Comagmatic A-Type Granophyre and Rhyolite from the Alid Volcanic Center , Eritrea , Northeast Africa

vermicular to cuneiform (Barker, 1970). Typically, the Granophyric blocks within late-Pleistocene pyroclastic flow ejecta from the Alid volcanic center, northeast Africa, are the rapidly groundmass feldspars radiate off pre-existing feldspar crystallized, intrusive equivalent of pumice from the pyroclastic flow. phenocrysts, with which they are in optical continuity Phenocryst compositions and geochemical characteristics of the (Dunham, 1965). Some workers prefer the term pumice and granophyre are virtually identical. Silicate melt inclusions ‘micrographic’ for highly regular, interlocking arand other geochemical and geological constraints reveal those processes rangements of quartz and feldspar, as in the coarser leading to development of the granophyric texture. Rhyolitic (Agraphic granite (Fenn, 1986; Lentz & Fowler, 1992). type) magma with ~2·6 wt % dissolved H2O and a temperature Though Schloemer (1964) has shown that some graphic near 870°C was intruded to within 2–4 km of the surface, causing textures can grow by replacement [see also Augustithis deformation and structural doming of shallow marine and subaerial (1973)], most workers agree that phenocryst-bearing strata. Eruptions of crystal-poor rhyolite from this shallow magma granophyres typically form by rapid and simultaneous chamber caused degassing, which forced undercooling and consequent crystallization of quartz and feldspar from a melt (Smith, granophyric crystallization of some of the magma remaining in the 1974). Such crystallization is generally believed to be due intrusion. The most recent eruption from Alid excavated the cryto pronounced undercooling of the silicate liquid (Vogt, stallized granitic wall of the magma chamber, bringing the grano1930; Dunham, 1965); not necessarily at eutectic temphyric clasts to the surface. peratures or compositions (Fenn, 1986; London et al., 1989; Lentz & Fowler, 1992). Granophyric textures are common in epizonal granitic bodies, particularly those associated with volcanic rocks (Buddington, 1959; Dunham, 1965). On occasion, they erupt as comagmatic ejecta in pyroclastic deposits (e.g.


INTRODUCTION
correlating these erupted blocks with related volcanic rocks and by using the geologic relationships available, Granophyric intergrowths are among the most impressive we provide compelling evidence for the pressure, temand beautiful of rock textures.Commonly, they consist perature and geologic conditions under which the granoof fine-grained intergrowths of quartz and feldspar arranged in patterns that can vary from irregular to phyric groundmass formed.

