Pacific Science (1992), vol. 46, no. 1: 1-10 © 1992 by University of Hawaii Press. All rights reserved
Puu Mahana Near South Point in Hawaii Is a Primary Surtseyan Ash Ring, Not a Sandhills-type Littoral Cone!
GEORGE P. L. WALKER 2
ABSTRACT: Puu Mahana has previously been interpreted to be a littoral cone, formed at a secondary rootless vent where lava flowed from land into the ocean, but a number oflines ofevidence point to it being a remnant ofa Surtseyan tuff ring built on a primary vent. The differences between it and littoral cones are highlighted by a comparison of Puu Mahana with the undoubted littoral cone ofthe Sandhills that was observed to form in the 1840 flank eruption ofKilauea Volcano. Puu Mahana contains abundant lithic debris and accretionary lapilli, absent in the Sandhills deposit. Compared with the Sandhills, the Puu Mahana pyroclastic deposit is finer grained and more poorly sorted, and its juvenile component is less dense and more highly vesiculated. Puu Mahana lies 3 to 4 km offMauna Loa's southwest rift zone. Identification ofit as a primary vent implies that the lower rift zones of Hawaiian volcanoes can be much wider and more diffuse or more mobile than is currently acknowledged. The olivine grains that compose the well-known green-sand beach at Puu Mahana are likely derived from the ash, strongly concentrated and somewhat abraded by wave action.
Puu MARANA, A SMALL coastal hill situated 5 occurrence was fully described by Wentworth km northeast of South Point on the island of (1938), who drew a contour map of the cone. Hawaii, is an upstanding portion ofthe rim of Historical littoral cones in Hawaii were de an ash cone that has been partly buried by scribed by Moore and Ault (1965), and a de basalt lavas. Previous workers considered that tailed description including grain size analyses Puu Mahana is a littoral cone formed by ex of Puu Hou, the littoral cone group formed plosions at a secondary or rootless vent where by the 1868 eruption ofMauna Loa, was given lava flowed from land into water, but a critical by Fisher (1968). A general discussion of the reexamination indicates that the cone was diagnostic features oflittoral cones was given formed by a Surtseyan-style eruption at a by Macdonald (1972). He pointed out that primary vent. Nearby Mahana Bay formed the ash grains are usually denser and contain by preferential erosion of the ash. It is well fewer bubble-wall shards than those ofprima known for its green olivine-rich beach sand, ry vents; how dense the fragments are depends perhaps the finest example in Hawaii. This on how far the lava has traveled before reach study includes data on the green sand. ing the sea. He also noted that the explosive Puu Mahana was identified as a littoral activity that generates littoral cones preferen cone by Wentworth (1938), Stearns and tially occurs on aa lava flows, and suggested Macdonald (1946), Wentworth and Mac that the rough rubbly surface of aa provides donald (1953), Macdonald and Abbott many channels for ingress of water and pro (1970), and Macdonald et al. (1983). The motes explosive activity. To illustrate the characteristics of primary Surtseyan deposits and the differences be 1 Fieldwork on the island of Hawaii was funded by the tween them and littoral cones, a direct com Jaggar Bequest Fund. This is School of Ocean and Earth parison is now made between Puu Mahana Science and Technology contribution number 2661. Manuscript accepted 26 April 1991. and the unambiguous littoral cone of the 2 Department of Geology and Geophysics, University Sandhills (Puu Nanawale) in the Puna District of Hawaii at Manoa, Honolulu, Hawaii 96822. on the north coast ofKilauea. Sandhills is the 2 PACIFIC SCIENCE, Volume 46, January 1992 most accessible littoral cone in Hawaii, formed Puu Mahana ash on both sides of Mahana by explosions where the lava flow of 1840 Bay. The upper lava on the southwestern side entered the ocean. Identification of Puu and both lavas on the northeastern side con Mahana as a primary vent structure has tain abundant tree molds up to 16 cm in diam. important implications regarding the position at or near their base, showing that a consider and width of Mauna Loa's southwest rift able time interval elapsed between the Puu zone. Mahana eruption and the formation of these basalts. Puu Mahana is moreover