a summary of the geology and petrology of the sierra la primavera, jalisco, mexico

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 86, NO. BII, PAGES 10137-10152, NOVEMBER 10, 1981

A Summary of the Geology and Petrology of the Sierra La Primavera, Jalisco, Mexico GAIL A. MAHOOD

Department of Geology, School of Earth Sciences, Stanford University, Stanford, California 94305

The Sierra La Primavera, near Guadalajara, Mexico, is a Late Pleistocene rhyolitic center consisting of lava flows and domes, ash flow tuff, air fall pumice, and caldera lake sediments. All eruptive units are high-silica rhyolites, but systematic compositional differences correlate with age and eruptive mode. The earliest lavas erupted approximately 145,000 years ago and were followed approximately 95,000 years ago by the eruption of about 20 kra 3 of magma as ash flows that form the Tala Tuff. The Tala Tuff is zoned from a mildly peralkaline first-erupted portion enriched in Na, Rb, Cs, CI, F, Zn, Y, Zr, Nb, Sb, HREE, Hf, Ta, Pb, Th, and U to a metaluminous last-erupted part enriched in K, LREE, Sc, and Ti; AI, Ca, Mg, Mn, Fe, and Eu are constant within analytical errors. Collapse of the roof zone of the magma chamber led to the formation of a shallow 1 l-km-diameter caldera in which lake sediments began to col- lect. The earliest postcaldera lava, the south-central dome, is nearly identical to the last-erupted portion of the Tala Tuff, whereas the slightly younger north-central dome is chemically transitional from the south-central dome to later, more marie, ring domes. This sequence of ash flow tuff and domes represents the tapping of progressively deeper levels of a zoned magma chamber 95,000 + 5,000 years ago. Sedi- mentation continued and a period of volcanic quiescence was marked by the deposition of some 30 m of fine-grained ashy sediments. Approximately 75,000 years ago a new group of ring domes erupted at the southern margin of the ltke. These domes are lapped by only 10-20 m of sediments as uplift resulting from renewed insurgence of magma brought an end to the lake. This uplift culminated in the eruption, beginning approximately 60,000 years ago, of aphyric lavas along a southern arc. The youngest of these lavas erupted approximately 30,000 years ago. The lavas that erupted 75,000, 60,000, and 30,000 years ago became decteasingly peralkaline and progressively enriched only in Si, Rb, Cs, and possibly U with time. They represent successive eruption of the uppermost magma in the postcaldera magma chamber. Eruptive units of La Primavera are either aphyric or contain up to 15% phenocrysts of sodic sanidine _> quartz >> ferrohedenbergite > fayalite > ilmenite + titanomagnetite. Major element compositions of sanidine, clinopyroxene, and fayalite phenocrysts vary only slightly between eruptive groups, but the concentrations of many trace elements change by factors of 5-10. This is reflected in phenocryst/glass partition coefficients that differ by factors of up to 20 between successively erupted units. Because the major element compositions of the phenocrysts and the pressure, temperature, and fo2 of the magmas were es- sentially constant, the large variations in partitioning behavior are thought to result from small changes in bulk composition of the melt. Crystal settling and incremental partial melting are by themselves in- capable of producing either the chemical gradients within the Tala Tuff magma chamber or the trends with time in the post-95,000-year lavas. Rather, diffusional processes in the silicate liquid are thought to have been the dominant differentiation mechanisms. The zonation in the Tala Tuff is attributed to transport of trace metals as volatile complexes within a thermal and gravitational gradient in a volatile-rich but water-undersaturated magma. The evolution of the postcaldera lavas with time is thought to involve the diffusive emigration of trace elements from a relatively dry magma as a decreasing proportion of network modifiers and/or a decreasing concentration of complexing ligands progressively reduced octahedral site availability in the silicate melt.

INTRODUCTION

One way to understand the differentiation processes operating in high-level magma chambers that solidify as granitic plutons is to study material erupted from such chambers. Ash flow tuffs are rapidly quenched voluminous samples of magma chambers; as such they have made possible the char- acterization of chemical and thermal gradients within the upper portions of silicic magma chambers just prior to eruption [Lipman et al., 1966; Smith and Bailey, 1966; Hildreth, 1977, 1979; Smith, 1979; Ritchie, 1979; Hildreth et al., 1980]. Com- plementary to studies of ash flow tuffs, which are inverted rec- ords of magma chambers at single points in time, is the infor- mation contained in a sequence of eruptive units, which record the chemical evolution of the upper portions of a silicic magma chamber through time. Making the reasonable as- sumption that a lava flow or ash flow taps the most differenti- ated uppermost magma in the system at a particular moment, a sequence of eruptive units provides progress reports on the differentiation mechanisms operating at depth. The main put-

Copyright 1981 by the American Geophysical Union.

pose of this study of the Sierra La Primavera, therefore, has been to trace the chemical evolution of a rhyolitic complex through time, interpreting it as the periodic sampling of an evolving magma chamber.

To determine the eruptive history, the Sierra La Primavera, located on the western outskirts of Guadalajara, Jalisco, Mex- ico (Figure 1), was mapped at a scale of 1'25,000. Eruptive units were classified into groups based on stratigraphic relations, and the eruptive sequence was calibrated with over 50 K-At dates performed at University of California, Berkeley [Mahood, 1980a]. This paper summarizes the geological and chemical evolution of the Primavera system; the geology of the Sierra La Primavera is described in more detail by Ma- hood [1980a, 1980b], whereas the petrology is discussed more fully by Mahood [1980a, 1981].

REGIONAL SETTING

The Sierra La Primavera lies at the intersection of the two major Cenozoic volcanic provinces of Mexico (Figure 1). Its mildly peralkaline rocks stand in marked chemical contrast to both the andesitic stratovolcanoes and basaltic cinder cones

Paper number IB0529. 0148-0227/81/001B-0529501.00

10137

ACER ASPIREResaltado

10138 MAHOOD: SUMMARY OF THE GEOLOGY AND PETROLOGY OF LA PRIMAVERA

0 km 50 I I VOLCAN DE

COLIMA

:Fig. . Rcgiona! location map (modified mm Mahood [980b] I. This map shows the location of the Sierra La Primavera (SLP) relative to the majo andesitic stratovolcanoes of the western portion of the Pliocene-Recent Mexican Neovolcanic Belt (stippled), the Ter- tiary Cordilleran Province (diagonal ruling), regional faulting, and the cities of Guadalajara (G) and M6xico (M). The Sierra La Prima- vera lies along a line parallel to the northwest trend of the western portion of the Mexican Neovolcanic Belt defined by Volc,n Tequila and ten basaltic andesite cinder cones (small circles southeast of Gua- dalajara).

that make up the bulk of the Pliocene to Recent Mexican Neovolcanic Belt and to the largely metaluminous silicic lavas and ash flows that dominate the Tertiary Cordilleran Province. Normal faulting has broken the area into a series of grabens, one of the largest being occupied by Lake Chapala. Volc,n Colima, south of La Primavera, sits within a large north-south graben and is on line with the east-west trend of the eastern portion of the Mexican Neovolcanic Belt. The intersection of extensional features near the Sierra La Primavera may have focused marie volcanism in the area, leading to large-scale melting of crustal material to give rise to Primavera magmas.

Little is known of the prevolcanic basement that might be partially melted to give rise to the Primavera magmas. Despite the nearly 1000 m of relief in the nearby canyon of the Rio Grande de Santiago, the oldest rocks exposed are Miocene ash flow tuffs [Watkins et al., 1971]. Much of the area now occupied by the Sierra La Primavera appears to be immediately underlain by andesitic and basaltic lava flows, which are locally exposed in canyons within the Sierra and at its margins and are common as accidental debris in air fall pumice deposits.

All recognized faults in the Sierra La Primavera are related either to caldera collapse or to uplift thought to be a result of renewed insurgence of magma. All faults begin and end within the environs of the complex; none extend beyond the Sierra to offset older rocks. This suggests that, while the loca-

tion of the Sierra La Primavera may be controlled by regional tectonics, faulting within the complex is a product of local hy- drostatic adjustments to changes in the level of the underlying magma chamber. This lack of regional faults in the Sierra La Primavera is circumstantial evidence for an underlying magma chamber that cannot sustain brittle fracture. SUMMARY OF THE GEOLOGIC HISTORY AND PETROLOGY

A brief history of the Sierra La Primavera follows and is summarized in Table 1. The earliest lavas erupted about 145,000-100,000 years ago. They were followed by eruption, approximately 95,000 years ago, of about 20 km 3 of magma as ash flows that formed the Tala Tuff. Collapse of the roof zone of the magma chamber upon eruption of the Tala Tuff produced an 11-km-diameter caldera that soon filled with water. Onset of lacustrine sedimentation was followed rapidly by eruption of two domes in the center of the lake (the south-central and north-central domes, which form the lower portions of Mesa E1 Nejahuete (I) and Cerro Alto (H), respectively, in Figure 2). The base of the south-central dome occurs at the same position within the sediments as the 'giant pumice horizon', an important stratigraphic marker, whereas the north- central dome overlies the giant pumice horizon. Ring domes representing some 5 km 3 of porphyritic magma then erupted along two concentric arcs: one along the northeast portion of the ring fracture and the other crossing the middle of the lake. A period of volcanic quiescence was marked by deposition of some 30 m of fine-grained ashy sediments. Approximately 75,000 years ago, activity resumed with the eruption of 3 km of aphyric and porphyritic magma at the southern margin of the lake to form a younger group of ring domes. Ensuing uplift, thought to result from renewed insurgence of magma, brought an end to the lake. This uplift culminated in the eruption, beginning approximately 60,000 years ago, of 7 km of aphyric lavas along a southern arc. The youngest of the southern arc lavas, Cerro E1 Colli (FF, Figure 2), erupted approximately 25,000-30,000 years ago.