BACKGROUND GEOLOGY Alid volcanic center and the Danakil Depression
The Alid volcanic center, Eritrea (Marini, 1938a(Marini, , 1938b)), is located along the axis of the Danakil Depression, the graben trace of a crustal spreading center that radiates NNW from a plate-tectonic triple junction within a complexly rifted and faulted basaltic lowland called the Afar Triangle (Fig. 1).The Danakil Depression is a subaerial segment of the spreading system that is opening to form the Red Sea.It lies near or below sea level for much of its extent and is surrounded by the Danakil Alps to the east and the Eritrean plateau or highland to the west, which rises to elevations of 2000-3000 m above sea level (masl).Both of these bordering regions are underlain primarily by Precambrian gneisses, granites and schists, locally covered by mid-Tertiary basalt.Much of the Afar lowland is covered with Pliocene and Quaternary lavas (CNR-CNRS, 1973).Erta Ale, one of the most active volcanoes in the world, lies about 100 km SE of Alid.
The Alid volcanic center is an elliptical structural dome formed during uplift caused by shallow intrusion of rhyolitic magma, some of which was erupted (Clynne et al., 1996a(Clynne et al., , 1996b)).The major axis of Alid is 7 km, elongate ENE-WSW, perpendicular to the trend of the graben (Fig. 2).The minor axis is about 5 km long, parallel to the graben.Alid rises ~700 m above a field of Quaternary basaltic lava that laps unconformably against the north and south flanks of the mountain.
The oldest rocks to crop out on Alid are Precambrian quartz-mica and kyanite schists found within a deep canyon, Sillalo ´, that drains the east side of the mountain.Overlying this basement rock is the 'sedimentary sequence', consisting of marine siltstone and sandstone, pillow basalt, subaerial basalt, anhydrite beds and fossiliferous limestone, all interpreted to be Pleistocene in age (Clynne et al., 1996a;Duffield et al., 1997).The Modified from Barberi & Varet (1977).'sedimentary sequence' (ss) is found on all parts of the mountain, and dips radially away from its geographic center.Stratigraphically above the 'sedimentary et al., 1997).Similar rhyolite was erupted as pyroclastic sequence' is the 'lava shell' (ls), which consists of basalt flows from the summit region and today forms blankets and basaltic andesite lava flows, pyroclastic deposits, of tephra up to 15 m thick in the summit basin and the amphibole-bearing rhyolite domes, and intercalated eowest flank of Alid ('pf unit' of Fig. 2).The final phase of lian sands.The shell rocks form most of the slopes of the this eruption, a lava dome, is dated at 23•5±1•9 ka mountain, with dips locally as steep as 65°.Such steep ['pxrhy' of Duffield et al., 1997)].All post-doming rhyolites dip slopes must have formed subsequent to emplacement have mineralogy (anorthoclase and Fe-rich clinoof these fluid lava flows, as they are far too steep to be pyroxene) and trace-element geochemistry typical of Aoriginal (Clynne et al., 1996a(Clynne et al., , 1996b)).
type granite systems found in similar non-arc geological Structural doming caused considerable distension of settings (Bowden, 1974;Whalen et al., 1987;Eby, 1990; the shell and sedimentary units, and effected landsliding Coleman et al., 1992).Similar rhyolites and granites, which are related to earlier periods of Red Sea spreading, and collapse of the central region of the mountain, resulting in a basin-like depression.Clinopyroxene-bear-have been described in Saudi Arabia and Yemen by Coleman et al. (1992) and Chazot & Bertrand (1995), ing rhyolite lavas, dated at 33•5±4•6 ka, were erupted subsequent to the initiation of structural doming (Duffield respectively.We infer that a body of granitic magma intruded to a summit basin and on Alid's north flank (Clynne et al., 1996a(Clynne et al., , 1996b;;Lowenstern et al., 1997).shallow level, caused structural doming, and erupted as lava flows and pyroclastic deposits.The lack of postpyroclastic-flow, basaltic vents on Alid, even though such vents are common to the north and south, may signify

Blocks within the pyroxene rhyolite tephra
that magma still resides beneath Alid and impedes the deposits ascent of mafic magma (Clynne et al., 1996a(Clynne et al., , 1996b)).Whether the magma is still partially molten or completely Lithic blocks are abundant throughout the thick pycrystallized, it continues to release enough heat to drive roclastic flow deposits ('pf unit') in the summit basin and west flanks of Alid: they consist (in order of decreasing a geothermal system, as fumaroles are found in the abundance) of amphibole rhyolite, lava-shell basalts, and view and contains euhedral rods and triangles of quartz Precambrian schists and granites.In the upper 1-2 m (Fig. 3b).These textures are intermediate between the of the deposit, blocks of a fine-grained, clinopyroxenecoarse-grained graphic intergrowths grown from hydrous bearing granite are markedly different in macroscopic pegmatitic melts (London, 1996) and the crude, poorly texture from Precambrian granite clasts, which are ancrystallized granophyres of intracaldera welded tuffs, gular, indurated and contain hydrous minerals.In conwhich are highly undercooled and degassed.trast, the pyroxene granite clasts are friable, low density Thick-sections, prepared as slabs of 100-200 m thick-(~2•45±0•07 g/cm 3 ), rounded and reach up to 1 m in ness, reveal details that are missed in 30 m thin sections.diameter.They were briefly described by Beyth (1994, Silicate melt inclusions are typically 10-20% crystalline 1996), who called them 'microgranite'.Clynne et al. andoccur in both feldspar andquartz phenocrysts in the (1996a, 1996b) used the term 'pyroxene granite', and granophyric block.Small, 2-20 m, partly crystallized below, we use the term 'granophyric blocks'.The blocks, melt inclusions also are present within the granophyric as discussed below, are interpreted to be comagmatic stringers.In addition, small faceted, vapor-rich fluid with the pumice, and represent part of the pyroxene inclusions (predominantly a vapor bubble plus 10-40% rhyolite intrusion that caused deformation at Alid.Apliquid) are located in both the phenocryst phases and parently, their granophyric textures formed during an groundmass stringers in the granophyric blocks (Fig. 3f ).early eruption from this shallow intrusion.They were These fluid inclusions appear to represent primary magsubsequently disaggregated from the magma chamber matic vapors, trapped during original crystal growth.wall and erupted during the most recent pyroclastic Individual vapor-rich inclusions are common at the tips episode, at ~23•5 ka.This study presents chemical and of feldspar stringers in the groundmass (Fig. 3g), indicating petrographic data from a single large (~1 m diameter) that crystallization was accompanied by vapor exsolution granophyric block, inferred to be representative of the (second boiling).Void space within the granophyre is many others found in the pyroclastic flow ejecta.
commonly seen along grain boundaries, which may have served as channelways for escaping vapor.Vapor-rich inclusions were not observed in phenocrysts from the volcanic 'pf unit'.

