Distribution and Characteristics of Dikes in the Southeast Partof the Koolau Range, Oahu1

GORDON E. BIGELOW2

ABSTRACT: Exposed dikes and sills trending southwest and roughly perpendicularto the primary Koolau rift zone in the Waialae-Palolo area of Oahu are hypersthene-bearing tholeiitic rocks, intrusive equivalents of the Koolau Series basalt flows. NoHonolulu Series intrusives were located along a line joining Kaau Crater, Mauumae,Kaimuki, and Diamond Head, a secondary rift of the Koolau volcano. Mineralogy ofthe Koolau intrusives displays a striking chemical and morphological constancy.Feldspar crystals in these rocks show antipathy to ore inclusions. Cumuloporphyritictexture is strongly developed and may be related to flow patterns evident in ground-mass minerals.

FIELD WORK

Jones, 1940:981-982). The exploitation of thesouthwestward zone of crustal weakness in theKoolau volcano by Honolulu Series eruptivessuggested that some of the intrusives now ex-posed in the deeply incised Palolo and Wai-alaenui valleys might belong to the HonoluluSeries.

This study is a petrographic examination of26 intrusives sampled along the trend of theKaau rift. Maps prepared by Stearns (1939) andby Wentworth and Jones (1940) indicate ap-proximately 100 intrusive bodies located in thisarea. Continuous exposures of individual in-trusive bodies seldom exceed 20 feet, however,and are often less than 10 feet. The markedparallelism of these outcrops suggests that manydikes mapped separately are actually continuousunder the mantle of soil and vegetation.

Dikes were mapped and sampled during thesummer of 1967 in the Pukele (west) andWaiomao (east) branches of the Palolo Valleyand in the Waialaenui Valley. Locations wereestablished by Brunton compass, using utilitypoles and power lines as reference. Dikes weresampled at centers and margins. For larger in-trusives several samples were taken at intervalsinward from margins.

Most of the intrusive outcrops found in thisstudy were in or near streams. The three ridgesadjoining the Palolo and Waialaenui valleyswere searched, but soil and vegetation cover all

496

1 Manuscript received April 14, 1969.2 Chaminade College, Honolulu, Hawaii.

INTRUSIVE BODIES in the leeward slope of theKoolau Range, Oahu, outcrop along trendsclosely parallel to the inferred secondary riftzones of Honolulu Series volcanics (Wentworthand Jones, 1940:981). The Kaau rift, subjectof this study, was held by Winchell (1947:19)to be an underdeveloped primary south rift ofthe Koolau volcano (Fig. 1). Four distinctHonolulu Series vents occur aligned S 25 oWfrom Kaau Crater through Mauumae, Kaimuki,and Diamond Head. Other sites on windwardOahu suggest continuation of the Kaau riftnortheastward.

Though conspicuous in topographic expres-sion, the Honolulu Series rifts do not seem tocontain substantial intrusive bodies. A Bouguergravity anomaly map of Oahu (Strange,Machesky, and Woollard, 1965:351) clearlysuggests concentration of high density materialalong the primary northwest-trending Koolaurift. This anomaly has been interpreted as avolcanic plug (Adams and Furumoto, 1965:296). The Koko, Kaau, Tantalus, and Haikurifts (Winchell, 1947:19-20) of the HonoluluSeries show no .corresponding gravity effects.The feeders of the Honolulu Series eruptionsthus were probably simple dike fissures.

Dikes in the Palolo and Waialaenui valleysconsistently strike southwestward along the Kaaurift or southeastward parallel to the primaryrift of the Koolau volcano (Wentworth and

Dikes in the Southeastern Koolaus-BIGELOW

except the expansive 'buds' (Wentworth andJones, 1940:989) which appear on the easternslope of the Waialaenui Valley. In most casesit was impossible to determine whether ex-posures located in this study correspond to thosemapped by Stearns and by Wentworth andJones. The vicinity of Kaau Crater was searchedwith particular thoroughness. Its proximity tothe primary Koolau rift and the abundance ofintrusives already mapped on its eastern flankmade it a promising site for exposures. Mau-umae and Kaimuki were examined, an east-westfreeway cut providing fresh exposure throughthe latter. The rim, crater, and flanks of Dia-mond Head were also searched. Abundant dikeexposures were found near Kaau Crater; butnone appeared near the other eruptive sitesmentioned, despite reports by Hitchcock (1900:45), Wentworth (1926:44-45), and Stearns(1939 :41) of substantial Honolulu Series dikeson the southeast side of Diamond Head. Much

497

of this area has now been enclosed by retainingwalls and other construction.

