T. L. Wright Differentiation and magma mixing on Kilauea’s ... · ity of Kilauea’s magma...
29
Bull Volcanol (1996) 57 : 602–630 Q Springer-Verlag 1996 ORIGINAL PAPER T. L. Wright 7 R. T. Helz Differentiation and magma mixing on Kilauea’s east rift zone: a further look at the eruptions of 1955 and 1960. Part II. The 1960 lavas Received: July 10, 1995 / Accepted: October 10, 1995 Editorial responsibility: M. Carroll T. L. Wright (Y) Smithsonian Institution, National Museum of Natural History, NHB-119, Washington, DC 20560, USA R. T. Helz U.S. Geological Survey, 107 National Center, Reston, Virginia 22092, USA Abstract New and detailed petrographic observations, mineral compositional data, and whole-rock vs glass compositional trends document magma mixing in lavas erupted from Kilauea’s lower east rift zone in 1960. Evidence includes the occurrence of heterogeneous phenocryst assemblages, including resorbed and re- versely zoned minerals in the lavas inferred to be hy- brids. Calculations suggest that this mixing, which is shown to have taken place within magma reservoirs re- charged at the end of the 1955 eruption, involved intro- duction of four different magmas. These magmas origi- nated beneath Kilauea’s summit and moved into the rift reservoirs beginning 10 days after the eruption be- gan. We used microprobe analyses of glass to calculate temperatures of liquids erupted in 1955 and 1960. We then used the calculated proportions of stored and re- charge components to estimate the temperature of the recharge components, and found those temperatures to be consistent with the temperature of the same magmas as they appeared at Kilauea’s summit. Our studies rein- force conclusions reached in previous studies of Ki- lauea’s magmatic plumbing. We infer that magma en- ters shallow storage beneath Kilauea’s summit and also moves laterally into the fluid core of the East rift zone. During this process, if magmas of distinctive chemistry are present, they retain their chemical identity and the amount of cooling is comparable for magma trans- ported either upward or laterally to eruption sites. In- trusions within a few kilometers of the surface cool and crystallize to produce fractionated magma. Magma mix- ing occurs both within bodies of previously fractionated magma and when new magma intersects a preexisting reservoir. Magma is otherwise prevented from mixing, either by wall-rock septa or by differing thermal and density characteristics of the successive magma batches. Key words Kilauea 7 Magma mixing 7 Magmatic temperatures 7 Glass composition 7 Basalt composition 7 Magmatic plumbing/storage/transport Introduction The 1955 and 1960 lavas, both of which were erupted on Kilauea’s lower east rift zone (Fig. 1), are of great interest to volcanologists because of the light they shed on processes of magmatic storage, differentiation, mag- matic resupply, and eruptive behavior in a rift environ- ment. The 1960 lavas were erupted from reservoir(s) that were the source of the 1955 lavas and which were recharged in 1955 with relatively low-magnesia magma originating beneath Kilauea’s summit (Wright and Fiske 1971; Helz and Wright 1992). A similar recharge occurred in 1960 involving more magnesian magma (Murata and Richter 1966; Wright and Fiske 1971). Be- cause of the conceptual significance of these two erup- tions, we have revisited the petrogenesis of their la- vas. We have observed several different kinds of magma mixing in the 1960 lavas. Mixing occurs between stored magmas (fractionated compositions in equilibrium with olivinecclinopyroxenecplagioclase) and more primi- tive magmas (in equilibrium with olivine only). Both magma components are evident petrographically even though one component may be volumetrically minor. Mixing can also occur within stored magmas (“self-mix- ing”). Mixing occurred penecontemporaneous with the 1960 eruption (“recent” mixing) and also earlier, prob- ably dating from 1955. We have made new calculations to quantify the com- positional end members contributing to the composi- tion of each hybrid 1960 lava. Changes in the propor-
Transcript of T. L. Wright Differentiation and magma mixing on Kilauea’s ... · ity of Kilauea’s magma...
Bull Volcanol (1996) 57 :602–630 Q Springer-Verlag 1996
ORIGINAL PAPER
T. L. Wright 7 R. T. Helz
Differentiation and magma mixing on Kilauea’s east rift zone:
a further look at the eruptions of 1955 and 1960.
Part II. The 1960 lavas
Received: July 10, 1995 / Accepted: October 10, 1995
Editorial responsibility: M. Carroll
T. L. Wright (Y)Smithsonian Institution, National Museum of Natural History,NHB-119, Washington, DC 20560, USA
R. T. HelzU.S. Geological Survey, 107 National Center, Reston,Virginia 22092, USA
Abstract New and detailed petrographic observations,mineral compositional data, and whole-rock vs glasscompositional trends document magma mixing in lavaserupted from Kilauea’s lower east rift zone in 1960.Evidence includes the occurrence of heterogeneousphenocryst assemblages, including resorbed and re-versely zoned minerals in the lavas inferred to be hy-brids. Calculations suggest that this mixing, which isshown to have taken place within magma reservoirs re-charged at the end of the 1955 eruption, involved intro-duction of four different magmas. These magmas origi-nated beneath Kilauea’s summit and moved into therift reservoirs beginning 10 days after the eruption be-gan. We used microprobe analyses of glass to calculatetemperatures of liquids erupted in 1955 and 1960. Wethen used the calculated proportions of stored and re-charge components to estimate the temperature of therecharge components, and found those temperatures tobe consistent with the temperature of the same magmasas they appeared at Kilauea’s summit. Our studies rein-force conclusions reached in previous studies of Ki-lauea’s magmatic plumbing. We infer that magma en-ters shallow storage beneath Kilauea’s summit and alsomoves laterally into the fluid core of the East rift zone.During this process, if magmas of distinctive chemistryare present, they retain their chemical identity and theamount of cooling is comparable for magma trans-ported either upward or laterally to eruption sites. In-trusions within a few kilometers of the surface cool andcrystallize to produce fractionated magma. Magma mix-ing occurs both within bodies of previously fractionatedmagma and when new magma intersects a preexisting
reservoir. Magma is otherwise prevented from mixing,either by wall-rock septa or by differing thermal anddensity characteristics of the successive magmabatches.
Key words Kilauea 7 Magma mixing 7 Magmatictemperatures 7 Glass composition 7 Basaltcomposition 7 Magmatic plumbing/storage/transport
Introduction
The 1955 and 1960 lavas, both of which were eruptedon Kilauea’s lower east rift zone (Fig. 1), are of greatinterest to volcanologists because of the light they shedon processes of magmatic storage, differentiation, mag-matic resupply, and eruptive behavior in a rift environ-ment. The 1960 lavas were erupted from reservoir(s)that were the source of the 1955 lavas and which wererecharged in 1955 with relatively low-magnesia magmaoriginating beneath Kilauea’s summit (Wright andFiske 1971; Helz and Wright 1992). A similar rechargeoccurred in 1960 involving more magnesian magma(Murata and Richter 1966; Wright and Fiske 1971). Be-cause of the conceptual significance of these two erup-tions, we have revisited the petrogenesis of their la-vas.
We have observed several different kinds of magmamixing in the 1960 lavas. Mixing occurs between storedmagmas (fractionated compositions in equilibrium witholivinecclinopyroxenecplagioclase) and more primi-tive magmas (in equilibrium with olivine only). Bothmagma components are evident petrographically eventhough one component may be volumetrically minor.Mixing can also occur within stored magmas (“self-mix-ing”). Mixing occurred penecontemporaneous with the1960 eruption (“recent” mixing) and also earlier, prob-ably dating from 1955.
We have made new calculations to quantify the com-positional end members contributing to the composi-tion of each hybrid 1960 lava. Changes in the propor-
603
Fig. 1 Index map showing thelower part of Kilauea’s eastrift zone and vents and lavaflows active in 1955 and 1960
tions of end members with time of eruption providesadditional insight into the spatial and temporal continu-ity of Kilauea’s magma reservoirs and pathways. Final-ly, we evaluate the thermal character of the magma res-ervoir that fed the 1955 and 1960 eruptions and pro-pose a general model for magma transport and storagebeneath Kilauea’s east rift zone.
Magma mixing: general background and previous work
at Kilauea
Many workers have contributed to the development ofcriteria for recognizing and quantifying magma mixingand hybridization in volcanic rocks. Classic studies ofmixing in calc-alkalic systems include Eichelberger(1975) and Anderson (1976), among others. Mixing hasalso been recognized and evaluated in mid-ocean ridgebasalts (e.g., Rhodes et al. 1979). In the absence ofgross visible heterogeneity of the rocks, petrographicfeatures which are commonly attributed to mixing in-clude (a) hybrid phenocryst assemblages, with toomany, or mutually incompatible, phases, (b) bimodalcompositional ranges in minerals capable of continuoussolid solution, (c) widespread occurrence of reverselyzoned minerals, (d) widespread occurrence of resorbedcrystals, and (e) occurrences of very disparate glasses asinclusions in phenocryst phases.
Recognition of the role that mixing and hybridiza-tion can play at basaltic volcanoes, such as Kilauea, has
been more difficult because the effects of mixing andhybridization are subtler. Nevertheless, we believe thatmagma mixing is as important in this setting as else-where. It is appropriate to reevaluate the 1960 eruptionbecause it was the first eruption at Kilauea recognizedas having produced hybrid lavas.
The 1960 east rift eruption was originally describedby Murata and Richter (Murata and Richter 1966;Richter and Murata 1966) who viewed it as essentiallycontinuous with the 1959 summit eruption. These work-ers invoked magma mixing to explain petrographic andcompositional features exhibited by the lavas producedin the middle of the eruption. They considered the ear-liest 1960 lavas to have been derived from magma leftunerupted in 1955, but changes in the petrographiccharacter and chemistry of the lavas erupted after 21January led them to infer the existence of a second,hotter magmatic component in the later erupted lavas.The middle stages of the eruption were interpreted tobe mixed magmas produced by combining the new andstored components.
Taking the observations of Richter and Murata onthe 1960 eruption, along with additional data on the1955 eruption, Wright and Fiske (1971) expanded onthe idea of hybridization, offering a different hypothe-sis to explain the compositional shifts in both of theseeruptions. Wright and Fiske proposed that several dif-ferent batches of magma from Kilauea’s summit wereneeded to produce the late 1960 lavas. Their study,
604
which used least-squares calculations, but which did notinclude petrographic and phase chemical data, pro-vided permissive, but not conclusive, evidence for theprocess. A slightly later paper by Anderson and Wright(1972) interpreted textures and mineral and glass com-positions as evidence for magma mixing in the early1955 lavas, but did not look at the late 1955 or any ofthe 1960 lavas.
More recent studies documenting magma mixing ef-fects at Kilauea include the work of Garcia et al. (1989,1992) on the Puu Oo-Kupaianaha eruption, Clague etal. (1995) on samples from the submarine extension ofKilauea’s east rift zone, and Helz and Wright (1992) onthe late 1955 east rift lavas (Part I of the present study).A key feature of this previous study is our examinationof glassy spatter samples, in preference to holocrystal-line flow samples, in order to see pre-eruptive pro-cesses as clearly as possible.
Compositions and textures of early-formed minerals
The 1960 lavas contain phenocrysts of olivine, augite,and plagioclase. In addition we have observed rare, mi-crophenocryst-sized crystals of hypersthene, which ap-pear to be xenocrystic to their immediate hosts, and il-menite, which occurs as an included phase in hyper-sthene. Again, as in Part I, we focus on the coarser,more texturally varied crystals in the samples investi-gated, because they record more prior history thansmaller and/or homogeneous crystals. For this study weexamined four samples of the early 1960 lavas (erupted13–21 January 1960), three intermediate samples(erupted 22–30 January 1960), and seven late samples(erupted 2–18 February 1960); most of these are near-vent scoria and spatter.
Plagioclase
Representative compositions of plagioclase pheno-crysts, microphenocrysts, and xenocrysts are tabulatedin Table A1 and plotted in Fig. 2. The compositionalrange seen in the 1955 and 1960 lavas (An51–80) is large,almost as large as the range (An47.7–81.3) observed byClague et al. (1995) in plagioclase phenocrysts in lavasfrom the entire submarine extension of Kilauea’s eastrift zone.
In the four early 1960 spatter samples, plagioclasephenocrysts are lath-like and mostly normally zoned,with few textural complications. One exception is a re-sorbed and moth-eaten crystal in KP-4 which showssome reverse zoning. Coarse phenocrysts are rare; thelargest plagioclase observed in our sections is a crystal4 mm in length in sample KP-8. Compositions of allplagioclase crystals (Fig. 2C) range from An76 to An56;the maximum range observed in a single phenocryst isAn76–63. Microlites show a more restricted range (An60.5
to An65.0), reflecting the narrow range of melt composi-
tions seen in these samples. The amount of phenocrys-tic and microphenocrystic plagioclase present in thesesamples is 9–10% by volume (data of Richter and Mu-rata 1966).
Plagioclase phenocrysts in the late 1960 samples arealmost all strongly resorbed (Fig. 3), as was also docu-mented by Richter and Murata (1966, Fig. 6a, b) intheir original description. The plagioclase phenocrystsare consistently calcic, ranging from An65 to An80 (Fig.2). Calcic plagioclase in the late 1960 samples also oc-curs as inclusions in phenocrysts of iron-rich (Fo76–78)olivine (see columns 7 and 10 in Table A1). Three ofthe latest samples examined have no phenocrystic orother early plagioclase at all; these are KP-25, the hotscoria in KP-26 (glass compositions in columns 3 and 4in Table 1b), and KP-27, erupted from 13 February1960 to 18 February 1960. The differentiated scoria inKP-26 (glass composition in column 5 in Table 1b) doescontain phenocrystic plagioclase, which shows markedreverse zoning (columns 8 and 9 in Table A1). Richterand Murata (1966) show plagioclase phenocryst contentdeclining from 4% in some intermediate samples to~1% in the late 1960 lavas, consistent with our obser-vations.
Figure 2 also contains data on plagioclase pheno-cryst compositions from samples of the early and late1955 lavas, including unpublished analyses of Ho andGarcia (1988), to facilitate comparison with the 1960data. The range of observed compositions for plagio-clase phenocrysts is quite large for both the early andlate 1955 lavas, with compositions for plagioclase in thelate 1955 lavas being clearly bimodal (Fig. 2B). Therange in groundmass compositions in the early 1955samples is An55–61, which lies within the low end of therange of phenocryst compositions (Fig. 2A). The pat-tern of compositional data for plagioclase in the early1955 lavas is thus similar to that of the early 1960 lavas.Although each has a continuous, rather than bimodal,range in plagioclase phenocryst compositions, thegroundmass plagioclase falls within the range of pheno-cryst compositions, rather than extending below it. Thisimplies that neither end of the phenocryst range is inequilibrium with the observed liquids in these sam-ples.