PETROGRAPHY Granophyric textures
Intergrowths of quartz and alkali feldspar form the Mineral compositions groundmass of the granophyric block (Fig. 3a and b), and All rhyolites erupted within the last 33•5 kyr are clinomake up 50-70 vol.% of the rock.The elongate alkali pyroxene bearing and contain alkali feldspar, magnetite, feldspar crystals, or stringers, nucleate directly from the zircon, apatite and pyrrhotite phenocrysts.Rhyolitic pre-existing feldspar phenocrysts [Fig.3c and d; Texture pumice lumps of the 'pf unit' also contain minor quartz.e, 'radiating fringe', in figs 2-7 of Smith (1974)].Often, Some pyroxene rhyolites on Alid's east flank (frhy2 and the stringers coarsen as they emanate from the seed -3 of Fig. 2) contain minor fayalite, as well as pyroxene.crystal and form fan-like splays.Groups of stringers that Hydrous minerals are absent in all extrusive rhyolites, emanate from the same seed phenocryst are always in except in the much older (~210 ka) amphibole rhyolites.optical continuity with each other, as well as the seed.
The compositions of feldspar phenocrysts from the 'pf In effect, the feldspar groundmass is an extension of the unit' pumice lumps range from K-poor anorthoclase original phenocryst, and grew by cellular, quasi-skeletal (An 10 Ab 73 Or 17 ) to Ca-poor sodic sanidine (An 2 Ab 60 Or 38 ) growth rather than planar crystallization (Fenn, 1986).
(Table 1).Maps of feldspar crystals, made by scanning the The groundmass quartz never connects directly to phenocrysts beneath the beam of an electron microprobe, either feldspar or quartz phenocrysts (Fig 3e).Instead, demonstrate that many crystals contain a core of K-poor the quartz apparently nucleated on the feldspar stringers anorthoclase that contains rounded edges and numerous themselves.Growth of the feldspar may have induced embayments (Fig. 4).Subsequent growth of a rim of sodic local supersaturation with quartz, which then fills in the sanidine resulted in euhedral phenocrysts with K-rich interstices left between the growing feldspars (Lentz & rims (Figs 4 and 5).BaO concentrations are high (0•4-0•6 Fowler, 1992).Though the quartz is discontinuous in wt %) in the anorthoclase cores, and decline to about plan view, filling in the space between the stringers of 0•15 wt % or less in the potassic rims (Fig. 6).No reverse feldspar, it shows optical continuity over large regions zoning of feldspar was observed for either CaO or BaO, (up to 1 mm) demonstrating that the quartz is continuous possibly signifying that the resorption event that caused in three dimensions as relatively large individual crystals.
rounding of the feldspar cores was not caused by mixing Many areas of the groundmass have micrographic texof the melt with a less evolved, more Ca-or Ba-rich endture, where the granophyre is extremely uniform in pattern, sometimes cuneiform, relatively isotropic in plan member.Instead, the resorption may have been caused by simple heating of the rhyolite by mafic magma, without sanidine (Table 1).Potassium-rich rims range up to An 1 Ab 55 Or 44 and stringers of feldspar in the granophyric chemical mixing.
As in the pumice, feldspar phenocrysts in the grano-groundmass are even slightly more K enriched (up to An 1 Ab 52 Or 47 ).Interestingly, the K/Na of the feldspar phyre are zoned from K-poor anorthoclase to sodic •2 * For whole rocks, Fe calculated as ferric iron.For melt inclusions, Fe calculated as ferrous iron.†Whole-rock major-element analyses and melt inclusion average are recalculated to 100% normalized.Original total (nonnormalized) listed in total row.Values for melt inclusions represent mean of 19 electron microprobe analyses.Whole rocks represent individual samples analyzed by X-ray fluorescence (XRF) (major-elements+some trace) and inductively coupled plasma mass spectrometry (ICPMS).Mineral analyses by electron microprobe.Ternary components in pyroxene (and Fe 2 O 3 concentration) according to methods of Papike et al. (1974).Except for pyroxene analyses, FeO concentrations assume all Fe as ferrous.Biotite end-member components calculated with code provided by F. Mun ˜oz (personal communication, 1988).Groundmass feldspar (gr) is from a micrographic intergrowth in the granophyric block. 18O reported in units in per mil deviation relative to VSMOW.(See Appendix for details on all analytical techniques.)n.d.not determined.stringers (Or 40-45 ) is similar to that of the glass matrix of Alid, range from ferrosalite to ferroaugite in composition (Table 1).Cores are typically close to Wo 45 En 27 Fs 28 , the pumice.The K/Na ratio (by weight) lies between 1•03 and 1•20 for silicate melt inclusions and pumiceous mantled by more Fe-rich pyroxene.Rims reach Wo 42 En 18 Fs 40 .Rhyolites on the east flank of Alid ('frhy2'), matrix: stringers in the granophyre have similar ratios (Table 1).
which were erupted before the 'pf unit', contain more Fe-rich pyroxenes that reach near-end-member hed-Pyroxenes in the 'pf unit' pumice lumps, as well as several other 'frhy' rhyolites from the west flank of enbergite in composition.Pyroxene rims are slightly more Fe enriched in the granophyre than in the host pumice, constrain the depth and temperature of the magma reaching Wo 45 En 10 Fs 45 , though they are otherwise similar chamber that fed the eruptions on Alid, and to learn about in both rock types.
the conditions related to formation of the granophyric Besides their groundmasses, the only obvious mintextures.eralogical differences between the granophyric blocks and the rhyolite pumice are the following: (1) the presence of minor amounts of fluorite and chevkinite at the margins of phenocrysts in the granophyre; (2) partial replacement