mantled by Pahalaash, suggesting that it is a relatively old feature. A carbonaceous soil at the base of THE PUU MARANA CONE one ofthe overlying lavas yielded a 14C age of Puu Mahana is about 38 m high, but this is 28,000 yr (sample W4368, Rubin et al. 1987). a minimum height value since the base of the Puu Mahana therefore exceeds 28,000 yr in deposit extends an unknown distance below age. At the known subsidence rate of Hawaii sea level. The Sandhills deposits are smaller: (Moore 1987), it has probably subsided by at they rest on a lava platform and rise only 25 m least 67 m since it formed. The visible part of above it. This is also a minimum value, Puu Mahana is thus the top of a much larger however, since the dip is inland and the structure. This is reasonable: the primary tuff centers ofthe original ash cones were situated rings of Diamond Head and Koko Crater on some distance offshore and have since Oahu are respectively 232 m and 368 m high. been eroded away. The classic line of evidence for a littoral As shown by the map ofWentworth (1938), cone, namely the occurrence of two part Puu Mahana is an ash ridge about 250 m long cones on either side of a lava corridor, with by 120 m wide, the long axis of which trends synchroneity of the part-cones and lava, is roughly northwest at right angles to the coast. not available at Puu Mahana because only The ridge is abruptly truncated by, and is well one cone remnant is visible. The question of exposed in cross section in, the coastal cliff. whether or not it is a littoral cone must, there The well-known gre~n sand beach occurs at fore, depend on other lines of evidence. the foot of this cliff. Younger basaltic lava Four channel samples and eight samples of flows overlap onto the ashes on both sides of single beds were collected from different levels the ridge. Mahana Bay is a broad slot between to characterize the deposit. Each channel sam these lavas and occupies the seaward exten ple was collected across about 2 m of deposit sion of the ridge where the ashes were prefer thickness from a narrow-cut channel of entially eroded out by the ocean. approximate uniform depth and width so as Puu Mahana consists of well-bedded to to be representative of that thickness. These laminated ash deposits having a southwest samples were sieved employing a set of sieves ward dip of 10° to 20° (less than the angle of having a I-phi aperture spacing (Inman 1952). repose of loose pyroclastic materials). It is Bulk-sample and class densities (the density of apparently a small segment ofan ash ring, the material in each grain size class) were ob remainder ofwhich is either buried or eroded tained by weighing the samples and estimating away. The boundary between ash and over their volumes in graduated cylinders; an em lapping basalt is steeper, about 40°, on the pirical correction to the class densities was northeast side of the ridge (Figure Ie). This applied for intergrain voids. Component anal steeper slope is regarded as the inner wall of yses were performed on some samples. a crater. Angular to subangular lithic frag The ash deposit is incipiently lithified. The ments consisting of various basalt types are finest beds and fine aggregates (e.g., accre dispersed widely through the deposit. The tionary lapilli) are most strongly lithified largest, about 50 cm, are concentrated on the probably because of the large surface area of steep northeastern side and tend to confirm small grains and the narrowness of the capil that this is the inner wall of a crater. laries between them. Complete disaggregation Two younger basaltic flows overlap onto ofsome samples for sieving was not achieved. b
sea level C PUU MAHANA
younger lavas
I II I, I IIIIIII ( d
km 30 I
FIGURE 1. a, Sketch map of the 1840 littoral cones of Sandhilis, Kilauea volcano; b, cliff section of the Sandhilis littoral cones, vertical exaggeration x 2; c, cliff section ofthe Puu Mahana cone, on the same scale as a and b; d, index map of part of the island of Hawaii showing the locations of sites mentioned in the text. Old south rift zone after Lipman et al. (1990). Dashed lines: isobaths, at IOOO-m intervals. 4 PACIFIC SCIENCE, Volume 46, January 1992