Pyroxenes
Augite is the only pyroxene phenocryst present in the1960 lavas, whereas the more magnesian late 1960 lavascontain larger crystals of both augite and hypersthene.Representative pyroxene compositions (Table A2) cov-er the range of composition previously observed in eastrift lavas (Anderson and Wright 1972; Clague et al.1995).
In the early 1960 lavas, augite phenocrysts (typically1–2 mm in length) are common (6–7% by volume, ac-cording to Richter and Murata 1966). In some of theearliest samples (KP-4, KP-35) augite is deep olive-
605
Fig. 2A–D Plagioclase compositions observed in samples fromthe early and late parts of the 1955 and 1960 eruptions. Each sym-bol corresponds to a single analyzed point. Arrows on early 1955and early 1960 plots show range of microlite compositions
Fig. 3A, B Effects of magmamixing on plagioclase in late1960 lavas. A Resorbed plag-ioclase phenocryst in sampleKP-22, erupted on 4 February1960. The grain is 0.5 mmlong, with the compositionshown in column 3 in TableA1. Nicols partly crossed.B Clump of augite c plagio-clase microphenocrysts insame sample, showing differ-ent extent of resorption onplagioclase and augite. Aggre-gate is 2 mm across
606
Tab
le1a
Com
posi
tion
of
glas
ses
from
spa
tter
sam
ples
fro
m t
he e
arly
and
mid
dle
part
s of
the
196
0 er
upti
on
Col
umn
no.
Fie
ld n
o.D
ate
erup
ted
No.
of
poin
ts
1 KP
-41/
17/6
0E
arly
(st
ored
)6
2 KP
-49
1/17
/60
Ear
ly (
stor
ed)
3 KP
-35
1/17
/60
Ear
ly (
stor
ed)
4
4 KP
-35
1/17
/60
Ear
ly (
stor
ed)
10
5 KP
-71/
21/6
0E
arly
(st
ored
)9
6 KP
-81/
21/6
0E
arly
(st
ored
)8
7 KP
-15
1/29
/60
Lat
e (h
ybri
d)7
8 KP
-16
1/29
/60
Lat
e (h
ybri
d)8
9 KP
-19
2/1/
60L
ate
(hyb
rid)
8
10 KP
-19g
2/1/
60L
ate
(hyb
rid)
SiO
2
TiO
2
Al 2
O3
Cr 2
O3
SF
eOa
MnO
MgO
CaO
Na 2
OK
2O
P2O
2
50.7 3.66
13.4 0.00
12.6 0.24
5.24
9.56
3.12
0.81
0.37
50.9
53.
5813
.58
P 12.4
40.
185.
779.
772.
610.
720.
39
50.6 3.61
13.4 0.00
12.5 0.20
5.42
9.71
2.78
0.77
0.38
50.6 3.62
13.4 0.00
12.4 0.20
5.49
9.82
2.77
0.77
0.36
50.4 3.68
13.5 0.00
12.4 0.19
5.57
9.84
2.77
0.76
0.39
50.4 4.24
12.4 0.00
13.7 0.19
5.00
9.38
2.64
0.90
0.48
50.8 3.59
13.2 0.01
12.4 0.20
5.90
9.99
2.69
0.73
0.35
50.4 3.16
13.9 0.03
11.5 0.16
6.15
10.7
42.
690.
660.
28
50.5 3.17
13.8 0.01
11.4 0.17
6.21
10.7
22.
550.
660.
28
50.6
22.
9614
.05
P 11.2
40.
176.
7410
.84
2.49
0.59
0.30
Sum
99.8
810
0.00
b99
.17
99.4
399
.50
99.3
399
.86
99.6
799
.47
100.
00b
TM
gO11
2211
2211
2311
2511
1311
3311
3711
38
Com
men
tsR
eplic
ate
anal
ysis
Abu
ndan
tcr
ysta
llite
sin
gla
ss
Mur
ata
and
Ric
hter
no.
F-3
F-3
gP
PF
-5F
-6F
-8P
F-1
1F
-11g
USN
M n
o.11
611
2.03
116
112.
3411
611
2.06
116
112.
0711
611
2.14
116
112.
1511
611
2.18
aT
otal
iro
n as
FeO
bG
lass
ana
lyse
s fr
om M
urat
a an
d R
icht
er (
1966
), w
ith
all
iron
as
FeO
, ren
orm
aliz
ed t
o 10
0%
607
Tab
le1b
Com
posi
tion
of
glas
ses
from
spa
tter
sam
ples
fro
m t
he l
ate
1960
eru
ptio
n
Col
umn
no.
Fie
ld n
o.D
ate
erup
ted
No.
of
poin
ts
1 KP
-25
2/13
/60
Lat
e (h
ybri
d)8
2 KP
-25
2/13
/60
Lat
e (h
ybri
d)5
3 KP
-26
2/16
/60
Lat
e (h
ybri
d)4
4 KP
-26
2/16
/60
Lat
e (h
ybri
d)7
5 KP
-26
2/16
/60
Lat
e (h
ybri
d)4
6 KP
-27
2/18
/60
Lat
e (h
ybri
d)6
7 KP
-27
2/18
/60
Lat
e (h
ybri
d)10
8 KP
-27g
2/18
/60
Lat
e (h
ybri
d)
SiO
2
TiO
2
Al 2
O3
Cr 2
O3
SF
eOa
MnO
MgO
CaO
Na 2
OK
2O
P2O
5
50.6 2.97
13.9 0.03
10.9 0.18
6.48
11.1
22.
570.
590.
27
50.5 3.01
13.9 0.01
10.8 0.17
6.47
11.0
92.
690.
630.
26
50.6 3.01
13.9 0.00
10.9 0.24
6.45
10.8
32.
740.
650.
28
50.6 3.00
13.9 0.00
10.9 0.17
6.42
10.9
32.
720.
640.
26
50.8 3.44
13.6 0.02
12.0 0.18
5.76
9.96
2.79
0.74
0.32
51.1 3.00
13.6 0.00
11.0 0.17
6.40
11.0
22.
800.
640.
30
51.3 3.03
13.8
P11
.0 0.16
6.39
10.9
72.
740.
650.
27
50.5
92.
9613
.97
11.0 0.17
6.97
10.8
12.
540.
600.
30
Sum
99.6
199
.53
99.6
199
.54
99.4
110
0.05
100.
3110
0.00
TM
gO11
4411
4411
4311
4211
2911
4211
42
Com
men
tsR
eplic
ate
anal
ysis
Scor
ia 1
Scor
ia 2
Scor
ia 3
diff
eren
tiat
ed
Mur
ata
and
Ric
hter
no.
F-1
7F
-17
F-1
8F
-18
F-1
8F
-19
F-1
9F
-19g
USN
M n
o.11
611
2.24
116
112.
2511
611
2.26
aT
otal
iro
n as
FeO
bG
lass
ana
lyse
s fr
om M
urat
a an
d R
icht
er (
1966
), w
ith
all
iron
as
FeO
, ren
orm
aliz
ed t
o 10
0%
608
Fig. 4 Scanning electron microscopy (SEM) image of a com-pound pyroxene phenocryst in sample KP-24. This 1-mm-longgrain consists of a thick blade of hypersthene, shown by arrow,surrounded by relatively Fe-rich augite. This core has undergonetwo distinct episodes of partial resorption, followed by growth ofa more Mg-rich rim. The microprobe traverse shown in Fig. 5 isshown by a dashed line. Compositions of the different parts of thisgrain are given in columns 7–10 in Table A2
green in color; in most other samples it is brownish, butstill fairly strongly colored. In contrast to the early 1955lavas, where the green augite is distinctly more ferroan(Anderson and Wright 1972; Table 2) than the pale au-gite found in the same samples, the differently coloredaugites in the early 1960 lavas do not have differentcompositions. Zoning, which can be either normal orreverse, is dominated by the Ca–Fe exchange, withoutlarge changes in MgO content. The zoned crystals (Ta-ble A2) show a substantial increase in Cr in the Fe-poor rims, consistent with the interpretation that this isreverse zoning, even though the MgO content of therims is similar to that of the cores.
Augite phenocrysts in the later 1960 lavas are stillabundant (3–5% according to Richter and Murata1966), and unlike plagioclase, rarely have resorbed out-lines. The contrast between euhedral augite and re-sorbed plagioclase in an augite–plagioclase clot isshown in Fig. 3B. More detailed examination of augitephenocrysts, involving electron-microprobe traversesand scanning electron microscopy (SEM) imaging, hasrevealed complex histories of multiple episodes of re-sorption and/or reverse zoning in individual augite phe-nocrysts, even those with euhedral outlines.
A compound phenocryst from sample KP-24 (Fig. 4)illustrates some of the complexities found in pyroxenephenocrysts from the stored magma. It consists of athick lamella of hypersthene surrounded by relativelybright Fe-rich augite (column 7 in Table A2). The rag-ged outline of the Fe-rich augite is surrounded by asomewhat more magnesian augite (column 8 in TableA2) which fills in the deepest embayments. That zone is
mostly euhedral, but with some local resorption, and isin turn surrounded by a distinctly darker rim of muchmore magnesian and Cr-rich augite (column 9 in TableA2) which gives the grain its present euhedral outlineagainst the groundmass. A microprobe traverse acrossthe left side of the grain (Fig. 5) shows the two stages ofreverse zoning in more detail.
The sector-zoned augite shown in Fig. 6 also shows acomplex history, even though it originated in a hotterrecharge component. The sectors in this crystal differ atboth major- and minor-element levels, as can be seen inthe microprobe traverse shown in Fig. 7. The smalldark trapezoidal area near the core (column 12 in Ta-ble A2) is subcalcic augite like that found in a sector-zoned augite in one of the late 1955 lavas (Helz andWright 1992, Fig. 4e). This composition is metastable,plotting within the pyroxene solvus as defined by Rossand Huebner (1979). The remainder of the dark sector(the part that is inclusion-ridden) is augite with Ca, Mg,and Fe contents very similar to those of the lighter,dominant sectors, but with Al and Ti contents likethose of the metastable, subcalcic augite. The sectoredcore has been slightly resorbed around its edge; thedarker, slightly more magnesian augite rim truncatesthe sectoring, and is uniform in composition around thecrystal (see Fig. 7). Clearly, this grain records an initiallarge departure from equilibrium, and has had a fairlycomplicated subsequent history as well. (Not all sec-tored augite in the late 1960 lavas is complex; the ana-lyses in columns 18 and 19 in Table A2 are from a sec-tored augite phenocryst where there is no composition-al difference between sectors. In this crystal the sector-ing must be due to slight differences in orientation ofthe crystal lattice only, and the departure from equili-brium during its growth is slight.)
Hypersthene as a phenocrystic mineral is rare at Ki-lauea. It is fairly common in the early 1955 lavas (Mac-donald and Eaton 1964; Anderson and Wright 1972),but has not been observed in the late 1955 samples. Itwas reported by Richter and Murata (1966) as rare inthe early 1960 lavas, detected only in heavy-mineralseparates; none occurs in any of our sections. However,we have found hypersthene in two of the late 1960 sam-ples, where it has not been observed previously.
Hypersthene in the late 1960 lavas has a uniformcomposition (Table A2), but its textural relationship tothe host varies. The thick blade of hypersthene en-closed in the complexly zoned augite described above(Fig. 4) is in equilibrium with the adjacent Fe-rich au-gite (columns 6 and 7 in Table A2). Hypersthene alsooccurs as clusters of microphenocryst-sized, resorbedcrystals surrounded by thin rims of magnesian augite(Fig. 8A; analyses in columns 3 and 4 in Table A2); therims are not, in this case, in equilibrium with the hy-persthene. One hypersthene crystal (Fig. 8A) containsan inclusion of ilmenite (column 5 in Table A2). Thepresence of the ilmenite inclusion, and the iron-richcharacter of the hypersthene itself, confirm that thesecrystals have come from a fractionated source. Hyper-
609
Fig. 5 Microprobe traverse of compound pyroxene shown in Fig.4. The Fe, Ti, and Cr profiles show best the two distinct rimsalong the upper edge of the grain. Ca, Mg, and Fe for the opxlamella have been omitted from this figure because they plot off-scale, but the Al and Ti data show the contrast between the mi-nor-element concentrations in opx and augite
Fig. 6 SEM image of a sector-zoned augite microphenocryst insample KP-25, erupted 13 February 1960. The crystal is 0.5 mmacross. Compositions of different parts of this grain are given incolumns 10–12 in Table 1b. The arrows indicate the orientation ofthe microprobe traverse shown in Fig. 7
Fig. 7 Microprobe traverse of the sector-zoned augite shown inFig. 6. Note the contrast in Al, Ti, and Cr contents of the augite inthe dark sector vs the augite in the light sector. The extremelylow-Ca cpx makes up only a very small region near the center ofthe grain
sthene is present also in the piece of differentiated sco-ria in sample KP-26. It is in direct contact with the melt,but is not in equilibrium with the melt or with any ofthe augite present in the same glass (columns 14–16 inTable A2).
Olivine
Olivine is the most abundant phenocrystic phase in the1960 lavas. In early 1960 lavas, olivine is relatively small(~2 mm in length) and sparse (1–2% by volume ac-cording to Richter and Murata 1966), euhedral in formand normally zoned (columns 1 and 2 in Table A3).The late 1960 lavas have abundant olivine phenocrysts(5–14% by volume according to Richter and Murata1966), with a variety of core compositions (Fig. 9), zon-ing patterns (Fig. 10), and textures (Figs. 11–12), whichprovide vivid documentation of recent magma mixingin the chamber from which the lavas were erupted.