All the eruptive units of the Sierra La Primavera are mildly peralkaline high-silica (>75% SiO2 on an anhydrous basis) rhyolites. All of them are similar in gross composition, but there are systematic chemical differences that correlate with both age and eruptive mode. Variations in major element compositions at La Primavera are small; in contrast, variations in the minor and trace elements are striking. Representa- tive analyses of major and minor elements by classical wet chemistry, and of minor and trace elements by instrumental neutron activation analysis (INAA) and X ray fluorescence are given in Table 2. Additional analyses can be found in works by Mahood [1980a, 1981]. In the following discussion, samples with variable degrees of posteruptive hydration have been compared on the basis of analyses recalculated free of water and normalized to 100%.

The proportion of phenocrysts in the porphyritic eruptive units ranges from less than 1% to 15% by volume. Sodic sanidine and quartz, the former generally more abundant, to- gether comprise approximately 97-99% of the phenocrysts, the remainder being ferrohedenbergite, fayalite, and ilmenite, typically in the proportions 50:10' 1. Titanomagnetite, zircon, and apatite occur in some units. Phenocryst assemblages of the various eruptive groups are summarized in Table 1, and representative analyses appear in Table 3. Sanidine, ferrohedenbergite, and ilmenite are unzoned and homogeneous. The compositions of the phenocrysts changed slightly with time,

MAHOOD.' SUMMARY OF THE GEOLOGY AND PETROLOGY OF LA PRIMAVERA 10139

TABLE 1. Summary of the Geologic History of the Sierra La Primavera

Event

Approximate Age Based on K-Ar Dates Magma Total Phenocrysts,

[Mahood, 1980a], years Volume, km 3 % Phenocryst Assemblage Eruption of precaldera lavas Eruption of the Tala Tuff Caldera collapse Eruption of central domes and

deposition of giant pumice horizon

Giant pumice horizon South central dome North central dome

Eruption of older ring domes Eruption of younger ring domes Uplift Eruption of southern arc lavas

144,000-100,000 2(?) 0 or 10-15 20 0-1

95,000 + 10,000

75,000

60,000 and 30,000

0.15 1 0.8 1 0.7 10 5 10-15 3 0or 10

San >_ Q >> Cpx >> Fa > Ilm > Mt San >_ Q

San > Q >> Cpx >> Ilm San > Q >> Cpx > Fa >> Ilm San > Q >> Cpx > Fa >> Ilm San >_ Q >> Cpx > F a >> Ilm San >_ Q >> Cpx > Fa >> Ilm

7 0 None

but within eruptive groups the phenocrysts are uniform in composition.

GEOLOGICAL AND CHEMICAL EVOLUTION OF THE SIERRA LA PRIMAVERA

Precaldera Lavas

About 145,000 years ago the first lavas of the Sierra La Primavera were erupted. These porphyritic and aphyric domes and flows presently crop out in two areas (Figures 2 and 3a), but there may well have been other lavas now covered by younger eruptive units. The precaldera lavas make up the compositionally most heterogeneous eruptive group. In the northern area, the porphyritic Rio Salado dome (A, Fig- ure 2) erupted first, followed by the aphyric Cation de las Flores flow (B, Figure 2). Both units are clearly overlain by

is characterized by white pumice lapilli, bearing less than 1% quartz and sanidine, set in a pink ashy matrix. In many places, the third member shows interlayering of ash flow and air fall layers, and the ash flow portions are laminated.

Compositional Zonation in the Tala Tuff The Tala Tuff is mildly zoned; most major elements remain

approximately constant through the erupted volume, and the most strongly zoned elements, Y and Sc, vary only by a little more than a factor of two. From the first-erupted through the last-erupted portion of the Tala Tuff, the molar ratio Na,_O/ (Na,_O + K,_O) drops from 0.6 to 0.55. Wet chemical analyses demonstrate that TiO,_ increases from 0.09 to 0.13 wt. %, while FeO* (total iron expressed as FeO), MnO, and MgO remain approximately constant at 1.70, 0.05, and 0.04 weight percent,

the porphyritic dome of Mesa E1 Le6n (C, Figure 2) and by respectively. Chlorine is strongly partitioned into water-rich the Tala Tuff. These three units show no systematic chemical vapor relative to a silicic liquid [Burnham, 1967]. Despite pos- trends with time. Because the southern group of precaldera lavas has been blanketed by Tala Tuff, minor ash flows, air fall, and alluvium, neither the number of separate units nor their relative ages are known.

Tala Tuff About 95,000 years ago, ash flows representing approxi-

mately 20 km 3 of magma erupted from the Sierra La Prima- vera and emptied into the surrounding basins, covering some 700 km'- (Figure 4). These ash flows are collectively named the

sible loss on eruption, C1, as well as F, is enriched in the first- empted portion of the Tala Tuff, which contains approximately 0.16% C1 and 0.12% F. C1 and F constitute about 0.11% and 0.09%, respectively, of the last-erupted portion of the Tala Tuff. Na, Rb, Cs, Sm, Gd, Tb, Tm, Yb, Lu, Y, Zr, Hf, Pb, Th, U, Nb, Ta, Sb, and Zn are all enriched in the first-erupted portion of the Tala Tuff, whereas K, Ca, La, Ce, Nd, Ti, and Sc are enriched in the last-erupted part. The Tala Tuff is zoned from a mildly peralkaline first-erupted portion (agpaitic index -- 1.10, Zr - 600 ppm) to a barely metaluminous last-

Tala Tuff for the village of Tala (Figure 1). The Tala Tuff erupted part (agpaitic index -- 0.99, Zr = 500 ppm). La Prima- consists of many small ash flows that are grouped into three easily recognized informal members:

1. The first member contains white aphyric pumice and makes up more than 90% of the volume of the tuff. It occurs as intracaldera ash flows within the Sierra La Primavera proper and as nonwelded, approximately 60-m-thick, outflow sheets filling the surrounding basins.

2. The second member of the Tala Tuff is about 10 m thick and is characterized by white aphyric pumice and grey pumice lapilli containing very sparse quartz and sanidine, as well as lapilli that represent the mixing of the two magmas. Despite their darker color, the grey pumice lapilli are the same composition as the white pumice lapilli in the overlying third member. The transition from the second member of the Tala Tuff to the underlying and overlying members is grada- tional but takes place within less than 1 m stratigraphically.

3. The third member of the Tala Tuff is 5-10 m thick and

vera is not unique in this regard. Some of the Tertiary caldera complexes of the western United States (e.g., the Black Moun- tain [Christiansen and Noble, 1965] and Silent Canyon [Noble et al., 1968] calderas) produced ash flows zoned from peralkaline to subalkalic compositions. Many ring dike complexes in New England [Billings, 1956] and northern Nigeria [Jacob- sen, 1977] show alternation of peralkaline and metaluminous granites in the same center.

Elemental abundances in the first-erupted portion of the Tala Tuff relative to those in the last-erupted portion are plotted in Figure 5, following Hildreth [1979]. This is equivalent to plotting the ratio of the abundance in th e roof of the magma chamber to that in the deepest level tapped. As is true for the Bishop Tuff [Hildreth, 1979], light rare earth elements (LREE) are depleted and heavy rare earth elements (HREE) are enriched roofward. Sr and Ba (among the most strongly zoned elements in the Bishop Tuff) do not appear in Figure 5


0 KM 5 ( l , , , I

Fig. 2. Location map of the Sierra La Primavera (modified from Mahood [1980b]). The Sierra La Primavera is located on the western outskirts of Guadalajara, Jalisco, Mexico. Diagonal ruling: pre-Primavera volcanic rocks; V pattern: precaldera lavas; unpatterned: Tala Tuff (also includes minor ash flow tuffs, primary and reworked air fall deposits, and alluvium, all generally underlain by the Tala Tuff); light stipple: lake sediments; double-dash pattern: older ring domes; rectilinear dots: younger ring domes; heavy stipple: southern arc lavas. Faults are shown as heavy dashed lines. Letters refer to the names of eruptive centers, whereas numbers refer to cultural features, as follows: A: Rio Salado dome; B: Cation de Las Flores flow; C: Mesa E1 Lon dome; D: Arroyo Saucillo group; E: Mesa E1 Chiquihuitillo; F: Mesa E1 Burro dome; G: Cerro Chato dome; H: Cerro Alto composite dome; I: Mesa E1 Nejahuete composite dome; J: Cerro E1 Tule dome; K: E1 Madr6n dome; L: Pinar de La Venta dome; M: Arroyo La Cuartilla dome; N: Mesa La Lobera dome; O: Cerro E1 Chap- ulin dome; P: Dos Coyotes dome; Q: Arroyo Las Pilas dome; R: Arroyo Ixtahuatonte dome; S: La Cuesta dome; T: Cerro E1 Culebreado dome; U: La Puerta dome; V: Arroyo Las Animas dome; W: Cerro E1 Pedernal center; AA: Cerro San Mi- guel center; BB: Llano Grande flow; CC: Cerros Las Planillas center; DD: Arroyo Colorado dome; EE: Cerro E1Tajo center; FF: Cerro E1 Colli dome; 1: Rio Caliente; 2: La Venta del Astillero; 3: Tierra Blanca.

