedral phenocrysts and appears in one lava as a large (-3 mm), inclusion-filled and corroded xenocryst, but is found most commonly in intergrowths with plagioclase. These intergrowths range from simple "bow ties" (Bryan 1979) to multiple, glomeroporphyritic varieties. The bow-tie intergrowths are notably
1-1 1-6 1-10 3-7" 3- 12' 5- 1 5-5 5-6 6-6 6- 10 7-4 7-7 7- 10 9- la 9-8 9- 10 9- 14' 10- l b 10-4 10-7 10- lob
3.2 1.6 4.4 0.5 2.8 9.5 10.2 10.6 5.4 7.2 6.0 3.8 5.9 9.2 12.4 17.9 12.1 0.8 2.6 4.8 2.5
1.6 0.2 0.8 0.7 4.2 2.8 2.9 5.2
0.1 0.0 0.3
95.1 98.2 94.5 98.8 92.3 86.4 82.9 81.7 94.4 92.5 91.4 93.5 90.8 80.8 80.2 73.8 82.6 99.0 96.3 93.4 96.9
0.2 2.4 2.6 2.3 8.0 4.2 2.7 2.8 tr 1.1 1.1 0.4
0.6 1.3 4.0 2.5 0.2 0.1 0.2 0.1 0.2 0.9 3.2 5.6 2.3 tr tr
NOTE:Secondary minerals were not counted. "Also has trace amounts of spinel. bLess than 1000 points were counted.
abundant, particularly as microphenocrysts, and are also present in the groundmass. Olivine-plagioclase, olivineplagioclase-clinopyroxene, and plagioclase intergrowths have also been encountered. The most widespread groundmass texture is intersertal: the
WALKER ET AL.
TABLE 2. Representative whole-rock analyses of basalts from 36'25'N 1-6
(wt.%) Si02 Ti02 A1203
FeO* MnO MgO CaO Na10 K20
P20s Total (PP~)
Zr Y Sr
Sm Eu Tb Yb
Th Hf crystalline portion is composed of intergrowths of acicular plagioclase and pyroxene. The groundmass of each of the basalts from station 6 consists almost entirely of such intergrowths, often in the form of "flowerlike" aggregates. Some of the basalts are vitrophyric to nearly vitrophyric. Olivines and very rarely oxides can also be present in the groundmass. The average volume percentage of vesicles is 2.7%. Basalts from stations 10 and 11 are notably more vesicular (averaging 4.1% by volume). All of the samples have experienced some patchy alteration. The alteration is usually centred on vesicles or glassy portions of the groundmass. The secondary minerals are brown and green clay (smectite), with the exception noted below. Radial aggregates of green and (or) brown clay are particularly prominent. A few olivines have been partially altered to brown clay around their margins and along fractures. A number of vesicles in sample 6-6 are filled with calcite. The degree of secondary mineral formation does not appear to correlate with depth or position on Mount Glooscap. Olivine-phyric basalts are common in the FAMOUS area and in the drill sites of legs 37 and 49 (Hekinian et al. 1976; Bryan and Moore 1977; Flower et al. 1977; Wood et al. 1979b). Picritic basalts or oceanites have also been found near the axis of rift valley 2 and in leg 37 drill sites (Hekinian et al. 1976; Blanchard et al. 1976). In contrast, olivine-phyric and picritic basalts have not been found on Mount Glooscap or in rift valley 3 (Stakes et al. 1981). Therefore, the most notable petrographic feature of basalts generated in rift valley 3 is the absence of olivine-phyric and picritic varieties, the former of which, at least, appears to be commonly produced in rift valley 2.
Whole-rock chemistry Two representative whole-rock analyses of basalt from each drill hole are presented in Table 2. In three of the holes, 1, 7,
and 10, most major- and trace-element variation is minor and probably within the analytical limits (Fig. 3). In drill hole 10, this negligible variation is inclusive of crystalline basalt units separated by intervening carbonate layers (Fig. 3). Compositional variation in the remaining four holes is small but notable. Major- and trace-element concentrations of the MG basalts as a group show a smaller variability than the wide compositional spectrum defined by basalts produced in rift valley 2 (Fig. 4). In an overall summary of the chemical variation in AMAR and FAMOUS basalts, Flower and Robinson (1979) showed that for a given MgO value, TiOz is more variable in basaltic glasses from the FAMOUS area, whereas CaO and A1203are higher in basaltic glasses from the AMAR region. In relation to FAMOUS basaltic glasses, AMAR basaltic glasses also have generally lower MgO contents (Flower and Robinson 1979; Shervais et al. 1983). Bryan (1979) emphasized that basaltic glasses from rift valley 3 reach distinctly higher FeO* and Ti02 contents than those from rift valley 2. Shervais et al. (1983) amended the findings of Flower and Robinson (1979) slightly by stating that basaltic glasses from the AMAR region have lower alkalis and incompatible elements and higher CaO/Al,03 and CaO/Na20 at equivalent measures of differentiation than basaltic glasses from the FAMOUS area. Figure 4 shows that when crystalline and older basalts are taken into account the comparison between basic rocks generated in rift valley 3 and those formed in rift valley 2 becomes more complex. For instance, there are important compositional differences between FAMOUS area basalts themselves (Fig. 4). Plagioclase-phyric and plagioclase-olivine-pyroxene basalts from the FAMOUS area, collectively referred to as plagioclase basalts henceforth, have higher CaO contents and lower concentrations of Ti02 and Zr than associated olivine basalts and most basaltic glasses. It could be that the compositions of the plagioclase basalts are largely influenced by accumulation of
CAN. 1. EARTH SCI. VOL. 21, 1984
I ' d
4 4 4 4
/~ aO 0
- ". s -a
& C .' $m
WALKER ET AL.
MgO content A
* A 0
6 1 3 9 10
-- 66 - 14 -
6.72 7.07 7.57 8.03 8.13
Mg # 54.5 58.0 58.1 66.4 67.4
plagioclase or a plagioclase-dominated assemblage. The A1203 contents of the plagioclase basalts, however, are not noticeably higher than those of the olivine basalts and basaltic glasses, as they should be if influenced by plagioclase accumulation (e.g., le Roex et al. 1981). In addition, many of the plagioclase basalts are aphyric (Bougault and Hekinian 1974; le Roex et al. 1981). Thus, the plagioclase basalts appear to represent magmas distinct from those producing the olivine basalts, as recognized by Bougault and Hekinian (1974) and le Roex et al. (1981) on the basis of trace-element systematics and minorelement variations in olivine phenocrysts. The plagioclase basalts are also distinct from most magmas represented by the basaltic glasses. Except for their unusually high and diverse K,O contents, which will be discussed below, MG basalts most closely resemble plagioclase basalts from the FAMOUS area in composition, with notably higher CaO contents and lower Ti02 and Zr contents than the olivine basalts and basaltic glasses at equivalent MgO concentrations. These compositional similarities, moreover, are enhanced once the effects of olivine accumulation on the MG basalts are removed (see below). The MG basalts, however, overlap with the "evolved" end of FAMOUS compositions, starting from and extending to lower MgO contents (Fig. 4). In addition, about 30% of the MG basalts are quartz normative, whereas few FAMOUS basalts are. Moreover, the A1203contents of MG basalts are generally higher than those of all FAMOUS basalts at equal MgO values. Therefore, it appears that magmas erupted in rift valley 3 as reflected in both -2 Ma crystalline basalts and "0-age" basaltic glasses (e.g., Shervais et al. 1983) have, on average, lower MgO contents and higher A1203than all FAMOUS magmas and higher contents of CaO and lower contents of Ti02,Zr, and possibly other incompatible elements than many. The -1 -3.5 Ma basalts generated at rift valley 2 do not exhibit the same compositional differences from MG basalts.