Core compositions (Fig. 9) range from Fo76 to Fo88,and fall into three compositional groups. These proba-
610
Fig. 8A, B Occurrence of hy-persthene in the 1960 lavas.A Cluster of hypersthene crys-tals with thin rims of magne-sian augite, in sample KP-24,erupted on 12 February 1960.One hypersthene grain con-tains an inclusion of ilmenite.Phase compositions given incolumns 3–5 in Table A2.Cluster is 1.5 mm across.B Hypersthene crystal (1 mmlong) in differentiated scoriain sample KP-26, erupted 16February 1960. The grain, thecomposition of which is givenin column 14 in Table A2, isresorbed, but lacks any augiterim
Fig. 9 Frequency distribution of core compositions of olivinephenocrysts, microphenocrysts, and xenocrysts in the 1960 lavas.Includes some data from early sample KP-4, as well as from thelate samples. The data fall into three groups corresponding tocooler magma stored in the rift, hotter magma introduced fromKilauea’s summit, and the magma hybridized by mixing
Fig. 10 Olivine zoning in samples of the late 1960 lavas (solidsymbols core compositions; open symbols rim composition). Thetwo vertical lines bracket the range of initial olivine compositionswhich would be in equilibrium with the liquid at the first appear-ance of plagioclase, assuming that the distribution for Fe–Mg ex-change between olivine and liquid is 0.27–0.30 (Roeder and Ems-lie 1970). Olivine compositions lying to the right or left of thisband must have crystallized from more Mg-rich or less Mg-richliquids, respectively. Note that grains with core compositions ly-ing outside the vertical lines are zoned toward intermediate com-positions
611
Fig. 11A, B Olivines inheritedfrom the stored component. ASEM image of a resorbed, re-versely zoned olivine pheno-cryst in sample KP-24, eruptedon 12 February 1960. Thisgrain is relatively Fe-rich; thisfact, plus the presence of plag-ioclase inclusions (black poly-gonal areas in core), and theabsence of chromite inclusionsall suggest that this grain crys-tallized from a differentiatedmagma. The grain is 1.2 mmacross. B Inclusion-rich olivinefragment in sample KP-25,erupted 12 February 1960.The grain is 1.5 mm long andhas been broken at both ends.It shows slight reverse zoning(compositions are given in co-lumns 6 and 7 in Table A3).The inclusions are compound,containing glass c chromiteB sulfide c vapor. Crossednicols
bly represent, from low to high Fo content, olivine crys-tallized from the stored component, olivine growing inthe newly created hybrid magma, and olivine broughtin with the various primitive components. The zoningpatterns (Fig. 10) support this inference. Five of the sixsamples examined contain olivine phenocrysts withboth Fe-rich and Mg-rich cores, neither of which werestable in the present (hybrid) melts. The zoning ofthese phenocrysts is toward a central range of olivinecompositions. This range (Fo79.5–82.5) corresponds tothe olivine which would be stable at the melt MgO con-tent where plagioclase begins to crystallize in Kilaueanmelts (approximately 7.0% MgO). Olivines with corecompositions in this range are mostly unzoned; reverse-ly zoned crystals in some of the latest samples may re-flect repeated recharge of the chamber. Sample KP-27,the last pumice sample collected at the very end of theeruption, is completely free of phenocrysts of Fe-richolivine, plagioclase, or augite.
The textures observed in the Fe-rich olivine pheno-crysts (Fig. 11) confirm the inferences drawn from thedistribution of core compositions and from zoning pat-terns. Figure 11A shows a very strongly resorbed grainwith a cluster of included plagioclase grains. The core isFe-rich (columns 4 and 5 in Table A3); the very narrow,more magnesian rim follows the elaborately resorbedoutline of the grain, showing unequivocally that the re-verse zoning postdates the resorption. The presence ofplagioclase inclusions in this grain provides additionalevidence that it grew in a relatively differentiated mag-ma.
Figure 11B shows an elongate olivine grain with aninclusion-rich core and clear rims, approximately30 mm thick, along its sides (compositions in columns 6and 7 in Table A3). The rim has been locally resorbed,as can be seen by its curving upper contact. In one
place along this side, the rim has been completely dis-solved away, and the inclusion-rich core is in direct con-tact with the melt. All zoning is cleanly truncated at thebroken ends of the fragment, showing that the break-age postdates the development of the reversely zonedrim, and (probably) the limited resorption observedalong it. There is no differential resorption of the dif-ferent zones along the broken surfaces, nor any devel-opment of more forsteritic rims on the core region ex-posed by the breaks, which suggests that the breakageis very recent. This grain thus shows evidence of twostages of hybridization and/or disruption in the magmachamber from which it was derived.
The very forsteritic olivines include euhedral crystalsand deformed grains, irregular in form, with multipleextinction discontinuities. The latter occur singly or inaggregates, measuring up to 5 mm across (Fig. 12).Their core compositions range from Fo84 to Fo88. Thesegrains and aggregates are identical in appearance andin their compositional range to the large, blocky, de-formed olivines and dunitic aggregates observed in the1959 summit lavas and in Kilauea Iki lava lake. (Rich-ter and Murata (1966) noted the coarse grain size andnoneuhedral forms of olivine phenocrysts in both the1959 and 1960 lavas, but did not mention deformationfeatures in either. Olivines like this are also quite com-mon in the submarine east rift lavas as stated in Clagueet al. 1995.) Helz (1987a) interpreted the deformed,blocky olivines in the 1959 lavas as xenocrysts, and sug-gested that they were derived from olivine cumulates,precipitated from earlier Kilauean lavas, but now trans-ected by Kilauea’s current plumbing system, an ideageneralized recently to apply to other Kilauean picrites(Clague and Denlinger 1994). We believe the coarse,deformed olivines in the late 1960 lavas have the sameorigin. The fact that their NiO contents cover the same
612
Fig. 12 Aggregate of mildly deformed olivine in sample KB-22.The aggregate is 5 mm across. Grains in the aggregate are forster-itic (composition given in column 3 in Table A3), except for avery thin rim next to the devitrified matrix, which reaches Fo 82.5locally. All grains in the aggregate are deformed, and almost allcoarse chromite (black) is found at grain boundaries within theaggregate, rather than as inclusions in the olivine. This aggregatevery closely resembles the commonest type of olivine aggregate inthe 1959 summit lavas (Fig. 5B and 7C in Helz 1987a)
range as those of the euhedral forsteritic phenocrystspresent in the same samples (see Table A3) is consist-ent with this suggestion.
Chemical variation: whole-rock and glass compositions
Murata and Richter (1966) published major-oxidechemical analyses of lava samples collected during theeruption. Their data is reproduced in Table 2, withtrace element data (Table A4) analyzed by nondisper-sive X-ray fluorescence as described in Part I (Helz andWright 1992, p. 374ff). Selected major oxides (Al2O3,total iron such as FeO, CaO, and TiO2) are plottedagainst MgO (Fig. 13A) and selected incompatible (Zr,Sr) and compatible (Cr, Ni) trace elements are alsoplotted against MgO (Fig. 13B). In addition, Fig. 13Ashows the compositions from selected summit eruptionswhich occurred close to, both before and after, the 1960east rift eruption. The lines extending back to 12%MgO are “olivine control” lines, which assume an aver-age olivine phenocryst composition of Fo87. Bulk com-positions falling along such control lines have been ob-served in the 1959 lavas (Wright 1971; Wright 1973)and have been inferred to exist for other Kilauean mag-mas (Wright 1971).
As can be seen in Fig. 13A, the early 1960 lavas(samples erupted from 13 January through 21 January)have a limited range of compositions (MgOp6.0–6.5%), similar to those of lavas erupted near the end ofthe 1955 eruption. The main difference between theearly 1960 lavas and the late 1955 lavas is that the 1960lavas are somewhat lower in FeO at a given MgO con-
tent. These samples define the “stored” component ofthe 1960 eruption. The MgO content of lava increasesafter 26 January, ranging between 7 and 13%.
The expected liquid line of descent for Kilaueancompositions should follow the “olivine control” slopeshown until three-phase fractionation begins at approx-imately 7.0% MgO. In two plots (those for CaO andFeO) the 1960 samples define slopes on the variationdiagrams which deviate from olivine-only fractionationat MgO contents where multiphase fractionation is notpossible [see Thompson and Tilley’s (1969) experimen-tal results on the crystallization behavior of the 1960lavas]. In addition, for certain oxides the 1960 bulkcompositions fall well outside the range defined byerupted Kilauean magmas or their inferred, more mag-nesian parents: In particular, TiO2 is too high and CaOtoo low for the 1960 lavas to have been derived by frac-tionation of any known Kilauean parental lava.
Even more telling evidence for the hybrid nature ofthe 1960 lavas can be found in looking at the actualglass compositions found in these samples (shown inTable 1 and in Fig. 13A). The glass analyses were ob-tained as described in Part I (Helz and Wright 1992);quenching temperatures (TMgO) are inferred using thecalibration of Helz and Thornber (1987). Replicateanalyses for several samples give some indication of theuncertainty in the analyses including both the analyticaluncertainty and the effects of slight heterogeneity inthe glasses. Most of the spatter samples analyzed wereuniform in composition and texture, even when multi-ple lapilli were examined. Sample KP-26, however, con-tained two lapilli with hotter, less-differentiated glass(columns 3 and 4 in Table 1b) and a third fragment ofcooler, more differentiated glass (column 5 in Table1b). The difference between these glass compositions isanalytically significant for all components.
The glass compositions unequivocally define the lo-cation of the liquid line of descent for the 1960 bulkcompositions. Figure 13A shows that the whole-rocktrends for the intermediate and late lavas diverge wide-ly from that line. This is especially conspicuous in theplot of CaO vs MgO, where the glasses continue toshow a positive correlation between the two oxides,with a peak at 11.2% CaO and 7.0% MgO, the point atwhich plagioclase begins to crystallize in Kilauean mag-mas. The whole-rock compositions, by contrast, showCaO peaking at ~10.5% CaO. The divergence be-tween melt and bulk-rock compositional trends is notpeculiar to the microprobe analyses: Three wet-chemi-cal analyses of glass separates from the 1960 eruption,presented by Murata and Richter (1966), and plotted inFig. 13A for comparison, are similar to the microprobeanalyses. (Their slightly higher MgO content mostprobably reflects the presence of very minor amountsof olivine in the glass separates.) These data provideunequivocal evidence that the later 1960 bulk composi-tional trends diverge widely from the true liquid line ofdescent of Kilauean liquids in general, and the 1960equilibrium liquids in particular.
613
Fig. 13A, B Selected MgOvariation diagrams. A repre-sentative liquid line of descentfor Kilauean magmas extendsfrom olivine control lines ex-trapolated to approximately7.0% MgO and the 1955 glass(open squares). The 1960 lavacompositions do not follow ol-ivine control lines; instead,they can be modeled as a mix-ture of one or more composi-tions lying on a liquid line ofdescent with olivine-rich li-quids falling on olivine controllines or their extrapolation tomore magnesian compositions.A Al2O3, total iron as FeO,CaO, and TiO2, showing rela-tionships between glass andwhole-rock compositions fromthe 1955 and 1960 eruptions.B Selected incompatible ele-ments (Zr, Sr) and compatible(Cr, Ni) trace elements
Discussion: petrographic evidence for mixing in the
1960 lavas
The early 1960 lavas
All of the minerals present in the early lavas of Ki-lauea’s 1960 east rift eruption record evidence for mag-ma mixing, despite the great uniformity in the glasscompositions in the samples investigated. This evidenceincludes the presence of reversely zoned augite andnormally zoned olivine phenocrysts/microphenocrysts
in the same sample (KP-4; columns 1 and 2 in TablesA2 and A3). The rim compositions from KP-4 are inequilibrium with each other, and are presumably inequilibrium with the present host melt (columns 1 and2 in Table 1a). This implies that the core compositionsof these same microphenocrysts are not in equilibriumwith each other, and that neither core is in equilibriumwith the host liquid. The narrowness of the zoned rimssuggest that the mixing occurred very close to the timeof eruption. The early 1960 lavas also contain resorbed,reversely zoned plagioclase microphenocrysts.
614
Tab
le2
Maj
or-o
xide
a an
d tr
ace
elem
entb
data
for
sam
ples
fro
m t
he 1
960
erup
tion
(F
-1 t
o F
-20)
and
fro
m K
ilaue
a’s
sum
mit
(19
52–1
967)
ID n
o.F
ield
no.
Eru
ptio
nda
te
F-1
1/13
F-6
KP
-08
1/21
F-7
KP
-13
1/26
F-9
KP
-17
1/30
F-1
1K
P-1
92/
1
F-1
2K
P-2
02/
2
F-1
3K
P-2
12/
4
F-1
4K
P-2
22/
4
F-1
6K
P-2
42/
12
F-1
7K
P-2
52/
13
F-1
8K
P-2
62/
16
F-1
9K
P-2
72/
18
F-2
0K
P-2
82/
18
1960
Ec
1952
c19
61c
1967
c19
59 S
-1c
SiO
2
Al 2
O3
FeO
tot
alM
gOC
aON
a 2O
K2O
TiO
2
P2O
5
MnO
50.6
813
.91
11.5
06.
5710
.45
2.59
0.65
3.12
0.35
0.18
50.9
813
.64
11.9
46.
199.
992.
700.
673.
370.
370.
18
50.5
813
.55
11.7
46.
9910
.21
2.60
0.62
3.18
0.34
0.18
49.7
412
.99
11.5
59.
4410
.03
2.34
0.57
2.84
0.31
0.18
49.3
712
.38
11.4
411
.31
9.77
2.17
0.53
2.60
0.27
0.17
49.4
812
.60
11.5
510
.61
9.83
2.25
0.54
2.70
0.28
0.17
49.0
212
.02
11.5
212
.15
9.54
2.16
0.50
2.61
0.31
0.18
49.3
012
.27
11.4
811
.31
9.71
2.26
0.52
2.70
0.28
0.18
49.1
811
.89
11.5
312
.34
9.45
2.10
0.50
2.56
0.26
0.18
48.9
511
.59
11.5
113
.18
9.32
2.09
0.49
2.43
0.26
0.18
49.4
412
.27
11.4
511
.15
9.88
2.23
0.51
2.62
0.28
0.17
49.0
411
.74
11.5
712
.49
9.49
2.17
0.48
2.56
0.28
0.18
49.2
711
.99
11.4
611
.85
9.61
2.20
0.50
2.66
0.27
0.18
50.8
13.7
711
.74
6.33
10.1
82.
670.
673.
280.
370.
19
50.3
213
.91
11.3
27.
1411
.36
2.26
0.53
2.72
0.27
0.17
50.3
813
.70
10.8
37.
6411
.36
2.32
0.55
2.77
0.27
0.18
50.4
513
.65
11.2
67.