THE SAND HILLS LITTORAL CONE flowed down and mantled a cliff. The pre-l840 coastline was thus slightly seaward ofthe pres The Puu Nanawale (Sandhills) ash deposit ent coastline at the western part-cone and extends about 400 m along the north Puna coincident with it at the eastern cone. The fact coast and consists of two asymmetric part that the lowest ash rests on this lava platform cones nearly equal in size separated by a nar and the centers of the part-cones are seaward row corridor ofbasalt lava (Figure la, b). The ofthe present coastline indicates that the main western cone rises to about 36 m above sea explosive activity did not begin until the lava level, and the ash attains a maximum thick had flowed some distance north ofthe present ness of about 25 m. The eastern part-cone is coastline. slightly lower, and the ash is slightly thinner. The ash deposit is rather poorly bedded and The ash ofboth part-cones rests on the lowest grain size variations between beds are rather flow units of the 1840 lava. small. In detail the thickness of some beds At each end of the coast section the ash is varies laterally, suggestive ofdeposition from seen to thin rapidly. Basalt flow units overlap weak base surges or in a strong wind. Some onto it, and some ash is interbedded with lava beds are predominantly made of ash-grade flow units. Interbedding with lava is also seen particles, but their content of fine ash is within part of the western cone. Basalt of very low. The coarsest beds contain abundant the corridor that separates the two part-cones basalt fragments several centimeters in size overlaps onto the ash with minor inter accompanied by much ash-grade material. All bedding. These overlapping contacts are the pyroclastic material is juvenile, and olivine much steeper than the outer ones, and the is a prominent constituent. eastern part-cone rises particularly steeply Five channel samples and eight samples of above the corridor, suggestive of erosion and single beds were collected to characterize the undercutting of ash by the flowing lava. The deposit. One channel sample each was collect ash is baked to a bright red color in places ed from the upper and lower part of each where basalt rests on it. part-cone, plus one from a finer-grained basal It is apparent from the 30° to 35° inland dip part ofthe eastern cone. An important feature of the ash that the centers of the part-cones of the pyroclastic material is its high density lay seaward ofthepresent coast and have been and weak vesicularity. The class-density and eroded away. The eyewitness account by the sample bulk density values of the Sandhilis Rev. Titus Coan (Brigham 1909) recorded that deposit are higher than for any sampled de "three hills of scoria and sand were (also) posits of primary-vent cinder cones or ash formed in the sea, the lowest about 200, and rings in Hawaii and the Azores (Figure 2). the highest about 300 ft. high.... The sand Bombs range in size up to about 35 cm and hills thrown up at this place were found to be have an irregular cauliflowerlike shape. The 150, and 250 ft. high eight months after their irregular shape results from centimeter-sized formation, but since then the sea has removed vesicles that are absent from most outcrops of the whole mass [sic]. Even in 1865 they were 1840 lava and appear to be due to the penetra not a third of the measured height." Survival tion into the lava of steam generated from of the part-cones that remain on this north seawater. The more general millimeter-sized ward-facing coast, exposed as it is to the trade vesicles are similar in size to those in the lava winds, is clearly due to the fact that the ash crust and were already present in the lava rests on a platform oferosion-resistant basalt when it entered the ocean. Some flow units of lava. The ash itself is completely loose and the lava are pahoehoe, and others are aa. shows no observed sign oflithification. The olivine-rich 1840 lava of the platform below the western part-cone overlies an DIFFERENCES BETWEEN THE PUU MARANA olivine-poor lava interpreted to be a pre-l840 AND SANDHILLS DEPOSITS flow. The 1840 lava below the eastern part cone locally has a steep seaward slope and The following differences are exhibited be smooth upper surface and appears to have tween the Puu Mahana and Sandhills deposits. Puu Mahana, a Primary Surtseyan Ash Ring-WALKER 5
3....------, superficial nature of littoral explosions and failure oflittoral vents to cut down into older rocks.