8 MgO , wt. %
FIG.5. Rare earth element contents of MG basalts. Normalizing values are from Hanson (1980). Shaded area shows range found in basalts from the FAMOUS area and DSDP hole 332. Range in (Ce/Yb)Nis 0.59-1.63. About 40% of the published analyses are LREE depleted. Sources of data: White and Bryan (1977), Langmuir et al. (1977), Blanchard et al. (1976), and O'Nions and Pankhurst (1976).
mw STATION 5
6 7 10 11
FIG.6. K20(wt.%)contents of MG basalts compared with those of olivine basalts from the FAMOUS area (dashed line), plagioclasephyric basalts and plagioclase-olivine-pyroxene from the FAMOUS area (dotted line), and Oceanographer Fracture Zone basalts (solid line). Symbols as in Fig. 4. Sources of data: Bougault and Hekinian (1974). le Roex et al. (1981), and Shibata and Fox (1975). The range of MgO and A1203contents in basalts from DSDP sites 332, 333, 41 1, and 412 is roughly that of the MG and FAMOUS basalts combined (Fig. 4; -3.5 Ma basalts from DSDP site 413 are not included in comparisons as they are geochemically atypical of FAMOUS (Wood et al. 1979b)). Moreover, although olivine basalts and "plagioclase" basalts from DSDP sites 332 and 333 (types IV and I, respectively, of Blanchard et al. 1976) generally separate in terms of CaO as do their FAMOUS counterparts, basaltic glasses from sites 332 and 333 and crystalline basalts from sites 41 1 and 412 have CaO contents equivalent to those of MG basalts at comparable MgO contents. In addition, Ti02 and to a lesser extent Zr contents of the "older" rift valley 2 basalts are often lower than both those of FAMOUS and MG basalts at equal MgO (see also Bryan and Thompson 1977; Flower and Robinson 1979). The absolute abundances of rare-earth elements in MG basalts fall, for the most part, at the high end of the wide limits defined by FAMOUS and site 332 basalts (Fig. 5). In contrast to these areas, however, where light rare earth element (LREE) -depleted basalts are common (Langmuir et al. 1977; Blanchard et al. 1976; O'Nions and Pankhurst 1976), all of the MG basalts have LREE-enriched to flat patterns ((Ce/Yb)N = 1.08- 1.65). Also, in contrast to basalts from site 332 and the FAMOUS area (e.g., Langmuir et al. 1977), the MG basalts, with one exception, show a positive correlation between LREE enrichment and heavy rare earth element (HREE) abundance. Basalts from station 1 have low rare earth contents given their degree of differentiation (Fig. 5; Table 2). They also have lower concentrations of Ti02, Zr, Hf, Sr, Y, and Th than other MG basalts at equivalent measures of differentiation (Fig. 4; Table 2). In addition, basalts from station 1 have somewhat low Ce and Nd contents and high Sm contents in light of their HREE abundances. These "anomalies," however, are within the estimated analytical limits, although the two analyses of station 1 basalts are very similar (Table 2). In summary, then, most of the compositional differences between basalts erupted in rift valley 3 and "0-age" (FAMOUS) basalts from rift valley 2 are not present in 1-3.5 Ma basalts from rift valley 2. The only consistent, compositional differences between basalts erupted in rift valley 3 and those pro-
CAN. J. EARTH SCI. VOL. 21, 1984
MG BASALTS STATION
5 6 7 10 11
. . ..A A A
K20, wt. %
K 2 0 , wt.%
FIG.7. Series of incompatible-element plots for MG basalts showing positive correlation between K20and Rb and lack of such between K20 and Rb and the immobile elements, Ti02 and Zr. Symbols as in Fig. 4. Ti02 (wt.%);Rb and Ce (ppm). (d) Within-hole variation in Ce which might reflect low-temperature alteration or, alternatively, the analytical limits of the analyses. duced in rift valley 2 is that high-MgO and LREE-depleted basalts appear to be absent in the former. This statement, of course, may only reflect a sampling bias and complements the petrographic observation that olivine-phyric and picritic basalts are absent in rift valley 3. Most recent basalts produced in the median valley of rift valley 2 (i.e., FAMOUS basalts) have notably lower contents of CaO and AI,O, and higher concentrations of some incompatible elements. Figures 4 and 5 and Table 2 indicate that basalts from the eastern flank of Mount Glooscap, that is, from stations 3, 6, and 7 (Fig. 2), are enriched in Ti02, Zr, and Hr and to a lesser extent in Sr, Th, Y, and total rare earths relative to the basalts from the core and western flank of the peak. This spatial subdivision, if taken to imply separate magmatic origins and histories for the two groups, is misleading as shown below. Instead, as already intimated, basalts from station 1 should be viewed separately in this regard. Nevertheless, a verification of these spatial characteristics and, if indeed present, a discussion of their nature will be one of the subjects of future investigations at Mount Glooscap.
Discussion Low-temperature alteration As was briefly alluded to above, the K 2 0contents of the MG basalts are unusually high and varied in comparison with those of FAMOUS basalts (Fig. 6). Although K 2 0 is present in less evolved basalts, the overall variation in K 2 0 is similir to that found in dredged basalts from the Oceanographer Fracture Zone at 35"N (Fig. 6). The high K20 contents in basalts from the Oceanographer Fracture Zone have been attributed both to a combination of extensive fractionation and sea-water alteration (Shibata and Fox 1975) and to unusual, enriched source characteristics, since incompatible-element ratios and isotopic values are also anomalous in these basalts (White and Schilling 1978). The high and varied K20 contents of the MG basalts, in general, are very likely the result of sea-water alteration, although in the case of the more evolved samples the high K20 contents may be partially ascribed to differentiation. As evidence, K20 in the MG basalts correlates positively with Rb
WALKER ET AL.