because in all samples they are below detection levels (10 and Caldera Collapse 20 ppm, respectively). The antithetic behavior of Na:O and K:O in the Tala Tuff is common to a number of high-silica The second and third members of the Tala Tuff occur only eruptive units, including the Bishop Tuff, the BandelJer Tuff in the central part of the Sierra La Primavera; they are not [Smith and Macdonald, 1979], and several ash flows from the found as outflow sheets in the surrounding basins. This sug- Yellowstone Plateau (W. Hildreth, personal communication, gests that the estimated 150-500 rn of collapse of the roof zone 1979). of the magma chamber began while the first member of the

MAHOOD.' SUMMARY OF THE GEOLOGY AND PETROLOGY OF LA PRIMAVERA 10141

TABLE 2. Representative Whole Rock Analyses

Eruptive Group PCD TT-E TT-L SCD NCD ORD YRD 60 30

SiO: 76.84 76.44 77.02 76.67 76.43 76.10 76.85 77.02 77.28 TiO2 0.09 0.09 0.13 0.13 0.18 0.17 0.12 0.09 0.06 A1203 11.79 11.70 11.70 11.67 11.66 11.71 11.44 11.71 11.91 FeO* 1.55 1.70 1.72 1.76 1.79 2.02 1.72 1.42 1.18 MnO 0.04 0.05 0.05 0.05 0.05 0.07 0.05 0.04 0.04 MgO 0.04 0.04 0.05 0.05 0.05 0.05 0.04 0.03 0.02 CaO 0.21 0.18 0.26 0.27 0.20 0.27 0.20 0.23 0.30 NaO 4.25 4.70 3.88 4.42 4.60 4.62 4.54 4.69 4.51 KO 4.89 4.72 4.81 4.66 4.80 4.83 4.71 4.44 4.47 F 0.11 0.12 0.09 0.09 0.08 0.09 0.09 0.10 0.10 CI 0.12 0.16 0.11 0.10 0.09 0.09 0.11 0.12 0.13 Agpaitic

index 1.04 1.10 0.99 1.06 1.09 1.10 1.10 1.07 1.03 La 48 41 66 67 81 81 57 39 35 Ce 112 92 141 135 171 172 124 89 79 Nd 46 50 56 54 67 6 52 39 35 Sm 9.6 13.1 11.9 10.9 11.3 11.3 10.3 8.4 8.0 Eu 0.07 0.09 0.06 0.06 0.12 0.12 0.09 0.06 0.03 Tb 1.54 2.58 1.63 1.63 1.59 1.48 1.55 1.36 1.25 Yb 6.4 10.8 6.2 6.0 5.6 5.9 6.1 5.6 5.4 Lu 0.96 1.41 0.90 0.86 0.86 0.84 0.90 0.80 0.79 Rb 188 275 171 172 141 144 163 167 195 Cs 3.7 6.4 4.2 3.6 3.4 2.9 3.3 3.6 4.4 Th 18.4 27.3 18.9 19.1 18.1 17.6 18.2 16.6 19.6 U 7.0 9.5 6.2 7.0 6.6 5.1 5.8 5.7 6.5 Hf 13.6 20.0 13.2 13.3 15.3 15.2 14.7 11.3 8.9 Y 65 139 54 76 51 47 53 50 47 Zr 454 615 499 526 683 682 584 379 228 Nb 66 112 69 71 70 66 65 68 68 Ta 4.3 6.7 3.9 3.9 3.9 3.5 3.8 3.6 3.4 Sc 0.96 0.41 0.95 0.99 0.83 1.06 0.71 0.69 1.10 Zn 116 179 114 106 100 106 108 100 84

To facilitate comparison of variably hydrated samples, all analyses have been recalculated to 100%, free of water.

PCD = precaldera dome, TT-E = early-erupted Tala Tuff, TT-L = late-erupted Tala Tuff, SCD = south-central dome, NCD = north-central dome, ORD = older (95,000-yr) ting domes, YRD = younger (75,000-yr) ring domes, 60 = 60,000-yr southern arc lavas, 30 = 30,000-yr-old youngest dome, Cerro El Colli.

Tala Tuff was still erupting, so that the two upper members were entirely ponded within the collapse basin. Topographic expression of a bounding ring fault is present only to the west of Rio Caliente (1 in Figure 2), where the precaldera Rio Sal- ado and Cation de las Flores lavas are truncated. The circular pattern defined by the vents for the northeastern arc of older ring domes and the younger ring domes indicates the location of the remainder of the master ting fault. Caldera Lake Sedimentation

The caldera rapidly filled with water, and sediment began to be deposited on the surface of the Tala Tuff almost immediately, protecting it from all but minor fluvial erosion. These sediments consist dominantly of 15- to 50-cm-thick layers of planar-bedded and low-angle cross-bedded coarse, ash and pumice lapilli that show repetitive normal grading with some erosion of the finely cross-bedded tops of the sequences. They seem to be saturated pumice turbidity flows. The source for most of this pumice appears to be erosion of small 'islands' of Tala Tuff within the shallow lake and reworking of immediately post-Tala air fall deposits, which are several meters thick outside the caldera. The thickness of this basal pumiceous section varies from 0 to 25 m due to pre-existing topography on

the floor of the caldera, but everywhere the top is marked by a 30-cm-thick fine white ashy layer.

Then a rather spectacular sedimentation event occurred. A 3- to 13-m-thick horizon of giant pumice blocks set in a diatomaceous ashy matrix was deposited over the entire 110 km 2 of the lake. The blocks of fully inflated pumice range from 0.3 to more than 6 m across and are commonly columnar-jointed in a crudely radial pattern. The ashy matrix is finely laminated, and the laminae are deformed around the pumice blocks. The distribution of particle sizes is bimodal; pumice lapilli of intermediate size are rare. The unique character of this deposit made it an important stratigraphic marker bed within the lake bed sequence.

It is envisioned that the giant pumice blocks formed when phenocryst-poor lavas erupted into the middle of the shallow lake. The blocks floated over the surface of the lake and even- tually saturating with water sank gently to the bottom of the lake, where fine wind-blown ash from the surrounding plains of the Tala Tuff and diatoms were accumulating as finely laminated sediments [Mahood, 1980b].

At the margins of the lake, the giant pumice horizon was deposited directly over the eroded Tala Tuff or on 'inselbergs' of pre-Tala rocks. Over most of the lake it was deposited on


top of the 30-cm-thick layer of fine white ash that caps the basal pumiceous sediments. The giant pumice horizon 'seals' faults in the Tala Tuff and the overlying pumiceous sediments, but is rarely cut by faults itself, showing that faulting related to caldera collapse had ceased by the time it was deposited.

Central Domes

A dome comprising the lower part of Mesa E1 Najahuete was eraplaced at the stratigraphic level of the giant pumice horizon in the south-central portion of the lake (Figure 3b). Both the giant pumice horizon and this south-central dome have less than 1% phenocrysts, in contrast to later domes with approximately 10% phenocrysts that stratigraphically overlie the giant pumice horizon.

The chemical similarity of the last-erupted portion of the Tala Tuff to the south central dome and the presence of a 'co- ignimbrite lag-fall deposit' [Wright and Walker, 1977] near the dome suggest that the Tala Tuff may have erupted from a central vent near the present site of the south-central dome [Mahood, 1980b].

Shortly after eraplacement of the south-central dome and deposition of the giant pumice horizon, a dome containing 10% phenocrysts erupted through the middle of the lake (Fig- ure 3b). The base of this north-central dome overlies subaerial pumice breccias on top of the giant pumice horizon. The whole rock and phenocryst compositions of the giant pumice horizon are generally intermediate between those of the south- and north-central domes [Mahood, 1981].

The sedimentary section above the giant pumice horizon is quite variable. Typically the giant pumice horizon is overlain by less than 10 m of stratified pumice before the section be- comes dominated by fine white ashy and diatomaceous beds. Locally, fine-grained deposits directly overlie the giant pumice horizon. The pumiceous lacustrine beds appear to be the product of air fall eruptions that preceded the eraplacement of a group of ring domes. Older Ring Domes

The older ring domes contain approximately 10% phenocrysts and represent some 5 km 3 of magma. They erupted along two concentric arcs: one along the ring fracture at the northeast margin of the lake, and the other through the middle of the lake (Figure 3c). The new magma erupted through the earlier central domes; thus Cerro Alto and Mesa E1 Nejahuete are composite domes, their lower and upper portions consisting of magmas of slightly different composition. The autobrecciated lower margins of older ring domes that erupted in the lake or at its margin are horizontal; the contacts generally occur within the lower portion of the fine- grained sediments, approximately 10 m above the giant pumice horizon. Locally (e.g., on the west flank of Mesa E1 Neja- huete) the intrusion of these domes deformed and uplifted the lake sediments, including the giant pumice horizon.

The cauldron block is cut by an arcuate system of normal faults that extends 9 km from Cerro Alto to just south of Cerro E1 Tule (Figure 2). This fault offsets Mesa E1 Nejahuete and seems to have provided structural control for the extrusion of a ridge of pumiceous porphyritic lava on its south flank, suggesting that faulting and eruption were closely related. This fault may represent a fracture developed in the roof zone of the underlying magma chamber on caldera collapse that was reactivated during eruption of the older ring

MAHOOD: SUMMARY OF THE GEOLOGY AND PETROLOGY OF LA PRIMAVERA 10143

Fig. 3. Palcogeography of the Sierra La Primavera [from Mahood, 1980b]. Pre-Primavera volcanic rocks: diagonal ruling; precaldera lavas: V pattern; Tala Tuff: blank; lake sediments: light stipple; central domes and older ring domes: paired dashes; younger ring domes: rectilinear dots; faults: dashed lines. (a) 100,000 years ago, following eruption of the precaldera lavas. There may have been other lavas belonging to this period, now covered by younger eruptive units. (b) 95,000 years ago, following eruption of the Tala Tuff, caldera collapse, and the eruption of the central domes through the middle of the lake. (c) 90,000 years ago, following eruption of the older ring domes. These ring domes erupted along two concentric arcs: one along the ring fracture at the northeast margin of the lake, and the other through the middle of the lake. (d) 65,000 years ago, following eruption of the younger ring domes. Both aphyric and porphyritic lavas erupted along the ring fracture at the southern margin of the lake.

domes. The presence of the older ring domes Cerro Chato, Mesa E1 Burro, and Mesa E1 Chiquihuitillo (G, F, E, Figure 2) on the extension of the arcuate trend of the fault beyond the margin of the lake suggests that, at that time, the underlying magma chamber was somewhat larger than the collapse basin.