6011
.13
2.31
0.54
2.64
0.26
0.17
50.1
012
.88
11.2
18.
1111
.97
2.14
0.55
2.63
0.25
0.17
Tot
al10
0.00
100.
0010
0.00
100.
0010
0.00
100.
0010
0.00
100.
0010
0.00
100.
0010
0.00
100.
0010
0.00
100.
010
0.00
100.
0010
0.00
100.
0
Rb
Sr Y Zr
Nb
Ba
Ni
Cu
Zn
Cr
10 418 33 218 24 176
118
140
129
222
11 428 34 229 26 173 91 134
134
149
11 408 34 211 24 170
140
143
136
247
938
6 28 187 23 149
270
133
126
507
836
3 25 169 20 152
312
126
122
686
P P P P P P P P P P
735
0 26 170 18 129
383
125
122
771
10 372 27 179 19 146
337
119
118
665
11 361 26 169 21 138
402
130
115
718
934
4 26 159 18 144
451
131
119
790
936
5 28 178 17 144
295
116
125
665
834
8 28 165 19 131
421
130
123
817
11 369 26 171 20 141
331
127
115
658
11 423 34 224 25 174
105
137
132
185
942
7 29 173 22 178
122
127
122
307
11 413 29 181 23 156
135
138
122
399
939
9 26 172 15 132
115
126
116
411
537
1 26 165 25 165
122
123
111
439
USN
Md
.07
.12
.16
.18
.20
.21
.23
.24
.25
.26
.27
NO
TE
:a
Dat
a fo
r th
e 19
60 s
ampl
es f
rom
Mur
ata
and
Ric
hter
(19
66; s
ee a
lso
Tab
le 1
). D
ata
are
norm
aliz
ed t
o 10
0% a
fter
con
vert
ing
iron
to
FeO
bD
ata
from
Tab
le A
4. N
orm
aliz
ed v
alue
s (e
.g.,
Rb
n) a
re t
aken
fro
m t
he “
Dec
” co
lum
nsc
Maj
or o
xide
ave
rage
s fo
r 19
52, 1
961,
and
196
7 fr
om W
righ
t an
d F
iske
(19
71; s
ee a
lso
Tab
le 8
); m
ajor
oxi
de d
ata
for
1959
S-1
fro
m M
urat
a an
d R
icht
er (
1966
, sam
ple
Iki-
58).
Tra
ceel
emen
t da
ta f
rom
Tab
le A
4d
Num
bers
und
er w
hich
the
sam
ples
are
cat
alog
ed a
nd s
tore
d at
the
U.S
. Nat
iona
l M
useu
m, W
ashi
ngto
n D
.C. (
USN
M 1
1611
2 nu
mbe
r gi
ven
in t
his
row
)
615
Lastly, some of the plagioclase phenocrysts are moresodic than the groundmass (microlitic) plagioclase pres-ent in the same glassy samples, again evidence that theearly lavas are at least slightly hybrid. The mixing ap-pears to have occurred within the magma chamber pri-or to eruption, without clear evidence for the admix-ture of hotter magma into the material being erupted;this type of mixing was also observed in the early 1955lavas (Anderson and Wright 1972; Helz and Wright1992).
When we compare the early 1960 lavas with the late1955 lavas, which they so closely resemble, we see thatconspicuous evidence for recent recharge and mixing inthe 1955 magma chamber, documented for the late1955 lavas in Helz and Wright (1992), has been largelyobliterated. The plagioclase phenocryst population inthe late 1955 lavas is bimodal in composition, whereasthat in the early 1960 lavas is narrower, and more con-tinuous in composition (Fig. 2). We interpret this nar-rowing of the range in plagioclase phenocryst composi-tion in the early 1960 lavas as resulting from re-equili-bration, followed by cooling and crystallization, duringthe 5 years between recharge and hybridization in 1955and renewed eruption from the same chamber in 1960.In addition, the more forsteritic olivine phenocrystsand the coarse, complexly zoned augite phenocrystsseen in the late 1955 lavas are absent from the early1960 samples we have examined. Again, we interpretthis as the effects of very minor crystal settling, re-equilibration and renewed cooling and crystallizationbetween 1955 and 1960.
The late 1960 lavas
The later 1960 lavas all contain very conspicuous evi-dence, not just for disruption and self-mixing of a pre-viously hybridized, stored magma, but also for the in-flux of multiple pulses of more primitive magma andthe incorporation of new differentiated magma into thestored body. The plagioclase phenocrysts are, almostwithout exception, resorbed and reversely zoned. Theabsence of euhedral overgrowths suggests that the re-sorption is quite recent. The restricted compositionalrange of plagioclase in the late 1960 lavas relative to theranges observed in the early 1955, late 1955, and early1960 lavas is consistent with a rise in the average tem-perature of the stored magma body, which is most rea-sonably attributed to the addition of hotter magma.
The augite phenocrysts frequently bear the marks ofmultiple reintrusion. One indicator is the appearance ofprogressively more magnesian rims on the phenocrysts,with each showing some signs of resorption before thenew growth occurred (as in Figs. 4 and 6). The presenceof some metastable subcalcic augite in the core of a sec-tor-zoned augite (Figs. 6 and 7) provides evidence forrecent, significant undercooling of at least part of thehotter component, during mixing of two magmas of dif-ferent temperature. Similar features have been docu-mented in augite phenocrysts in the late 1955 lavas
(Helz and Wright 1992, their Figs. 4e and f, and pp 368–369), where they were also attributed to recent mixingof thermally disparate magmas.
The mere presence of augite and plagioclase in thelate lavas is anomalous, because the late 1960 lavas allhave bulk MgO contents well above the 7.0–7.5% atwhich augite and plagioclase begin to crystallize fromKilauean liquids; summit lavas of such compositionscontain only olivine as a phenocryst. We suggest thatthese crystals have come from the lower part of a dif-ferentiated magma body, of which the early 1960 lavasrepresent the upper part. The overlap in augite compo-sitions (Table A2) between the early and late 1960 la-vas suggests that this slightly more crystal-rich lowerlayer was not much hotter than the upper part of thechamber prior to recharge.
The rare hypersthene reported in the early 1960 la-vas (Richter and Murata 1966) must be xenocrystic tothem, because the quenching temperature of all knownearly 1960 glasses (1122–1125 7C; see Table 5) is abovehypersthene-in for Kilauean lavas (ca. 1090 7C; Helzand Thornber 1987; Helz 1987b). The hypersthenefound in some of the late 1960 lavas is also clearly xe-nocrystic to its immediate host: It is resorbed and oftenrimmed with magnesian augite with which it is not inequilibrium, and is enclosed in glasses quenched fromeven higher temperatures than those in the early 1960lavas.
The varying extent of augite overgrowth on hyper-sthene in the late 1960 samples provides strong evi-dence for repeated inmixing of highly differentiated li-quids, similar in composition to the early 1955 lavas,during the latter part of the 1960 eruption. The occur-rence of hypersthene c ilmenite confirms that onepocket of differentiated magma had reached a tempera-ture of 1100 7C or less (Helz and Thornber 1987) by thetime of its entrainment. The fact that the hypersthenein the late 1960 samples is slightly more Fe-rich thansome of that in the early 1955 lavas is consistent withthese pockets having undergone 5 more years of cool-ing and fractionation than had been achieved in 1955.
The olivine phenocryst population in the late 1960samples, with its trimodal distribution of phenocryst-core compositions (Fig. 9) and convergent zoning (Fig.10) strongly supports the idea that the late 1960 lavasare hybrids. Because olivine re-equilibrates fairlyquickly, the preservation of the variant zoning patternsin crystals within single thin sections suggests that themixing was recent. Following the arguments of Maaloeand Hansen (1982), it appears that some mixing eventsmight have occurred only a few days before eruptionand quenching, where the zoned margin is very narrow(Fig. 11A); the time gap would be longer, perhapsweeks to a few months in cases where the zoned marginis wider (Fig. 11B).
The presence of highly magnesian (Fo87–89) olivine,both euhedral and kink-banded, provides direct evi-dence for the influx of primitive magma into the erupt-ing chamber, as no olivine of this composition is pres-
616
ent in the 1955 or earlier 1960 lavas. The most magne-sian olivine composition observed in the late 1955 lavaswas Fo84–85 (Helz and Wright 1992), which would bethe liquidus olivine composition in a Kilauean meltcontaining at least 8.0–8.5% MgO, using the olivine-melt equilibrium data of Roeder and Emslie (1970).Using the same assumptions (KDp0.30; ferrous ironcontent of the melts 9.50–9.75%), the presence of euhe-dral Fo87–89 olivine in the late 1960 lavas implies thattheir original host melts contained 12–13% MgO. Suchmelts would have quenching temperatures of 1255–1275 7C, much hotter than any of the observed glasses(see Table 1).
Glass compositions and quenching temperatures vs
magma mixing
We have noted already the divergence between theglass compositions and bulk compositions in Fig. 13A,which is prima facie evidence for magma mixing. In thissection we consider the implications of the inferredquenching temperatures of those glasses for the 1955and 1960 eruption processes. The quenching tempera-tures of the early 1960 glasses are 1122–1125 7C; thoseinferred for glass and mesostasis compositions in thelate 1955 samples are 1127–1132 7C. The sample base issmall, but the difference in temperature is consistentwith the early 1960 lavas having been produced by ahybridization event occurring in 1955, followed by 5–6years of subsequent cooling and crystallization, duringwhich (apparently) no further recharge occurred.
The composition of the late 1960 glasses gives a min-imum estimate for the dominant postmixing tempera-ture in the upper part of the magma chamber at the endof the 1960 eruption was 1142–1144 7C, 10 degrees be-low plagioclase-in (Thompson and Tilley 1969; Helzand Wright 1992), and 18–227 above the temperaturesof the early 1960 melts. This rise in eruption tempera-ture is consistent with considerable further recharge byhotter magma. Excavation of a zoned chamber is ruledout by the abundant petrographic evidence for very re-cent mixing, and by the paucity of melts of intermediatecomposition and temperature.
The contrast in glass composition (and quenchingtemperature) of the three lapilli in sample KP-26 im-plies additional complexity in the mixing process. Thissample is described by Murata and Richter (1966) ashaving been collected at 8 :30 p.m. on 18 February, andis apparently an observed fall. The three scoriae are allequally fresh, with no obvious evidence that one mighthave been erupted earlier than the other two. Two ofthe scoriae are among the hottest erupted; the third,differentiated scoria is cooler, but is hotter than theearly 1960 pumices, and itself shows petrographic evi-dence of recent mixing. The heterogeneity suggests tap-ping of segmented chambers not recharged during the1955 eruption, some of which had remained isolatedthroughout most of the 1960 eruption.
1 The 1959E composition was calculated as an average of olivine-controlled lavas erupted early in 1959 (Wright and Fiske 1971;Table 8, footnote). Later, it was shown that the chemistry of allsamples from the 1959 eruption could be expressed by addingvariable amounts of olivine to the chemistry of samples S-1 andS-2, erupted respectively from the easternmost and westernmostvents active on the first day of eruption (Wright 1973). In the mix-ing calculations of this paper (Table 3 and Fig. 14) we use thechemistry of S-1, rather than 1959E.
Magnesian recharge components of the later 1960
lavas
The variation in the 1960 lava compositions shown inFig. 13 suggests that the magnesian lavas were pro-duced by mixing a differentiated liquid that had under-gone three-phase fractionation with a more magnesianliquid representative of magmas erupted at Kilauea’ssummit. Wright and Fiske evaluated this possibilitywith a series of mixing calculations (Table 11 of Wrightand Fiske 1971), using average compositions for theearly and late 1960 lavas. Several different summit mag-mas were tested; the best match involved three differ-ent summit compositions, 1959E1 (lava from the earlypart of the 1959 eruption), 1961 (lava from the 1961eruption in Halemaumau Crater), and 1967 (lavaerupted from the 1967–1968 eruption of HalemaumauCrater). We have redone these calculations, takingeach intermediate and late 1960 lava composition inturn, to verify the identity of the mixing components,and to look for systematic changes in their identity withtime of eruption.
We use the following mixing equation for each sam-ple erupted from 26 January to the end of the eruption,solving it by a least-squares method to identify the mix-ing components that yield the lowest residuals:
Hybrid magma(FP7 to FP20)p1960EcRechargemagma (1952, 1961, 1967, 1959)cOlivine(Fo90, Fo70)
The 1960E composition is the average of the early 1960lavas, and represents magma stored beneath the regionof the 1960 vents prior to the beginning of the eruption.When lava compositions from eruptions earlier than1952 or later than 1967 have been tested as possible re-charge magmas, they have invariably resulted in unac-ceptably high residuals in one or more oxides, and/ornegative mixing coefficients.
Diagnostic oxide and trace element characteristicsfor the preferred summit mixing components includethe following:
Mixingcomponent
Distinctive chemical components (com-pared at the same MgO content)
1952 high Al2O3, Sr; low Na2O, K2O1961 low “FeO”; high Zr1967–1968 low CaO, TiO2, Nb1959 high CaO; low Al2O3, Sr
The bulk compositions used are shown in Table 2. Eachcalculation was made with different combinations ofthese magmatic components until a solution was ob-
617
Table 3a Results of major-oxide mixing calculations for the 1960 eruption
Sample no.a
Field No.Date (1960)a
Active ventb
MgO contenta
F-7KP-1326 JanB6.97
F-8KP-1529 JanB8.27
F-9KP-1730 JanB9.39
F-11KP-191 FebB11.27
F-12KP-202 FebL10.58
F-13KP-214 FebL12.11
F-14KP-224 FebL11.28
F-16KP-2412 FebB12.29
F-17KP-2513 FebJ13.13
F-18KP-2616 FebJ11.12
F-19KP-2718 FebJ12.46
F-20KP-2818 FebJ11.82
Mixing results1960E195219611967–19681959 S-1
Olivine(Fo content)
86.312.2PPP1.5
(84)
55.440.6PPP4.0
(88)
49.4P43.9PP6.7
(84)
32.0P57.1PP10.9(85)
41.2P49.3PP9.5
(85)
34.1P50.1P2.7
13.1(85)
42.5P39.3P7.3
10.9(86)
39.8PP32.714.313.2(87.5)
33.9PP33.517.515.1(87.5)
38.2PP31.420.59.9
(88)
44.4PP14.527.513.6(87)
49.2PP12.026.911.9(88.5)
ResidualsLargestAverage
MgO of magmaadded to 1960Ec
0.040.02
11.1
0.090.02
10.7
0.110.02
12.5
0.050.02
13.7
0.080.02
13.6
0.070.03
15.2
0.070.02
15.0
0.050.02
16.3
0.040.02
16.7
0.050.02
14.2
0.060.02
17.4
0.050.02
17.2
a See Murata and Richter (1966; see also Table 2)b See Richter et al. (1970)
c Calculated by combining olivine with olivine-controlled summitcompositions in the proportions shown
Table 3b Trace element balances using the mixing solutions of Table 3a and the trace element contents in Table 2
Sample no. F-7 F-9 F-11 F-13 F-14 F-16 F-17 F-18 F-19 F-20
Rb (2 s.d.p4 ppm)obs.a
calc.b
Sr (2 s.d.p11 ppm)obs.calc.