(Fig. 7), which like K20 increases significantly during lowtemperature alteration of sea-floor basalts (e.g., Thompson 1973; Robinson et al. 1977). Neither K 2 0 nor Rb, however, shows positive correlations with any of the so-called immobile incompatible elements (Fig. 7). These results illustrate the danger in attributing absolute K 2 0 and Rb abundances solely to primary igneous processes. The intensity of weathering as estimated by the absolute K,O and Rb contents shows no consistent down-hole variation or between-hole differences, such as between cores with a carbonate cover and those without. Low-temperature alteration may also explain some of the variability in the light rare earth elements evident within the same drill hole (Table 2) (Ludden and Thompson 1979), although Ce shows a good correlation with Zr (Fig. 7). Along with increases in K20, Rb, and the light rare earths, lowtemperature weathering can also cause simultaneous loss of CaO and MgO (e.g., Thompson 1973). As was pointed out above, the MgO contents of the MG basalts are characteristically lower than those of the younger and less altered FAMOUS basalts. Nevertheless, the CaO contents of the MG basalts are characteristically higher, so that low-temperature alteration probably has little to do with the differences in major elements between FAMOUS and MG basalts. Carbonate contamination can also cause important changes in CaO and MgO, but again these changes should be in the same direction (e.g., Robinson et al. 1977). The anomalously high CaO content of sample 3-7 (e.g., Fig. 4) is, however, likely the result of carbonate contamination, as carbonate veins were quite common in this section of the core. Within-site compositional variation and phenocryst accumulation After the effects of alteration are accounted for, much of the remaining within-site compositional variation in MG basalts is explicable in terms of minor accumulation of phenocryst minerals. This is illustrated in Table 3. Shown are the results of a least-squares mixing calculation (e.g., Bryan et al. 1969), which taken together with the modal data (Table 1) indicates that basalt 5-5 has accumulated small amounts of olivine, clinopyroxene, and plagioclase relative to basalt 5-1. The remaining within-site compositional variation can be explained by minor amounts of fractional crystallization as shown in Table 4. (Partition coefficients used for all accompanying traceelement modelling of fractional crystallization (e.g., Neumann et al. 1954) are given in Table 5.) Phenocryst accumulation is also likely an important part of interstation compositional variability. Basalts from station 5 are phyric with notable quantities of mafic phenocrysts (Table 1). Both Figs. 8 and 9 suggest that the compositions of these basalts, relative to MG basalts as a whole, are influenced by olivine (+ clinopyroxene) accumulation. The compositions of some or all of the basalts from stations 3, 7, and 11 may also show the consequences of mafic phenocryst accumulation, although some of these basalts are aphyric to nearly aphyric (Table 1). Similarly, the compositions of a few of the basalts may be influenced by plagioclase accumulation, although slight carbonate contamination, as was mentioned above, may also have produced the few anomalously high CaO contents (Fig. 9). Making allowances for the compositional changes attributable to mineral accumulation, a projected cotectic was estimated for the MG basalts and is shown by the dashed lines in Figs. 8 and 9. In the case of Fig. 8, the hypothesized cotectic
TABLE3. Results of least-squares modelling, 5-1 + plag + cpx + 01, 5-5 Fractions
0.01 0.01 0.02 0.97
0.14 0.24 0.62
Variables Plag An8/ Cpx En4, ~0~~~ 01 F090' 5.1
Major elements Si02 Ti02 A120, FeO MnO MgO CaO NazO Observed 5-5 50.8 1.03 Calculated 5-5 50.1 1.03
14.9 9.74 0.17 9.20 11.9 1.89 15.0 9.80 0.15 9.21 12.0 1.89
NOTES:Criteria for acceptance of a given least-squares solution are those suggested by Wright (1974). Minerals used in all least-squares calculations are from Bryan (1979). Olivines are required to be within 5 mol% of the average of the equilibrium olivine compositions for both rocks using a K, = 0.27 (Bryan 1979) and the approach of Roeder and Emslie (1970); Z R Z = 0.03. "From IB, Table 8 (Bryan 1979). hFrom 5 , Table 9 (Bryan 1979). 'From IOA, Table 10 (Bryan 1979).
is coincident with the experimentally determined trend for midocean ridge basalts of Walker et al. (1979) and the trend defined by glass compositions from rift valley 3 (analyses from Bryan 1979). In Fig. 9, thesetrends are parallel. It is important to note that basalts from station 10, which are the most porphyritic of MG basalts with significant quantities of mafic phenocrysts (Table l), lie on or close to the proposed cotectics in both figures. Indeed, it has been suggested that some phyric mid-Atlantic basalts from 22-25"N represent parental liquid compositions in relation to their quenched glass margins, which are believed to be residual liquids resulting from variable amounts of closed-system fractional crystallization (Bryan 1981; Bryan et al. 1981). Table 6, moreover, shows through a least-squares analysis that the average composition of basalts from station 10 could by almost 40% crystallization and separation of plagioclase, clinopyroxene, and olivine produce a major-element composition matching that of the most differentiated MG basalt, 6- 10. Simultaneous trace-element modelling, however, fails to match some of the trace-element contents of 6-10. Although it could be that the basalts from stations 10 and 6 have separate parent magmas with different traceelement characteristics, a simpler explanation is that the compositions of basalts from station 10 reflect accumulation of all three of their phenocryst phases and only fortuitously lie on the suggested cotectic. To illustrate, addition of 18% plagioclase + 4% olivine + 6% clinopyroxene (percentages that are within the bounds of the modal proportions in station 10 basalts) to the average composition of basalts from station 1 produces a basalt very like those from station 10 (Fig. 9). Biggar (1983) pointed out that if the compositional effects of phenocryst accumulation are not taken into account, poor leastsquares solutions will likely result, perhaps masking true petrogenetic relationships. The results in Table 6 indicate that phenocryst accumulation in basalts may also lead to acceptable least-squares solutions but erroneous petrogenetic relationships. Trace-element modelling is necessary to test proposed fractionation paths for phyric samples. Fractional crystallization, partial melting, and source heterogeneity Because the basalts from station 10, whether they represent
CAN. J. EARTH SCI. VOL. 21. 1984
TABLE 4. Results of least-squares modelling, (6-2 + 6-17),,. + 6-10 + plag + cpx + 01 Fractions
0.01 0.01 0.01 0.98
0.31 0.40 0.29
Variables Plag An," Cpx En4, Woab 01 F08; Residual liquid, 6- 10 Major elements
Observed(6-2+6-17),,. Calculated (6-2 + 6-17),,,
Si02 TiOz A1203
Zr Y Sr
0.09 0.37 0.61
98 27 118
100 29 113
NOTE:E R 2 = 0.01. "From IB, Table 8 (Bryan 1979). bFrom 5, Table 9 (Bryan 1979). "From 3A, Table 10 (Bryan 1979). TABLE5. Mineral-liquid partition coefficients used in fractional crystallization modelling
Ce Nd Sm Eu Tb Yb Hf Zr Y Sr
0.12" 0.10" 0.08" 0.232' 0.08"* 0.037'* 0.02d 0.01' 0. 18R 1.80h
0.09Sb 0.21h 0.26b O.3lb O.4Ob* O.3Ob* O.3Od 0.20f 0.70 0. 12h
0.009' 0.01' 0.01 1' 0.01' 0.013'* 0.023'* 0.00 0.01' 0.13' 0.01'
*Interpolated. "Drake and Weill (1975), T = 1235°C. bGrutzeck er al. (1974). 'Schnetzler and Philpotts (1970). dDostal el a[. (1983). 'Pearce and Nony (1979). 'McCallum and Charette (1978). *le Roex et al. (1981). 'McKay and Weill (1977). hPhilpotts and Schnetzler (1970).
