Regional Variation and Petrogenesis of Basalt Glasses from the FAMOUS Area, Mid-Atlantic Ridge

Transcription

1 Reginal Variatin and Petrgenesis f Basalt Glasses frm the FAMOUS Area, Mid-Atlantic Ridge by W. B. BRYAN Wds Hle Oceangraphic Institutin, Wds Hle, Massachusetts (Received 5 July 1977; in revised frm 26 April 1978) ABSTRACT Basalt glasses and minerals frm precisely lcated dredge and submersible statins in the FAMOUS area have been analysed by electrn micrprbe, and their reginal relatinships and petrgraphic features are described. Samples frm the median valley suth f fracture zne B, and frm the walls f the valleys nrth and suth f fracture zne B, tend t shw higher K 2 O, TiO 2, and / than samples frm the active vlcanic centers in the nrthern median valley. Sme fracture zne samples are high in K 2 O relative t median valley basalts. All glasses define an apparent binary ctectic in the nrmative plagiclase-pyrxene-livine ternary, suggesting cmpsitinal cntrl by lw-pressure crystal-melt equilibria. Crrelatins f / between pyrxene and glass and livine and glass suggest that mst crystals are in equilibrium with their liquids; a few xencrysts depart significantly frm equilibrium. Plagiclase cmpsitins are much less well crrelated with liquid cmpsitin. The ttal range f variatin f FAMOUS basalt glasses appraches that f all glass samples frm spreading centers in the Atlantic. The high /, high TiO 2 glasses have been thught t be typical f the East Pacific Rise, but ther examples are knwn frm the Atlantic at N. The submersible samples, which have been described in ther papers, define the lw TiO 2, lw / end f the ttal distributin. Mdel calculatins indicate that high level fractinal crystallizatin can accunt fr thse subsets f parent-residual liquids which shw a rati f increase f / t TiO 2 f abut 1-5 t 1-0. Pstulated parent-residual pairs in which TiO 2 enrichment is greater than / enrichment als shw excess K 2 O enrichment; this feature is characteristic f many ther tested parentresidual pairs amng seaflr basalts, and f submersible samples frm the FAMOUS dive area. These TiO 2 -enriched samples may be derived frm a zned magma chamber, as previusly suggested, r may have been prduced directly by mantle-related prcesses nt yet well-defined. The K 2 O-enriched basalts frm fracture zne B als may represent a distinct magma type erupted within the fracture zne. INTRODUCTION FOR three years preceding the 1974 submersible prgram, the FAMOUS area was intensively studied by a variety f surface ship techniques. Cruises by the R.V. Atlantis II in 1972 and 1973, by the R.V. Charct in 1972, and by the R.V. Knrr and R.V. Nrit during the submersible prgram in 1974 included dredging and cring prgrams which have prvided rck samples extending frm fracture zne A t the center f the median valley segment suth f fracture zne B, and as far west as site 332 f the Deep Sea Drilling Prject (DSDP) Leg 37 (Fig. 1, Table 1). The Leg 37 drilling sites are psitined in crust inferred t have been derived riginally frm the vicinity f the main dive area between fracture znes A and B. Petrgraphic and chemical data frm this extensive cllectin f dredge and cre samples, which varies widely in space and time, prvides the reginal cmpsitinal setting within which the mre clsely spaced and temprally restricted Uurnml f Petrlgy, Vl. 20, Put 2, pp J, 1979]

2 294 W. B. BRYAN 37'00' 33"3O' Fi. 1. Sample lcatins in FAMOUS area. Cntur interval is 200 fathms. Numbers adjacent t symbls indicate statins listed in Table 1. Symbls indicate cruise and cllecting vessel as fllws: Atlantis II, Cruise 73, clsed circle; Atlantis II, Cruise 77, pen circle; Knrr, Cruise 42, slid star; Charct, Cruise 31, pen square; Archimede, 1973 dives, pen star; Archimde, 1974 dives, slid triangle; Alvin, 1974 dives, star in circle; DSDP Leg 37, dt in circle. 36*30' sample set recvered by the submersibles may be interpreted. The purpse f this paper is t discuss these reginal variatins, their relatin t particular gelgic settings, and t define the petrgenic prblems which they present Samples frm the FAMOUS area can be related t exceptinally detailed bathymetry btained by narrw-beam ech sunding. Many f the dredge lcatins were precisely psitined using bttm-mred acustic transpnders n Cruises 73 and 77 f the R.V. Atlantis II and n R.V. Knrr Cruise 42. Mst f

3 BASALT GLASSES FROM THE FAMOUS AREA 295 TABLE 1 Lcatins f sample statins discussed in this paper. Fr dredge statins, psitin is best estimate f lcatin frm which rck was recvered Atlantis II, Cruise 73 (AII-73): Statin ' N, ' W Statin ' N, ' W Statin ' N, ' W Statin ' N, ' W Statin ' N, ' W Statin ' N, ' W Atlantis II, Cruise 77 (AII-77): Statin 19 Statin 23 Statin 26 Statin 28 Statin ' 34-5' 250' 340' 43-0' N, ' W N, ' W N, ' W N, 'W N, 'W Statin ' N, ' W Statin 52 Statin 58 Statin 64 Statin 67 Statin 71 Statin 76 Knrr, Cruise 42 (KN42): Statin 28 Statin 77 Statin ' 27-5' 260' 26-5' 27-0' 26-5' 540' 37-0' 390' N, 'W N, ' W N, 'W N, 'W N, 'W N, 'W N, ' W N, CC W N, 'W Statin ' N, ' W Statin 108 Statin 121 Statin 122 Statin 123 Statin ' 37-5' 37-2' 490' 36-5' N, ' W N, 'W N, 'W N, 'W N, 'W Charct, Cruise 31 (CH31): Statin 3 Statin 11 Archimede 1973 (ARP73): Statin 10-2 Statin 10-3 Archimede 1974 (ARP74): Statin Statin Statin Alvin 1974 (ALV525): Statin DSDP Leg 37 Site332A Site332B Site 334 Site ' 49-5' 50-0' 500' 57-0' 57-0' 51-5' 52-7' 52-8' 02-1' 17-7' N, 'W N, ' W N, 'W N, ' W N, 'W N, ' W N, ' W 48-6' N, ' W N, ' W N, 'W N, 'W N, ' W the ther dredge sites were psitined relative t surface-mred radar transpnder buys. The least precisely lcated dredge sites are thse in the western rift; muntains, which were lcated by satellite navigatin. In general, satellite navigated psitins are crrect t ±1 km, radar navigated psitins t ±0-5 km, and acustically navigated psitins are crrect t ±0-1 km.