Tala Tuff Magma Chamber

Ar dating, about 5,000-10,000 years [Mahood, 1980a, b]. The last-erupted portion of the Tala Tuff is compositionally nearly identical to the south-central dome, whereas the north- central dome and older ring domes are slightly more mafic in composition (Tables 2 and 3). This suggests continuity in compositional gradients from that part of the magma chamber that erupted explosively as the Tala Tuff to that which erupted as early postcaldera lavas. Similar relations have been

The eruption of the Tala Tuff, caldera collapse, inception of documented in several large silicic systems, where the imme- lake sedimentation, deposition of the giant pumice horizon, diately post ash flow lavas are similar to or lie on slightly more and eruption of the two central domes followed by the older mafic extensions of the chemical trends in the ash flows (e.g., ring domes all occurred within a time span unresolvable by K- Long Valley, California [Bailey et al., 1976], the Valles cal-


Ta G

0 km 20

Fig. 4. Distribution of the Tala Tuff (modified from Mahood [1980b]). The Tala Tuff (pebble pattern) buried approximately 700

dera, New Mexico [Smith, 1979], and Yellowstone [Hildreth et al., 1980]). In all these examples, ring domes presumably represent the tapping of slightly deeper portions of the compositionally zoned magma chamber from which the ash flow erupted.

Most elements show unidirectional changes from the first- erupted portions of the Tala Tuff to the older ring domes (Fig- ure 6). Certain elements, however, show a reversal in trend. Zr and Hf are most depleted in the last-erupted portion of the Tala Tuff and the south-central dome but increase again in the north-central and older ring domes. Likewise, Na20 is most depleted in the last-erupted portion of the Tala Tuff but is significantly higher in the south-central dome. Whereas K20 and Na:O behave antithetically in the ash flow, they both increase in the central domes and older ring domes. Iron appears unzoned in the ash flow, but like Zr, Hf, and Na, it is more abundant in the north-central dome and older ring domes. The reversals in Na, Zr, and Hf parallel changes in the agpaitic index, which is 1.10 in the first-erupted portion of the Tala Tuff, falls to approximately 0.99 in the last erupted portion, and then rises to approximately 1.10 in the older ring domes.

The rapid evacuation of the Tala Tuff magma would lead to a rapid reorganization of the magma chamber, with magma

from deeper levels becoming the most roofward. During this interval immediately following caldera collapse, the new roof zone magma may have been temporarily out of equilibrium with its surroundings. Additionally, a free vapor phase may have been present as a result of vesiculation during the caldera-producing eruption. During this period of re-estab- lishment of magma chamber equilibrium, transient mechanisms may have produced the increase in Na, Fe, Zr, and Hf in the magma that erupted as the north-central dome and older ring domes. If so, the mechanisms must have operated on these elements alone or much more rapidly on them than on other elements. Alternatively these reversals may have been present within a continuously zoned magma chamber and reflect contrasting differentiation mechanisms and condi- tions in the volatile-enriched roof zone that erupted as the Tala Tuff and in the relatively drier, slightly hotter, deeper magma that erupted later as the older ring domes.

Younger Ring Domes Caldera lake sedimentation continued after the emplace-

ment of the older ring domes, with fine ashy and diatomaceous material dominating the section. The deposition of some 30 m of these fine-grained sediments indicates a period of volcanic quiescence. (The fine-grained nature of the lake sediments and the virtual absence of lithie debris derived from outside the caldera suggest that, prior to formation of the cal.- dera, the whole region was elevated so that streams drained away from the future site of the Sierra La Primavera. This may represent an example of regional intumescence above a shallow silicic magma chamber, Stage 1 of the resurgent cauldron cycle [Smith and Bailey, 1968].) About 75,000 years ago, activity resumed with the eruption of a new group of ring domes at the southern margin of the lake (Figure 3d). Both aphyric and porphyritic lavas were erupted, totalling approximately 3 km 3 of magma. The younger K-Ar dates on these domes are corroborated by their position higher within the lake bed sequence [Mahood, 1980b]. The younger ring domes are lapped by only 10-20 m of sediment. Fine-grained sediments give way to fluvially reworked air fall pumice, and the uppermost part of the section consists of subaerially deposited tephra units.

ENRICHED

t3 Sb Tm Yb Ta

CI Zn Rb Cs Tb Na Gd

Mn Fe Co Sm Eu CONSTANT 1

K Nd

Ca Ti La Ce

DEPLETED

I , Sc _ 1/2

1/3

Z

Fig. 5. Tala Tuff enrichment and depletion factors. The elemental abundances in the first-erupted portion of the Tala Tuff are ratioed to those in the last-erupted part. If an element plots above the line it is enriched roofward, below the line it is depleted roofward, in each case by the factor on the right.


200 -

150 -

100 -

1000 -

600 -

200 -

1.0-

0.8-

0.6-

0.4-

2.0- %

1.5

700

600

500

-10

-8

-6

-2000

-1000

-7

-5

-3

-1.1

-1.0

77- %

76-

- 20

-18

-16

-14

4.5 '=

% 4.0

250 -

200 -

150 -

30-

25-

20-

15-

SiO21---'--I1- 203

20 "K20 Na Rb

Th

_ Zr 180- Zn Hf 140 lOO

I I I I I I I I I I I E M L SCD NCD ORD E M L SCD NCD ORD

(-95) (-95)

-12 %

-11

-8

-6

-4

-10

-8

-6

- 150

- 100

50

Fig. 6. Elemental trends in the Tala Tuff and immediately postcaldera lavas. E, M, and L stand for the early-, middle-, and late-erupted portions of the Tala Tuff, respectively, while SCD, NCD, and ORD refer to the south-central, north-central, and older (95,000-yr) ring domes, respectively. A.I. is the agpaitic index.

Uplift that produced the topographic Sierra La Primavera brought an end to lacustrine sedimentation approximately 60,000 years ago. Much of the geomorphic expression of the caldera was destroyed because the hinge line for this uplift nearly coincided with the ring fracture of the caldera. The largest fault associated with this uplift parallels the caldera margin and can be traced for approximately 8 km from south- west of La Venta del Astillero (2 in Figure 2) to Cerro E1 Ped- ernal. The fault consists of two en echelon segments; the maximum displacement is found at Rio Caliente (1 in Figure 2), . where the Tala Tuff and overlying lacustrine deposits are cut by a 100-m scarp.

Deformation during uplift was concentrated at the margins of the lake. Radial dips of 10-20 on sediments at the margins of the former lake flatten to approximately 2 within a short distance towards its center. Uplift was assymetric to the caldera and was greatest at the southern margin of the lake; thus a gentle northerly component is superimposed on the radial dips of the lake sediments. Uplift of at least 250 m is indicated [Mahood, 1980b]. The notable scarcity of faults that cut sediments above the giant pumice horizon within the central portion of the former lake indicates that the uplift was piston- like; the former caldera block apparently rose as a single unit except at its margins. Southern Arc Lavas

Uplift culminated in the eruption, beginning approximately 60,000 years ago, of 7 km 3 of magma as aphyric flows and domes. Contemporaneous with extrusion of these domes and flows were eruptions of air fall pumice, which caps the lake bed section, and a small ash flow that spread over approximately 2.5 km ' near Tierra Bianca (3 in Figure 2). These tuffs have compositions relating them to the 60,000-yr-old southern arc lavas [Mahood, 1980a].

The lavas of the southern arc are generally younger toward the east. Cerro E1 Tajo and Cerro E1 Colli (EE, FF, Figure 2),

the easternmost centers, give K-Ar dates of 25,000-30,000 years, which are in agreement with the youthful topographic expression of these units. Cerro E1 Colli is chemically distinct from the rest of the southern arc lavas, and its position in the compositional trend defined by the postcaldera domes indicates that it is the youngest eruptive unit in the Sierra La Primavera.

Post-95,000-Year Evolution of the Primavera Magma Chamber

After the eruption of the older (95,000-yr) ring domes, the lavas became progressively more silicic and less peralkaline (Table 2, Figure 7). Values for FeO*, TiO,, MgO, and MnO in the older ring domes are approximately twice those in the most recently erupted unit, Cerro E1 Colli. There is some scat- ter in the alkali data due to mobility on posteruptive hydration, but Na,O remained approximately constant while K,O decreased from about 4.8 to 4.45% with time (Table 2). C1 concentrations appear to have increased slightly from 0.9 wt. % in the perlitic and pumiceous older ring domes to 0.13% in the youngest of the southern arc obsidians, although this may reflect better retention of C1 by obsidian samples. F remained approximately constant at about 0.10%. Rb, Cs, and possibly U are the only trace elements that increased monotonically with time in the lavas (Figure 7); nearly all other elements decreased (REE, Zr, Hf, K, Mg, Ti, Mn, Fe, and Zn) or remained approximately constant (Ta, Nb, Sb, Th, Pb, and Y). Sc, like Ca, is unsystematic in behavior, and Th, which remained approximately constant from the older ring domes to the 60,000-yr southern arc lavas, increased again in Cerro E1 Colli, the youngest dome.