FIELD STRUCTURES AND RELATIONS

The deep dissection of the southern slope ofthe Koolau shield volcano has formed a series ofsteep-sided, southwest-trending ridges with in-tervening valleys concentrating drainage abruptlyin the uplands (Fig. 2). The shield consists oflava flows, usually 4 to 12 feet in thickness,dipping gently southward in the area studiedand showing little evidence of weathering be-tween flows. The texture varies from coarselyvesicular and clinkery material on the tops andbottoms of flows to dense, vertically jointedblocks in the centers. Lava tubes are common,but are generally small and only a few feet inlength. None was found filled by intrusion, al-though such fillings have been observed else-where in the Koolau Range.

FIG. 1. Aerial view southward across Kaau Crater (center) toward Diamond Head (upper right). Thelow shield of Kaimuki is not distinguishable at this angle. (Photograph by A. T. Abbott.)

498

0)::-'----rl;~~----:; '4ll[

FIG. 2. Eruptive features and main streams of theWaialae-Palolo area, southeastern Oahu.

Dikes commonly intersect the flows at highangles, near the normal to the flow plane. Ofthe intrusives observed, 13 dip less than 65°,however, and 5 were below 35°. Dike trendsare typically near S 45-55°W and S 40-50 0 E,but vary widely. Limited exposures preventedprecise determination of the dip of many dikes.

The relatively unconsolidated clinker betweenflows was not observed to offer a plane of weak-ness to the intrusive bodies. Thus the interpreta-tion of apparently conformable intrusives as sillsrequired extreme caution in following the con-tact, and often remained in doubt. The dips offive intrusives tentatively identified as sillsranged from 5° to 14°. Wider samplings byWentworth and Jones (1940 :981) and Stearns(1939) remain the basis for judgments on theattitudes of leeward Koolau intrusives.

The dikes studied ranged in thickness fromglassy stringers a fraction of an inch across andlensing out completely, to a dense, massivelyjointed holocrystalline body 78 inches wide. The'bud' sampled is an intrusive body with con-

PACIFIC SCIENCE, Vol. XXIII, October 1969

centric platy jointing, several tens of feet acrossand of indeterminate height (26 feet exposedwith bottom apparently converging downward) .These buds have been described in detail else-where (Wentworth and Jones, 1940 :988-990) .

The Palolo Quarry intrusive is not includedin the scope of this study, although it is withinthe area here described. Detailed studies appearelsewhere of its structure (Wentworth andJones, 1940 :986-988) and its petrogenesis(Kuno, et aI., 1957: 181-187).

LABORATORY PROCEDURE

Five thin sections were prepared from sam-plings of each intrusive. These were petro-graphically described and nine modal analyses(1000 points each) were completed, includingtwo on different thin sections of the same dikeas a measure of sampling error. Composition ofphenocryst feldspar, ferromagnesian minerals,and interstitial glass was determined by oil im-mersion of powdered samples. Matrix feldsparcomposition was determined by maximum ex-tinction angles of microlites.

Textural descriptions are as defined byJohannsen (1931). and Williams, Turner, andGilbert (1954). Mineral data follow Heinrich( 1965) and Winchell and Winchell (1937).

Measurements of optic axial angle (2V)were by Tobi's method (Bloss, 1961 :205) forclinopyroxenes, Wright's method (Bloss, 1961:203) for orthopyroxene and Kanib's method(Bloss, 1961: 179-180) for olivine. Individualvalues of these readings commonly vary -t- 3-5percent, and so ranges based on a probabilitydistribution of several hundred readings havebeen reported.

PETROGRAPHY

The dikes and sills of this area are typicallyporphyritic with a fine-grained, hypocrystallinematrix, commonly intersertal in the coarsercenters to hyaloophitic at the chilled margins.

Phenocrysts

Phenocrysts of olivine, orthopyroxene, andplagioclase constitute 1 to 40 percent of therock in widely varying combinations.

Dikes in the Southeastern Koolaus-BIGELOW

OLIVINE: Olivine comprises 1 to 4 percent ofthe dikes and sills, although it sometimes wasnot observed in entire sections. Some larger in-trusives in this area contain 20 percent olivineby volume. Crystals range in size from matrixgrains 0.02 to 0.1 mm across to tabular pheno-crysts 3.0 mm long. Individuals 1.0 to 1.5 mmlong are common. Y-indices of 1.686 to 1.692indicate compositions in the range F080-85.