4 296 W. B. BRYAN Submarine basalts ffer a unique pprtunity t examine liquidus variatin trends and slid-liquid equilibria because fresh r mderately weathered samples almst invariably include prtins f a thick glass rind that represents the quenched margin f a lava pillw r tube. These glasses may be analyzed quickly and simply by the electrn micrprbe, yielding cmpsitins free f the effects f crystal accumulatin. All chemical data reprted here were btained with the cmputer autmated electrn micrprbe in the Department f Earth and Planetary Sciences, Massachusetts Institute f Technlgy. Precisin and accuracy f the methd has been discussed by Frey et al. (1974), Bryan et al. (1977), and Melsn et al. (1976). In general, agreement between the M.I.T. and Smithsnian data is gd t excellent; there has been sme tendency fr the M.I.T. prbe data t give slightly high values fr and, and precisin fr Cr 2 O 3 is pr fr values belw 0-05 wt. per cent. All sample designatins fllw ur usual three-part numbering system. The first cmbined letter and number designatin indicates the ship and cruise number, as listed in Table 1. The secnd number refers t the statin, and the third number is the sample number within that statin. In the tables f analyses, shuld be understd t represent ttal irn expressed as ; Fe 2 O 3 and P 2 O 3 were nt determined and n allwance fr these cmpnents was made in cmputing the nrms. The effect f these missins, thugh small, is t increase slightly the amunt f nrmative livine and, f curse, n magnetite appears in the nrm. Fr sme purpses, the 'Mg number' expressed as the mlecular rati Mg/(Mg + Fe) is useful, and is given in the data tables. The rati / is preferred fr graphical presentatins because its greater numerical range prvides a mre sensitive discriminatin between data pints. DISTRIBUTION OF VOLCANIC AND INTRUSIVE ACTIVITY The general nature f recvered samples shws that present-day active vlcanism is limited t the flrs f the median valley segments nrth and suth f fracture zne B (Fig. 1). N evidence has been fund fr recent vlcanism in the rift muntains r in the fracture znes. Basaltic rck is, hwever, the mst cmmn material recvered frm the fracture znes. Only a very few samples f intrusive mafic r ultramafic rck (serpentinite and meta-gabbr) have been recvered frm fracture zne B. Rcks dredged in and near the central median valley are generally very fresh in appearance, with glass crusts nearly intact and shwing nly incipient surface alteratin t palagnite. Nrth f fracture zne B, the freshest samples were recvered alng the western side f the valley near the base f the west wall. Suth f the fracture zne B, the freshest samples were recvered frm ridges and small hills alng the eastern side f the valley flr. Samples recvered n the walls f the central valley all shw varying degrees f palagnitizatin and thin manganese catings, and may shw cnsiderable amunts f attached cral and wrm tubes. Samples frm the rift muntains and within the fracture znes als shw mderate palagnitizatin and manganese catings, and much r all f the glass may be spalled ff. In general, the amunt f weathering, the thickness f manganese and palagnite catings, and the amunt f attached rganisms

5 BASALT GLASSES FROM THE FAMOUS AREA 297 increase regularly with distance frm the median valley (Hekinian & Hffert, 1975; Bryan & Mre, 1977). Phtgraphy, bathymetry, and the nature f dredged samples all shw that recent vlcanic activity extends almst exactly t the limits f the active transfrm fault in fracture zne B, which is defined by a narrw trugh and by micr-earthquake epicenters (Reid & Macdnald, 1973; Spindel et al., 1974). Fresh basalt was recvered at statin 86, Knrr Cruise 42, in the flr f the median valley just t the nrth f the eastern end f the transfrm, and frm statins 23 and 28, R.V. Atlantis II, Cruise 77, just t the suth f the western end f the transfrm. The verlap f the nrthern and suthern median valleys which extend beynd the active transfrm, suggest that faulting may tend t migrate nrth and suth within a zne abut 20 km wide between 30' N and 37 40' N (Fig. 1). Tectnically defrmed and brecciated greenstnes and hydrthermally-altered basalts have been recvered thrughut the width f this zne. Relatively fresh vlcanic rck with abundant glassy selvedges was recvered thrughut the 600 m basement penetratin at DSDP Leg 37, site 332B, and relatively fresh basalt was recvered as far west as site 335. Thus, the cnversin f basalt t greenstne is nt a simple functin f age and depth. It appears that hydrthermal activity favring cnversin f basalt t greenstne is lcalized alng fracture znes, r at least is especially favred by the cmplex interactin between spreading and transfrm mvements in fracture zne B. Preliminary petrgraphic and chemical study f samples frm the DSDP Leg 37 drill hles (Bryan & Thmpsn, 1977; Bryan et al., 1977) suggests that mst f the material riginated as extrusive basalt r as shallw thin sills, and thus des nt prvide any direct evidence fr extensive intrusive activity in the rift muntains. Carse plutnic gabbr was penetrated at site 334, which may lie n the extensin f fracture zne A (P. J. Fx, persnal cmmunicatin, 1975). Other than the very limited amunt f serpentinite and pyrxenite recvered in fracture zne B dredge hauls, these are the nly bviusly plutnic rcks recvered in the FAMOUS area. CHEMICAL VARIATIONS Analyses f 71 basalt glasses are presented in Tables 2 t 6, gruped accrding t their gelgic r gegraphic assciatin. Inspectin f the tables shw cnsiderable verlap in the data sets. This may partly reflect the smewhat arbitrary nature f the distinctins between the prvinces. Fr example, the bundary between rift valley walls and flrs is nt well-defined; extensins f flr vlcanism int fracture zne B may be influenced in subtle ways by the fracture zne; and lder basalt recvered within fracture znes and at the leg 37 sites may have riginated in ne f the median valleys. Nevertheless this smewhat arbitrary subdivisin des reveal sme useful generalizatins. Samples cntaining unusually lw TiO 2 and and relatively high appear in the nrthern valleyflrand amng basalts at DSDP site 332 B (Table 2, cl. 15, 16; Table 6, cls. 5 and 6). This type f material is knwn t be especially characteristic f the central vlcanic highs such as Mt. Plut and Mt. Venus (Bryan & Mre, 1977). Hwever, similar material is at least sparingly present in the fracture znes and in the suthern rift

6 N, 33 15' W (Hekinian el at., 1973) ' W (Hekinian el at., 1973). a DO < SiO, TiOj A1 2 O 3 K 2 O Cr 2 O, TOTAL Q Or Ab An Di Hy 01 Mt 11 Ap Cr / Mg# TABLE 2 Electrn micrprbe analyses f basalt glass frm the nrthern rift valley m at Penn State University (Patern Lib) n September 18, ** n.d n.d * 1 -Q? ** n.d n.d * O. 1 Q {.' \y Sum includes 0-18 P 2 O 3, cls. 11 and 12; lss n ignitin 0-65 cl. 11,0-52 cl. 12. ** All Fe recalculated as. Nrms calculated frm riginal analysis, retaining and Fe 2 O 3 reprted by authrs. 1. AII lb, central valley suth f Mt. Plut. 9. AII , base f west wall. 2. AI , central valley suth f Mt. Plut. 10. AII , base f west wall. 3. AII , central valley suth f ML Plut. 11. KNR42-86, average f samples 6 and 83, west side f valley. 4. AII , central valley suth f ML Plut. 12. CH31-DR11-315XR.V. Charct cllectin, 'ppping rck' 49-3' 5. AII , central valley suth f Mt. Plut. 13. CH31-DR11-315Y R.V. Charct cllectin, glassy slab, 49-3' N, 6. AII , central valley suth f Mt. Plut. 14. ARP , nrth f Mt. Venus. 7. AII A, central valley suth f ML Plut. 15. ARP73-1O-3, east flank f Mt. Venus. 8. AII , central valley suth f ML Plut. 16. ALV , west flank f Mt. Plut (Bryan & Mre, 1977).

7 CO > in > -1 O r> GO m VI -n JO O H I m -d > 2 c m SiO 2 TiOj AljOj K 2 O Cr 2 O 3 TOTAL Q Or Ab An Di Hy 01 Cr 11 / Mg# / TABLE 3 Electrn micrprbe analyses f basalt glasses frm rift valley walls at Penn State University (Patern Lib) n September 18, AI lb, east flank f nrthern valley. 2. AII , east flank f nrthern valley. 3. AII , upper west wall, nrthern valley. 4. KN42-88, crest f west wall (rck drill) nrthern valley. 5. AII , west wall, suthern valley. 6. All , west wall, suthern valley. 7. KN , rift muntains suth f fracture zne A. 8. KN , rift muntains suth f fracture zne A. 9. KN , rift muntains suth f fracture zne A. 10. K.N42-123, base f west wall. 11. CH31-DR3-356, east f Mt. Venus. 12. ARP73-1O-2, east f Mt. Venus. 13. AI , east flank f suthern valley. 14. AlI , east flank f suthern valley. 15. KN , rift muntains west f nrthern valley. 16. K.N , rift muntains west f nrthern valley.