Dissolved volatile concentrations
of clinopyroxene in the granophyre by Fe-rich biotite The euhedral quartz phenocrysts in both the 'pf unit' (Annite 59 Phlogopite 41 Siderophyllite 0 ), apparently before pumice lumps and granophyre block contain numerous the crystallization of the groundmass; (3) minor amounts silicate melt inclusions (MI; Fig. 7); the following inof clay and chlorite in cracks and small miarolitic cavities.
formation is based on analyses of silicate melt inclusions The analyzed biotites (Table 1) are uniform in comchosen from quartz separates from pumice lumps of the position, not varying more than a few mol %, and are 'pf unit'.Melt inclusions from the granophyric block more Fe rich than biotites in the amphibole-bearing were not studied because they are more crystallized than rhyolites (Annite 31 Phlogopite 67 Siderophyllite 2 ) that were MI in the 'pf unit' and all large (>25 m) inclusions in erupted at ~210 ka, long before deformation of Alid.
the granophyric sample appeared to have leaked.Many The annite-rich biotites are characteristic of micas found studies have shown that MI provide a record of the in A-type granites (Abdel-Rahman, 1994).

TRACE-ELEMENT AND ISOTOPE
the 'pf unit' pumice are <100 m in diameter, and most are <50 m.They are mostly glass, but contain ~5% GEOCHEMISTRY crystals; wiry Fe-Ti oxides, pyroxenes, biotite(?)and The granophyric block and the pumice from the pyroother (unidentified) minerals.The MI contain one or clastic flow are nearly identical in major-element, tracemore small vapor bubbles, generally <10 m in diameter, element and isotopic compositions (Table 1).Nearly all which typically make up <2 vol % of the inclusions.By analyzed trace elements are the same, within 5% relative, heating the MI-bearing quartz grains in a muffle furnace in the two rocks.Most of the other pyroxene rhyolites at 925°C for 4 h, the inclusions were remelted and share these similarities in major-, trace-element and isohomogenized to an optically homogeneous glass.Electron topic compositions with the 'pf unit' pumice and granomicroprobe analyses of different regions in the inclusions phyric blocks, but have more restricted phenocryst revealed no major-element heterogeneities in the hocompositions than those discussed above.The granomogenized inclusions.phyric block appears to be the crystallized (closed-system) Fourier-transform infrared (FTIR) spectroscopic anaequivalent of the 'pf unit' pumice.
lyses of six remelted, doubly polished MI ( (mean 2•56 wt %) using the 3570 cm -1 peak.Quanment compositions of the basement granites (e.g.Rb= tification with the near-IR peaks yielded an average of 214 ppm; Ba=148 ppm; U=1•36 ppm) are also unlike 2•70 wt %.Water was also estimated by the summation the granophyric blocks.It is clear that the friable, granodeficit of electron microprobe analyses, assuming that phyric blocks are distinct texturally, mineralogically, and the difference between the analysis total and 100% was chemically from the Precambrian basement lithic madue to the presence of water (Devine et al., 1995).Using terial that was ejected during eruption of the 'pf unit'.
this technique, the mean of the six MI was 2•62 wt %, consistent with the FTIR results.The mean deficit for 19 inclusions, 13 not analyzed by FTIR spectroscopy,