In Figure 8, the elemental values for Cerro E1 Colli are shown in ratio to those in the older ring domes. This diagram is not strictly comparable to the enrichment factor diagram for the Tala Tuff (Figure 5) because it does not represent the chemical zonation present in the magma chamber at any one moment in time; rather, it is thought to illustrate the chemical


175 -

125

75-

1000-

600 -

200 -

1.0- 0.8-

2.0- %

1.5-

600 -

400 -

200 -

Ce ====== Yb ,,

Ti Ca Mn

FeO A.I. Zr

r 30

-8

6

-2000

-1000

-4 %77- -3 76-

4.75 - % 4.50 -

-1.1

175 - -- 1.0

150 -

20- -14

15- -12

100 - -10

80-

,SiO2A203 Na2"-'_K20 Rb

Zn

i i i i 95 75 60 30

-12 %

11

-4 -3

-6 -5

- 100

- 50

(=ORD) (=ORD) Fig. 7. Elemental trends in the post-9S,000-yr lavas. This figure is similar to Figure 6. The 95: older ting domes (ORD

of Figure 6); 75: younger ring domes; 60: 60,000-yr southern arc lavas; 30: the youngest southern arc lava, Cerro El Colli, erupted approximately 30,000 years ago.

evolution through time of the uppermost magma in the chamber. The difference between the two enrichment patterns is striking; with increasing differentiation, the ash flow magma was enriched in the HREE and many of the multivalent cations, whereas the post-95,000-yr lavas were enriched system- aticaBy only in Si, Rb, Cs, and possibly U. (Th plots above the no-enrichment line only because of its increase in Cerro E1 Colli.)

INTENSIVE PARAMETERS IN PRIMAVERA MAGMAS

Temperature and Oxygen Fugacity Neither the aphyric nature of many Primavera eruptive

units nor the phenocryst assemblages in most others lend themselves to thermodynamic determination of crystal-liquid equilibration temperatures. Hence it remains uncertain whether the chemical evolution documented for the postcaldera lavas took place under a waxing, waning, or fluctuat- ing thermal regime.

Microprobe analyses of titanomagnetite and ilmenite from the Rio Salado dome (A, Figure 2) were recalculated using the method of Carmichael [1967], and the recast analyses were used to determine the temperature and oxygen fugacity using Buddington and Lindsley's [1964] calibration of the Fe-Ti oxides. A temperature of 850C and a log fo: of-13.8 were obtained, Which are appropriate to a magma on the FMQ buf- fer. The small variations between eruptive units of the major element compositions of phenocrysts suggest only small differences in'magma temperatures. This fact, coupled with the results of two geothermometers [Mahood, 1980a], leads to the conclusion that eruptive temperatures of porphyritic Prima- vera magmas probably varied by no more than 30C.

Pressure

There is little direct evidence for the depth to the roof of the Primavera magma chamber. Formation of a caldera 11 km in diameter on eruption of only 20 km 3 of magma as the Tala

ENRICHED Cs

i CI Rb U F Na AI Si Ca Sc Y Nb Sb Ta Pb Th CONSTANT

Go Mn Fe HI

1/2

La

DEPLETED Mg

Ti 1/3 Zr

1/4

Fig. 8. Enrichment and depletion factors in the post-95,000-yr lavas. The elemental value in the youngest southern arc lava, Cerro El Colli, is ratioed to the average of the older ring domes. This figure is not directly analogous to Figure 5, as it does not represent the zonation present in the magma chamber at any one point in time; rather it presumably illustrates the evolution through time of the most roofward magma.


Tuff would seem to require a magma chamber with a fairly shallow roof. By analogy with eroded cauldrons and ring complexes of similar mineralogy and bulk composition, the roof of the Primavera magma chamber was probably between 2 and 6 km beneath the surface. Assuming that the phenocrysts precipitated within the upper portion of a high-level magma chamber, the small variation between eruptive groups of the major element compositions of the phenocrysts suggests

the computer program of Wright and Doherty [1970] require 1-30% fractionation. Sums of the least squares residuals are quite small (a necessary consequence of the slight compositional differences between parent and daughter magmas), but more than half the solutions require both the addition and subtraction of phases, a result that seems geologically improb- able, especially in crystal-poor rhyolites.

The best computer solutions calculated on the basis of ma- that the depth to the roof of this magma chamber changed jor elements were tested to see if they predicted correct trace little after eruption of the Tala Tuff. element concentrations in the daughter magmas. Poor agree-

EVALUATION OF DIFFERENTIATION MECHANISMS

The homogeneity of phenocryst and whole rock compositions within each postcaldera eruptive group, along with the arcuate arrangement of vents, is strong evidence that these lavas erupted from a single unified magma chamber. Studies of large-volume zoned ash flow eruptions have shown that magma at the roof of the chamber is generally more differen- tiated and poorer in phenocrysts than at deeper levels. The trend in Primavera post-95,000-yr lavas is opposite to that expected on tapping successively deeper levels of a compositionally zoned magma chamber; the last lavas erupted are phenocryst-free and are the most siliceous. Thus it seems more likely that the Tala Tuff and all the postcaldera lavas erupted froma magma chamber that was evolving chemically with time. This is not to say that the geometry of the magma chamber did not change; indeed, the southern arc lavas may represent a southward expansion of the magma chamber. However, the magma chamber was (is) presumably large enough that magma in its roof zone did not immediately 'feel' the effects of new additions below, and thus changes in composition within the sequence of lavas reflect high-level differentiation mechanisms. Since there is no positive evidence of a unitary magma chamber during eruption of the precaldera lavas, they will not be considered further.

Crystal Fractionation Most of the Tala Tuff and many of the Primavera lavas are

aphyric. If crystal settling operated in these magmas, either all the phenocrysts settled out, or the magmas became super- heated prior to eruption and the phenocrysts were resorbed. Phenocrysts show no evidence of having settled: they are hedral, unzoned, and homogeneous. Repeated tapping of a cooling magma chamber that is crystallizing inward from its margins is equivalent geochemically to settling of the observed phenocryst phases; thus the chemical contraints on both of these differentiation mechanisms are similar.

The compositional zonation present in the magma chamber just prior to the eruption of the Tala Tuff was not the product of crystal fractionation alone for the following reasons:

1. Zn is enriched and Sc and Ti are depleted roofward, while Fe, Mg, and Mn are unzoned (Figure 5). These divergent trends are incompatible with the removal of mafic phenocrysts, which are strongly enriched relative to glass in these elements. '

2. To produce the nearly two-fold roofward enrichment of elements such as Nb, Y, Ta, U, Rb, and Cs would require removal of at least half the magma as phenocrysts. But this would have profound effects on major elements such as Ca and Fe, effects that are not observed.

Separation of phenocrysts in their modal proportions does not account for the chemical trends in the post-95,000-yr lavas [Mahood, 1981]. Crystal fractionation models calculated using

ment was obtained between the predicted and measured concentrations of most trace elements in the daughter magmas [Mahood, 1981]. Most trace elements decrease or remain approximately constant with time in the postcaldera lavas (Fig- ure 7), but the calculations predict increasing concentrations. This is a result of the large proportion of sanidine and quartz in the separating phases which, despite the strong enrichment of elements such as the LREE and Zn in the marie phases, causes the bulk distribution coefficient D for these elements to be less than one. In all the calculated models the removal of phenocrysts does not produce the observed decrease in LREE with time. Also, despite the dominant role of sanidine in the calculated models and D.u for all cases being greater than one, the predicted decrease in Eu is always smaller than that observed.

Partial Melting Incremental partial melting of a source region would not

produce the trends observed in the post-95,000-yr lavas of La Primavera. The first partial melts would be enriched in Si, Rb, and Cs and depleted in Fe, Mg, and Ti relative to subsequent melt fractions. The reverse is observed in the post-95,000-yr lavas: they become increasingly siliceous and depleted in the marie elements with time.

The chemical zonation in the Tala Tuff did not result from accretion of successive partial melt increments. Incremental partial melting does not readily explain the crossover in REE on the enrichment factor diagram (Figure 5) nor the divergent enrichment patterns of the first transition series elements [Ma. hood, 1980a; see also Hildreth, 1977].

PHENOCRYST/OLASS PARTITIONING AND MELT STRUCTURE

Crystal fractionation, whether during partial melting or due to crystal settling, cannot alone account for the compositional gradient in the Tala Tuff or the trends with time in the post- 95,000-yr lavas. On the other hand, the progressive increase in the ratio of network-forming elements to network-modifying elements in the post-95,000-yr lavas suggests that trace-element evolution is associated with changing melt structure. (The term 'melt structure' is used here to signify the mutual spatial relationship between atoms in a silicate liquid. Since melts lack long-range order, a melt structural state is characterized by an average concentration of variously coordinated cations (e.g., a highly polymerized melt has, on the average, fewer octahedral 'sites' than a less polymerized melt). In this sense, at any given pressure, temperature, volatile content, and bulk composition, the melt structure is 'fixed' in a statisti- cal sense.)