8 Oil = JO "< z SiO, TiO, A1A K 2 O Cr 2 O 3 TOTAL Q Or Ab An Di Hy 01 Cr 11 / Mg# / L TABLE 4 Electrn micrprbe analyses f basalt glasses frm the suthern rift valley U _ ' _ rm at Penn State University (Patern Lib) n September 18, A1I-73-5O-IC, valley center. 2. A U, nrth end f valley. 3. AII lb, nrth end f valley. 4. All , nrth end f valley. 5. Al , west side f valley center. 6. AII , east side f valley center. 7. Al , east side f valley center. 8. All ,cast side f valley center. 9. AII , east side f valley center. 10. AII , east side f valley center. 11. AII , east side f valley center. 12. All , east side f valley center. 13. A1I , east side f valley center. 14. AII , center f valley. 15. AII , center f valley. 16. AI , center f valley.

9 ' > t~ H O r~ m n 70 O 2 H X m > c tn > 70 m SiO 2 TiO 2 A1 2 O 3 K 2 O Cr 2 O 3 TOTAL Q Or Ab Ne An Di Hy 01 Cr II / Mg# / TABLE 5 Electrn micrprbe analyses f basalt glasses frm fracture znes at Penn State University (Patern Lib) n September 18, AII , west flank f'b-ridge'. 2. All , west flank f 'B-Rklge'. 3. All , west flank f'b-ridge'. 4. All , west flank f'b-ridge'. 5. AU-77-52P, west flank f'b-ridge'. 6. K.N , nrth flank f transfrm fault. 7. KN , nrth flank f transfrm faull 8. KN , nrth flank f transfrm fault. 9. ARP , west end f transfrm, fracture zne A. 10. ARP , west end f transfrm, fracture zne A. 11. KN , suth flank f transfrm fault, fracture zne B. 12. K.N , suth flank f transfrm fault, fracture zne B. 13. KN , suth flank f transfrm fault, fracture zne B. 14. KN , suth flank f transfrm fault, fracture zne B.

10 302 W. B. BRYAN TABLE 6 Electrn micrprbe analyses f basalt glasses frm DSDP Leg 37. Data frm Bryan etai, A 332B 335 SiO 2 TiO 2 MA K 2 O TOTAL Q Or Ab An Di Hy Ol 11 / Mg# _ _ valley (Table 5, cls. 5, 9, and 10; Table 4, cls. 8 and 9). The suthern valley, hwever, appears t be characterized by basalt shwing distinct enrichment in and TiO 2. Basalt resembling the mst enriched samples als appears in fracture / A F / A V A 7 60 FIG. 2. A prtin f the AFM ternary shwingfieldf glass data f tables 2-6 (dashed line) and f whle rck data (Bugault & Hekinian, 1974; dtted line).

11 BASALT GLASSES FROM THE FAMOUS AREA 303 zne B. The rift walls as a grup appear t be less diverse, being characterized by mderate enrichment in TiO 2 and. The DSDP Leg 37 basalts tend tward relatively high with lw t mderate TiO 2. A grup f samples characterized by unusually high K 2 O was recvered frm the vicinity f the active transfrm in fracture zne B (Table 5, cls. 6-8 and 11-14). The relatively subtle chemical variatins can be visualized mst readily with the aid f graphical presentatins. The usual AFM plt (Fig. 2) shws that data frm Atlantis II, Knrr and DSDP Leg 37 cllectins cluster in a tight gruping between I N. rift valley S. rift valley Fracture znes A Rift walls D DSDP TiO 2 15 FIG. 3. Variatin f / as a functin f TiO r Dashed line enclses field f all basalt glass data frm the Alvin cllectin in the nrthern median valley (Bryan & Mre, 1977). H = high-livine average; L = lwlivine average (Table 11). Numbered vectrs crrespnd t least-squares calculatins in Tables 'A' per cent and 'F' per cent. The least irn enrichment is shwn by the basalt glass frm Mt. Venus cllected by Archimede (Mg# = 0-70; Table 2, cl. 15), and the greatest irn enrichment is shwn by glasses frm AH-77 statin 76 in the suthern rift valley (Mg# = 0-50; Table 4, cls. 12 and 13). Whle rck analyses frm the Charct cllectin (Bugault & Hekinian, 1974) clearly shw the effect f livine accumulatin, with a distinct elngatin f the field tward the psitin f analyzed livine. Three f the mst imprtant variables are TiO 2, K 2 O, and the / rati. A plt f / as a functin f TiO 2 (Fig. 3) shws a general tendency fr a psitive crrelatin but with cnsiderable scatter. The very lw TiO 2 and / in the samples frm Mt. Venus and the fracture znes are clsely apprached by sme Leg 37 samples and are typical f Alvin samples cllected frm ML Plut (Bryan & Mre, 1977). The high TiO^ high / basalts cme frm fracture zne B, statin 52; the suthern rift valley, statin 76, and frm the Archimede 1973 dive statin 10 n the east flank f the nrthern rift valley.

12 304 W. B. BRYAN The range f variatin in the FAMOUS area basalt glasses appraches that which s far has been demnstrated fr the Atlantic Ocean as a whle. They cver almst the whle range frm the lw TiO^ lw / glasses at sites 14 and 18 f DSDP Leg 3, t the cmpsitin f the glass frm site 15, which was cnsidered t be a highly fractinated derivative f these glasses (Frey et al., 1974). In the summary by Melsn et al. (1976) glass analyses frm the Atlantic with less than 1-00 wt. per cent TiO 2 are very rare. The East Pacific Rise and Juan de Fuca Ridge are characterized by high TiO 2, high glasses resembling thse at statin 76 f Atlantis II Cruise 77, but als extend t varieties even mre enriched in TiO 2 and. The similarities are striking fr mst elements, as shwn in Table 7. SiO 2 TiO 2 A1 2 O 3 K 2 O TOTAL Mg# DSDP TABLE 7 Cmparisns f basalt glass analyses ARP JFR VG AII DSDP AVE MAR Aim DSDP 3-18 and 3-15, Suth Atlantic, and JFR VG-2, Juan de Fuca Ridge, frm Frey el al, AVE.MAR = average f Mid-Atlantic Ridge basalt glasses, excluding fracture znes and ff-ridge surces (Melsn et al., 1976). ARP73-1O-3, Mt Venus; AII , median valley suth f fracture zne B; AII , west wall f suthern median valley. Further imprtant similarities and differences are brught ut in plts f K 2 O vs. TiO 2 (Fig. 4). K 2 O tends t 'peak ut' at abut 1-50 TiO 2. The high TiO 2 glasses frm Atlantis II statin 76 frm a distinct grup ff the main trend f the ther analyses. The DSDP Leg 3 analyses als diverge frm this trend, as might be expected frm their distinctly lwer K 2 O cntents. Glasses frm Knrr 42 statins 77, 121, and 129 in fracture zne B als diverge significantly ff the trend, due t their higher K 2 O cntents. Excluding these bviusly divergent grups, there is, verall, a distinct psitive crrelatin f K 2 O with TiO 2. There als is systematic variatin in the prprtins f the three principal nrmative minerals, with the analyses clustering alng a line f almst cnstant plagiclase prprtin in the plagiclase-pyrxene-livine ternary (Fig. 5). This trend resembles the livine-plagiclase ctectic in the analgus An-Di-F experimental system (Osbrn & Tait, 1952), but is ffset t the plagiclase side f the ctectic. This trend is matched almst exactly by glasses recvered by DSDP frm the three majr cean basins (Bryan et al., 1976). Althugh ne might expect 0-58

13 BASALT GLASSES FROM THE FAMOUS AREA 305 CD O D O O» O *O CD cdn» O CD OO O ) O Wt.% TIO 2 OOOO OO <D O Fi. 4. Variatin f K 2 O as a functin f TiO 2. Slid circles are high K 2 O glasses frm fracture zne B. Slid triangles are samples frm Leg 37 drill sites. Open triangles are glasses frm DSPD sites 14 and 15 (Frey et al., 1974). 60, Wt.% FIG. 5. A prtin f the nrmative ternary plagiclase-pyrxene-livine. Dashed phase bundaries are thse fr the analgus anrthite-dipside-frsterite system (Osbrn & Tait, 1952). Triangles represent samples with phencryst pyrxene. Variables are ttal nrmative pyrxene, livine, and plagiclase. Pyrxene is the sum f nrmative dipside, enstatite, and ferrsilite; plagiclase is the sum f nrmative albite and anrthite; and livine is the sum f nrmative frsterite and fayalite; all nrmalized t 100%.