THE ALID MAGMA RESERVOIR
was 2•8 wt %.Assuming that the MI compositions represent those of the silicate melt at the time of inclusion The mineralogical and compositional similarities between entrapment (Roedder, 1984), the rhyolitic magma apthe pumice and granophyric blocks demonstrate that parently contained lower dissolved H 2 O concentrations these two rocks share a similar origin.We can, therefore, than most other rhyolitic systems discussed in the lituse the 'pf unit' pumice to infer characteristics of the erature ( Johnson et al., 1994;Lowenstern, 1995).magma chamber at the time of granophyre crystallization.
Chlorine concentrations in MI were very consistent, The following sections utilize silicate melt inclusions, trace elements, and other geological information to help and averaged 3800 ppm (2 =400 ppm, n=19).Fluorine •6 * The inclusion-bearing quartz grains were selected individually from crystal separates.†See Appendix for analytical details on FTIR spectroscopy.EPMA, electron microprobe (see text and Appendix).P sat (saturation pressure) and % CO 2 v (mol % CO 2 ) in vapor are discussed in text and were calculated with procedure outlined by Holloway & Blank (1994).Errors for FTIR results were calculated as discussed by Lowenstern et al. (1994), using propagation method of Bevington (1969).Reported errors for mean value row are 1 SD around the mean of six inclusions from six quartz grains found within a single pumice lump.
concentrations were more scattered, ranging from 1300 makes it likely that significant amounts of CO 2 were present in this magmatic system.Even today, fumaroles to 5200 ppm and averaging 3300 ppm (2 = 2400 ppm).Pumiceous matrix glass retained its chlorine, though venting from the summit and flanks of Alid are notably CO 2 rich, and have magmatic carbon isotope signatures fluorine was generally absent.This could indicate either preferential degassing of F during eruption, or al- (Lowenstern et al., 1997).Therefore, the low CO 2 concentrations in the melt do not necessitate low CO 2 in ternatively, leaching of the F during secular hydration of the pumice.the Alid magmatic system, but rather may be due to the low pressure at the time of MI entrapment (and thus the Dissolved carbon dioxide concentrations, also measured by FTIR spectroscopy, were 28-58 ppm for the six low solubility for CO 2 in the melt).As discussed below, the assumption that the H 2 O and CO 2 concentrations MI (Table 2).Many researchers have noted high CO 2 in A-type magmatic systems (Bailey & Macdonald, 1987).were controlled by melt-vapor equilibria allows one to calculate pressure of the magma at the time of inclusion Association of the Alid rhyolites with numerous younger and older basaltic units (Clynne et al., 1996a(Clynne et al., , 1996b)), entrapment.may therefore directly reflect volatile solubilities in a H 2 O-CO 2 -melt equilibria: implications for shallow magma chamber, at 2-4 km beneath the Alid magma depth structural dome.As demonstrated by Anderson et al. (1989), the concentrations of H 2 O and CO 2 in silicate MI can be used to calculate a minimum pressure at which the inclusion Magmatic temperatures was trapped, as well as the composition of the vapor that The presence of a single alkali feldspar is typical of would coexist with the melt at that pressure (P sat of hypersolvus igneous systems where temperature is high Table 2).Holloway & Blank (1994) outlined a procedure and water pressure low.However, it is difficult to quantify that calculates these parameters by using solubility models temperature simply by the presence of a single alkali for CO 2 and H 2 O in rhyolite along with a Redlichfeldspar, other than to say that temperature must be Kwong model for components of the vapor phase.Using above ~680°C (Tuttle & Bowen, 1958).Lack of ilmenite an estimated magma temperature of 875°C (see below), in the Alid rhyolites precludes use of Fe-Ti oxide geowe calculate saturation pressures between 380 and 925 thermometry to calculate pre-eruptive magma tembars, corresponding to depths of 1•4-3•4 km, assuming peratures.Similarly, these rhyolites contain only one a lithostatic pressure gradient and rock density of pyroxene, so that other common pyroxene geo-2700 kg/m 3 .The vapor in equilibrium with the silicate thermometers cannot be used.melt would be composed of ~10-18% CO 2 , the rest Elevated Zr concentrations in these rhyolites (Clynne being water.Other gas components not included in the et al., 1996a), hint at a high magma temperature.The calculation (HCl, HF, SO 2 , H 2 S) would further dilute the Zr geothermometer (Watson & Harrison, 1983) relates CO 2 component in the vapor, and permit the observed zircon solubility to melt composition and the temperature H 2 O and CO 2 concentrations in the melt at slightly of rhyolitic melts.The Alid rhyolites are all similar greater pressures.