liquid compositions or not, are among the "least evolved" MG basalts, the successful least-squares solution in Table 6 suggests that the entire major-element variation in MG basalts, along the proposed cotectic, is explicable by simple, closedsystem fractional crystallization. A satisfactory least-squares solution is also obtained with a relatively unevolved, nearly aphyric basalt, 10-7, as a parental composition (Table 7). Sample 10-7 also lies close to the estimated cotectics of Figs. 8 and 9. The trace-element variation between 10-7 and 6-10 is also successfully modelled (Table 7). Therefore, it appears that fractional crystallization plus phenocryst accumulation can explain much of the compositional variation in MG basalts. The overall increase in Ti02 with fractional crystal-
FIG.8. CMAS projection of major-element compositions of MG basalts from diopside into CA-M-S (wt.%) after O'Hara (1968). Most symbols as in Fig. 4. Solid line is suggested 1 atm (101.3 kPa) cotectic for mid-ocean ridge basalts from Walker et al. (1979). Dashed line is suggested cotectic for MG basalts; in this case it is coincident with the experimentally determined cotectic. The qualitative compositional effects of plagioclase and olivine accumulation are shown. lization is shown in Fig. 10. The increases in TiOz and FeO*/MgO are 0.47 wt.% and 0.78, respectively. The change in FeO*/MgO relative to TiOz is therefore in the ratio of 1.66: 1, supportive of the least-squares results of Bryan (1979), which suggest that fractional crystallization of mid-ocean ridge basalts always leads to a greater change in FeO*/MgO than in Ti02. In the calculation shown in Table 7, almost equivalent proportions of plagioclase and clinopyroxene are removed with only an accessory amount of olivine. The necessity of significant amounts of clinopyroxene fractionation may indicate that differentiation occurred at low pressure and relatively low temperature, as clinopyroxene generally crystallizes after olivine
WALKER ET AL.
TABLE 6. Results of least-squares modelling, 9,,,* + 6-10
+ plag + cpx + ol
0.20 0.15 0.02 0.62
0.53 0.41 0.06
Variables Plag An,," Cpx Ens3 Wo,,b 01 FowC Residual liquid, 9.". Maior elements
Observed9,,. Calculated 9.".
Ced ~d~ Smd Eud ybd Hfd
zr' Yf Sr'
Calculated. 6- 10
Observed, 6- 10
0.10 0.14 0.15 0.25 0.21 0.14 0.13 0.09 0.39 1.00
19.36 12.81 3.92 1.22 0.87 3.16 2.23 97 27 116
16.85' 10.94' 3.68' 1.15' 0.89' 3.13' 2.30' 100 29 113
NOTE:ER' = 0.04. "From I A, Table 8 (Bryan 1979). 'From 4A, Table 9 (Bryan 1979). "From IOA, Table 10 (Bryan 1979). "verage of data for 9-8 and 9-14. 'Average of data for 6-6 and 6-17. 'Average of data for all basalts from station 9.
FIG.9. Enlarged CaO vs. MgO for MG basalts. Symbols, most lines, etc. as in Figs. 4 and 8. Dotted line shows trend for basaltic glasses from rift valley 3 (Bryan 1979); X shows result of addition of 4% olivine 6% clinopyroxene to average of 18% plagioclase station I basalts.
and plagioclase in mid-ocean ridge basalts at 1 atm (101.3 kPa) (e.g., Walker et al. 1979). Alternatively, differentiation at higher pressure (i.e., 5- 10 kbar (500- 1000 MPa)) may have occurred where clinopyroxene replaces olivine and plagioclase on the liquidus (Kushiro and Thompson 1972; Bender et al. 1978; Fisk et al. 1980). These two possibilities for the physical
conditions prevailing during crystallization are also indicated by the relatively sodic plagioclase required to balance the Na20 contents in the least-squares calculation (Table 7) (Green 1969; Bender et al. 1978). In a CMAS ternary projection from plagioclase into olivine-silica-diopside (after Walker et al. 1979), all of the MG basalts fall on or to the concave side of the experimentally established 1 atm (101.3 kPa) cotectic for mid-ocean ridge basalts (Fig. 11). This is indeed true of many pyroxene-phyric mid-ocean ridge basalts, as shown by Bryan (1983). If differentiation occurred at higher pressure the MG basalts should define a relict cotectic to the convex side of the 1 atm (101.3 kPa) cotectic (e.g., Walker 1981). In addition, many mid-ocean ridge basalts that display the compositional signature of higher pressure clinopyroxene fractionation, in contrast to MG basalts, do not contain clinopyroxene phenocrysts or xenocrysts because clinopyroxene is either efficiently removed or resorbed before surface eruption (e.g., Thompson et al. 1980; Fisk et al. 1982). Therefore, it is probable that MG basalts have experienced fractional crystallization at low pressures and relatively low temperatures. Most of the MG basalts, including all those believed to define a cotectic, fall on the concave side of the 1 atm (101.3 kPa) cotectic in Fig. 11. This could indicate an important role for magma mixing in the production of MG basalts (e.g., Walker et al. 1979). Basalts from rift valley 3 are said to have mineralogical evidence for magma mixing (Stakes et al. 1981). An important role for magma mixing as envisioned by Walker et al. (1979) and O'Hara (1977), however,
CAN. I. EARTH SCI. VOL. 21. 1984
TABLE7. Results of least-squares modelling, 10-7 -+ 6- 10 + plag + cpx + ol Fractions
0.20 0.19 0.01 0.59
0.49 0.48 0.04
Variables Plag An,,," Cpx Ens3Wo,,b 01 Fad' Residual liquid, 6-10 Major elements
Observed10-7 51.2 Calculated 10-7 50.8
Ced Nd" Smd Eud Tbd Ybd
Zr Y Sr
Calculated, 6- 10
0.11 0.15 0.16 0.26 0.23 0.16 0.15 0.10 0.43 0.94
17.02 10.51 3.40 1.21 0.88 3.25 2.34 95 27 113
16.85' 10.94' 3.68' 1.15' 0.89' 3.13' 2.30' 100 29 113
NOTE:Z R 2 = 0.01. "From IA, Table 8 (Bryan 1979). 'From 4A, Table 9 (Bryan 1979). 'From IOA, Table 10 (Bryan 1979). "Data for 10-4 used for 10-7. 'Average of data for 6-6 and 6-17.