14 306 W. B. BRYAN a negative crrelatin between nrmative livine and the / rati, there is a curius dispersin f / values with decreasing livine (Fig. 6), such that the lw / lavas are represented by bth lw and high-livine varieties. Inspectin f the figure shws that the apparently anmalus lw-livine, lw / grup cmes frm a variety f lcatins, including the western flr and wall f the nrthern rift valley, fracture zne B, and statin 76 n the east side f O) r 3 3 O O 5 10 Nrmative 15 Olivine FIG. 6. Variatin f / as a functin f nrmative livine. Symbls represent gegraphic lcatins and are the same as in Fig. 3. the suthern rift valley. This latter statin als includes high TiO 2, high / samples at the lw-livine end f the data distributin. The high-livine glasses d, hwever, all have lw TiO 2 and lw /. They are characteristic f the central vlcanic highs Mt. Plut and Mt. Venus in the nrthern rift valley, sampled in mre detail by the submersibles (Bryan & Mre, 1977), and f certain intervals at DSDP drill site 332 B. The mst undersaturated liquids are frm fracture zne A (Table 5), cls. 9 and 10). These are als the mst TiO 2 -pr examples, althugh they have slightly higher / ratis than the. Mt. Plut and Mt. Venus liquids.

15 BASALT GLASSES FROM THE FAMOUS AREA 307 PETROGRAPHY AND MINERALOGY Rck glasses are by definitin uncrystallized and, therefre, cannt be characterized in terms f their mineral assemblages and textures. The nrmative cmpsitin f these glasses as reprted in Tables 2-6 prvides an indicatin f the prprtins and cmpsitins f mineral phases which culd be expected t appear in the hlcrystalline equivalents f the glasses. Althugh phencrysts are, f curse, nt included in the analyses, their cmpsitins and prprtins might be expected t be systematically related t cmpsitin f the enclsing glass, if they are indeed in equilibrium, and if there has been little relative mvement between crystals and liquid. Micrphencrysts and micrlites in the glass can mre cnfidently be assumed t have been precipitated frm the glass, and thus might be used as in labratry quenching experiments t determine the relatin f the glass cmpsitin t the pssible phase bundaries, in the natural basalt system. Unfrtunately, these crystals ften are t small fr reliable micrprbe analysis, and because cling rates are unknwn and uncntrlled, they culd represent metastable assemblages. Further prblems in petrgraphic characterizatin are f a practical and peratinal nature. Where ne is dealing with small amunts (5 per cent r less) f dispersed mineral phases, as is ften the case in these samples, a single small thin sectin may nt be an adequate sample f the rck. Further, phencryst prprtins vary significantly within a single sample, evidently due t flw differentiatin within the tubular extrusins which are s cmmn in the median valley. These matters have been discussed in mre detail in relatin t quantitative mdal estimates in samples recvered by R.V. Alvin (Bryan & Mre, 1977). It was shwn in particular that there are wide variatins in the relative prprtins f livine and plagiclase phencrysts, in spite f the nearly cnstant plagiclase prprtins implied by the nrmative data. It is these large variatins in phencryst prprtins which lead t the principal plagiclasebasalt and livine-basalt variants recgnized by Miyashir et al. (1969), by Bugault & Hekinian (1974), and by ARCYANA (1977) in their classificatins based n bulk rck chemical and petrgraphic data. Frm the pint f view f descriptive classificatin, the glass data are clearly less cmplex than the whle rck data. Of the mre than 120 glass analyses nw available fr the FAMOUS area basalts, nne lies appreciably away frm the implied livine-plagiclase phase bundary (Fig. 5), while their cmpsitins extend frm the intersectin f that bundary with the spinel field t, but nt appreciably beynd, the pyrxene-plagiclase jin. Any subdivisin within this cntinuum clearly wuld be arbitrary. The mst lgical nmenclature wuld be based n the phase bundaries as deduced frm the analgus experimental system, and frm quench phases in the glass. Glasses with mre than abut 10 per cent nrmative livine wuld be livine basalts, and thse with 10 per cent r less nrmative livine (r n quench livine) wuld be pyrxene basalts, with a virtually cnstant prprtin f plagiclase. Plagiclase-rich r livine-rich variants in the FAMOUS area, and pssibly amng sea-flr basalts in general, appear t be simply the result f secular accumulatin f plagiclase r livine

16 308 W. B. BRYAN phencrysts. Hwever, a few glasses enriched in nrmative plagiclase have been dcumented amng samples recvered by DSDP (Bryan et al., 1976). Althugh the glass cannt display a texture, the shapes, sizes, and intergrwths f dispersed mineral phases may be relevant t their interpretatin in terms f crystal-melt equilibria. In subsequent discussins, and in the tables f mineral analyses, distinctins are made between xencrysts, phencrysts, micrphencrysts, and micrlites. Within a given sample, xencrysts are generally larger than phencrysts and tend t be subhedral t anhedral with evidence f marginal crrsin and reverse zning. Phencrysts are generally subhedral t euhedral and shw little r n marginal reactin with the magma. Micrphencrysts are relatively smaller than phencrysts f the same phase and are generally euhedral, but may shw sme tendency tward skeletal frms. Micrlites tend t be skeletal; 'belt-buckle' plagiclase and 'lantern' livine are typical examples. Size alne cannt be a basis fr these distinctins. A xencrystal spinel greater than 0-5 mm in diameter wuld be exceptinal, while plagiclase phencrysts may exceed 6 mm. Olivine phencrysts tend t be in the range 1-3 mm. Pyrxene xencrysts are very rare, but may equal r exceed plagiclase in size when they d ccur. Pyrxene micrphencrysts are cmmnly intergrwn with plagiclase in 'bw-tie' clusters, in which pyrxene frms the knt, and plagiclase the bw (Plate 1, Fig. A). Plagiclase-livine glmercrysts are cmmn, and livine plagiclase 'bw-ties' smetimes ccur, as with pyrxene. These bservatins have physical significance, in that xencrysts are likely t be derived by slw plutnic crystallizatin and may r may nt be in equilibrium with the glass in which they ccur, while phencrysts and micrphencrysts prbably are in equilibrium and are prducts f crystallizatin during r shrtly befre extrusin. Micrlites may be grwn after extrusin frm a super-cled melt, and thus may be nt strictly f equilibrium cmpsitin r may be related t a mre advanced stage f magma fractinatin. 'Bw-tie' intergrwths appear t result frm simultaneus grwth utward frm a cmmn nucleus, and are cnsidered t be textural evidence f ctectic crystallizatin. Plagiclase is ubiquitus as phencrysts r micrphencrysts, ranging frm a few per cent by vlume in the nearly aphyric samples t mre than 40 per cent in sme f the highly prphyritic varieties. Plagiclase cumulates are sparingly present thrughut the suthern and nrthern median valleys but seem especially characteristic f samples recvered frm the west wall f the nrthern valley. There appears t be n simple relatin between plagiclase abundance and glass cmpsitin, as previusly nted, althugh plagiclase cumulates shw sme tendency t be assciated with pyrxene-basalt glasses. Plagiclase cmpsitins range frm near An 90 t a lw f An 62 (Table 8). Within the individual samples, there may be a cnsiderable difference between phencrysts and micrlites r xencrysts. This is illustrated by the analyses in cls. 1A, B and 2A, B, C f Table 8. As shwn here, the xencryst plagiclase may be mre sdic than the phencrysts and appraches the cmpsitin f micrlites in the glass. Xencryst margins are runded, crrded, and may be reverse-zned t cmpsitins appraching that f the phencrysts. Hwever, there is little