Y (2 s.d.p3 ppm)obs.calc.
Zr (2 s.d.p7 ppm)obs.calc.
Nb (2 s.d.p4 ppm)obs.calc.
Ba (2 s.d.p24 ppm)obs.calc.
Ni (2 s.d.p10 ppm)c
obs.calc.
Cu (2 s.d.p8 ppm)obs.calc.
Zn (2 s.d.p6 ppm)c
obs.calc.
Cr (2 s.d.p18 ppm)c
obs.calc.
1110
408417
3432
211214
2425
170172
140137
143134
136130
247252
910
386390
2829
187190
2323
149154
270253
133128
126125
507512
810
363371
2527
169175
2021
152145
312341
126123
122122
686686
79
350361
2627
170171
1821
129142
383384
125119
122122
771755
109
372369
2728
179178
1922
146147
337337
119121
118123
665667
118
361351
2625
169169
2119
138136
402376
130113
115119
718754
98
344342
2624
159162
1818
144132
451415
131110
119117
790830
98
365363
2826
178173
1720
144142
295311
116117
125119
665652
87
348348
2826
165170
1920
131142
421405
130113
123120
817797
118
369356
2627
171175
2021
141146
331350
127116
115120
658694
NOTE! Calculated data shown in bold/italic (e.g., 354) differfrom the observed by more than two standard deviationsa Data from Table 2
b Calculated from major oxide mixing solution given in Table 3ac Calculated assuming the following chemistry for olivine(BchromiteBsulfide): 2114 ppm Ni; 3660 ppm Cr; 96 ppm Zn
tained with positive coefficients for all mixing compo-nents. (Negative coefficients are not physically mean-ingful, and the components with which they are asso-ciated are considered not relevant to the mixing processin question.) The Fo content of olivine in the mix is
allowed to vary, as the least-squares program calculateswhatever composition fits best, in the range Fo 90 to70.
Two sets of calculations were tried; the first, as de-scribed above, used the early 1960 composition as the
618
starting point, in agreement with the petrographic evi-dence discussed previously. A second set was tried, inwhich the early 1955 composition was substituted forthe early 1960 composition, in order to see how sensi-tive the results were to the composition of the storedcomponent. This alternate set of calculations still re-quired the same four recharge components, and in sim-ilar proportions and identical sequence. The onlychange was that the 1960E fraction was replaced by thecombination 1955E c 1961, representing recharge ofthe 1955 magma chamber in calculations made pre-viously (Table 11 of Wright and Fiske 1971).
The solutions with the lowest residuals are displayedin Table 3a and are shown in a bar graph series in Fig.14. These solutions are well constrained, and the pat-tern seen in Table 2 is stable, even when the composi-tion of the stored component is varied, as discussedabove. Although small amounts (~10%) of rejectedcompositions cannot be ruled out, the solutions are dis-tinctly worse when the components present in amountsgreater than 10% are left out of the calculation. Thecalculated Fo contents of olivine are reasonable, fallingwithin the range both of those observed in the 1960 ol-ivines (see Fig. 9 and accompanying discussion) andthose typical of lavas erupted at Kilauea’s summit andeast rift zone.
Three features of the results plotted in Fig. 14 standout. Firstly, the compositions of the preferred inputsrepresent magmas erupted within a few years of 1960,i.e., those which could have been present in Kilauea’splumbing in early 1960. Secondly, the magnesian re-charge magmas required by the least-squares solutionsvary systematically when the hybrid compositions areplotted in their observed time sequence. Finally, the or-der in which recharge magmas are accepted as mixingcomponents matches the order in which the magmascan be inferred to have arrived in the shallow Kilaueaplumbing. These features of the results suggest that the
mixing calculations are successfully modeling a realprocess, in which one recharge magma has followed an-other into the chamber which fed the later 1960 lavas.
We are aware of the fact that it is possible to explainthe chemical variation alone by fractionation of olivine,clinopyroxene, and plagioclase, as was demonstratedfor the 1960 lavas by Russell and Stanley (1990, theirFig. 10, p 5034–5035). However, the petrographic evi-dence for mixing, and the large divergence between thesequence of 1960 bulk compositions and the actual 1960Kilauean liquid descent line are more than sufficient todemonstrate that such calculations, no matter how suc-cessful mathematically, do not correspond to physicalreality.
Trace element balances (Table 3b) are broadly con-sistent with the proposed mixing, with elements ex-cluded from olivine, chromite, and sulfide in excellentagreement. Balances for the compatible elements (Cr,Ni, Cu) show minor discrepancies. We believe that thepoorer fits for these elements reflect small variations inmineral composition and proportions [e.g., chromite/ol-ivine ratios, variable nickel contents in olivine at thesame Fo content (e.g., Clague et al. 1995, their Fig. 7, p.313), and the presence of minor sulfide], rather thanproblems with the magmatic components chosen by themixing calculations.
The variation of trace elements with MgO (Fig. 13B)illustrate the complexity of the mixing process. None ofthe four elements plotted for the hybrid samples followan olivine control line (represented by the 1959 pair).Cr for some of the hybrid lavas falls slightly below anypossible linear mixing trend, suggesting that some ofthe unfractionated magmas used as mixing componentshad lost chromite relative to olivine during storage atthe summit and transport in the rift. This was also infer-red to have occurred in the magnesian lavas which re-charged the 1955 chamber (Helz and Wright 1992).
Temperatures of magmas involved in the mixing
process
A minimum temperature of 1122–1125 7C for magmaresiding in the reservoir which fed the 1960 eruption iscalculated from compositions of the early 1960 glasses(Table 1) using the calibration of Helz and Thornber(1987). After recharge during the 1960 eruption, glasstemperatures as high as 1144 7C are observed. From theabsence of phenocrystic plagioclase in several of thelatest erupted samples, we can infer that the highestpostmixing temperatures were closer to 1155 7C, thetemperature at which plagioclase begins to crystallize in1960 compositions (Thompson and Tilley 1969). Thetemperatures inferred from the glass compositions are11–13 7C lower, and the glasses contain tiny microlitesof plagioclase, consistent with the lower temperatures.Growth of groundmass crystals presumably occurredafter mixing, most likely during eruption and subse-quent cooling at the surface.
Fig. 14 Results of mixing calculations (Table 3a) for lavas fromthe 1960 eruption (see text for further explanation)
619
Table 4 Estimated temperatures (7C) for the various primitive components, required by least-squares calculations to make up the late1960 lavas, assuming that the stored component had the composition and temperature of the early 1960 lavas
Sample no. (hybrid) KP-15F-8
KP-19F-11
KP-25F-17
KP-26F-18
KP-27F-19
Stored componentWeight fractiona
TMgO
1960E0.581123c
1960E0.361123c
1960E0.401123c
1960E0.421123c
1960E0.511123c
HybridTMgO (Table 1b)T (plag-in)b
1133P
1138P
11441155
11431155
11421155
Primitive componentType requiredWeight fractiona
Tcalc (minimum)Tcalc (plag-in)
19520.421144
19610.641146
1967c19590.39c0.211154c1174d
1169c1242d
1967c19590.35c0.231152c1172e
1167c1239e
1967c19590.17c0.32
a Weight fractions from Table 3a, renormalized to exclude oliv-ineb From Thompson and Tilley’s (1969) experiments on 1960 lavacompositions. Consistent with TMgO values in Table 1bc Average of early 1960 glasses (columns 1–5 in Table 1a)
d Calculated from solving simultaneous equations for mixing solu-tions for KP-27 and KP-25 (Table 3a)e Calculated from solving simultaneous equations for mixing solu-tions for KP-27 and KP-26 (Table 3a)
We have used the TMgO values of the early 1960glasses and the various later hybrids, plus the propor-tions of primitive and stored magmas taken from themixing results (Table 2) to calculate a minimum tem-perature for the magmas added to the 1960 reservoirduring the 1960 eruption (Table 4). The equation usedis:
(Xstored)(Tstored)c(Xrecharge)(Trecharge)pThybrid; (1)
where X is the mass fraction of the magmatic compo-nents and T is the temperature. The only unknown isthe temperature of the recharge magma. We assumethat there are no nonideal heat-of-mixing effects whenone Kilauean lava mixes with another, which seemsreasonable, for such closely related melts.
Equation (1) applies when both stored and newcomponents are crystallizing the same assemblage, inthis case, the three-phase assemblage olivine c augitec plagioclase, which occurs below 1155 7C. However,for higher-temperature melts, crystallizing olivine only,the amount of heat released per degree is approximate-ly 56% of that released during crystallization of thethree-phase assemblage (Ghiorso 1985; Helz et al.1993). If Trecharge is `1155 7C, while Tstored is less, thenan adjustment to Trecharge is necessary. During mixing ofmagmas crystallizing the two different assemblages, theextra heat extracted from the hotter component is con-sumed remelting plagioclase B augite in the coolercomponent. Therefore, if the initial temperature calcu-lated for the recharge magma is above 1155 7C, we mustadjust the result as follows:
Tfinalp1155c1.78 (TrechargeP1155) (2)
The results of these calculations are shown in Table 4as Tcalc (minimum).
The minimum temperature of the 1952 componentmixed with the 1960 stored magma is 1144 7C, based on
the mixing proportions obtained for sample KP-15 (F-8). This value may be low, as the glass in this sample issomewhat devitrified. The temperature of the 1961component, based on sample KP-19 (F-11), is 1146 7C.
The other two components (1959 and 1967) arefound only in combination. For these components weset up two equations of the form of Eq. (1) above, withtwo unknowns, and solved them simultaneously. Theminimum temperature calculated for the 1959 compo-nent in KP-25, KP-26, and KP-27 (1172–1174 7C) is con-siderably hotter than 1155 7C (plag-in), and so was ad-justed using Eq. (2). The minimum temperature of the1967–1968 component is 1152–1154 7C, calculated fromthe data for the same three samples.
The late samples KP-25, KP-26, and KP-27 are thosewhere the available thin sections lack any phenocrysticplagioclase. For these the temperature of the chamberimmediately after mixing was arguably at least 1155 7C.Table 4 therefore includes the results of alternative cal-culations for the temperatures of the 1959 and 1967–1968 components, based on the assumption that thetrue temperatures at the time of mixing were 1155 7C.The resulting temperatures are 1239–1242 7C for the1959 component and 1167–1169 7C for the 1967–1968component. These temperatures represent the maxi-mum inferable temperatures for these two components,consistent with the available petrographic and chemicaldata. The question is, Are these higher estimates stillrealistic, or are the minimum estimates based on actualglass compositions better?
We can evaluate the calculated temperatures ofmagma inputs by comparing them with the eruptiontemperatures estimated from glasses analyzed from the1952, 1961, 1967–1968, and 1959 eruptions at Kilauea’ssummit and with liquidus temperatures calculated forbulk compositions of the samples from which glass wasanalyzed; these data are shown in Table 5. Our inter-
620
Table 5 Composition and temperatures of glasses and bulk rocks from olivine-controlled eruptions at Kilauea’s summit
Column no. 1 2 3 4 5 6 7 8
1952 glassa 1961 glassa 1967 glassb Ik-22 glassc 1952 1961d 1967 1959 S1SiO2
Al2O3
FeO (total)MgOCaONa2OK2OTiO2
P2O5
MnOCr2O3
Sum
50.114.011.306.91
11.32.360.542.710.260.180.01
99.67
50.713.810.86.99
11.202.490.572.730.260.190.02
99.85
50.9013.412.006.44
10.82.680.653.130.230.180.01
100.42
49.1812.1011.3310.0311.172.330.522.490.220.130.08
99.58
50.3213.9111.327.14
11.362.260.532.720.270.17
P100.00
50.3813.7010.837.64
11.362.320.552.770.270.18
P100.00
50.4513.6511.267.60
11.132.310.542.640.260.17
P100.00
50.1012.8811.218.11
11.972.140.552.630.250.17
P100.0
TMgO (glass)TMgO (bulk)Comments
1152 1154 1143
Abundant, well-formed crystal-lites
12161158 1168 1167 1177
a Analyses from Helz et al. (1995)b New datac Hottest sample collected during the 1959 eruption and richest inthe 1959E component (Table 25.4; in Helz 1987a). See text for
further explanationd Average of two samples, K-61-4 and K-61-22 (Richter et al.1964)
pretations are limited by two factors:1. All observed glass temperatures are minimum, be-
cause of imperfect quenching.2. Each summit magma may have been thermally
zoned in storage beneath Kilauea’s summit, with theerupted fraction representing only the cooler, upperparts of the magma body.The glass-quenching temperatures for samples from
the summit eruptions in 1952 and 1961 are 1152 and1154 7C, respectively (Table 5) and represent minimumtemperatures for those magmas in storage beneath Ki-lauea’s summit. The liquidus temperatures of the bulkcompositions used in the calculations are 1158 and1168 7C, respectively, and represent the maximum tem-perature of these components for which we have directevidence. The inferred temperatures for 1952 and 1961magmas that entered the 1960 chamber are close to thetemperatures of the same magmas erupted at Kilauea’ssummit, given that the uncertainty in TMgO is B10 7C(Helz and Thornber 1987). If one accepts the results atface value, they suggest that the magma which re-charged the chamber in 1960 was similar in tempera-ture and composition to that erupted at the summit in1952 and 1961.