Crystal/liquid partition coefficients are thought to be func- tions of (1) temperature [Hkli and Wright, 1967; Leemah and Lindstrom, 1978], (2) fo [Drake, 1975; Sun et al., 1974], (3) pressure [Mysen and Kushiro, 1978], (4) composition of the


TABLE 4. Summary of Phenocryst/Glass Partition Coefficients

Eruptive Group

Sanidine Ferrohedenbergite Fayalite Ilmenite PCD TT ORD YRD PCD SCD ORD YRD SCD ORD YRD ORD

La 0.06 0.037 0.057 0.11 23 2.9 3.5 24 1.08 2.0 23 1.31 Ce 0.033 0.031 0.043 0.095 20.9 3.5 4.0 21.6 0.93 1.78 20.4 1.19 Nd 0.01 0.02 0.04 0.09 16 5.3 5.9 17 0.77 1.5 14 0.96 Sm 0.019 0.020 0.029 0.946 14 6.2 6.2 13 0.496 1.060 8.4 0.684 Eu 2.3 2.3 2.0 2.2 12 5 6 10 0.1 0.4 5.8 0.4 Tb 0.012 0.016 0.023 0.022 6.9 5.0 4.0 6.1 0.31 0.54 2.8 0.36 Dy 0.01 0.02 0.02 0.02 6 4.5 4.8 1 1 3 0.4 Yb 0.005 0.014 0.016 0.018 6.4 5.0 4.5 5.5 0.92 0.98 1.96 0.55 Lu 0.008 0.010 0.015 0.014 8 6.9 6.1 7 1.3 1.3 2.2 0.72 Th 0.010 0.017 0.021 0.025 3.44 0.36 0.77 2.89 0.276 0.438 3.17 0.43 U 0.008 0.021 0.018 0.014 1.0 0.12 0.33 0.87 0.12 0.13 0.69 0.06 Hf 0.005 0.018 0.015 0.015 1.52 0.30 0.39 2.04 0.12 0.28 1.78 0.65 Ta 0.004 0.018 0.018 0.015 0.69 0.10 0.97 0.49 0.098 0.090 0.575 18.0 Sc 0.04 0.04 0.05 0.06 133 110 109 172 5.1 4.8 9.3 18 Ti 0.1 0.05 0.11 0.11 2.6 4 2.5 2.5 0.3 0.4 0.5 380 Mn 0.011 0.020 0.018 0.020 34 30 33 63 56 77 27 Fe 0.14 0.14 0.11 0.14 20 18 15 18 37 33 40 24 Zn 0.04 0.06 0.06 0.07 21 8.9 7.2 14.8 11.1 10.4 10.7 7.8 Rb 0.39 0.31 0.36 0.34 Cs 0.012 0.023 0.009 0.028

The calculated errors on the partition coefficients, the method of correcting for impurities in the mineral separates, and the INAA analyses of the separated glass and phenocrysts are given by Mahood [1980a]. Values are stated such that uncertainty is in the last digit. See footnote to Table 2 for the Eruptive Group abbreviations.

phenocryst phase [Schnetzler and Philpotts, 1970], and (5) bulk composition of the melt [Watson, 1977; Takahashi, 1978; Hart and Davis, 1978; Mysen and Virgo, 1980]. The temperature, oxygen fugacity, and pressure of the Primavera magmas are thought to have varied little between eruptive groups. Compo- sitions of the marie phenocrysts in the different groups are nearly identical, and the range of sanidine compositions is small. Thus variations in partition coefficients between the older and younger ring domes or between them and the Tala Tuff and its outgassed equivalent, the south-central dome, reflect changes in melt structure resulting from changes in bulk

tion coefficients for all analyzed elements with the exception of Ta in ferrohedenbergite remain constant or increase from the 95,000- to the 75,000-yr dome. The single exception of Ta, however, might simply be an artifact of an improper correc- tion for ilmenite inclusions in the clinopyroxenes.

The observed decrease in Fe, Mn, Mg, and K in the sequence of post-95,000-yr lavas would be expected to foster increasing polymerization of the melts and, as a result, increasing crystal/liquid partition coefficients. The changes in crystal/glass partition coefficients between the older and younger ring domes, however, seem remarkably large consid-

composition of the magma. A decrease in the concentration of ering the small changes in major element composition. Such the network modifiers, Fe, Mn, Mg, and Ca (and Na and K in peralkaline melts), or in dissolved H,_O or F would promote a higher proportion of bridging oxygen bonds and an increase in the polymerization of the melt [Burnham, 1975], with a cor- responding increase in crystal-liquid partition coefficients.

With this in mind, seventeen mineral and glass separates from five representative Primavera samples were analyzed at Lawrence Berkeley Laboratory by INAA. The resulting analyses and partitioning patterns are discussed elsewhere [Ma- hood, 1980a; G. A. Mahood and W. Hildreth, manuscript in preparation, 1981]; the pertinent partition coefficients are summarized in Table 4. Despite only minor differences in the major element compositions of phenocrysts, the concentra-

behavior suggests one or more of the following conclusions: 1. High-silica rhyolitic melts are so strongly polymerized

that small changes in major element composition can greatly affect the number of sites in the liquid suitable for most trace elements. Small distortions of the silicate network may make already marginal sites even less energetically favored.

2. The increase in partition coefficients may reflect increasing melt polymerization and resultant decrease in trace metal sites caused chiefly by a decrease in the dissolved volatile content of the melt. Although the absolute difference in volatile content between the magmas of the older and younger ring domes could not have been large, at low water (or fluorine) concentrations, small variations can have great effects, as

tions of many of the trace elements vary by factors of 5-10 illustrated by the rapid drop in liquidus temperature on add- among analyzed samples. This is reflected in crystal/glass partition coefficients that differ by as much as a factor of 20.

Phenocryst/Glass Partitioning in the Older and Younger Ring Domes

All the southern arc lavas are aphyric, so any discussion of the chemical evolution of the post-95,000-yr lavas as reflected in trace element partitioning is necessarily restricted to the older and younger ring domes. The behavior of trace element partition coefficients in the ring domes is systematic; the parti-

ing small amounts of water or fluorine to the anhydrous granitic system [Naney and Swanson, 1980; Wyllie and Tuttle, 19611. '

3. If elements that are not easily accommodated by the silicate network are complexed by hydroxyl, fluoride, chloride, phosphate, sulfate, or carbonate ligands, any decrease in concentration of these complexing agents would increase the activity of the trace elements in the melt, favoring increased concentrations in the crystalline phases. Complexing could lead to discontinuities in crystal/liquid partitioning behavior; partition coefficients might remain low as long as the concen-


tration of the complexing ligands is sufficient to complex most of the trace metals, but once the concentration of complexing

'ligands falls below that threshold, the partition coefficients will rise. This mechanism might be most effective in relatively volatile-poor high-silica magmas where the total concentration of possible complexing ligands is of the same order as the sum of the trace metals. Other types of complexing may also be important in increasing the solubility of trace metals in melts. For example, Watson [1979] has suggested, on the basis of measurements demonstrating an increase in the solubility

sociated with the alkalies as complexes.

Migration of Elements to Structurally More Favorable Melts

The combined effects of gradients in temperature, bulk composition, and volatile content should logically result in a magma chamber with a melt structural gradient. In the relatively dry post-95,000-yr system, the combined gradients could have been such that the roof was more polymerized than the more mafic, hotter, deeper levels. The progressive decrease in whole rock (and glass) abundances of most trace ele-

Fe, Mn, and Mg than the older ring dome. On the contrary, for the elements and phases analyzed, partition coefficients for most trace elements in the Tala Tuff magma (Table 4) are significantly smaller than in the younger ring domes and, in fact, are smaller than or similar to those in the more mafic older ring dome.

Limited analytical data, thermodynamic calculations, and mineralogical evidence [Smith and Bailey, 1966; Gibson, 1970; Noble and Parker, 1974; $ommer, 1978; Hildreth, 1977, 1979; R. L. Smith, personal communication, 1979; this study] in-

in H20, F, and C1 relative to the last-erupted portions, and, by inference, the roofward magmas are enriched in volatiles. An increase in the water and/or halogen content of a rhyolitic melt, like an increase in concentration of the cationic network modifiers, should depolymerize the magma [Burnham, 1975]. One expression of such depolymerization would be smaller phenocryst/glass partition coefficients, as indeed are found in the Tala Tuff and its outgassed equivalent, the south-central dome. The contrast in elemental enrichment patterns between the Tala Tuff and the post-95,000-yr lavas (Figures 5, 6, 7, and 8) suggests that mechanisms other than melt structural control

ments with time in the post-95,000-yr eruptive groups (each of produced the trace element gradients in the Tala Tuff magma. which presumably represented the uppermost and most differ- entiated magma in the Primavera magma chamber at the moment of its eruption) may have been accomplished by migration of these elements out of the roofward magma into deeper, less polymerized levels of the magma chamber.

The two-liquid partitioning studies of Watson [1976] and Ryerson and Hess [1978] in immiscible mafic and silicic melts are examples of diffusion into a structurally more favorable melt. They found that Ti, Fe, Mn, Mg, Ca, P, Zr, Ta, Cr, REE, Sr, and Ba were partitioned into the less polymerized marie melt and that A1, Na, K, and Cs were preferentially concentrated in the silicic melt. Aluminum apparently sub- stitutes for silicon in tetrahedral coordination, which requires the coupled substitution of the alkalies to maintain local charge balance. The systematic enrichment of Primavera post-95,000-yr magmas in only Si, Rb, and Cs, while all other elements remain approximately constant or decrease in concentration, is analogous to the partitioning of all cations but aluminum and the alkalies into the less polymerized marie melts in the two-liquid experiments. The observed decrease in total mole percent alkalies is consistent with the implications

If melt structure had been the dominant control on trace element variations in the Tala Tuff, enrichment of those elements (Si, A1, and the alkalies) which have been shown exper- imentally to partition into the more strongly polymerized melt would be antithetic to those elements (Fe, Ti, REE, Zr, Ta, etc.) that partition into the less polymerized melts. The verti- cal trace element gradients in the Tala Tuff magma chamber are not in accord with such a model. Na and K behave anti-

thetically, as do the HREE and LREE. Elements of the first transition series do not behave consistently: Sc and Ti are depleted and Zn is enriched roofward, whereas Fe is unzoned. These data suggest that melt structure did not exert as strong a control on trace element behavior in the ash flow magma as in the magmas of the ring domes, due to the depolymerizing effects of elevated volatile concentrations.