17 BASALT GLASSES FROM THE FAMOUS AREA I a in 5s f? ^ M O 00 e K c»t.a a «3 3 5 > 8» J & J l:l pll >>J O J= cu < UU < CQ (J Q

18 310 W. B. BRYAN TABLE 8 Electrn micrprbe analyses fplagiclase 1A IB 2A 2B 2C SiO, TiO, Al 2 Oj K.,0 TOTAL Si Ti Al Fe Mn Mg Ca Na K. TOTAL An = SiO 2 TiO; A1 2 O, Na,O K 2 O TOTAL Si Ti Al Fe Mn Mg Ca Na K TOTAL An = A B O-OOO O-OOO ISA B AII : A, runded and crrded xencryst; B, subhedral phencryst. 2. AII : A, runded xencryst; B, phencryst; C, grundmass micrlite. 3. AII : phencryst 4. AII C: phencryst intergrwn with pyrxene. 5. AII : phencryst in glass. 6. AII-77-23: phencryst. 7. AH : phencryst. 8. CH31-DR3-356: A, micrphencryst in glass; B, micrlite in glass. 9. ARP : micrphencryst in glass. 10. ARP : micrphencryst in glass. 13. AII : phencryst in glass. 11. ARP : micrphencryst in glass. 14. All : phencryst in glass. 12. AII : phencryst in glass. 15. AII : A, phencryst; B, grundmass micrlite.

19 BASALT GLASSES FROM THE FAMOUS AREA 311 difference in the / ratis, suggesting that all crystals have all grwn frm liquids f similar / rati. This, and the similarity in anrthite cntent between xencryst and micrlite plagiclase, suggests that these are cgnate xencrysts derived frm a cler and mre crystalline prtin f the same magma chamber, r alternatively, that the xencrysts riginally equilibrated at significantly higher pressures (Bender et ai, 1978). l O c < ? 0 a? 65 D A 2 O 9 D * Plagiclase 1 i I l Wt.% An in Liquid FIG. 7. Variatin in plagiclase anrthite cntent as a functin f nrmative anrthite in enclsing basalt glass. Slid circles are phencrysts, pen squares are xencrysts, pen circles are micrphencrysts, and pen triangles are micrlites. In Fig. 7, the anrthite prprtin in the analyzed crystals is pltted against anrthite cntent f nrmative plagiclase in the glass fr thse samples fr which data fr c-existing glass are available. There is cnsiderable scatter due t the wide range between feldspars in a given sample. Hwever, sme f the scatter is als prduced by distinct differences in phencryst cmpsitins in liquids which are similar in nrmative feldspar cntent, but which differ in ther cmpsitinal parameters. In these cases, the mre sdic plagiclase is assciated with the mre irn-rich and mre titanium-rich liquid. Pyrxene is relatively cmmn in a wide variety f basalts, and it is difficult t generalize its distributin either in relatin t magma cmpsitin r t gelgic setting. Grundmass pyrxene is readily identified in the hlcrystalline grundmass f mst f the basalts. As phencrysts, micrphencrysts, r xencrysts in 0

20 312 W. B. BRYAN TABLE 9 Electrn micrprbe analyses f pyrxene 1A IB 1C 2A 2B 2C 3 4A 4B SiO 2 TiO 2 A1 2 O, K 2 O Cr 2 O 3 TOTAL Si Tl Al Fe Mn Mg Ca Na K Cr TOTAL SiO 2 TiO 2 A1 2 O 3 K 2 O Cr 2 O, TOTAL Si Ti AJ Fe Mn Mg Ca Na K Cr A B C A B TOTAL 4-0O AII : A, average f sectr-zned micrphencryst; B, large grundmass pyrxene; C, grundmass micrlite. 2. AII : A, average f sectr-zned micrphencryst; B, micrphencryst; C, micrlite. 3. AII : phitic intergrwth with plagiclase. 4. AII-73-5O-1C: A, phencryst; B, grundmass pyrxene.

21 BASALT GLASSES FROM THE FAMOUS AREA 313 quenched glass r aphanitic basalt, it tends t ccur in liquids with less than 80 per cent nrmative livine, with TiO 2 and / greater than 1-0, and K 2 O greater than Hwever, neither the abundance f these crystals, their size and mrphlgy, r their cmpsitin appear t be simply related t the liquid in which they ccur. Fr example, prminent runded pyrxene xencrysts (r phencrysts?) are present in CH frm the east wall f the nrthern rift valley. This glass has TiO 2 = 1-05, / = 1-02, and ver 7-5 per cent nrmative livine. Althugh the runded frm and size f these crystals suggests. they are xencrysts, they d have a relatively lw (0-22) / rati which is cmpatible with that f the glass in which they ccur, as discussed belw. At the ther extreme, AII shws abundant well-frmed pyrxene micrphencrysts in bw-tie intergrwth with plagiclase. Althugh these pyrxenes shw n bvius disequilibrium features, they have / f -0-22, in cntrast t the glass which averages / = 1-68 and t quench pyrxene in this glass with / = 0-40 (Table 9, cls. 6A, B, C). Very large, blcky, presumably xencrystal, pyrxene ccurs in AII ; as shwn in Table 9, 1A, B, C there is little difference in cmpsitin between the suppsed xencrystal pyrxene and grundmass micrlites. The same is true f AII (Table 9, cls. 2A, B, C); in bth examples there is a tendency fr the quenched pyrxene t shw distinctly higher TiO 2 and especially higher A1 2 O 3. Even this relatin des nt hld fr pyrxene in AII-73-5O-1C (Table 9, cls. 4A, B), hwever. Such generalizatins are cmplicated by the patchy zning which is s evident ptically, and which micrprbe pint analyses shw may be characterized by distinct variability in A1 2 O 3 and TiO 2 as well as / ratis. Basalt glass ARP , characterized by TiO 2 = 1-59, / = 1-59 and K 2 O = 0-26, shws nly rare micrphencrysts f pyrxene; the analyzed example (Table 9, cl. 8) shws apprpriately high /. Plutnic pyrxene ccurs in fracture zne B in ne f the few mderately altered mafic rcks recvered there. It ccurs as relatively large, anhedral crystals and als as a finer intergranular matrix in assciatin with ilmenite and large patches f a white granular material which is high in Ca and Al and may be granulated and altered plagiclase. This pyrxene (Table 9, cl. 5) is ne f the mst irn-rich that has been analyzed. There is little cmpsitin difference between the carse and fine varieties in this rck. Cmpsitin relatinships are summarized in Fig. 8, in which pyrxene / ratis are cmpared t / in the glass with which they are assciated. The phencryst data shw a gd psitive crrelatin, and the implied K D fr (/) crystal/(/) liquid averages 0-26, with a range frm 0-22 t The field assciatin f pyrxene is equally ambiguus. In the nrthern rift valley it appears t be absent as a liquidus phase in lavas frm the central valley flr, but is cmmn in lavas n the valley walls (Bryan & Mre, 1977). In the 5. AII : large andedral crystal in gabbr. 6. AII : A,B. lw-irn micrphencrysts; C, small sectr-zned micrphencryst intergrwn with plagiclase. 7. AII-77-76X: A, grundmass pyrxene; B, quench pyrxene in glass. 8. ARP : micrphencryst in glass. 9. CH31ODR3-356: micrphencryst in glass.