The most magnesian glass erupted in 1959 has aquenching temperature of 1216 7C (Table 5 of Helz1987a). This is well above the liquidus temperature ofsample 1959 S-1 (1177 7C), which represents the 1959component in the mixing calculations. The sample inwhich this most magnesian glass occurs, Iki-22 (sampleS-5; Murata and Richter 1966), is itself a hybrid, con-taining 68% of the juvenile 1959 component and 32%of the stored 1959 component (Wright 1973; Helz1987a). The temperature of the 1959 juvenile compo-
nent in Iki-22, calculated using Eq. (1), is 1231 7C, as-suming a glass quenching temperature of 1185 7C forsamples consisting entirely of the stored 1959 compo-nent (Helz 1987a). The temperature of the 1959 juve-nile component estimated from the 1960 samples(1239–1242 7C) and from Iki-22 (1231 7C) are remarka-bly close, given the assumptions involved in the calcula-tions. Hence, we suggest that the higher temperaturesare reasonable for this component. We mentioned pre-viously that the presence of euhedral Fo87–89 in some ofthe later 1960 lavas implied that those grains had crys-tallized, fairly recently, from magmas having tempera-tures of 1255–1275 7C; clearly, this would be the compo-nent most likely to be responsible for the presence ofthese olivines in the late 1960 lavas.
For the last component (the 1967 magma) the hot-test actual spatter sample found for that eruption hasglass with a quenching temperature of 1143 7C, and theliquidus temperature is 1167 7C (Table 5), both not un-like the value of 1152–1154 7C and 1167–1169 7C calcu-lated as the minimum and maximum temperature forthat component in Table 4. Thus, three of the four com-ponents identified by least-squares calculations as pos-sible new inputs arrived with minimum temperatures of1144–1154 7C, essentially identical to the temperaturesas observed in the equivalent summit lavas (1143–1154 7C), within the uncertainty of the glass geother-mometer (B10 7C; Helz and Thornber 1987). Thefourth, the 1959 component, is distinctly hotter, and ar-rived crystallizing olivine only, regardless of which cal-culated result (1173 vs 1224 7C) we consider best. Themost straightforward interpretation of these results, forall four components, is that transport of magma fromthe main reservoir to the east rift zone involves no
621
more cooling than transport from the reservoir to thesurface at the summit.
Magnesia contents of the bulk compositions (olivinec liquid) added to the 1960 stored magma are tabu-lated in Table 3a. These range from just under 11 tonearly 18%. The temperatures calculated above fortransport in the rift zone correspond to melt MgO con-tents of 6.5–10.3%, much lower than the magnesia con-tents indicated in the mixing calculations. The two setsof calculations can be reconciled if olivine was en-trained during lateral transport in the rift zone.
Comparison of the 1955 and 1960 eruptions
The 1955 and 1960 eruptions were fed from the samereservoir complex underlying the easternmost subaerialpart of Kilauea’s east rift zone. To complete our inter-pretation of the 1960 eruption we contrast and compareevents of 1960 with the events associated with the pre-ceding 1955 eruption.
Eruption chronology
The chronology of events associated with the 1955eruption were summarized in Part I (Helz and Wright1992, their Table 1, Figs.1 and 2). The chronology ofevents preceding, accompanying, and following the1960 eruption are shown in Fig. 15 and Table 6, ex-tracted from the summary of Richter et al. (1970). Theeruption began on 13 January and ended 5.5 weeks lat-er, on 20 February. Eruption was initiated along a 1-km-long fissure northwest of Kapoho Crater (Fig. 1).Vents opened first down-rift, then up-rift (Richter et al.1970, their Fig. 43), similar to the 1955 sequence. Laterin the eruption, two additional vents opened within ap-proximately 0.5 km to the east of the eastern end of theoriginal line of vents. Temperatures, measured by opti-cal pyrometer during the eruption, increased markedlyon 31 January (Table 6; Richter et al. 1970, p. E58).The last 2 weeks of eruption were marked by steady,low-level activity without any obvious changes in thecharacter of the effusion.
The 1960 eruption was preceded by several weeks ofheightened seismic activity near the eventual site oferuption, culminating in an intense earthquake swarm(Fig. 15) and subsidence of the Kapoho Graben (Fig. 1)on the day before lava reached the surface. Five daysafter the eruption began, Kilauea’s summit began tocollapse; a delay between rift eruption and summit col-lapse also occurred in 1955. The collapse was accompa-nied by a swarm of long-period earthquakes beneathKilauea’s summit. Seismicity at the eruption site diedout quickly, reaching background levels within a fewdays of the arrival of lava at the surface. During theeruption there was one short burst of increased seismi-city lasting 2 days. Seismicity at Kilauea’s summit grad-ually declined after the end of the collapse. Seismicityon the rift zone remained at background levels throughthe end of the eruption.
Essential features of the 1955 and 1960 eruptions arecompared in Table 7. The two eruptions share the fol-lowing characteristics:1. Each eruption was preceded by several weeks of in-
creasing seismicity near the eruption site, culminat-ing in several days of felt earthquakes, ground crack-ing, and local subsidence.
2. Eruption began several days before collapse of Ki-lauea’s summit.The two eruptions also differ in the following impor-
tant respects:1. Resupply was more rapid in 1960, as evidenced by
the shorter time separating summit collapse andchange of eruption chemistry.
2. The 1960 eruption was more vigorous, as indicatedby higher fountaining, with more olivine entrain-ment during mixing.We infer from these data that the rift plumbing was
more open in 1960 than in 1955, and that the travel ofmagma to the surface was more forceful and turbulentin 1960 than in 1955.
Fig. 15 Geophysical and geological observations made before,during, and following the 1960 eruption. Top: Tilting at Uweka-huna vault, Kilauea’s summit, showing timing of summit collapserelative to beginning and end of eruption on the rift zone. Bot-tom: Seismicity at Kilauea’s summit and near the site of eruption,compared with beginning and end of eruption and changes inchemistry during the eruption (see text for further explanation).Unpublished data, US Geological Survey Hawaiian Volcano Ob-servatory. (Used with permission)
622
Table 6 Chronology of significant events during the 1960 eruption. [Eruption observations taken from Richter et al. (1970). Mixingchemistry taken from calculations presented in this paper (Table 3)]
Date (1960) Active vents Chemistry/field-temperature
1/131/141/151/171/181/211/241/261/271/291/301/312/1–32/42/52/62/72/12–182/20
Eruption begins at vents A–JVents B–E active; vent J sporadically activeB now principal vent
Highest fountaining of eruption
Vent J reactivated
New vents L and K open east of J
Rift line from E–H reopens; activity at L and B decreasesActivity at L and B ceases; reactivated vents also declineWeak activity at B, E, and J to end of eruption
Eruption ends
1960E
1960E
1960E
1960Ec1952
1960Ec19521960Ec1961Temperature increases from 1050–1080 to 1100 7C1960Ec19611960Ec1961c1959
1960Ec1967c1959
Table 7 Comparison of the 1955 and 1960 eruptions
Observations 1955 1960
Pre-eruptionSeismicity beneath eruption site Seismicity increased several weeks be-
fore eruption; intense seismic swarm 4days before eruption
Seismicity increased several weeks be-fore eruption; intense seismic swarm1 day before eruption
Ground deformation near eruptionsite
Inflationary tilt change at Pahoa seis-mometer 4 days before eruption
Kapoho graben collapsed 1 day beforeeruption
During eruptionCollapse of Kilauea’s summit 7 days after eruption began 4 days after eruption began
Initial change of lava composition 13–18 days after summit collapse 5–8 days after summit collapse
Character of eruptionLava chemistryTemperature (glass MgO)Fountain heights
Low MgO (5–7%)1099–1132 7C30–130 m (average 50–60 m)
High MgO (6–14%)1122–1144 7C150–500 m (average 300 m)
Changes in mixing components related to eruptiveevents
In 1955, as documented by Helz and Wright (1992) andWright and Fiske (1971), stored magma was hybridized,apparently by magma of 1952 composition, evident inlavas erupted approximately 2 weeks after the eruptionbegan. During the eruption, magma of 1961 composi-tion also moved from beneath Kilauea’s summit to be-come the dominant component of the recharged reser-voir which supplied magma to the 1960 eruption.
The pattern of mixing in 1960 is demonstrated bymixing calculations (Table 3a), ordered in Fig. 15 bythe date of eruption. The only mixing componentswhich yield low residuals in the mixing calculationshave the chemistry of four summit lavas erupted within10 years before or after the 1960 eruption. During the
course of eruption the preferred magnesian rechargecomponent changes systematically from one having thechemical peculiarities of the 1952 summit lava (26–29January) through one having those of the 1961 summitlava (30 January to 4 February) to one having the char-acter of the 1967 Halemaumau lava (12–18 February).The temporal sequence matches the order in whichthese compositions were erupted at Kilauea’s summit.
One additional composition, 1959, becomes an im-portant component of the mix beginning 4 Februaryand remains so to the end of eruption. It appears witheither the 1961 or 1967 components, as indicated.
Many of the compositional changes indicated by themixing calculations can be correlated with specificevents noted in the eruption chronology (Table 6).During the earliest part of the eruption, lava composi-tions suggest that only magma which had been hybrid-
623
ized in 1955 was actually erupted, although recharge ofthe chamber may have begun.
The first change of chemistry in the 1960 eruptionoccurred 1 week after Kilauea’s summit began to col-lapse. The 1952 component (needed in samples KP-13and KP-15; see Fig. 15) was already present in the riftzone, as it had been the early and dominant mixingcomponent in 1955 (Table 6 of Helz and Wright1992).
A temperature increase measured in the field (Table6) and by increase of MgO in glass (Table 1a) corre-lates with opening of new vents east of the original cur-tain of fire, an increase in the ratio of juvenile to storedmagma (KP-15; Table 3a) and a change in the rechargecomposition from 1952 to 1961 (sample KP-19; Table3a). (Fountain temperatures obtained by optical pyrom-etry are known to be underestimated due to cooler ma-terial between the instrument and the source. Relativetemperature increases or decreases are valid undercomparable observing conditions. Temperatures ob-tained near this time from glass chemistry indicate thatthe field temperatures are approximately 35–40 7C toolow.) The 1961 magma was also present in the rift, iden-tified as the dominant magma introduced toward theend of the 1955 eruption (Wright and Fiske 1971). Afurther increase in temperature noted on 13 February(KP-25-27; Table 1a) followed a flurry of seismic activi-ty near the eruption site on 4 February (Fig. 15) and thefirst appearance of the 1959 composition (Table 3a),the mantle derived component in the 1959 eruption(Helz 1987a).
Two independent lines of evidence are consistentwith involvement of the 1959 magma in the 1960 erup-tion. We noted above that kink-banded olivines, singlyor in aggregates, are observed in some of the late 1960samples investigated. Kink-banded olivines are presentin 5 of 7 samples which take the 1959 composition inthe mixing solutions of Table 3. By contrast, equallymagnesian 1960 lavas (F-11, F-12) which do not takethe 1959 component do not have the deformed olivines.From the observations of Helz (1987a), this componentof the 1959 eruption is the only known historic Ki-lauean summit magma to contain an abundance ofkink-banded olivines. The association of the 1959 com-position and this class of olivine in both the 1959 and1960 eruptions suggests that there was something dis-tinctive about the transport of this magma in Kilauea’splumbing. It may have moved faster than other mag-mas, whether vertically or horizontally, consistent withthe very high temperatures observed and inferred forthis component in both the 1959 and late 1960 lavas.
An additional piece of circumstantial evidence iscontained in the report by Clague et al. (1995, p. 325) ofa peculiar inclusion of hornfels-textured troctolite in a1960 flow collected underwater off Cape Kumukahi.The host flow, dredged from the surface, would mostlikely be one of the latest flows to be erupted in 1960.Clague and coworkers (1995) interpreted the inclusionas coming from layer 3 of the oceanic crust under Ki-
lauea. Xenoliths are extremely rare in Hawaiian tho-leiitic lavas (Jackson 1968). It may be coincidental, butthe 1959 magma, erupted immediately before the 1960eruption, is one of the few other Kilauean lavas whichcontains xenolithic inclusions, some of which probablycome from below the volcanic edifice (see discussion ofannealed dunites in Helz 1987a).
The appearance of the 1967 component, in samplesKP-24 through KP-28, is the most intriguing, as samplescontaining this component were erupted with no unu-sual rift events immediately preceding or accompanyingthe change in composition. Both of the hypersthene-bearing 1960 samples (KP-24, KP-26) require the 1967–1968 summit composition as a component of their make-up (Fig. 14), implying that the path traced by the 1967–1968 summit magma was offset from that of the 1952and 1961 magmas responsible for previous mixing epi-sodes in 1955 and 1960.
Additional calculations were made to test whetherother mixing components (e.g., the 1955E chemistry in-stead of 1960E) obviated the need for the 1967 compo-sition. In every case the residuals were unacceptablyhigh when the 1967 component was absent. Inclusion ofa small percentage of 1955E composition did not affectthe residuals significantly, although the amount takenby the calculations is too small to be well constrained.We conclude that the 1967 component moved from be-neath Kilauea’s summit to the lower east rift zone by1960, at least 2 years earlier than previous estimates(Wright et al. 1975; Wright and Tilling 1980; Wrightand Heliker 1987). The calculated temperatures of mix-ing components summarized in Table 5 indicates thatthe 1967 component was hotter than the 1952 and 1961components. This is consistent with its more recent ap-pearance in the magmatic plumbing (i.e., 1960 com-pared with 1952 and 1955, respectively).