in alkali to alumina ratios, and have a subaluminous The low calculated crystallization depths (1•4-3•4 km) composition.Whole-rock Zr concentrations, as deare consistent with a variety of geological and mintermined by X-ray fluorescence analysis, range from 362 eralogical constraints.First, Alid is a relatively small, to 414 ppm for all pyroxene rhyolites from Alid (Duffield steep-sided structural dome formed as a result of intrusion et al., 1997), including the aphyric units.Rhyolitic pumice of rhyolitic magma (Clynne et al., 1996a(Clynne et al., , 1996b)).It is from the 'pf unit' pumice and the granophyric block unlikely that such a feature could form by an intrusion at gave concentrations toward the top of this range.Asdepths much greater than 5 km.For example, laccoliths, suming that the accessory zircon crystals contain 65% which form by similar mechanisms, typically have dia-ZrO 2 , and that the molar (Na + K + 2Ca)/(Al × Si) is meters that are a direct function of their depth of inequal to 1•07, temperatures calculated with the aforetrusion, with the ratio being close to unity (Price & mentioned Zr concentrations yield temperatures between Cosgrove, 1990).Given the diameter of the mountain 860 and 880°C (Watson & Harrison, 1983).Reasonable itself (5-7 km), the intrusion might then have reached expected variations in melt peralkalinity or zircon com-<4 km below the surface before erosion, and today would positions would not affect the calculated temperatures be located 1500-3000 m below Alid's depressed summit more than 20°C.We also used the phosphorus (Duffield et al., 1996).
geothermometer of Harrison & Watson (1984), which Plausibly, the MI could have been trapped before yielded similar temperatures (~850°C) for the 'pf unit' ascent and shallow emplacement of the rhyolitic magma.pumice and granophyric blocks.Analytical uncertainty However, several lines of evidence are consistent with in P 2 O 5 analyses (±0•01 wt %) results in uncertainties inclusion entrapment after emplacement.The first rhyof ±20°C.Because the Zr and P 2 O 5 geothermometry is olites erupted after deformation are aphyric, and are based on analysis of whole rocks and not glass separates, strikingly similar in bulk composition to the later porthe calculated temperatures are maxima.The rhyolites phyritic rhyolites (Clynne et al., 1996a).Moreover, the do not contain abundant zircon crystals, but rocks with MI were trapped in small, euhedral quartz grains (Fig. 7a higher modal zircon may have erupted at somewhat and b) that apparently were the last phase to crystallize lower temperatures than units with fewer zircon grains from the melt.They do not appear to be long lived and and similar Zr concentrations.show no evidence for decompression-related resorption As mentioned above, crystals and bubbles in silicate (Whitney, 1988), which might occur if they had risen from melt inclusions were homogenized at temperatures significant depth.Moreover, quartz was not observed in around 925°C.Temperatures lower than 850°C were any of the earlier 'frhy' units.
insufficient to cause the crystals to melt, and this probably Finally, as discussed above, the low CO 2 in the MI indicates that the original entrapment temperature for would be surprising for an A-type rhyolite associated the inclusions was higher than 850°C, in accordance with the Zr geothermometer.with abundant rift basalts.The low volatile concentrations Oxygen fugacity is likely to be close to the quartz-would be expected for systems where diffusion cannot fayalite-magnetite (QFM) buffer, given that some of the keep pace with the rapid rate of crystal growth.Alid pyroxene rhyolites also contain fayalite, and those The most recent eruption (23•5±1•9 ka; Duffield et magmas, therefore, contained all three minerals of that al., 1997), from the summit basin, resulted in deposition of buffer.The presence of ferroaugite rather than fayalite, a thick blanket of pyroclastic flow materials.Granophyric though, permits slightly higher oxidation states, perhaps blocks within the upper part of this deposit must have up to a log unit above QFM.The low presumed oxygen been excavated from the crystalline carapace at the fugacity is consistent with the very low calculated ferric magma chamber walls or roof and were carried to the iron concentrations in the clinopyroxene.
surface by the erupting jet.