22 314 W. B. BRYAN PYROXENE O :* * '/ Liquid FIG. 8. Variatin f / in pyrxene and enclsing basalt glass. Slid circles are phencrysts, pen circles are xencrysts; and the pen triangle is a micrlite. suthern rift valley, xencryst and phencryst pyrxene are cnspicuus in basalts frm the central highs (AII and 28) at the nrthern end f the valley flr in fracture zne B, and in the high Fe, high Ti basalts frm AII n the east side f the valley center. Pyrxene is als prminent at statins 26 and 58 n the west and east sides f the valley, respectively. The relative abundance f pyrxene in the suthern valley lavas is cnsistent with their general tendency t saturated r versaturated cmpsitins. N rthpyrxene r pigenite has been detected in any f these lavas, althugh a trend tward subcalcic grundmass augite is evident in samples frm AII (Table 9, cl. 4A, B). Phase equilibria presented by Bender et al. (1978) indicate that pyrxene is the liquidus phase abve 10 kb, but wuld be unstable in all but the mst fractinated liquids at lw pressures. These experimental results are cnsistent with the petrgraphic evidence cited here that xencrystal pyrxene, presumably f relatively deep rigin, is being resrbed in many f the basalts in which it ccurs. Olivine appears either as a liquidus phase, r as quenched grundmass crystals in the majrity f the samples, as might be predicted frm the nrmative data (Table 10). Like plagiclase, it varies greatly in prprtin; lcal cumulates give rise t picritic basalt varieties which were especially cmmn in the Charct cllectin (Bugault & Hekinian, 1974). A typical example is ARP frm the east flank f Mt. Venus; like ther glasses frm ML Plut (Hekinian et al., 1976), this glass shws lw TiO 2, lw /, and high nrmative livine. In general, the abundance and cmpsitin f livine seems much mre clsely related t the cmpsitin f its assciated glass than was the case fr pyrxene (Fig. 9). Unlike the pyrxene data, that fr livine des nt shw severe departures frm the trend. The exceptins which fall belw the trend are xencrysts frm AII and AII The mean K D fr these livine-glass pairs is 0-27, with a range f 0-24 t 0-33.

23 BASALT GLASSES FROM THE FAMOUS AREA O ON Q O O O ON O ON O O O 8 s 1.b s c '2? ON O O r*4 VI (N r~ 6 ON 6 6 ON ^5 VO (N ON V-I O O ON O > m O 39 (N wi «<N O m d d d <i> d c> d OO <N NO OO (N O I CJ ON m O O O - O O <N O O v> v~i m r^ r<i O OO O ~^ O ON "^ ON O f*^ O VO O O O O O O i n r^ ^ OO O Q O Q O ON <N O 00 O O O O O O O ( ON (!8 VO m Tf 1^ <N O ON ' 8 >n <N O : ; > (N -^ O» VO O ^ ON NO rn V) 00 «OOOMSfNQO ON 5 (N O r- O O O O O O -- O O ON * (N-OOO r- O Q (N O Q ON {N O OO O O I 6 I : H O O O O O ~- <N vi m O ON -H 88^ u a.5 a O a> a. J3 O C g 8 c a li *3> a C til 8-00 O. 5 w S "fi ec».a ««x E S S g S ; O O <s m > O Q Q rn fn s^ NO r*- ^ j i i ^ v v v v 2 2 ' : < < < u <

24 316 W. B. BRYAN 3 V> O O CD 5 OLIVINE > Liquid Fi. 9. Variatin f / in livine and enclsing basalt glass. Slid circles are phencrysts and micrphencrysts, pen circles are xencrysts. Nt all glasses with high nrmative livine shw high cncentratins f livine; in particular, thse frm Mt. Plut (Table 2, cl. 16) tend t be aphyric (Bryan & Mre, 1977). Typically, these glasses d shw quench livine as minute 'spinifex'- like grwths which increase in size and abundance tward the hlcrystalline pillw interir (Plate 1, Fig. B). In these rcks livine is almst invariably assciated with brwn spinel, which is present bth as runded t subhedral inclusins in livine, and as free-flating ctahedral micrphencrysts. In the lw nrmativelivine and in the mre irn-rich liquids, spinel is rare r absent, and livine tends t cexist with clinpyrxene. Gd examples f this assciatin appear amng the high Fe, high Ti lavas at AII-77-76, and at DSDP Leg 37, site 332A (Bryan et al, 1977). In these examples, euhedral t subhedral livine and pyrxene ccur in clse assciatin with n petrgraphic evidence f reactin, and bth livine and pyrxene frm 'bw-tie' intergrwths with plagiclase (Plate 1, Fig. C). This is cnsistent with nrmative cmpsitins which place these liquids clse t the livine-plagiclasepyrxene ternary. At AII-77-28, large skeletal livines are clsely assciated with large pyrxene phencrysts r xencrysts in a liquid which cntains a small amunt f nrmative quartz (Table 4 cls. 2 and 3). There is cnsiderable variatin in livine mrphlgy. Olivines in the cumulates tend t be large, up t 1 cm r mre, with runded anhedral utlines and a 'spngy' internal appearance with many runded inclusins f glass. Small phencrysts and micrphencrysts usually ccur as well-frmed bi-pyramidal crystals, ften with chevrn-like internal glass inclusins r 'christmas-tree' extensins in the C-axis directin (Plate 1, Fig. D). Quench crystals tend tward rectangular utlines filled in by 'spinifex' lattice-wrk with mre substantial extensins alng the diagnals. Anther cmmn variant is the 'lantern-and-chain' in which there is extreme elngatin in the C-axis directin. Vesicles are cmmn in all the lavas and tend t increase in size and abundance frm the glassy surface tward the crystalline interir f pillws. Centers f pillws rt

25 BASALT GLASSES FROM THE FAMOUS AREA 317 and tubes are cmmnly hllw, but this is nly partly due t cncentratin f gas tward their centers. Observatins frm Alvin (Ballard et ai, 1975) have shwn that lava 'drain-away' is als a significant factr in prducing these hllw cres. Vesicles in the glasses tend t be 1020 fi in diameter and are lined with glbules, presumably sulphide as described by Mre & Calk (1971). These seem especially cmmn in the glasses frm Mt. Venus and Mt. Plut. Bryan & Mre (1977) have discussed the abundance f vesicles, water, and sulphides in Alvin samples frm the nrthern median valley. Overall, there is reasnable cnsistency between petrgraphic characteristics f the lavas and their chemical (including nrmative) cmpsitins. Olivine cmpsitins shw high crrelatin with liquid cmpsitins; fr pyrxene and especially fr plagiclase this relatin is nt as bvius. Spinel is cnspicuus in liquids which plt clse t the spinel field in the livine-plagiclase-pyrxene ternary. Pyrxene is a typical liquidus phase in liquids which have little r n nrmative livine and in which TiO 2 and / are greater than 1-0. Ttal phencryst abundances may be highly variable even in a single sample and are evidently very susceptible t secular variatin. Much f the variatin reprted fr the whle-rck basalt cmpsitins appears t be due t this secular accumulatin. The glass cmpsitins are much mre systematic and can be classified as either livine- r pyrxene-basalt depending n their psitin in the three-phase ternary system. PETROGENESIS Certain critical features f the basalt liquids prvide cnstraints n pssible chices f petrgenetic mdels. Samples which vary in cntent f /, K 2 O, and TiO 2, suggestive f variatin in degree f differentiatin, are nearly identical in La/Sm and in strntium istpe ratis (Schilling, 1975; White & Bryan, 1977), and thus mixing f magmas related t different mantle surce areas appears unlikely. Rather, the bserved variatins may be related t varying degrees f partial melting f a hmgeneus surce, r t high-level prcesses perating n already hmgenized mantle-derived magma. The adherence f the nrmative liquid cmpsitins t a ctectic-like linear trend in the pyrxene-plagiclase-livine ternary strngly suggests cmpsitinal variatin cntrlled by crystal-melt equilibria at lw pressure. The divergence f the natural glass data frm the experimental ctectic may simply be an artifact f prjecting the mre cmplex natural rck data int this simplified system, but cncentratin f water r ther vlatiles in the mre differentiated liquids als might cause divergence f the trend int the plagiclase field, relative t the dry system. Other peculiarities f the data are nt s easily ratinalized. It has been shwn that the livine-pr pyrxene basalt liquids shw a wide range in / (Fig. 6). Certain lw-livine liquids are almst as lw in, TiO 2 and K 2 O as the high-livine liquids. Again, DSDP basalt glasses shw a similar distributin (Bryan et ai, 1976). Average cmpsitins f these lw- and highlivine candidate parent liquids are cmpared in Table 11. The lw-livine liquid is distinctly higher in SiO 2 and lwer in A1 2 O 3 and, and slightly lwer in ttal