MgO, wt. %
FIG. 10. Effects of fractional crystallization, as in Table 7 (10-7, 6- 10). and olivine accumulation on TiOz (wt.%) and MgO contents of MG basalts. Note position of basalts from station 1. Olivine
implies the workings of a large, steady-state magma chamber whose existence beneath this section of the slow-spreading Mid-Atlantic Ridge has been seriously questioned on the basis of various geophysical data (e.g., Sleep 1975; Fowler 1976). Therefore, a large role for magma mixing seems unlikely. The suggested 1 atm (101.3 kPa) cotectic for mid-ocean ridge basalts shown in Fig. I1 is based on experiments on unusually alkalic Oceanographer Fracture Zone (OFZ) basalts (Shibata and Fox 1975; Walker et al. 1979). The preferred explanation for the presence of MG basalts to the concave side of this
FIG. 11. Major-element compositions of MG basalts projected from plagioclase into olivine-silica-diopside (mol%) after Walker et al. (1979). Symbols as in Fig. 4. Lines as in Fig. 8.
cotectic is the lower average Na,O contents of the MG basalts relative to OFZ basalts (Hoover and Presnall 1981; Biggar 1983). This explanation is also indicated by the slightly reduced plagioclase-phase volume of the liquid trend for MG basalts relative to the OFZ 1 atm (101.3 kPa) cotectic in a
TABLE8. Results of least-squares modelling, 10-7 4 I.,.* + plag + cpx + ol Fractions
0.15 0.15 0.01 0.69
0.50 0.48 0.03
Plag An,,;' Cpx Ens3 Wo,,b 01 Fowc Residual liquid, I.,, Major elements
Observed 10-7 51.2 Calculated 10-7 51.5
Ced Ndd Smd E U ~ Tbd ybd
Hp Zr Y Sr
Observed, 1 ,,
0.11 0.15 0.17 0.27 0.23 0.16 0.15 0.10 0.43 0.96
14.79 9.19 2.99 1.08 0.78 2.85 2.05 82 25 111
11.96' 7.96' 3.58' 0.95' 0.71' 2.63' 1.68' 70' 22f 102'
NOTE:ZR' = 0.03. "From IA, Table 8 (Bryan 1979). %om 4A, Table 9 (Bryan 1979). "From IOA, Table 10 (Bryan 1979). dData for 10-4 used for 10-7. 'Average of data for 1-6 and 1 - I I . 'Average of data for all basalts from station 1 .
FIG. 12. Major-element compositions of MG basalts projected from diopside into olivine-plagioclase-silica (mol%) after Walker et al. (1979). Symbols as in Fig. 4. Lines as in Fig. 8.
diopside projection into olivine-plagioclase-silica (Fig. 12) (Biggar 1983). As discussed in the previous section, basalts that lie to the right of the fractionation trend in Fig. 10 show the compositional effects of olivine accumulation. The vectors in Fig. 10 point out the significant changes in MgO produced by as little as 2.5% added olivine (Fogs)by volume, which is roughly
the modal percentage of olivine in these rocks (Table 1). Basalts from station 1, on the other hand, fall below or to the left of the fractionation trend in Fig. 10. This could be indicative of plagioclase accumulation, but station 1 basalts show no modal enrichment in plagioclase (Table 1). In addition, station 1 basalts lie along the proposed cotectics (Figs. 8, 9). Table 8 shows that, except for Ti02,the major-element compositions of station 1 basalts are successfully modelled by fractional crystallization of plagioclase, clinopyroxene, and minor olivine, with sample 10-7 again as a parental composition. TiOz and the other trace elements do not substantiate this fractionation step. All of the predicted trace-element contents except Sm are from 9 to 25% too high. The predicted Sm content, in contrast, is about 17% too low. Sm, like the other trace elements used in the modelling, should behave as an incompatible element during batch melting or fractional fusion of a "normal" mantle assemblage (e.g., Sun et al. 1979; Wood et al. 1979a). Thus, the results of Table 8 may not only exclude a fractional crystallization relationship between station 1 and other MG basalts, but also a relationship based on partial melting of a mantle source homogeneous in trace elements, since Sm varies in the opposite sense from the other trace elements. However, as discussed above, the "anomalously" high Sm or, alternatively, low Ce and Nd, although present in both station I basalts analyzed, could be the result of analytical error since neighboring rare earths are behaving unsystematically. If so, a kinship founded on partial melting of a common source cannot be excluded. The lower incompatible-element contents of station
CAN. 1. EARTH SCI. VOL. 21, 1984
1 basalts relative to other MG basalts at the same MgO level would indicate relatively larger degrees of melting of such a common source. Similar petrogenetic conclusions have been reached by many of the geochemical studies of the "FAMOUS" or rift valley 2 ridge segment, namely, that many erupted basalts, even those closely related in time and space, reflect separate parent magma batches with different trace-element characteristics (Bougault and Hekinian 1974; Blanchard et al. 1976; Flower et al. 1977; Langmuir et al. 1977; Byerly and Wright 1978; le Roex et al. 1981; Shervais et al. 1983). This picture of magma genesis may be relevant to slow-spreading, submarine ridge segments in general, as implied by Nisbet and Fowler (1978). Volcano evolution
It may be possible to interpret Mount Glooscap as a single volcano or volcanic complex. If so, comparisons with Mont de VCnus are enlightening. Mont de VCnus, besides being the largest active volcano in the FAMOUS area, is one of its youngest volcanic features (Bellaiche et al. 1974). The lava flow stratigraphy at Mont de Vknus is picritic basalts overlain by olivine basalts (Arcyana 1977). These lavas from Mont de VCnus have the thinnest palagonite rims of FAMOUS lavas, suggesting that they are the youngest flows in the rift valley (Hekinian and Hoffert 1975). Lavas on Mount Glooscap are pyroxene bearing and, in general, more evolved than FAMOUS lavas. Combining the results from Mount Glooscap and Mont de VCnus, therefore, suggests that lavas forming central volcanoes at spreading centres will become increasingly differentiated with time. A necessary rider to this suggestion is that successive lavas may be unrelated to one another by fractional crystallization or phenocryst accumulation as brought out at the end of the preceding subsection. In DSDP hole 332B, the leg 37 drill site with the deepest basement penetration, olivine-phyric and picritic basalts are more common near the bottom, whereas plagioclase-phyric and aphyric basalts are more abundant near the top (Byerly and Wright 1978; Flower and Robinson 1979). This overall lithologic and chemical stratigraphy is grossly consistent with the suggested temporal evolution of lavas at central volcanoes, if indeed the 583 m of 332B core represents a single volcano. Again it appears that this 583 m of basalt constitutes many magma batches, unrelated by fractional crystallization or phenocryst accumulation (e. g . , Byerly and Wright 1978). The trend toward more evolved compositions with growth of a central volcano may reflect the increasing amount of differentiation required for magma ascent in areas of thicker crust in accordance with the volcano growth models of Eaton and Murata (1960) and Ben-Avraham and Nur (1980). This argument has been used to explain a specific example of the commonly observed trend of subduction zone volcanoes toward eruption of more differentiated compositions with time (e-g., Rose et at. 1977). A similar theme was implied by Nisbet and Fowler (1978) in explaining the striking petrological and geochemical zonation of basaIts acmss the median valley of the FAMOUS area where contemporaneous flank lavas, which presumably ascend through thicker crust, are relatively evolved compared with axial lavas. However, application of this simpIe hydrostatic model to spreading cenrre: volcanoes may conflict with the predicted density increase of mid-ocean ridge basalrs with low-pressure cotectic fractionation (Stolper and Walker 1980; Sparks et 01. 1980). As an alternative. the trend toward evolved compositions may only be a manifestation of the
increased crustal residence time of magmas during ascent. The predicted evolution of spreading centre volcanoes toward more differentiated lavas with time is not envisaged as a straight-line process but as a circuitous one with probable reversals and interruptions from both magmatic processes, such as magma mixing, and tectonic processes, such as shifting of the axis of volcanism (MacDonald 1977). A circuitous path is also most compatible with the lithologic and chemical stratigraphy of nearby DSDP drill holes. Therefore, the existing lack of olivine-rich (pyroxene-poor), high-MgO, and possible LREE-depleted basalts in samples from rift valley 3 may only be the result of the current volcanic quiescence in the median valley of rift valley 3 (Stakes et al. 1981) and their lack of near-surface exposure on Mount Glooscap.