Iddingsite rims are common, but they aregenerally not well developed and frequently donot appear in an entire section. They manifestno relation to crystal size. In the Waialaenui bud,iddingsite formation was more advanced, some-times consuming entire large olivine phenocrysts.Hematite flakes sometimes appear in these rimsin parallel arrangement or each with its long di-mension normal to the olivine margins. Parts ofthese rims showing variable coloration have in-dices considerably below olivine, perhaps analteration product of the olivine (serpentine?).Overgrowths of pyroxene are rare and were notobserved to accompany iddingsite. Second gen-erations of late-growth olivine beyond iddingsiterims do not occur.

Some olivine phenocrysts are euhedral, espe-cially in dike margins, but more commonly,rounded or deeply embayed crystals contain in-vasions of late-formed matrix minerals and glass.Infrequent cores of olivine in orthopyroxenephenocrysts have been almost completely re-sorbed. Long, very thin survivals of largerolivine phenocrysts are common.

ORTHOPYROXENE: Orthopyroxene shares witholivine the early phenocryst role (Cross, 1915:19). Sizes range from 0.3 mm to 1.5 mm, ap-parently not extending into the groundmasssize. 2Vx range of 80° to 85 ° and Z-indices of1.685 to 1.690 indicate compositions of En79-84'

Orthopyroxene phenocrysts comprise 0.2 to6.5 percent of the volume of dikes and sills,usually exceeding the olivine fraction. Theseorthopyroxenes are typically clear and opticallyhomogeneous. Inclusions of magnetite are notcommon. Exsolution lamellae were not ob-served. In thin sections 0.04 to 0.05 mm thickdistinct pleochroism was noted, from brownishpink in the fast direction to bluish green in theslow direction. Orthopyroxene phenocrysts alsoare conspicuously rounded by resorption, but

499

are only occasionally embayed. They frequentlyform cumuloporphyritic clusters alone or withfeldspar phenocrysts of comparable size, oftenin subradial arrangement. Some stand distinctlyapart from such clots, however, as do all ofthose with olivine cores. Olivine phenocrysts donot enter the glomerocrysts.

Orthopyroxene phenocrysts with clinopy-roxene jackets rarely occur. This is the onlyassociation of clinopyroxene with the phenocrystphase, and these thin rims undoubtedly formedduring the period of groundmass crystallization.

FELDSPAR: Phenocrysts of plagioclase feldsparare evident in most of the intrusive bodies ex-amined. Individuals 1.3 to 1.5 mm long arenot unusual in the cumuloporphyry. Y-indicesof 1.560 to 1.567 indicate compositions An55-68with the most frequent range 1.562 to 1.563(An58-62) corresponding to an intermediatelabradorite. Larger phenocrysts show normalzoning in a thin outer rim. Carlsbad and peri-cline twinning are common, whereas albitetwins are infrequently and poorly developed.Phenocryst size grades continuously into thegroundmass laths. Total feldspar comprises 32to 50 percent by volume in these rocks.

Grolmdmass

The groundmass consists essentially of a net-work of plagioclase laths with interstitial clino-pyroxene crystals and glass containing includedores.

CLINOPYROXENES: Augite with 2Vz of 57° to62°, pigeonite with 2Vz of 13° to 27° and'subca1cic augite' (Kuno and Nagashima, 1952:1000) with 2Vz of 31 ° to 37° have beenidentified as grains 0.005 to 0.15 mm across,but individuals with measurable optical prop-erties are rare. Grains of clinopyroxene arecharacteristically pale brown, anhedral, and in-terstitial to the feldspars. Clinopyroxenes con-stitute 29 to 39 percent of the volume. Becausethey are so small and inclusion-ridden and donot manifest definable linear properties, com-position of the clinopyroxenes remains indoubt.

The most useful data were obtained by choos-ing crystals with highest apparent birefringenceand determining maximum indices in oil im-mersion. This procedure yielded maximum slow

500

direction refractive index values of 1.706 to1.723. Without basis to determine calcium con-tent, these data are not sufficient to establish themagnesium-to-iron ratio. Judgments cannot bemade, therefore, as to the proportions ofpigeonite, augite, and possible pyroxenes of in-termediate compositions in these rocks.

FELDSPAR: Any effort to distinguish pheno-cryst from matrix feldspar would have beenarbitrary with the data available. The con-tinuous gradation in feldspar crystal size sug-gests continuous separation of feldspar as theserocks formed. Extinction angles of groundmassmicrolites (method described in Heinrich, 1965:362-364) indicate compositions of An56-65'

OTHER CONSTITUENTS: Angular intersticesin the groundmass are filled by amber to darkbrown tachylite with index from 1.576 to 1.588,indicating compositions of 49 to 51 percentsilica (George, 1924: 365). In the chilledmargins this glass sometimes exceeds 80 percentof the bulk volume. Glass comprises 6 to 12percent of the total volume of the dikes.