Mixing elsewhere in the east rift zone
Magma mixing has been documented chemically byWright and coworkers for eruptions elsewhere on theeast rift zone (Wright and Fiske 1971; Wright et al.1975; Wright and Tilling 1980). More recently, Garciaet al. (1989; 1992) has made similar interpretations forthe Puu Oo-Kupaianaha eruption of Kilauea which be-gan in 1983, and Clague et al. (1995) have published astudy of lavas erupted on the submarine part of Ki-lauea’s east rift zone. The range of mineral composi-tions for the submarine lavas is shown in Table 8. Therange of mineral compositions observed in the 1955and 1960 lavas is almost as great as the total ranges ofcompositions for minerals in the submarine lavas, eventhough the latter cover lavas erupted over a long periodof time and on a longer segment of the rift zone. Notsurprisingly, abundant petrographic and chemical evi-dence for magma mixing is documented for the sub-marine lavas. We conclude that magma mixing is ubi-quitous on Kilauea’s east rift zone and is a necessary
624
Table 8 Comparison of mineral compositions (mol percent) in the 1955–1960 lavas with those collected from submarine lavas onKilauea’s east rift zone
Mineral 1955 lavas (Anderson andWright 1972; Helz and Wright1992)
1960 lavas (this paper) Submarine lavas (Clague et al.1995)
Olivine Core Fo76–85 Core Fo76–88 Core Fo78–91
Hypersthene En75 En70–71 En66–81
Augite Mg/(MgcFe)p0.83–0.69Wop0.28P0.42
Mg/(MgcFe)p0.84–0.74Wop0.37–0.43
Mg/(MgcFe)p0.84–0.66Wop0.32–0.43
Plagioclase An51–78 An57–80 An47.7–81.3
Fig. 16 Model for magma mixing in 1960. Temperatures on theleft are estimated for magma moving upward from the mantle tosupply Kilauea’s shallow summit reservoir. The temperatures atthe time of mixing in 1960 are given above ovals labeled with thecomposition of the various summit components involved in themixing. The 1959 magma traveled along a deeper path than theother summit components. Approximately 60 7C of cooling occursduring storage before eruption at Kilauea’s summit or east riftzone, regardless of the path traveled
consequence of the model for magma transport andstorage presented in the final section of this paper.
Magma transport and storage at Kilauea
General model and the concept of magma batches
A broad framework for magma storage and transportat Kilauea has been developed from consideration ofseismic and geodetic data. The distribution of magmabeneath Kilauea’s summit and upper east rift zone hasbeen illustrated and discussed most comprehensivelyby Ryan (1988). Subsequent authors have confirmedand enlarged upon the great vertical extent of the riftmagma plumbing (e.g., Delaney et al. 1990) and thepresence of olivine cumulates at its base (Clague andDenlinger 1994). Our preferred interpretation of therift plumbing, consistent with but not identical to thosein the cited references, consists of three- tiers (Fiske et1993). The uppermost tier, extending from the surfaceto depths of 3–4 km, has a high ratio of rock to magma.The middle tier (4–6 km) is the upper part of Ryan’saseismic “deep molten core.” This tier is dominated byrelatively crystal-poor magma. The lowest tier (6 km tothe base of the volcanic pile at 10 km) is marked by in-creasing ratios of olivine to liquid, the deepest part cor-responding to Clague and Denlinger’s zone of dunitecumulate.
The upper zone is marked by intrusion of dikes, of-ten accompanied by earthquake swarms (Klein et al.1987). Many dikes reach the surface to feed eruptions;others do not and provide secondary bodies of storedmagma which cool and crystallize, providing sources offractionated liquids. The deeper parts of the rift zoneare nearly aseismic (Klein et al. 1987).
Magma from the Earth’s mantle enters Kilauea’sshallow plumbing in batches of discretely differentcomposition. A magma batch has been defined as achemically distinctive group of lavas, from a singleeruption, which are either uniform within analytical er-ror or differ only by addition or subtraction of olivine(Wright 1971). The distinctive chemistry of the magmabatches seen in 1952–1968 at Kilauea has provided aunique way of tracking movement of magma through
the Kilauea plumbing, and led to the startling conclu-sion that a magma batch may appear in a hybrid lavaerupted on Kilauea’s east rift zone some years before itis seen in its unhybridized form at Kilauea’s summit(Wright and Fiske 1971). The origin of inter-batchchemical differences is not known, but must be deeperthan the shallow plumbing system, based on the obser-vation that the chemistry of different batches cannot berelated by fractionation of observed phenocrysts.
Model based on the 1960 eruption
Thermal structure of the rift plumbing
Figure 16 summarizes our inferences regarding thethermal structure of Kilauea’s plumbing. Magma ar-rives from the mantle with a temperature at least as hotas 1300 7C (see also Clague et al. 1995). Cooling duringstorage at 2–6 km depth beneath Kilauea’s summit re-
625
sults in a thermally zoned chamber whose top is ap-proximately 1170 7C. We estimate that an additional10–157 cooling usually occurs during upward transportto eruption at Kilauea’s summit. Recharge of lowereast rift reservoirs during the 1955 and 1960 eruptionswas from magma that moved from the direction of Ki-lauea’s summit unaccompanied by earthquakes directlyassociated with the transport path. The recharge mag-ma arrived, as described above, at a temperatures simi-lar to those observed in the corresponding eruptions atKilauea’s summit. These two facts argue very stronglythat recharge occurs in a largely molten layer of theeast rift zone, and that this zone retains a high temper-ature and a fluid connectivity along the entire distance,even during periods such as 1924–1952, when there isrelatively little activity at Kilauea. These inferences areconsistent with the geodetic evidence for a deep magmabody within the rift zone (Ryan 1988; Delaney et al.1990), as well as with the continuity of deep seismic ac-tivity on Kilauea’s south flank, even in the absence ofshallow rift intrusions or eruptions.
Water content and degassing of Kilauean magmas
A general model for degassing of Kilauea magma hasbeen presented by Gerlach and Graeber (1985) andGreenland et al. (1985). These models have been ap-plied to the origin of submarine lavas from Kilauea’seast rift zone (Dixon et al. 1991; Clague et al. 1995). Wecannot apply Dixon and Clague’s model of mixing ofundegassed and degassed magmas in the rift zone tothe 1960 eruption, as we lack data on volatile contentsin either lava glasses or in glass and fluid inclusions inminerals from the 1960 lavas. However, we can say thatthe volatile contents are low, as the phase relations forthe 1960 lavas closely resemble those established forthe Kilauea lava lakes. The derivation of internally con-sistent temperatures for magmas which had beenerupted prior to 1960 and magmas not yet eruptedmakes it highly unlikely that the volatile contents of thedifferent magmas used as mixing components differ sig-nificantly from each other or from the average volatilecontent of magma in shallow storage at Kilauea.
Rates of crystallization and cooling during storage inthe east rift zone
It is possible using the results of the present study tofurther quantify the rate of crystallization of storedmagma in the cooler upper tier of the rift zone. Wrightand Tilling (1980) had inferred earlier that magmastored in the upper east rift, during a period when thatsegment of the rift had been very active, crystallized athree-phase assemblage of plagioclase-augite-Fe-richolivine at a rate of 1–2% by weight per year. Modeldata for the late 1955 lavas (Table 2, average of Y andZ spatter from Wright and Fiske 1971) as compared
with the early 1960 lavas (Table 2, average of F-1, -4,and -6) show approximately 8.1% crystallization by vol-ume (8.5 % by weight) over the 5 years separating thetwo eruptions. As discussed above, the contrast in glasscompositions between the late 1955 (1127–11327; Table2 of Helz and Wright 1992) and early 1960 glasses(1122–11257; columns 1–5 in Table 1a), each of whichlies in the three-phase crystallization regime, is consis-tent with cooling of 2–10 degrees over the 5-year inter-val between those eruptions. The crystallization andcooling rates can be related by thermodynamic model-ing of the crystallization of Kilauean basalt (Ghiorso1985), which yields 1–2.5 7C of cooling per year forthree-phase crystallization and to 2–5 7C for crystalliza-tion of olivine B chromite only (Helz et al. 1993). Thevarious estimates of cooling and crystallization rates formagma stored in the east rift zone are mutually consis-tent.
Summary
The 1955 and 1960 eruptions demonstrate the impor-tance of magma mixing within Kilauea’s east rift zoneand provide constraints on magma transport and stor-age at Kilauea. Kilauea magmas arriving from the man-tle are stored before eruption in a reservoir located 2–6 km beneath Kilauea’s summit. One component of the1959 eruption in Kilauea Iki crater apparently followeda different path, arriving at the surface without inter-secting Kilauea’s summit reservoir and at a much high-er temperature than that of most Kilauea eruptions.Three different lavas originating from Kilauea’s summitreservoir traveled to the lower east rift zone in 1955and 1960 to mix with cooler magma already storedthere. The 1959 component was erupted in hybrid lavaslate in the 1960 eruption, and likewise appears to havebeen much hotter than the other summit componentsmixed in 1960.
We estimate that most magma feeding rift eruptionsleaves the main conduit near the base of shallow stor-age, approximately 5–6 km beneath Kilauea’s summit.Magma progresses aseismically through the fluid coreof the rift zone, cooling as it moves away from the sum-mit. During intrusions magma moves upward from themain rift conduit into the upper few kilometers of therift zone and becomes isolated from the main transportpath. The reservoir which fed the 1955 and 1960 erup-tions represent one such intrusion. The intruded mag-mas cool, crystallize, and differentiate at a rate of ap-proximately 1–2% by weight per year. As in 1960 theymay be intersected at any time by hotter magmas mov-ing from the main rift conduit to the surface, thus pro-ducing eruptions in which magma mixing is important.
The amount of cooling that a magma undergoes insummit storage and between the summit reservoir andthe point of eruption, i.e., transport through the summitreservoir or through the main east rift conduit, aver-ages approximately 60 7C, regardless of the path taken.
626
The temperature at the time of eruption or mixing justprior to eruption may vary with time in storage, hottermagmas having spent less time in storage.
Appendix
Table A1 Compositions of plagioclase in 1960 spatter samples
Column no.Field no.
1KP-4
2KP-4
3KP-22
4KP-22
5KP-22
6KP-22
7KP-24
8KP-26
9KP-26
10KP-28
Type Microphenocrysts Resorbed Euhedralaverage
Phenocryst Inclusions Phenocryst Inclusions
Grain Cores Rims Core Tip Core TipNo. of points 5 4 6 9 3 1 3 4 1 4
SiO2
TiO2
Al2O3
SFeOMnOMgOCaONa2OK2O
52.3P30.20.73
P0.15
13.403.510.12
52.9P29.50.97
P0.22
12.293.950.14
52.0P30.60.64
P0.12
13.133.580.12
49.1P32.20.71
P0.23
14.952.540.07
50.7P31.50.65
P0.24
14.282.930.09
48.1P32.90.88
P0.23
16.022.180.07
51.8P30.40.78
P0.17
13.353.470.10
51.0P31.10.69
P0.20
13.933.120.09
48.6P32.80.75
P0.23
15.252.320.06
48.9P32.20.92
P0.16
15.052.830.08
Sum 100.41 99.97 100.19 99.80 100.39 100.38 100.01 100.13 100.01 100.14
An (mol %)Ab (mol %)Or (mol %)
67.531.90.6
62.636.60.9
66.532.80.7
78.521.20.3
72.926.60.5
79.919.80.3
67.631.80.6
70.728.80.6
78.221.60.3
74.225.20.6
1–2: Average core and rim compositions for normally zoned, eu-hedral, lathy plagioclase microphenocrysts in early 1960 pumice3: Unzoned, strongly resorbed plagioclase phenocryst in late 1960pumice, illustrated in Fig. 3A4–6: Reversely zoned, euhedral plagioclase phenocryst, fromsame section as feldspar in column 37: Average composition of inclusions in the resorbed, reverselyzoned olivine shown in Fig. 11A
8–9: Composition of slightly resorbed and reversely zoned plagio-clase phenocryst in the more differentiated scoria in sample KP-2610: Average composition of inclusions in a resorbed, reverselyzoned olivine phenocryst in sample KP-28
Table A2 Compositions of pyroxenes and ilmenite in 1960 spatter samples
Column no.Field no.TypeGrainNo. of points
1KP-4AugiteCore3
2KP-4AugiteRim2
3KP-24Opx
3
4KP-24AugiteRim2
5KP-24IlmeniteIn opx2
6KP-24Opx
4
7KP-24AugiteCore6
8KP-24AugiteRim 13
9KP-24AugiteRim 24
10KP-24AugiteDarkest5
SiO2
TiO2
Al2O3
Cr2O3
SFeOMnONiOMgOCaONa2O
51.11.042.170.398.490.210.05
16.919.00.23
50.91.082.740.677.180.190.07
16.520.20.22
53.20.611.390.08
16.40.34
P25.22.360.03
51.90.952.710.667.250.18
P17.818.70.26
0.0045.60.000.21
46.80.30
P6.250.000.00
52.60.671.280.00
16.40.250.03
25.92.280.00
50.81.402.390.00
10.20.180.02
16.018.90.24
50.81.062.210.079.610.180.01
17.218.70.21
51.80.892.490.626.850.100.05
17.519.60.17
52.70.621.580.406.490.170.01
18.918.90.10
Sum 99.56 99.75 99.67 100.41 99.16 99.41 100.13 100.05 100.07 99.87
En (mol %)Fs (mol %)Wo (mol %)
47.913.438.7
47.111.541.4
69.825.54.7
50.511.538.0
70.525.04.5
45.316.238.5
47.815.037.2
49.410.839.8
52.310.037.6
(Continued)
627
Table A2 Continued
Column no.Field no.Type
11KP-25Augite
12KP-25Subcalcicaugite
13KP-25Augite core
14KP-26Opx
15KP-26Augite core
16KP-26Augite rim
17KP-28
18KP-28Augitephenocryst
19KP-28
Grain Dark sector Light sector Core Rim Core RimNo. of points 6 2 6 8 2 1 7 5 3
SiO2
TiO2
Al2O3
Cr2O3
SFeOMnONiOMgOCaONa2O
51.21.022.340.028.290.150.01
17.619.80.18
53.80.591.030.02
11.20.200.00
23.010.50.05
49.41.333.780.307.870.160.01
16.919.70.16
53.80.661.020.05
16.70.300.07
25.42.370.05
50.71.362.970.278.320.200.10
16.419.60.26
51.61.032.490.517.730.150.12
17.319.40.25
51.20.942.580.740.670.160.08
16.720.60.25
51.70.931.850.218.40.220.06
17.119.00.24
51.50.952.410.667.310.180.07
17.618.80.26
Sum 100.42 100.37 99.65 100.42 100.18 100.58 99.52 99.71 99.74
En (mol %)Fs (mol %)Wo (mol %)
48.212.839.0
62.317.220.6
47.612.539.9
69.725.74.6
46.613.340.1
48.612.239.2
47.710.042.3
48.213.338.5
49.911.738.4
1–2: Core and rim composition for augite microphenocryst in KP-4; crystals is next to olivine phenocrysts in Table A3, columns 1and 23: Hypersthene in cluster shown in Fig. 8A4: Augite rim on hypersthene in cluster shown in Fig. 8A5: Ilmenite inclusion in hyperstehene (Fig. 8A)6–10: Compound grain shown in Fig. 4. Compositions in columns8–11 taken along line of traverse marked in that figure. Augite incolumn 11 is from darkest area adjacent to opx lamella
11–13: Sector-zone augite from sample KP-25 shown in Fig. 6.Compositions taken along line of traverse14–16: Hypersthene xenocryst (Fig. 8B) and augite micropheno-crysts from differentiated scoria in sample KP-26 (see glass analy-sis in column 5 in Table 1b)17: Euhedral augite phenocryst (unzoned) in sample KP-2818–19: Reversely zoned augite with phantom sector zoning (butno preserved compositional differences) in sample KP-28
Table A3 Compositions of olivine in 1960 spatter samples
Column no.Field No.