OCCURRENCE OF ACCRETIONARY LAPILLI. Pea-sized accretionary lapilli are abundant at all levels in the Puu Mahana deposit, but ap pear to be absent from the Sandhills deposit. I follow Moore and Peck (1962) in interpreting "'2 ...... E most accretionary lapilli, including many of OJ the Puu Mahana ones, as resulting from the ~ capture and accretion of ash by raindrops ~ falling through an ash cloud. Accretionary 'en-c lapilli having this origin require a combina (]) tion of three circumstances, namely, a plenti "0 (J) ful supply of fine ash, rain, and an ash cloud (J) sufficiently high or sufficiently dense to enable CI1 (31 falling raindrops to capture enough ash parti cles. To survive, accretionary lapilli must also be fairly dry when they land, otherwise they splash or break up on impact with the ground. Rain may well have fallen during both the Puu Mahana and Sandhills eruptions, but it is surmised that a low content of fine ash and inadequate cloud height account for the lack ofaccretionary lapilli in the Sandhills deposit. Q..J-+--t--j--t---t---t----:l--+-+__' 4 1 14 116 16 VESICULARITY OF JUVENILE CLASTS. Juvenile Sieve aperture size, mm material in the Puu Mahana cone includes
FIGURE 2. Class densities of sieved samples, as they fragile pumice up to about 10 cm having a 3 vary with grain size, for Puu Mahana, Sandhills, and a density ofO.7mg/m orless. Dense and poorly number ofother basaltic pyroclastic deposits. Solid lines: vesiculated blocky clasts also occur. The oc Surtseyanash rings (Capelinhos 1957-1958, Faial, Azores; currence of highly inflated pumice is charac Velas, San Jorge, Azores; Karl, Reykjanes, Iceland). Dashed lines: Strombolian cinder zones and basaltic teristic ofSurtseyan tuffrings and is consistent plinian deposits (Tantalus, Oahu, Hawaii; Paricutin, with the maintenance ofexplosive eruption by Mexico; Tarawera 1886, New Zealand; Fogo scoria cone, internal vesiculation of the magma. Pumice San Miguel, Azores; Etna 1974 flank vent; Mt. Rossi is absent from the Sandhills deposit, and the 1969, Etna; Heimaey 1973, Iceland). uniformly high density ofthe ejecta (Figure 2) It is not known to what extent these differ is consistent with their derivation from lava ences are specific to these two particular oc that had earlier passed through a stage of currences or are applicable to primary Surt intense vesiculation (at its primary vent, in seyan and rootless littoral cones in general. land) and had lost most ofthese vesicles; it was vesiculating only slightly ifat all at its second LITHIC CONTENT. Puu Mahana contains ary vent, although centimeter-sized vesicles conspicuous lithics, whereas Sandhills con were forming where vaporized seawater en tains none. The lithics are angular and include tered the lava. a variety ofbasalt types. They occur mostly as blocks up to 50 cm, and ash-grade lithics con DENSITY VARIATION WITH CLAST SIZE. When situte only a very small proportion (about 2%) pyroclastic deposits are sieved, the class den of the deposit. Surtseyan tuff rings invariably sities increase as the grain size decreases (Fig contain lithics, presumed to have been derived ure 2). The main reason is that the vesicle from vent walls, and the absence of lithics population has a wide size range, and as clasts from Sandhills is consistent with the very decrease in size they are capable of enclosing 6 PACIFIC SCIENCE, Volume 46, January 1992 only the vesicles in the ever-finer fine tail of difficult, and the amount of fine ash in ana this population. The rate of change of class lyzed samples is underestimated. density cruoely reflects the vesicle population, Channel samples reveal that the Puu except that when the erupted magma is por Mahana deposit is consistently finer grained phyritic a density peak may occur in the than the Sandhills deposit (Figures 3,4). grain size classes in which the dense crystals In windy weather a proportion of the fine occur. erupted ash is blown away and