A number of elements that partition into less polymerized melts in the two-liquid experiments have opposite enrichment directions in different ash flow systems, despite the roofward depletion in all systems of the elements Ca, Mg, and Ti. Zr, Hf, Zn, and Fe are enriched roofward in moderately to strongly peralkaline systems (e.g., Fantale Tuff [Gibson, 1970];

of the two-liquid data, because the measured decrease in. Grouse Canyon member of the Belted Range Tuff[Noble and magma viscosity with increasing peralkalinity [Riebling, 1966] indicates that alkalies in molecular excess of alumina are net-

work modifiers and should, therefore, migrate to a less polymerized melt. At Primavera this less polymerized melt presumably lies at a slightly deeper level of the magma chamber.

Phenocryst/Glass Partitioning in the As Flow Magma and the Origin of Compositional Zoning in the Tala Tuff

If bulk composition had been the major control on partition coefficients in the Tala Tuff magma, the last-erupted portion of the Tala Tuff and its compositionally identical, outgassed equivalent, the south-central dome, could be expected to have partition coefficients similar to those of the younger ring dome, which, in turn, should be higher than those for the older ring dome. This is because the last-erupted Tala Tuff, the south-central dome, and the younger ring dome are similar in major element composition and all three are lower in Ti,

Parker, 1974]; Soldier Meadow Tuff [Korringa, 1973]), whereas these same elements are unzoned or depleted roofward in 'calc-alkalic' systems (e.g., Bishop Tuff [Hildreth, 1979]; Topopah Springs member of the Paintbrush Tuff [Lip- man et al., 1966]; Rainier Mesa member of the Timber Moun- tain Tuff [Christiansen et al., 1977]). In mildly peralkaline or alkali-rhyolite systems (e.g., La Primavera; Bandelier Tuff (R. L. Smith, personal communication, 1979); Spearhead member of the Thirsty Canyon Tuff [Noble and Parker, 1974]) these elements are variously unzoned or slightly enriched roofward.

Such behavior seems inconsistent with melt structural control, and may be better explained by a mechanism in which trace elements migrate as volatile complexes in a thermal and gravitational gradient. As volatiles move roofward, they may preferentially complex with different trace metals; thus, systems with different volatile contents or ratios of H20, F, C1, and CO2 will have different elemental enrichment patterns. In this way the trace element enrichment pattern in the Tala


EFFECTS OF GRADIENTS SUM OF DOMINANT CONTROL ON ON MELT POLYMERIZATION EFFECTS TRACE-ELEMEMENT CONCENTRATIONS

INCREASING POLYMERIZATION

TRACE-ELEMENT VOLATiLE-COMPLEXES (tends to erupt explosively as ash flows and major air falls) POLYMERIZATION MAXIMUM---'-'

MELT STRUCTURE

(tends to erupt as lavas and minor air falls)

A. MAGMA CHAMBER WITH A VOLATILE-RICH ROOF

INCREASING POLYMERIZATION

POLYMERIZATION MAXIMUM

MELT STRUCTURE

(tends to erupt as lavas and minor air falls)

B. MAGMA CHAMBER LACKING A VOLATILE-RICH ROOF

Fig. 9. Diagrammatic representation of melt polymerization gradients within silicic magma chambers. (a) Silicic magma chamber with a volatile-rich roof zone and maximum melt polymerization at an intermediate level. In the volatile- rich zone above the polymerization maximum, transport of trace metals as dissolved volatile complexes may be the dominant differentiation mechanism, while below the polymerization maximum, the distribution of trace metals may be determined by melt structure. (b) Silicic magma chamber lacking a volatile-rich roof zone. The roofward magma might be the most polymerized, so most trace metals would diffuse to deeper, hotter, more mafic levels of the chamber where the melt is less polymerized.

Tuff (and in other ash flow systems) may be the natural result of development of a volatile-enriched cap on a dominantly drier magma chamber. Roofward enrichment in trace metals may be linked to roofward enrichment in the volstiles re- quired to produce an ash flow eruption.

DISCUSSION AND CONCLUSIONS

The structure of a silicate melt is determined by pressure, temperature, major element composition, and volatile content. Gradients in temperature and phenocryst and hole rock compositions preserved in ash flow tuffs are the main evidence for such gradients within silicic magma chambers and, inferentially, for the melt structural gradients that result from

gradient fiattens, increasing temperature and the more mafic composition of the magma with increasing depth in the chamber may overcome the roofward volatile gradient and reverse the polymerization gradient. The result could be a magma chamber with maximum melt polymerization at some intermediate level. In the volatile-rich zone above the polymerization maximum, transport of trace metals as dissolved volatile complexes may be the dominant differentiation mechanism, while below the polymerization maximum (which would roughly coincide with the leveling of the water gradient), the distribution of trace metals may be determined by melt structure. This concept is shown diagra- matically in Figure 9a. If an eruption were to tap through the

such zoning. The depolymerizing effect of a roofward increase polymerization maximum, it would sample magmas that ob- in dissolved volatile content of a magma is counteracted by tained their trace element abundance patterns through at least the roofward decrease in temperature and concentration of two different mechanisms; such an event might be manifested network-modifying cations. In different systems, the sum of by discontinuities in elemental trends. Such a change in the these effects could result in maximum polymerization occur- dominant mechanism of differentiation might account for sp- ring at the roof, at deep levels, or at some intermediate posi- parent reversals in the trends of Zr, Hf, Ns, and the agpaitic tion in a magma chamber, depending on the relative magni- index between the Tala Tuff and older ring domes (although a tudes of the individual gradients. transient vapor phase produced on eruption may have been

In chambers in which the roof zone is strongly enriched in involved). Magma that erupted as the south-central dome may water (or other volstiles), the melt may initially be increas- have occupied the boundary between the two differentiation ingly polymerized with depth in the chamber, but as the water regimes. The change in trace element trends coincides with


the end of the ash flow eruption, which is consistent with the eruption ceasing on exhaustion of the volatile-enriched magma.

In a silicic magma chamber that lacks a strongly volatile- enriched roof zone, the roofward magma might be the most polymerized, so most trace metals would diffuse to deeper, hotter, more mafic levels of the chamber where the melt is less polymerized (Figure 9b). Such a configuration might produce the trends observed in the post-95,000-yr lavas, each eruptive group representing successive samples of roof zone magma in

......................... uxy ystcm mat occamc progressive a tcxauvxy more S'llicic and more polymerized with time.

12782 and EAR-78-03648 to Carmichael and EAR-76-21833 to G. Curtis, and by two University of California at Berkeley Graduate Di- vision Grants-in-Aid and a N. S. F. Fellowship to the author. Early versions of this manuscript benefited from suggestions by Gilbert and Hildreth, and the final version from a review by J. Ratt6.

REFERENCES

Bailey, R. A., B. Dalrymple, and M. A. Lanphere, Volcanism, structure, and geochronology of Long Valley caldera, Mono County, California, J. Geophys. Res., 81, 725-744, 1976.

Billings, M.P., The geology of New Hampshire, 2, Bedrock geology, vo,, ,,. li. oau *a,'"'-t ,.mm-., ,.mtmau, :,. n., 1956.

Buddington, A. F., and D. H. Lindsley, Iron-titaMum oxide mMerals and synthetic equivalents, J. PetroL, 5, 310-357, 1964.

The presence of gradients implies that the upper portion of Burnham, C. W., Hydrothermal fluids at the magmatic stage, in Geo- the magma chamber that contains them is either static or, if chemistry of Hydrothermal Ore Deposits, pp. 34-76, edited by H. L. convecting, consists of many convection cells with small verti- cal dimensions. In the latter case, migration of trace metals is facilitated, because diffusion is the rate-limiting step only at interfaces between cells of contrasting melt structure, and the chemical gradients across the interfaces are constantly replen- ished by convection [Turner, 1973, 1974; McBirney and Noyes, 1979; Chen and Turner, 1980; Turner, 1980].

The preceding discussion of the origin of the zonation in the Tala Tuff and the evolution of the postcaldera lavas with time has emphasized two contrasting modes of liquid state differentiation: one involves transport of trace metals as volatile complexes within a thermal and gravitational gradient in a volatile-rich magma undersaturated with respect to water; the second involves emigration of trace elements in response to changes in melt structure produced by minor decreases in the concentration of network modifiers, dissolved volatiles, and/ or complexing ligands in a relatively 'dry' system. Both mechanisms could operate simultaneously and, at intermediate volatile concentrations, be of similar importance. Other differentiation mechanisms (e.g., crystal fractionation and contamination by assimilation of crustal material or by ad- mixing of more mafic magmas from below) presumably influ- enced the compositions of magmas erupted at La Primavera, but their contribution has been largely obscured by the over- print of liquid state differentiation mechanisms which oper-

Barnes, Holt, Rinehart, and Winston, New York, 1967. Burnham, C. W., Thermodynamics of melting in experimental sili-

cate-volatile systems, Fortschr. Mineral., 52, 101-118, 1975. Carmichael, I. S. E., The iron-titanium oxides of salic volcanic rocks

and their associated ferromagnesian silicates, Contrib. Mineral. Pet- rol., 14, 36-64, 1967.