1KP-4
2KP-4
3KP-22
4KP-24
5KP-24
6KP-25
7KP-25
8KP-25
9KP-25
Type Euhedral Deformed Euhedral Olivine fragment Deformed
GrainNo. of point
Core3
Rim2 3
Core4
Rim3
Core4
Rim2
Core3
Rim1
SiO2
TiO2
Al2O3
Cr2O3
SFeOMnONiOMgOCaONa2O
39.20.020.040.03
17.90.230.31
41.60.210.02
38.90.020.040.03
20.30.260.18
39.60.250.03
40.90.020.000.10
11.90.19
P46.10.250.00
39.40.040.020.02
20.50.280.15
39.60.230.00
39.00.050.010.06
18.00.270.22
41.50.300.02
38.90.010.060.03
20.20.290.25
40.20.280.03
38.60.020.060.04
18.10.270.28
41.80.320.02
40.50.060.000.07
13.60.19
P44.60.280.04
40.10.070.000.04
15.80.20
P43.50.310.03
Sum 99.56 99.61 99.46 100.24 99.43 100.22 99.45 99.34 100.05
Fo (mol %) 80.6 77.7 87.3 77.5 80.4 78.0 80.5 85.4 83.0
1–2: Olivine phenocryst from early sample KP-4. Euhedral, nor-mal zoning3: Cores of several olivines from the cluster of deformed olivinesillustrated in Fig. 124–5: Resorbed reversely zoned olivine in KP-24, shown in Fig.11A. Black rectangles are plagioclase inclusions. (Compositiongiven in column 7 in Table A1)
6–7: Inclusion-rich fragment of olivine from KP-25, shown in Fig.11B8–9: Deformed olivine in KP-25; this grain is right next to the Fe-rich olivine fragment (columns 6 and 7). The glass surroundingthese two grains is shown in Table 1b, columns 1 and 2
(Continued)
628
Table A3 Continued
Column no.Field no.
10KP-26
11KP-26
12KP-26
13KP-26
14KP-26
15KP-28
16KP-28
Type Deformed Euhedral Resorbed
GrainNo. of points
Core6
Rim1
Core5
Rim2 4
Core4
Rim1
SiO2
TiO2
Al2O3
Cr2O3
SFeOMnONiOMgOCaONa2O
40.20.020.080.10
11.90.190.32
46.70.300.02
39.50.050.000.04
16.80.28
P42.60.280.04
38.80.040.040.02
21.60.270.18
39.30.260.04
39.10.040.030.03
19.50.250.20
40.80.320.02
38.80.080.050.03
22.10.270.10
38.60.270.02
38.70.010.070.02
21.70.290.21
39.00.280.02
38.60.030.090.03
19.30.230.29
41.50.360.02
Sum 99.83 99.59 100.55 100.29 100.32 100.30 100.42
Fo (mol %) 87.3 81.9 76.4 78.9 75.6 76.2 79.3
10–11: Olivine xenocryst in hot scoria in sample KP-26 (glasscomposition shown in Table 1b, column 3)12–13: Euhedral, reversely zoned olivine microphenocryst in hotscoria in sample KP-2614: Olivine microphenocryst in differentiated scoria in sampleKP-26 (glass analysis shown in Table 1b, column 5)
15–16: Resorbed, coarsely skeletal (?) olivine, with plagioclase in-clusions, from sample KP-28. Inclusion compositions shown inTable 1b, column 10
Table A4 Trace element data for the 1960 and related eruptions
Sample Eruption MayRb n
DecRb n
MaySr n
DecSr n
MayY n
DecY n
MayZr n
DecZr n
MayNb n
DecNb n
52-5-32Iki-22Iki-58F-01F-06F-07F-09F-11F-13F-14F-16F-17F-18F-19F-19GLF-20K-61-01K-61-17K-61-22HM68-02HM68-15
19521959E1959E196019601960196019601960196019601960196019601960196019611961196119671967SD (%)a
SD (ppm)a
6
131214111112898
109
118
1010122
965
101111987
1011998
10111411999
285
424420417378365358362349341363348393360
374400
1.55
427284371418428408386363350372361344365348412369420416404392407
19
29313026242425252425252724
232742
291826333434282526272626282830263030272724
122
211222214183173175175166160173165189170
166173
24
173119165218229211187169170179169159178165192171180183180171173
17
24222519202017192020172120
15179.92
221625242624232018192118171921202321241714
NOTE: Samples were analyzed by XRF twice, in May and De-cember, 1989, and values were normalized to standard sampleBHVO-1, analyzed at the same time. The normalized data (e.g.,Rb n) are tabulated in this table. The December analyses areused in Tables 2 and 3b.a Standard deviations are calculated for replicated analyses of thesame sample done at the same time. The values given are the
average for all samples replicated and are used to estimate theprecision of analysis. The intra-sample variation is not known.The upper row gives the analytical precision as a percentage ofthe average value of the analyzed element; the lower row givesthe analytical precision in parts per million
(Continued)
629
Table A4 Continued
Sample Eruption MayBa n
DecBa n
MayNi n
DecNi n
MayCu n
DecCu n
MayZn n
DecZn n
MayCr n
DecCr n
52-5-32Iki-22Iki-58F-01F-06F-07F-09F-11F-13F-14F-16F-17F-18F-19F-19GLF-20K-61-01K-61-17K-61-22HM68-02HM68-15
19521959E1959E196019601960196019601960196019601960196019601960196019611961196119671967SD (%)a
SD (ppm)a
106
168166148152142126134135126136140144131
121130
3.512
178110165176173170149152129146138144144131173141165157145147119
985
127103159372421532483538622377544137394
126129
2.75
12273112211891
140270312383337402451295421132331141131133115116
134
145146148140134128130127126120114129110
125125
34
127126123140134143133126125119130131116130135127140143130126126
119
135141135129126122117119122121124122113
127123
33
122120111129134136126122122118115119125123123115123126118115118
1486
204144229444656686618681732614728338590
345347
39
3071586439222149247507686771665718790665817377658431381386413409
Acknowledgements We thank M. Garcia for providing unpub-lished microprobe data for plagioclase phenocrysts in the 1955lavas. David Clague and Elizabeth Moll-Stalcup made thoroughand incisive reviews of an early version of the manuscript. S. Maa-loe, M. Carroll, and an anonymous reviewer also contributed sig-nificantly to improving the final interpretations and their presen-tation.
References
Anderson AT (1976) Magma mixing: petrological process andvolcanological tool. J Volcanol Geotherm Res 1 :3–33
Anderson AT, Wright TL (1972) Phenocrysts and glass inclusionsand their bearing on oxidation and mixing of basaltic magmas,Kilauea volcano, Hawaii. Am Mineral 57 :188–216
Clague DA, Denlinger RP (1994) Role of olivine cumulates indestabilizing the flanks of Hawaiian volcanoes. Bull Volcanol56 :425–434
Clague DA, Moore JG, Dixon JE, Friesen WB (1995) Petrologyof submarine lavas from Kilauea’s Puna Ridge, Hawaii. J Pe-trol 36 :299–349
Delaney PT, Fiske RS, Miklius A, Okamura AT, Sako MK (1990)Deep magma body beneath the summit and rift zones of Ki-lauea Volcano, Hawaii. Science 247 :1311–1316
Dixon JE, Clague DA, Stolper EM (1991) Degassing history ofwater, sulfur, and carbon in submarine lavas from KilaueaVolcano, Hawaii. J Geol 99 :371–394
Eichelberger JC (1975) Origin of andesite and dacite: evidence ofmixing at Glass Mountain in California and at other circum-Pacific volcanoes. Geol Soc Am Bull 86 :1381–1391
Fiske RS, Swanson DA, Wright TL (1993) A model of KilaueaVolcano’s rift-zone magma system (abstract). Eos, Trans AmGeophys Union (Suppl) 74 :646
Garcia MO, Ho RA, Rhodes JM, Wolfe EW (1989) Petrologicconstraints on rift-zone processes. Bull Volcanol 52 :81–96
Garcia MO, Wolfe EW, Ulrich GE, Ho RA (1992) Petrology oflavas from episodes 2–47 of the Puu Oo eruption of KilaueaVolcano, Hawaii: evaluation of magmatic processes. Bull Vol-canol 55 :1–16
Gerlach TM, Graeber EJ (1985) Volatile budget of Kilauea volca-no. Nature 313 :273–277
Ghiorso MS (1985) Chemical mass transfer in magmatic processesI. Thermodynamic relations and numerical algorithms ContribMineral Petrol 90 :107–120
Greenland P, Rose WI, Stokes JB (1985) An estimate of gasemissions and magmatic gas content from Kilauea Volcano.Geochim Cosmochim Acta 49 :125–129
Helz RT (1987a) Diverse olivine types in lava of the 1959 erup-tion of Kilauea Volcano and their bearing on eruption dynam-ics. U.S. Geol Surv Professional Paper 1350 1 :691–722
Helz RT (1987b) Differentiation behavior of Kilauea Iki lavalake, Kilauea Volcano, Hawaii: an overview of past and cur-rent work. Magmatic processes: physicochemical principles.Geochem Soc Spec Publ 1 : 241–258
Helz RT, Thornber CR (1987) Geothermometry of Kilauea Ikilava lake, Hawaii. Bull Volcanol 49 :651–668
Helz RT, Wright TL (1992) Differentiation and magma mixing onKilauea’s east zone. Bull Volcanol 54 :361–384
Helz RT, Hon K, Heliker C (1993) Thermal efficiency of lavatubes at Kilauea Volcano, Hawaii. In: IAVCEI General As-sembly on Ancient Volcanism and Modern Analogues, Can-berra, Australia, Sept 1993 (abstract) :p. 47
Helz RT, Banks NG, Heliker C, Neal C, Wolfe EW (1995) Com-parative geothermometry of recent Hawaiian eruptions. JGeophys Res 100 :17637–17657)
Ho RA, Garcia MO (1988) Origin of differentiated lavas at Ki-lauea Volcano, Hawaii: implications from the 1955 eruption.Bull Volcanol 50 :35–46
630
Jackson ED (1968) The character of the lower crust and uppermantle beneath the Hawaii Islands. In: 23rd InternationalGeological Congress, Prague, 1968. 23rd Proc Int Geol Congrsec 1, Upper mantle (geological processes), pp 135–150
Klein FW, Koyanagi RY, Nakata JS, Tanigawa WR (1987) Theseismicity of Kilauea’s magma system. U.S. Geol Surv Profes-sional Paper 1350 2 :1019–1185
Maaloe S, Hansen B (1982) Olivine phenocrysts of Hawaiian oli-vine tholeiite and oceanite. Contrib Mineral Petrol 81 :203–211
Macdonald GA, Eaton JP (1964) Hawaiian volcanoes during1955. US Geol Surv Bull 1171 :1–170
Murata KJ, Richter DH (1966) Chemistry of the lavas of the1959–1960 eruption of Kilauea volcano, Hawaii. US Geol Pro-fessional Paper 537-A:A1–A26
Rhodes JM, Dungan MA, Blanchard DP, Long PE (1979) Magmamixing at mid-ocean ridges: evidence from basalts drilled near22 7N on the Mid-Atlantic ridge. Tectonophysics 55 :35–61
Richter DH, Murata KJ (1966) Petrography of the lavas of the1959–1960 eruption of Kilauea volcano, Hawaii. US Geol SurvProfessional Paper 537-D:D1–D12
Richter DH, Ault WU, Eaton JP, Moore JG (1964) The 1961eruption of Kilauea Volcano, Hawaii. US Geol Surv Profes-sional Paper 474-D:D1–D34
Richter DH, Eaton JP, Murata KJ, Ault WU, Krivoy HL (1970)Chronological narrative of the 1959–1960 eruption of Kilaueavolcano, Hawaii. US Geol Surv Professional Paper 537-E :E1–E73
Roeder PL, Emslie RF (1970) Olivine-liquid equilibrium. ContribMineral Petrol 29 :275–289
Ross M, Huebner JS (1979) Temperature-composition relation-ships between naturally occurring augite, pigeonite, and or-thopyroxene at one bar pressure. Am Mineral 64 :1133–1155
Russell JK, H Stanley CR (1990) Origins of the 1954–1960 lavas,Kilauea Volcano, Hawaii: major element constraints on shal-low reservoir magmatic processes. J Geophys Res 95 :5021–5047
Ryan MP (1988) The mechanics and three-dimensional internalstructure of active magmatic systems: Kilauea Volcano, Ha-waii. J Geophys Res 93 :4213–4248
Thompson RN, Tilley CE (1969) Melting and crystallisation rela-tions of Kilauean basalts of Hawaii. The lavas of the 1959–1960 Kilauea eruption. Earth Planet Sci Lett 5 :469–477
Wright TL (1971) Chemistry of Kilauea and Mauna Loa lava inspace and time. US Geol Surv Professional Paper 735 :1–40
Wright TL (1973) Magma mixing as illustrated by the 1959 erup-tion, Kilauea volcano, Hawaii. Geol Soc Am Bull 84 :849–858
Wright TL, Fiske RS (1971) Origin of the differentiated and hy-brid lavas of Kilauea volcano, Hawaii. J Petrol 12 :1–65
Wright TL, Heliker CC (1987) Magma storage and differentiationin the Kilauea east rift zone prior to the Puu Oo eruption(January 1983–present). In: Geol Soc Am 83rd Annual Meet-ing, Cordilleran Section, Hilo, Hawaii, 20–22 May, abstractswith programs, p. 465
Wright TL, Swanson DA, Duffield WA (1975) Chemical compo-sitions of Kilauea east-rift lava, 1968–1971. J Petrol 16 :110–133
Wright TL, Tilling RI (1980) Chemical variation in Kilauea erup-tions 1971–1974. Am J Sci 280-A:777–793