deposited The rate ofchange of class density for Puu outside the limit of the cone; no information Mahana is normal for pyroclastic deposits at is available on these distal deposits of either primary vents where the magma has a youth Puu Mahana or Sandhills, and whole-deposit ful vesicle population. The maximum rate grain size populations cannot therefore be ofchange for Sandhills occurs at larger grain determined. sizes than is normal for deposits of primary vents, reflecting derivation of the clasts from GRAIN SIZE PARAMETERS OF PYROCLASTIC a lava flow that had a mature vesicle popula BEDS. Each bed in a pyroclastic deposit has tion but gained large vesicles from vaporized a grain size distribution that can be character seawater. ized by such parameters (following Inman 1952) as the median grain diameter (Md) and graphic standard deviation (0), the values of STATE OF LIBERATED CRYSTALS. Both Puu which have some diagnostic value. Walker Mahana and Sandhills contain abundant and Croasdale (1972) showed that Surtseyan olivine crystals liberated from the magma and Strombolian beds plot on different fields (Figure 3). In the Puu Mahana ash about half on an MdrjJj(JrjJ plot, reflecting significant ofthem are euhedral and unbroken and retain differences in the grain size populations and a thin layer of glass with bubble-wall texture depositional mechanisms (Figure 5). on their surface. In contrast, practically all in The Puu Mahana samples plot in the Surt the Sandhills deposit are broken grains, con seyan field, whereas Sandhills samples (and sistent with fragmentation of less vesicular also samples from the Puu Hou littoral cone: lava. The difference could alternatively be Fisher 1968) plot astride the Surtseyanj due to a higher viscosity ofthe 1840 Sandhills Strombolian field boundary. lava. Trusdell (1991) commented on the uni form content of olivine phenocrysts in this CONE SLOPE ANGLE. In weak explosive erup lava, a feature that implies a relatively high tions the pyroclasts fall preponderantly very lava viscosity (Rowland and Walker 1988). close to the eruptive vent; their redistribution by gravity then generates steep-angle cones CONTENT OF FINE ASH. The content of fine sloping at the angle of repose (33°-36°) of ash (finer than 63 j.Lm) in basaltic pyroclastic loose materials, exemplified by typical cinder deposits depends partly on how much external cones of Strombolian-style eruptions. As water participates. It is typically very low in eruption power increases, the fallout is more Strombolian-style deposits and high in Surt widely dispersed and low-angle cones stand seyan ones (Walker 1973). The Sandhills, al ing at less than the angle ofrepose result. Puu though not a typical Strombolian deposit, re Mahana is such a low-angle cone: its eruption sembles one in conta,ining a negligible propor was more powerful than that of Sandhills, tion (about 1% wt in channel samples) offine consistent with it being a primary vent ash. Channel samples from Puu Mahana have structure. much more (average nearly 10%), typical of but somewhat less than for other Surtseyan ash rings. It compares with average values of ORIGIN OF THE GREEN-SAND BEACH 13% for the Karl ash ring in Reykjanes, Ice land, and 19% for the 1957-1958 Capelinhos Sieve and component analyses of two ring in Faial, Azores. Puu Mahana is slightly samples of the Mahana Bay beach sand give lithified, so that complete disaggregation is Md of +0.6 and + 1.1 phi (0.66 and 0.47 mm, Puu Mahana, a Primary Surtseyan Ash Ring-WALKER 7
60 wt% ·::·· scoria olivine crystals lithics Ibiogenic grains ...... :::. GREEN . .' .... BEACH SAND
20 wt%
PUU MAHANA 10
0~·4·. 16 4 116 16 4 1 4 Grainsize in mm