Conclusions Approximately 2 Ma crystalline basalts from a large peak, Mount Glooscap, in the crestal mountains of the Mid-Atlantic Ridge at 36"25'N, south of fracture zone B, have a narrower petrologic and compositional range than 0-3.5 Ma basalts generated in rift valley 2, north of fracture zone B. Olivinephyric, picritic, high-MgO, and LREE-depleted basalts are notably absent from Mount Glooscap. MG basalts are quite distinct from many "0-age" rift valley 2 (FAMOUS) basalts with, on average, lower contents of MgO. In addition, MG basalts have higher concentrations of Al2O3and CaO and lower contents of TiOz and Zr than many FAMOUS basalts at equivalent MgO values. Basalts from station 1 have lower contents of incompatible elements than other MG basalts at a given measure of differentiation and possibly anomalously high Sm concentrations. All of the MG basalts show the petrographic evidence of low-temperature alteration, including the presence of common smectites and uncommon calcite. K,O and Rb contents of MG basalts are quite variable, fail to correlate positively with other incompatible elements, including Zr and Ti02, and thus have probably been largely affected by low-temperature alteration. All of the within-hole compositional variation is explicable in terms of olivine ? plagioclase 2 clinopyroxene accumulation and minor amounts of fractional crystallization. Much of the compositional variability between drill sites can also be explained by accumulation of observed phenocryst phases, particularly olivine. Approximately 40% crystallization and separation of a 50% plagioclase + 50% clinopyroxene (with minor olivine) satisfactorily accounts for the total compositional range in TiOz and MgO. This fractionation likely took place close to the surface at relatively low temperatures. The incompatible-element contents of basalts from station 1 require a larger degree of melting of the mantle source of the MG basalts or a source with a different trace-element signature. Through comparisons of the suggested lava stratigraphy of Mont de Vtnus in rift valley 2 and the analyses of basalts from Mount Glooscap, it is tentatively proposed that central volcanoes along mid-ocean ridges may commonly erupt increasingly differentiated lavas with growth of the central cone, as is the common trend at subduction zone volcanoes. This suggestion is consistent with the volcano growth models of Eaton and Murata (1960) and Ben-Avraham and Nur (1980). If the proposed temporal evolution of lava compositions is applicable to Mount Glooscap, then the observed compositional differences between basalts generated at rift valley 2 and those generated at rift valley 3 are only a consequence of the current volcanic
WALKER ET AL.
quiescence in rift valley 3 and sampling bias on Mount Glooscap.
During the completion of this project, the senior author held a Killam postdoctoral fellowship at Dalhousie University. Additional financial support was provided by Dalhousie University (Research Development Fund grant 601089 (Walker)), and by the Natural Sciences and Engineering Research Council of Canada (NSERC) (strategic grants G0343, 0 3 4 5 , and GO742 (Ryall) and operating grant A-9036 (Zentilli)). Many thanks to: Capt. F. Mauger, the officers, and crew of C S S Hudson; Mr. George Fowler, designer of the drill; Roger Cassivi and Bill Whiteway for their invaluable assistance in obtaining the cores; P. Webster and J. M . Peckenham for help with sample preparation and analysis; R. M. McKay for the microprobe analyses; D . Van d e Rijt for typing the manuscript; and D. Meggison and S. Walker for drafting the figures. Careful reviews by W . B . Bryan and an anonymous reviewer greatly improved the paper. ARCYANA.1977. Rocks collected by bathyscaph and diving saucer in the FAMOUS area of the Mid-Atlantic Rift valley: petrological diversity and structural setting. Deep-Sea Research, 24, pp. 565-589. BELLAICHE, G., CHEMINEE, J.-L., FRANCHETEAU, J., HEKINIAN, R., H. D., and BALLARD, R. D. 1974. Inner LEPICHON,X., NEEDHAM, floor of the Rift Valley: first submersible study. Nature, 250, pp. 558-560. BEN-AVRAHAM, Z., and NUR, A. 1980. The elevation of volcanoes and their edifice heights at subduction zones. Journal of Geophysical Research, 85, pp. 4325-4335. BENDER,J. F., HODGES,F. N., and BENCE,A. E. 1978. Petrogenesis of basalts from the project FAMOUS area: experimental study from 0 to 15 Kbars. Earth and Planetary Science Letters, 41, pp. 277-302. BIGGAR,G. M. 1983. Crystallization of plagioclase, augite and olivine in synthetic systems and in tholeiites. Mineralogical Magazine, 47, pp. 161-176. BLANCHARD, D. P., RHODES,J . M., DUNGAN, M. A., RODGERS, K. V., DONALDSON, C. H., BRANNON, J. C., JACOBS,J. W., and GIBSON,E. K. 1976. The chemistry and petrology of basalts from Leg 37 of the Deep-Sea Drilling Project. Journal of Geophysical Research, 81, pp. 4231 -4246. BOUGAULT, H., and HEKINIAN, R. 1974. Rift valley in the Atlantic Ocean near 36"501N: petrology and geochemistry of basaltic rocks. Earth and Planetary Science Letters, 24, pp. 249-261. BRYAN,W. B. 1979. Regional variation and petrogenesis of basalt glasses from the FAMOUS area, Mid-Atlantic Ridge. Journal of Petrology, 20, pp. 293-325. 1981. The role and limitations of simple fractionation and mixing processes in the genesis of basalts from the Mid-Atlantic Ridge. In The generation of the oceanic lithosphere. Abstract Volume. American Geophysical Union Chapman Conference, Warrenton, VA. 1983. Systematics of modal phenocryst assemblages in submarine basalts: petrologic implications. Contributions to Mineralogy and Petrology, 83, pp. 62-74. BRYAN,W. B., and MOORE,J. G. 1977. Compositional variations of young basalts in the Mid-Atlantic Ridge rift valley near lat. 36"49'N. Geological Society of America Bulletin, 88, pp. 556-570. BRYAN,W. B., and THOMPSON, G. 1977. Basalts from DSDP Leg 37 and the FAMOUS area: compositional and petrogenetic comparisons. Canadian Journal of Earth Sciences, 14, pp. 875-885. BRYAN,W. B., FINGER,L. W., and CHAYES,F. 1969. Estimating proportions in petrogenetic mixing equations by least squares approximation. Science, 163, pp. 926-927.