26 318 W. B. BRYAN TABLE 11 Average cmpsitins f lw- and high-livine candidate parent liquids 1 2 SiO 2 TiO 2 A1 2 O 3 KjO Cr 2 O 3 TOTAL High nrmative livine basalt glass, average f ARP and 33, ARP , and ALV Lw nrmative livine basalt glass, average f AII-77-52P, A1I , AII and 71. and, features which clearly accunt fr the nrmative differences. It is evident that fr the data set as a whle, there are nt systematic changes in /, TiOj, K 2 O, and ther cmpsitinal parameters sensitive t livine/plagiclase crystallizatin, as might be expected if these liquids represent a simple variatin series alng a ctectic bundary, starting frm a cmmn parent liquid cmpsitin. It is pssible that the apparent ctectic (Fig. 5) cnsists f distinct fractinatin trends which result frm crystallizatin f varying prprtins f livine, plagiclase, and pyrxene, and which therefre tend t prject ut f the plane f the figure and t verlap alng a cmmn ctectic. Tables shw examples f TABLE 12 Least-squares test f crystallizatin in lw-irn dredge samples SiO 2 TiO 2 A1 2 O, K 2 O Calculated Observed Variable Residual Plagiclase 1 Olivine 2 Pyrxene 3 y 2 Weight fractin Parent is average lw-livine basalt (Table 11), residual liquid is average f glasses 2, 3 and 4 frm AH-7 7 statin 52. 'Cl. IB,Table8. 2 Cl. 1 A, Table 10. 'Cl. IB,Table9.

27 BASALT GLASSES FROM THE FAMOUS AREA 319 TABLE 13 Least-squares test f crystallizatin in high-irn dredge samples Parent calculated Parent bserved Variable Weight fractin SiO 2 TiOj A1 2 O 3 K 2 O Residual Plagiclase 1 Olivine 2 Pyrxene = Parent is average high-fe glass frm KN-42 statin 28; residual liquid is average f high-fe glasses frm AII-77 statin 76. 'Cl. 1A, Table 8. 2 Cl. IB, Table Cl. 4A, Table 9. unweighted least-squares calculatins, which test the ability f simple crystal melt fractinatin t prduce the verall majr element variatin bserved amng specific subsets f the basalt glasses. In these calculatins, the 'best' apprximatin t the inferred parent cmpsitin is calculated as a linear cmbinatin f inferred residual liquid and the apprpriate mineral phases as deduced frm petrgraphic data. Frm data expressed this way, the 'per cent crystallized' is easily calculated as 100 x (sum f mineral weight fractins). The cmpsitin vectrs in /-TiO 2 space are shwn in Fig. 3, numbered t agree with the crrespnding table. These and ther similar calculatins cmpleted t date reveal several significant relatinships. In general, agreement is gd between calculated and bserved parent cmpsitins fr mst majr element xides. Hwever, as shwn in Tables TABLE 14 Least-squares test f crystallizatin in lw-titanium dredge samples SiO 2 TiO 2 Al 2 Oj K 2 O Calculated Observed Variable Residual Olivine 1 Plagiclase 2 Pyrxene 3 y 2 Weight fractin = 0058 Parent is average f lw-livine basalt f Table 11; residual liquid is average high-irn glass frm KN-42 statin Average f Cls. 4 and 5, Table Cl. IB,Table8. 3 Cl. IB, Table 9.

28 320 W. B. BRYAN TABLE 15 Least-squares test f crystallizatin in high-titanium dredge samples. Parent calculated Parent bserved Variable Weight fractin SiO 2 TiO 2 A1 2 O 3 K 2 O Residual Plagiclase 1 Olivine 2 Pyrxene 3 y j O.98Q Parent is average high titanium glass frm AII-77 statin 52 (2, 3, and 4); inferred residual is average f high-fe glasses frm AII-77 statin 76. 'Cl. 14, Table 8. 2 Cl. IB, Table Average f cls. 7A, 7B, Table 9. 12, 13 and 16, and Fig. 10, thse residual-parent sets which require relatively large increases in TiO 2 relative t / shw systematic misfits fr TiO 2, apparently requiring a parent with abut 1-5 times the amunt f TiO 2 actually present in the assumed parent. Many, but nt all f the same calculatins shw a similar effect fr K 2 O. This same relatinship has been discussed in mre detail elsewhere (Bryan & Mre, 1977; White & Bryan, 1977) in relatin t petrgenesis f the A Ivin data set frm the nrthern rift valley. Systematically high TiO 2, K 2 O, and 'magmaphile' trace elements have als been demnstrated in a variety f fractinatin calculatins based n basalt glass data frm the Deep Sea Drilling Prgram (Bryan et al, 1976). In cntrast, these systematic misfits are nt encuntered when the rati f increase in / t TiO 2 is abut 1-5 (Tables Hand 15; Fig. 11). TABLE 16 Least-squares test f crystallizatin in basalt glasses frm AII-77 statin 76 SiO 2 TiO 2 A1 2 O 3 K 2 O Parent calculated Parent bserved Variable Residual Plagiclase 1 Olivine 2 Pyrxene 3 Weight fractin = - U- lt>4 Parent is average f lw-tio 2 glasses (61, 66, and 71); inferred residual liquid is average f high-tio 2 glasses (53, 59 and 60). 1 Cl. 11, Table 8. 2 Cl. IB, Table Average f cls. 7 A, 7B, Table 9.

29 BASALT GLASSES FROM THE FAMOUS AREA 321 O x) e - Fe Enriched Suites 80 8> w 9 a 0 0 0) 8 SiO 2 Aip 3 Nap Kp &? c ' 9 0 y 0 TiO 2 Fi. 10. Graphical summary f residuals fr seven least-squares slutins relating irn-enriched residual liquids t lw-irn parents. Rati f increase f / t TiO 2 is abut 1-5 fr each pair. A perfect fit (zer residuals fr all xides) wuld plt n central line. Nte tight clustering and very small residuals fr K 2 O and ; rather large dispersin f residuals fr and. The calculatins d reveal that livine must play a very minr rle in prducing the variatin trends bserved within the basalt liquids, cntrary t what wuld be expected frm the data prjectins in Fig. 2 and Fig. 5. Fr mst calculatins livine is assigned very small psitive r negative weight fractins which in mst cases d nt differ significantly frm zer. The lw-livine glass average (Table 11) als has generally prved t be a mre tractable candidate parent liquid than the high-livine glass. Fr example, substituting the high-livine liquid fr the lwlivine liquid parent in the calculatin f Table 14 gives a sum f squares f 1-79, and this can be imprved nly t a sum f squares f 0-53, using a mre irn-rich livine, still a significantly wrse result than that shwn in Table 14. Attempts t relate the lw-livine basalt liquid t the high-livine basalt liquid encunters similar prblems, with an especially big misfit fr (Fig. 11). SiO 2 Al 2 q, Nap Kp TiOj Fi. 11. Graphical summary f residuals fr eight least-squares slutins relating titanium-enriched residual liquids t lw-titanium parents. Rati f increase f / t TiO 2 is «1 0 fr each pair. Nte cnsistently psitive residuals fr K 2 O and TiO 2, and tendency t rather large negative residuals fr,, and. Slid circles are residuals fr a slutin relating the lw-livine average as residual t the high-livine average as parent. Triangles represent tw slutins fr Alvin data frm the nrthern valley, with flank magma as residual and Mt. Plut central magma as parent Cmpare t Bryan & Mre, 1977, Fig. 15.