Fig. 2 .
Fig.2.Generalized geologic map of Alid volcanic center.Because the map was traced from lines on air photographs, the scale is approximate and varies somewhat across the map [modified fromClynne et al. (1996a)].

Fig. 3 .
Fig. 3. Photomicrographs of areas within granophyric block found as a lithic in the 'pf unit' at the Alid volcanic center.Figure parts (a)-(e) with crossed Nicols.(a) Micrographic groundmass surrounds feldspar phenocryst fragment.(b) Micrographic groundmass around clinopyroxene phenocryst.Quartz is white and feldspar is at extinction.(c, d) Radiating fringe of feldspar in groundmass is in optical continuity with phenocryst, from which it nucleated.(e) Quartz phenocryst is surrounded by feldspar groundmass.Quartz in groundmass does not appear to have nucleated directly on phenocryst, but is in optical continuity with it.(f ) Primary vapor-rich fluid inclusion in phenocrystic quartz.Liquid (l) is seen at the bottom right, the rest being vapor (v).(g) Primary vapor-rich inclusion at the tip of a quartz stringer (outlined for clarity) in the granophyric groundmass.

Fig. 4 .
Fig. 4. Maps of K distribution in alkali feldspars.Red represents highest concentrations, followed by yellow, green and blue.See Appendix for details.Scales as indicated in figure parts.(a) Feldspar in 'pf unit' pumice, surrounded by pumiceous glass.The feldspar core is a K-poor anorthoclase that has been resorbed.Embayments have been filled with more K-rich sodic sanidine.(b) Feldspar in granophyric block.Feldspar shows same general features as in (a).Feldspar stringers in granophyre are approximately the same composition as the K-rich rim of sodic sanidine.

Fig. 5 .
Fig. 5. Ternary diagram showing mol % of feldspar components (Or, Orthoclase; An, Anorthite; Ab, Albite) in feldspars from granophyric block (gr) and pumice from 'pf unit' (see Table 1 and Appendix).Feldspars in both rock types are similar, with cores of K-poor anorthoclase and rims of sodic sanidine.Stringers of feldspar in the granophyric groundmass reach values of about Or 50 .

Fig. 6 .
Fig. 6.Wt % BaO and CaO vs K 2 O for feldspars from the 'pf unit' and granophyric block (gr), as determined by electron microprobe.

Fig. 7 .
Fig. 7. Quartz phenocrysts in pumice of the 'pf unit' are euhedral (a) and contain abundant melt inclusions (b).Scale for (a), a scanning electron micrograph, is shown by bar representing 100 m.Quartz grain in (b) is 650 m in diameter.Photomicrograph taken with crossed Nicols.

Table 1 :
Chemical and isotopic analyses of pyroxenes, alkali feldspars, biotite, glass, and whole rocks from granophyric block (gr) and rhyolitic pumice from the 'pf unit'(pf), Alid volcanic center, Eritrea

Table 2 :
H 2 O and CO 2 contents of melt inclusions in quartz from 'pf unit' pumice