FIGURE 3. Grain size histograms ofchannel samples from Puu Mahana and Sandhills (note lesser amount offines in the latter) and two samples of olivine-rich beach sand from Puu Mahana. The bars for i mm and coarser are subdivided to show the proportions of their components. respectively) and a graphic standard deviation nents (enrichment factor = C2/P2 x PrlCI, (0') of 0.55 to 0.5 phi (Figures 3,4). These where CI and PI are weight percentages of values are normal for beach sands. Crystals crystals and pumice in crushed pumice, and (mostly olivine; also some pyroxene) consti C2 and P2 are weight percentages ofthese com tute 88 and 92% wt, respectively, of these ponents in the concentrate). The lava flows on samples. They are well rounded. Coarse pum either side of Mahana Bay are rich in olivine ice from the Puu Mahana ash when crushed and may have contributed to the beach sand. yields 17% wt of olivine having a maximum diameter of 3 mm and a median of +0.3 phi (0.81 mm). An almost identical olivine population occurs in channel samples on IMPLICAnONS which component analyses were performed The foregoing datacollectively point to Puu (Figure 4). Mahana as being a remnant of an ash cone This is consistent with the olivine in the resulting from a Surtseyan-style eruption at beach sand being derived from erosion of the a primary vent. A number of differences ash, suffering a general reduction in diameter between Puu Mahana and Sandhills are docu of20 to 40% by abrasion, and being enriched mented. Some may be specific to these two by a factor of 44 as prolonged wave action particular occurrences and may not be general selectively removed the less-dense compo- differences diagnostic of primary Surtseyan 8 PACIFIC SCIENCE, Volume 46, January 1992 98,-.------....,....,....-----, -,------,-----,----,
95
~ 90 .r::. +- ~ 80 Q) C/) ~ ~ o () ~50 +- ~ Q) > +- ~ 20 ;:,
E;:, 10 () 5
2-+--+----l--+--+--+------1i---+--+---+--i+-+--,-L---r--,---j -4 o 4phi 2 16 8 4 2 1 12 14 18 116 mm 2 14 Sieve aperture size
FIGURE 4. Plots on probability graph paper showing grain size distributions. Left, channel samples from Puu Mahana and Sandhills; the latter are systematically finer. Right, A and B, two samples ofolivine-rich beach sand from Puu Mahana as in Figure 3; C, olivine population separated from crushed pumice of Puu Mahana; D, olivine population in two channel samples (averaged) ofPuu Mahana ash.
versus rootless littoral ongm. Probably a and Lipman et al. (1990) marked by a progres spectrum of types with gradational charac sive westward lateral shift of the rift zone teristics exists, depending on the volumetric (away from Kilauea, inferred to have acted flux of lava, the relative fluxes of lava and as a barrier to symmetrical growth). If, as was water (Wohletz 1983, 1986), and how inti suggested by Lipman (1980a), the anomalous mately lava and water are intermingled. Ninole Hills mark an earlier flank of Mauna Recognition of the primary-vent origin of Loa, Puu Mahana represents a westward mi Puu Mahana has important implications re gration by about halfofthe distance from the garding the position and width of Mauna Ninole Hills to the present rift zone. Loa's southwest rift zone. Puu Mahana lies 3 Puu Mahana is also about 17 km farther to 4 km east ofthe rift zone and may indicate south than the southernmost documented his that the rift zone is much wider than is cur torical primary vent on Mauna Loa (that of rently recognized. It is, however, consistent the 1868 eruption). This is consistent with with a trend pointed out by Lipman (l980a, b) another trend, documented by Moore et al. Puu Mahana, a Primary Surtseyan Ash Ring-WALKER 9
4,-.------, .SURTSEYAN X STROMBOLIAN • ..c •• .:... 1& • Q. ."". 52 . .. . ~-...'.~ ~.-a: . +oJ -x:-.--x: •• • ~ XX~>-
FIGURE 5. Plots of graphic standard deviation (0") against median grain diameters (Md) ofsieved samples ofsingle beds. Above, from typical Surtseyan ash rings and Strombolian cinder zones (Walker and Croasdale 1972); dashed line delineates approximate boundary between fields. Below, Puu Mahana samples plot in the Surtseyan field; Sandhills and Puu Hou (Fisher 1968) littoral cone samples straddle the Surtseyan/Strombolian field boundary.