BRYAN,W. B., THOMPSON, G., and LUDDEN, J. N. 1981. Compositional variations in normal MORB from 22"-25"N: Mid-Atlantic Ridge and Kane Fracture Zone. Journal of Geophysical Research, 86, pp. 11 815 - 11 836. BYERLY,G., and WRIGHT,T. L. 1978. Origin of major-element chemical trends in DSDP Leg 37 basalts, Mid-Atlantic Ridge. Journal of Volcanology and Geothermal Research, 3, pp. 229-279. DONALDSON, C. H. 1976. An experimental investigation of olivine morphology. Contributions to Mineralogy and Petrology, 57, pp. 187-213. M., DOSTAL,J., DUPUY,C., CARRON, J. P., LE GUENDE KERNEIZON, and MAURY,R. C. 1983. Partition coefficients of trace elements: application to volcanic rocks of St. Vincent, West Indies. Geochimica et Cosmochimica Acta, 47, pp. 525-533. DRAKE,M. J., and WEILL,D. F. 1975. Partition of Sr, Ba, Ca, Y, Eu2+,Eu3+, and other REE between plagioclase feldspar and magmatic liquid: an experimental study. Geochimica et Cosmochimica Acta, 39, pp. 689-712. DUPUY,C., DOSTAL,J., GIROD,M., and LIOTARD, M. 1981. Origin of volcanic rocks from Stromboli (Italy). Journal of Volcanology and Geothermal Research, 10, pp. 49-65. EATON,J. P., and MURATA,K. J. 1960. How volcanoes grow. Science, 132, pp. 925-938. H. 1980. An experiFISK,M. R., SCHILLING, J.-G., and SICURDSSON, mental investigation of Iceland Reykjanes Ridge tholeiites. Contributions to Mineralogy and Petrology, 74, pp. 361-374. FISK, M. R., BENCE,A. E., and SCHILLING, J.-G. 1982. Major element chemistry of Galapagos Rift Zone magmas and their phenocrysts. Earth and Planetary Science Letters, 61, pp. 171- 189. FLOWER,M. F. J., and ROBINSON, P. T. 1979. Evolution of the "FAMOUS" ocean ridge segment: evidence from submarine and deep sea drilling investigations. In Deep drilling results in the Atlantic Ocean: ocean crust. Edited by M. Talwani, C. G. Harrison, and D. E. Hayes. American Geophysical Union, Washington, DC, pp. 314-330. FLOWER,M. F. J., ROBINSON,P. T., SCHMINCKE, H.-U., and OHNMACHT, W. 1977. Magma fractionation systems beneath the Mid-Atlantic Ridge at 36-37ON. Contributions to Mineralogy and Petrology, 64, pp. 167- 195. FOWLER,C. M. R. 1976. Crustal structure of the Mid-Atlantic Ridge crest at 37"N. Geophysical Journal of the Royal Astronomical Society, 47, pp. 459-491. FOWLER,G. A., and KINGSTON,P. 1975. An underwater drill for continental shelf exploration. Journal of the Society of Underwater Technology, 1, pp. 18-22. R. J., EMMERMAN, R., SCHMINCKE, GIBSON,I. L., KIRKPATRICK, H.-U., PRITCHARD,G., OAKLEY,P. J., THORPE,R. S., and MARRINER, G. F. 1982. The trace element composition of the lavas and dikes from a 3-km vertical section through the lava pile of eastern Iceland. Journal of Geophysical Research, 87, pp. 6532-6546. GREEN, T. H. 1969. High-pressure experimental studies in the origin of anotthosite. Canadian Journal of Earth Sciences, 6, pp. 427-440. GRUTZECK, M. W., KRIDELBAUGH, S. J., and WEILL,D. F. 1974. The distribution of Sr and the REE between diopside and silicate liquid. Geophysical Research Letters, 1, pp. 273-275. HEKINIAN,R., and HOFFERT,M. 1975. Rate of palagonitization and manganese coating of basaltic rocks from the rift valley in the Atlantic Ocean near 35"50fN. Marine Geology, 19, pp. 91 - 109. HEKINIAN,R., MOORE,J. G., and BRYAN,W. B. 1976. Volcanic rocks and processes of the Mid-Atlantic Ridge Rift Valley near 36"49'N. Contributions to Mineralogy and Petrology, 58, pp. 83- 110. HOOVER,J. D., and PRESNALL, D. C. 1981. Partial melting of simplified lherzolite in the system CaO-Mg0-A1203-SiOz-Na20 from 1 atm to 20 Kb. In The generation of the oceanic lithosphere. Abstract Volume. American Geophysical Union Chapman Conference, Warrenton, VA.
CAN. I. EARTH SCI.