30 322 W. B. BRYAN The calculatin in Table 16 is an attempt t relate lw-tio 2 and high-tio 2 glasses recvered in a single dredge haul frm All 77 statin 76 in the suthern rift valley. The lw-tio 2 glass is als a lw-livine type and was included in the average given in Table 11. As already nted, it shws the systematic misfit fr TiO 2 bserved in ther calculatins invlving large differences in TiO 2 cntent between inferred parent and residual liquids. Apparently, clse spatial relatins between basalts des nt guarantee a simple genetic relatin between them. TABLE 17 Least-squares test f crystallizatin relating DSDP 37 and FAMOUS samples SiO 2 A1 2 O 3 K 2 O TiO 2 Parent calculated Parent bserved Variable Glass 332A Olivine F,, Pyrxene Plagiclase An,, IR! = Weight fractin Parent is average lw-livine basalt (Table 11), residual liquid is average high-fe glass frm site 332A (Table 6). TABLE 18 Least-squares test f crystallizatin relating DSDP 37 and FAMOUS samples SiO 2 A1A K 2 0 TiOj Parent calculated Parent bserved Variable Glass 332A Olivine F,, Pyrxene Plagiclase An,, y Weight fractin n ni U'JJI Parent is average high-livine basalt (Table 11), residual liquid is average high-fe glass frm site 332A (Table 6). Tables 17 and 18 shw calculatins relating irn-enriched glass frm DSDP site 332 A t the lw- and high-livine candidate parent liquids frm the FAMOUS area. Thefitis gd fr mst xides and it is nly slightly better fr the lw-livine parent than fr the high-livine parent. In particular, the fit is gd nt nly fr K 2 O and TiO^ but als fr mst trace elements (Bryan & Thmpsn, 1977), in agreement with ther calculatins where the change in / is relatively large cmpared t TiO 2. Even the high-livine parent requires crystallizatin f mre pyrxene than livine.

31 BASALT GLASSES FROM THE FAMOUS AREA 323 Overall, the calculatins shw the need fr great cautin in deducing genetic relatins frm graphic prjectins r small subsets f the data. Simple crystal-melt equilibria can accunt fr mst majr element differences in all slutins. Parentresidual pairs which require a rate f increase f / t TiO 2 f abut 1-5 t 1, rughly fllwing the shrt bundaries f the parallelgram-shaped data field f Fig. 3, are largely free f the systematic misfits fr TiO 2 and K 2 O. The dispersin f / relative t TiO 2 can be visualized as a result f high-level fractinatin superimpsed n a range f parent liquids arranged alng the lwer bundary f the data field, as utlined in Fig <D i TiO 2 Fi. 12. Summary f inferred variatin paths f basalt liquids in /TiO 2 space. Stippled arrw represents pssible parent magmas with variable TiO 2 enrichment, resulting frm variable mantle melting, diffusin r mixing in intermediate-level chambers, r ther undefined prcesses. Slid arrws are cmpsitin trends prduced by small degrees (abut 20 per cent) f crystallizatin in high-level vents r chambers. Pyrxene dminates livine as the mafic phase in all least-squares slutins. This reflects the fact that mst f the parent liquids are lw-livine basalts, and it must be recgnized that the implied cmpsitin vectrs fr these slutins in Fig. 5 are nt alng the ctectic parallel t the plane f the paper. The mst imprtant cmpsitinal changes during high-level crystallizatin invlve variatins in the cmpsitin f plagiclase and pyrxene, rather than changes in the relative prprtins f these phases. Thus, the ttal variatin during crystallizatin is multidimensinal in cmpsitin space, and is nt easily represented in simple twdimensinal diagrams. Bender et al. (1978) nte that a small increase in vlatile cntent wuld tend t expand the field f pyrxene stability defined by their experiments. Bryan & Mre (1977) shwed a 2 t 3-fld increase in H 2 O frm the relatively unfractinated central valley lavas t the mre fractinated flank lavas. Pssibly, accumulatin f vlatile cmpnents during crystallizatin causes earlier appearance f pyrxene, as cmpared t a ttally dry system. This effect wuld be enhanced by lcal

32 324 W. B. BRYAN accumulatin f vlatiles in cupla znes r by vlatiles acquired frm interactin with wall rcks. Unlike livine and plagiclase, pyrxene crystallized under these cnditins wuld tend t be resrbed in cntact with htter, dryer magma except at relatively high pressures. This may accunt fr the absence f pyrxene cumulates in the FAMOUS basalts. An imprtant implicatin f the experimental data f Bender et al. (1978) is that either high-pressure fractinatin f pyrxene, r resrptin f pyrxene cmbined with crystallizatin f livine and plagiclase, may play an imprtant rle in the genesis f sme f the variants in basalt cmpsitins. Sme f these effects may be t subtle t deduce frm majr element data alne, and mre cmprehensive trace element and istpe data will be required t test these alternative mdels. The mst fundamental prblems f magma genesis amng FAMOUS basalts thus relate t the rigin and inter-relatinships f the varius candidate parent liquids, shwn schematically by the stippled arrw in Fig. 12. These liquids are characterized by rather large increases in TiO 2 relative t /, and represent an extensin f the variatin trends f the submersible data set. The enrichments in TiO 2 and ther incmpatible elements exceed thse which can be attained with simple single-stage crystallizatin by a factr f tw r mre. At present, a number f prcesses appear pssible, and include variable degrees f partial melting, (White & Bryan, 1977; Langmuir et al, 1977); diffusin and vlatile transfer (Bryan & Mre, 1977); and multiple episdes f injectin and crystallizatin in an pen chamber (O'Hara, 1977; Bryan et al., 1979). A cmbinatin f these prcesses is likely t be required t explain all variatins. The K 2 O-enriched glasses frm fracture zne B may represent still anther mantle-derived magma type representing a different surce r degree f melting, and may represent a distinctive type f liquid erupted within the fracture zne, as cmpared t ther basalt tectnically transprted int it. Similar glasses frm fracture znes have been dcumented by Melsn et al., N attempt has been made t mdel the genesis f these liquids here. Prper understanding f their genesis undubtedly will require supprting trace element and istpic data. A mre cmplete discussin f fracture zne lithlgies, their gechemistry, and petrgenesis, is being prepared fr separate publicatin. ACKNOWLEDGEMENTS Many individuals have participated in the cruises t the FAMOUS area and have assisted in sample cllectin and analysis. Dr. J. R. Heirtzler served as prject chief scientist and als as chief scientist n Knrr Cruise 42, n which the authr was assisted in sample cllectin and curating by T. Atwater, P. Jhnsn, and K. Sullivan. Dr. J. D. Phillips served with the authr as c-chief scientist n Atlantis II Cruises 73 and 77. A. Driscll, C. Grant, P. Jezek and T. Stetsn assisted in the dredging prgram n these cruises. M. L. Adams, J. Guertler and P. Jezek assisted with petrgraphic studies, micrprbe analyses and calculatins. R. Hekinian prvided samples frm the Charct and Archimede cllectins. I am especially grateful t J. G. Mre, R. Hekinian, and ther members f the French and U.S. scientific grups fr many helpful discussins ffieldand petrgraphic relatins f