Chen, C. F., and J. S. Turner, Crystallization in a double-diffusive system, J. Geophys. Res., 85, 2573-2593, 1980.

Christiansen, R. L., and D.C. Noble, Black Mountain volcanism of southern Nevada, Spec. Pap. Geol. Soc. Am., 82, 246, 1965.

Christiansen, R. L., P. W. Lipman, W. J. Cart, F. M. Byers, Jr., P. P. Orkild, and K. A. Sargent, Timber Mountain-Oasis Valley caldera complex of southern Nevada, Geol. Soc. Am. Bull., 88, 943-959, 1977.

Drake, M. F., The oxidation state of europium as an indicator of oxygen fugacity, Geochim. Cosmochim. Acta, 39, 55-64, 1975.

Gibson, I. L., A pantelleritic welded ash-flow tuff from the Ethiopian Rift Valley, Contrib. Mineral. Petrol., 28, 89-111, 1970.

Hikli, T. A., and T. L. Wright, The fractionation of nickel between olivine and augite as a geothermometer, Geochim. Cosmochim. Acta, 31, 877-884, 1967.

Hart, S. R., and K. Davis, Nickel partitioning between olivine and silicate melt, Earth Planet. Sci. Lett., 40, 203-219, 1978.

Hildreth, E. W., The magma chamber of the Bishop Tuff: Gradients in temperature, pressure, and composition, Ph.D. dissertation, Univ. of Calif., Berkeley, 1977.

Hildreth, W., The Bishop Tuff: Evidence for the origin of compositional zoning in silicic magma chambers, Spec. Pap. Geol. Soc. Am., 180, 43-75, 1979.

Hildreth, W., R. L. Christiansen, and J. R. O'Neil, Catastrophic iso- topic modification of rhyolitic magma by creation of the Yellow-

ated in the high-level magma chamber. stone caldera, GeoL Soc. Am. Abstr. Programs, 12, 111, 1980. The composition of any magma is controlled by the relative Jacobsen, R. R. E., Geology of the Liruei, Bante, and adjacent

efficiency of simultaneously operating differentiation mechanisms; for example, in highly polymerized high-silica rhyolites, melt structural effects may dominate, but in low-viscosity andesites and basalts, crystal fractionation may be more important. The relative efficiency of different mechanisms is controlled by physical properties of the melt (such as viscosity, density, and melt structure), the temperature gradient within the magma chamber, and the absolute abundances and relative proportions of volatile components. Because diffusive transport relies on limited convection to speed the process and the nature of convection is determined by density as well as temperature gradients, the chemical evolution and theological evolution of a magma chamber are intertwined.

Acknowledgments. Discussions with E. W. Hildreth and I. S. E. Carmichael contributed to the petrological ideas presented in this paper, and the help of C. M. Gilbert in the field made the geological synthesis possible. The assistance of S. Anderson-Nosker, Z. de- Cserna, R. Drake, W. Duffield, A. Grunder, D. Kosco, and A. Razo at various stages in this study is greatly appreciated. I am grateful to R. L. Smith and R. Macdonald for arranging additional INAA and XRF analyses. Material support was provided by N. S. F. Grants EAR-74-

Younger Granite ring complexes, GeoL Surv. Niger. BulL, 33, 177 pp., 1977.

Korringa, M. K., Linear vent area of the Soldier Meadow Tuff, an ash-flow sheet in northwestern Nevada, Geol. Soc. Am. Bull., 84,

,3849-3866, 1973. Leeman, W. P., and D. J. Lindstrom, Partitioning of Ni :+ between

basaltic and synthetic melts and olivines--An experimental study, Geochim. Cosmochim. Acta, 42, 801-816, 1978.

Lipman, P. W., R. L. Christiansen, and J. T. O'Connor, A compositionally zoned ash-flow sheet in southern Nevada, Geol. Surv. Prof lap. U.S., 524-F, 1-47, 1966.

Mahood, G. A., The geological and chemical evolution of a Late Pleistocene rhyolitic center: The Sierra La Primavera, Jalisco, M6x- ico, Ph.D. dissertation, Univ. of Calif., Berkeley, 1980a.

Mahood, G. A., Geological evolution of a Pleistocene rhyolitic center: Sierra La Primavera, Jalisco, M6xico, J. VolcanoL Geotherm. Res., 8, 199-230, 1980b.

Mahood, G. A., Chemical evolution of a Pleistocene rhyolitic center: Sierra La Primavera, Jalisco, M6xico, Contrib. Mineral. Petrol., in press, 1981.

McBirney, A. R., and R. M. Noyes, Crystallization and layering of the Skaergaard intrusion, J. Petrol., 20, 487-554, 1979.

Mysen, B. O., and I. Kushiro, The effect of pressure on the partitioning of nickel between olivine and aluminous silicate melt, Year Book Carnegie Inst. Washington, 77, 706-707, 1978.


Mysen, B. O., and D. Virgo, Trace element partitioning and melt structures: An experimental study at 1 atm pressure, Geochim. Cos- mochim. lcta, 44, 1917-1930, 1980.

Nancy, M. T., and S. E. Swanson, The effect of Fe and Mg on crystallization in granitic systems, lm. Mineral., 65, 639-653, 1980.

Noble, D.C., and D. F. Parker, Peralkaline silicic volcanic rocks of the western United States, Bull. Volcanol., 38, 803-827, 1974.

Noble, D.C., K. A. Sargent, H. H. Mehnert, E. B. Ekren, and F. M. Byers, Jr, Silent Canyon volcanic center, Nye County, Nevada, Mere. Geol. $oc. lm., 110, 65-76, 1968.

Riebling, E. F., Structure of sodium aluminosilicate melts containing at least 50 mole % SiO2 at 1500C, J. Chem. Phys., 44, 2857-2865, 1966.

Ritchie, J. L., Origin of divergent magmas at Crater Lake, Oregon, Ph.D. dissertation, Univ. of Oregon, Eugene, 1979.

Ryerson, F. J., and P. C. Hess, Implications of liquid-liquid distribution coefficients to mineral-liquid partitioning, Geochim. Cosmo- chim. lcta, 42, 921-932, 1978.

Schnetzler, C. C., and J. A. Philpotts, Partition coefficients of rare- earth elements between igneous matrix material and rock-forming mineral phenocrysts, 2, Geochim. Cosmochim. lcta, 34, 331-340, 1970.

Smith, R. L., Ash-flow magmatism, Spec. Pap. Geol. $oc. lm., 180, 5- 27, 1979.

Smith, R. L., and R. A. Bailey, The Bandelier Tuff: A study of ash- flow eruption cycles from zoned magma chambers, Bull. Volcanol., 29, 83-104, 1966.

Smith, R. L., and R. A. Bailey, Resurgent cauldrons, Mere. Geol. $oc. lm., 116, 613-662, 1968.

Smith, R. L., and R. Macdonald, Rhyolitic volcanism and its relationship to granitic plutonism, Geol. $oc. lm. lbstr. Programs, 11, 520, 1979.

Sommer, M. A., II, Volatiles in silicate melt inclusions from the air- fall and ash-flow tuffs of the Bandelier Tuff, Jemez Mountains, New Mexico, Geol. Soc. lm. lbstr. Programs, 10, 495, 1978.

Sun, C. O., R. J. Williams, and S.S. Sun, Distribution coefficients of Eu and Sr for plagioclase-liquid and clinopyroxene-liquid equi-

libria in oceanic ridge basalt: An experimental study, Geochim. Cos-. mochim. 4cta, 38, 1415-1433, 1974.

Takahashi, E., Partitioning of Ni 2+, Co 2+, Fe 2+, Mn 2+, and Mg 2+ between olivine and silicate melts: Compositional dependence of partition coefficients, Geochim. Cosmochim.4cta, 42, 1829-1844, 1978.

Turner, J. S., Buoyancy Effects in Fluids, Cambridge University Press, London, 1973.

Turner, J. S., Double-diffusive phenomena, 4nn. Rev. Fluid Mech., 6, 37-56, 1974.

Turner, J. S., A fluid dynamical model of differentiation and layering in magma chambers, Nature, 285, 213-215, 1980.

Watkins, N. D., B. M. Guam, A. K. Baksi, D. York, and J. Ade-Hall, Palcomagnetism, geochemistry, and potassium-argon ages of the Rio Grande de Santiago volcanics, central Mexico, Geol. Soc. 4m. Bull., 82, 1955-1968, 1971.

Watson, E. B., Two-liquid partition coefficients: Experimental data and geochemical implications, Contrib. Mineral. Petrol., 61, 35-48, 1976.

Watson, E. B., Partitioning of manganese between fosteritc and silicate liquid, Geochim. Cosmochim. 4cta, 41, 1363-1374, 1977.

Watson, E. B., Zircon saturation in felsic liquids: Experimental results and applications to trace element geochemistry, Contrib. Mineral. Petrol., 70, 407-420, 1979.

Wright, J. V., and G. P. L. Walker, The ignimbrite source problem: Significance of a co-ignimbrite lag-fall deposit, Geology, 5, 729-732, 1977.

Wright, T. L., and P. C. Doherty, A linear programming and least squares computer method for solving petrologic mixing problems, Geol. Soc. 4m. Bull., 81, 1995-2008, 1970.

Wyllie, P. J., and D. F. Tuttle, Experimental investigations of silicate systems containing two volatile components, 2, The effects of NH3 and HF, in addition to H20, on the melting temperatures of albite and granite, 4m. J. Sci., 259, 128-143, 1961.

(Received December 8, 1980; revised March 12, 1981;

accepted March 25, 1981.)

a summary of the geology and petrology of the sierra la primavera, jalisco, mexico

Documents