KUSHIRO,I., and THOMPSON, R. N. 1972. Origin of some abyssal tholeiites from the Mid-Atlantic Ridge. Carnegie Institution of Washington, Yearbook, 71, pp. 403-406. J. F., BENCE,A. E., HANSON, G. N., and LANGMUIR, C. H., BENDER, TAYLOR,S. R. 1977. Petrogenesis of basalts from the FAMOUS area, Mid-Atlantic Ridge. Earth and Planetary Science Letters, 36, pp. 133-156. LE ROEX,A. P., ERLANK, A. J., and NEEDHAM,H. D. 1981. Geochemical and mineralogical evidence for the occurrence of at least three distinct magma types in the "FAMOUS" region. Contributions to Mineralogy and Petrology, 77, pp. 24-37. LUDDEN,J. N., and THOMPSON,G. 1979. An evaluation of the behaviour of the rare-earths during the weathering of sea-floor basalts. Earth and Planetary Science Letters, 43, pp. 85-92. MACDONALD, K. C. 1977. Near-bottom magnetic anomalies, asymmetric spreading, oblique spreading and tectonics of the MidAtlantic Ridge near 37"N. Geological Society of America Bulletin, 88, pp. 541-555. MCCALLUM, I. S., and CHARE'ITE,M. P. 1978. Zr and Nb partition coefficients: implications for the genesis of mare basalts, KREEP and sea floor basalts. Geochimica et Cosmochimica Acta, 42, pp. 859-869. MCKAY, G. A,, and WEILL, D. F. 1977. KREEP petrogenesis revisited. Proceedings, 8th Lunar Science Conference, Houston, TX, pp. 2339-2355. MCKAY,R. M. 1981. Whole rock analysis using the electron microprobe. B.Sc. (Hons) thesis, Dalhousie University, Halifax, N.S. C. J. 1954. Trace eleNEUMANN, H., MEAD, J., and VITALIANO, ment variation during fractional crystallization as calculated from the distribution law. Geochimica et Cosmochimica Acta, 6, pp. 90-99. NISBET,E. G., and FOWLER,C. M. R. 1978. The Mid-Atlantic Ridge at 37" and 45"N: some geophysical and petrological constraints. Geophysical Journal of the Royal Astronomical Society, 54, pp. 631 -660. O'HARA, M. J. 1968. The bearing of phase equilibria studies in synthetic and natural systems on the origin and evolution of basic and ultrabasic rocks. Earth Science Reviews, 4, pp. 69-133. 1977. Geochemical evolution during fractional crystallization of a periodically refilled magma chamber. Nature, 266, pp. 503-507. O'NIONS,R. K., and PANKHURST, R. 1. 1976. Sr isotopic and rareearth element geochemistry of DSDP Leg 37 basalts. Earth and Planetary Science Letters, 31, pp 255 -261. PEARCE,J. A , , and NORRY,M. J. 1979. Petrogenetic implications of Ti, Zr, Y and Nb variations in volcanic rocks. Contributions to Mineralogy and Petrology, 69, pp. 33 -47. P ~ r ~ ~ o rJ.r A,, s , and SCHNETZLER, C. C. 1970. Phenocryst-matrix partition coefficients for K, Rb, Sr and Ba, with applications to anorthosite and basalt genesis. Geochimica et Cosmochimica Acta, 34, pp. 307-322. ROBINSON,P. T., FLOWER,M. F. J., SCHMINKE,H.-U., and OHNMACHT,W. 1977. Low temperature alteration of oceanic basalts, DSDP Leg 37. In Initial reports of the Deep Sea Drilling Project. Edited by F. Aumento, W. G. Melson, et al. United States Government Printing Office, Washington, DC, Vol. 37, pp. 775-793. ROEDER,P. L., and EMSLIE,E. F. 1970. Olivine-liquid equilibrium. Contributions to Mineralogy and Petrology, 29, pp. 275-289. ROSE,W. I., GRANT,N. K., HAHN,G. A., LANCE,I. M., POWELL, J. L., EASTER,J., and DEGRAAF,J. M. 1977. The evolution of Santa Maria volcano, Guatemala. Journal of Geology, 85, pp. 63-88. RYALL,P. J. C. 1982. Sampling MORB's using an electric rock core drill (abstract). Eos, 63, p. 473. RYALL,P. J. C., FOWLER,G. A,, and MANCHESTER, K. A. 1981. An electric rock core drill for deep ocean use. Proceedings, 13th Annual Offshore Technology Conference, Houston, TX. SCHNETZLER, C. C., and PHILPO'ITS,J. A. 1970. Partition coefficients
of rare-earth elements and barium between igneous matrix material and rock-forming mineral phenocrysts-11. Geochimica et Cosmochimica Acta, 34, pp. 331 -340. SHERVAIS, J. W., STAKES,D. S . , and HOPSON,C. A. 1983. Petrogenesis of MORB: evidence from MAR 37ON (AMAR) (abstract). Eos, 64, p. 332. SHIBATA, T., and FOX,P. J . 1975. Fractionation of abyssal tholeiites: samples from the Oceanographer Fracture Zone (35"N, 35"W). Earth and Planetary Science Letters, 27, pp. 62-72. SLEEP,N. H. 1975. Formation of oceanic crust: some thermal constraints. Journal of Geophysical Research, 80, pp. 4037-4042. SPARKS,R. S. J., MEYER,P., and SIGURDSSON, H. 1980. Density variation amongst Mid-Ocean Ridge basalts: implications for magma mixing and the scarcity of primitive lavas. Earth and Planetary Science Letters, 46, pp. 419-430. STAKES,D., SHERVAIS,J. W., and HOPSON,C. A. 1981. Compositional variations of basaltic liquids from the AMAR and FAMOUS valleys, MAR: evidence of magma chamber processes. In The generation of the oceanic lithosphere. Abstract Volume. American Geophysical Union Chapman Conference, Warrenton, VA. STOLPER,E., and WALKER, D. 1980. Melt density and average composition of basalt. Contributions to Mineralogy and Petrology, 74, pp. 7-12. SUN, S.-S., NESBIT,R. W., and SHARASKIN, A. Y. 1979. Geochemical characteristics of mid-ocean ridge basalts. Earth and Planetary Science Letters, 44, pp. 119- 138. TARNEY,J., SAUNDERS,A. D., WEAVER,S. D., DONNELLAN, N. C. B., and HENDRY,G. L. 1979. Minor-element geochemistry of basalts from Leg 49, North Atlantic Ocean. In Initial reports of the Deep Sea Drilling Project. Edited by B. P. Luyendyk, J. R. Cann, et al. United States Government Printing Office, Washington, DC, Vol. 49, pp. 657-691. THOMPSON, G. 1973. A geochemical study of the low-temperature interaction of sea-water and oceanic igneous rocks. Eos, 54, pp. 1015-1019. THOMPSON,G., BRYAN,W. B., and MELSON,W. G. 1980. Geological and geophysical investigation of the Mid-Cayman Rise spreading center: geochemical variation and petrogenesis of basalt glasses. Journal of Geology, 88, pp. 41-55. WALKER,D., SHIBATA,T., and DELONG,S. E. 1979. Abyssal tholeiites from the Oceanographer Fracture Zone. 11. Phase equilibria and mixing. Contributions to Mineralogy and Petrology, 70, pp. 111-125. WALKER,J. A. 1981. Petrogenesis of lavas from cinder cone fields from behind the volcanic front of Central America. Journal of Geology, 89, pp. 721 -739. WHITE,W. M., and BRYAN,W. B. 1977. Sr-isotope, K, Rb, Cs, Sr, Ba and rare-earth geochemistry of basalts from the FAMOUS area. Geological Society of America Bulletin, 88, pp. 571 -576. WHITE,W. M., and SCHILLING, J.-G. 1978. The nature and origin of geochemical variation in Mid-Atlantic Ridge basalts from the central North Atlantic. Geochimica et Cosmochimica Acta, 42, pp. 1501-1516. WOOD,D. A., JORON, J.-L., TREUIL,M., NORRY,M., and TARNEY, J . 1979a. Elemental and Sr isotope variations in basic lavas from Iceland and the surrounding ocean floor: the nature of mantle source heterogeneities. Contributions to Mineralogy and Petrology, 70, pp. 319-339. H., CORRE,O., JORON,J.-L., WOOD,D. A., VARET,J., BOUGAULT, TREUIL,M., BIZOUARD, H., NORRY,N. J., HAWKESWORTH, C. J., and RODDICK,J. C. 1979b. The petrology, geochemistry and mineralogy of North Atlantic basalts: a discussion based on IPOD Leg 49. In Initial reports of the Deep Sea Drilling Project. Edited by B. P. Luyendyk, J. R. Cann, et al. United States Government Printing Office, Washington, DC, Vol. 49, pp. 597-655. WRIGHT,T. L. 1974. Presentation and interpretation of chemical data