(1990), ofrift zone activity becoming increas LITERATURE CITED ingly restricted toward the upper part of Mauna Loa (a condition possibly heralding BRIGHAM, W. T. 1909. The volcanoes of the end of the shield-building stage of Kilauea and Mauna Loa on the island of activity). Hawaii. Bishop Museum Press, Honolulu. FISHER, R. V. 1968. Puu Hou littoral cones, Hawaii. Geol. Rundsch. 57: 837-864. INMAN, D. L. 1952. Measures ofdescribing the ACKNOWLEDGMENTS size distribution of sediments. J. Sediment. Petrol. 22: 125-145. Thanks are given to Peter W. Lipman for LIPMAN, P. W. 1980a. Rates ofvolcanicactivi helpful comments. ty along the southwest rift zone of Mauna 10 PACIFIC SCIENCE, Volume 46, January 1992
Loa volcano, Hawaii. Bull. Volcanol. 43: flows. J. Volcanol. Geotherm. Res. 35: 703-725. 55-66. ---. 1980b. The southwest rift zone of RUBIN, M., L. K. GARGULINSKI, and J. Mauna Loa: Implications for structural MCGEEHIN. 1987. Volcanism in Hawaii. evolution of Hawaiian volcanoes. Am. J. Hawaiian radiocarbon dates. U.S. Geol. Sci. 280-A: 752-776. Surv. Prof. Pap. 1350: 213-242. LIPMAN, P. W., J. M. RHODES, and G. B. STEARNS, H. T., and G. A. MACDONALD. 1946. DALRYMPLE. 1990. The Ninole Basalt Geology and ground-water resources ofthe Implications for the structural evolution island of Hawaii. Hawaii Div. Hydrogr. of Mauna Loa volcano, Hawaii. Bull. Bull. 9. Volcanol. 53: 1-19. TRUSDELL, F. A. 1991. The 1840 eruption of MACDONALD, G. A. 1972. Volcanoes. Prentice Kilauea Volcano: Petrologic and volcano Hall, Englewood Cliffs, New Jersey. logic constraints on rift zone processes. MACDONALD, G. A., and A. T. ABBOTT. 1970. M.S. thesis, University ofHawaii at Manoa, Volcanoes in the sea. University of Hawaii Honolulu. Press, Honolulu. WALKER, G. P. L. 1973. Explosive volcanic MACDONALD, G. A., A. T. ABBOTT, and F. L. eruptions-a new classification scheme. PETERSON. 1983. Volcanoes in the sea. 2d Geol. Rundsch. 62:431-446. ed. University of Hawaii Press, Honolulu. WALKER, G. P. L., and R. CROASDALE. 1972. MOORE, J. G. 1987. Volcanism in Hawaii. Characteristics of some basaltic pyroclas Subsidence of the Hawaiian Ridge. U.S. tics. Bull. Volcanol. 35:303-317. Geol. Surv. Prof. Pap. 1350: 85-100. WENTWORTH, C. K. 1938. Ash formations of MOORE, J. G., and W. U. AULT. 1965. Historic the island of Hawaii. Hawaiian Volcanoes littoral cones in Hawaii. Pac. Sci. 19: 3-11. Research Association, Honolulu. MOORE, J. G., and D. L. PECK. 1962. Accre WENTWORTH, C. K., and G. A. MACDONALD. tionary lapilli in volcanic rocks ofthe west 1953. Structures and forms ofbasaltic rocks ern continental United States. J. Geol. 70: in Hawaii. U.S. Geol. Surv. Bull. 994. 182-193. WOHLETZ, K. H. 1983. Mechanisms ofhydro MOORE, J. G., W. R. NORMARK, and B. J. volcanic pyroclastic formation: Grain-size, SZABo. 1990. Reef growth and volcanism scanning electron microscopy, and experi on the submarine southwest rift zone of mental studies. J. Volcanol. Geotherm. Mauna Loa, Hawaii. Bull. Volcanol. 52: Res. 17:31-63. 375-380. ---. 1986. Explosive magma-water inter ROWLAND, S. K., and G. P. L. WALKER. 1988. actions: Thermodynamics, explosion mech Mafic-crystal distributions, viscosities, and anisms, and field studies. Bull. Volcanol. lava structures of some Hawaiian lava 48: 245-264.