Minute dust and skeletal crystals of magne-tite are abundant in interstitial glass andmatrix pyroxene, often rendering these hostsvirtually opaque. Subordinate ilmenite mayusually be identified. The larger olivine andorthopyroxene phenocrysts contain few or noore inclusions. Smaller (presumably later-formed) individuals in the same section includean increasing proportion of these ores. Thereis a conspicuous absence of ore inclusions infeldspars. Even the microlites, whose growthmay be marginally impaired by preexistingmagnetite grains, stand out boldly in the

PACIFIC SCIENCE, Vol. XXIII, October 1969

groundmass between pyroxene and tachyliteareas overrun by abundant octahedrons of mag-netite and flakes of ilmenite.

The largest magnetite grains occur in theWaialaenui bud, where they reach 0.4 mm. Intypical dike rocks ore grains seldom exceed 0.1mm across. Ores make up 8 to 17 percent of thevolume of these rocks, the most common countsbeing in the 11 to 14 percent range. Ore con-tent may increase with elevation in these in-trusives (Table 1). Further modal counts wouldbe required to confirm this. Acicular apatite in-clusions usually are recognizable in larger feld-spar phenocrysts, especially in the Waialaenuibud. Apatite never exceeds 1 percent modalvolume.

Hematite occurs as minute red and blackflakes along occasional joint fractures whereminor alteration has taken place. It also appearsas oriented inclusions in iddingsite rims onolivine phenocrysts. These inclusions are elon-gate flakes or shreds, black in reflected light.Total hematite exceeds 1 percent of the rockvolume only in weathered material.

Traces of alteration products, mostly of lowbirefringence or isotropic and with indices 1.48to 1.55, occur in vesicles and along infrequentfractures and joint planes. These occurrences aretoo small for critical determinations. Fibrousdevelopment often suggests zeolites or chalce-dony. Bright red and green patches and stripsare probably palagonatized glass.

Vesicles comprise 1 to 10 percent of thetotal bulk, but may be entirely absent fromdenser portions. Most are consistently circularin outline.

TABLE 1MAGNETITE-ILMENITE CONTENT OF KAAU RIFT INTRUSIVES

SAMPLE

NO.

2624103

217

1917a17b

ELEVATION

(feet)

320 (±1O)560830920

10701200148015001500

ORE

(modal %)

7.8 (±0.2)7.7

11.39.1

11.612.117.114.514.6

DESCRIPTION

Dense inner part of Waialaenui budPegmatite from Palolo Quarry intrusive48"-57" dike on Waialae stream18"-26" dike on Waiomao stream26"-35" sill (?) on Waialae stream36"-78" dike on Waiomao stream16"-20" dike on Kaau outlet stream14"-16" dike on Kaau outlet streamSame dike as 17a

Dikes in the Southeastern Koolaus-BIGELOW

COMMENTARY

Source Magma

All intrusives sampled in this study aretholeiitic. The dikes and sills are tholeiiticbasalts in the sense defined by Macdonald andKatsura (1964 :89). Olivine is consistently lessthan 5 percent of the bulk. This is in contrastto the more olivine-rich rocks of the HonoluluSeries. No occurrences of nepheline, melilite, oraugite phenocrysts, diagnostic of HonoluluSeries mineralogy, were found.

Marginal glass in one dike containing onlyminor olivine as the crystalline phase has index1.576, indicating about 51 percent silica(George, 1924:365), which includes slightsilica enrichment resulting from separation ofolivine. Another dike has traces of olivine andapproximately 5 percent hypersthene in a glassbasis. The index of the glass is 1.588, indicatingabout 49 percent silica. Glassy margins of theseintrusives had refractive indices consistently be-low 1.59, or more than 48 percent silica accord-ing to the George curve.

Numerous analyses of Koolau basalts haveshown silica contents of 48 to 52 percent(Wentworth and Winchell, 1947: 71). Hono-lulu Series rocks range from 36 percent to 45percent silica (Winchell, 1947: 30) .

The petrography of the intrusives studiedhere closely resembles descriptions of Koolauextrusives given by other authors (Cross, 1915:18-20; Stearns and Vaksvik, 1935 :93; Went-worth and Winchell, 1947:67-70). The greatervesiculation of the latter may be attributed tothe effects of extrusion.

Aside from structural relations of material inplace, dike rocks are distinguished from flowfragments in the field largely by the platy orsplintery jointing and lack of conspicuous vesic-ulation in dike material. As previously noted,iddingsite rims are never succeeded by fresholivine growth in the intrusives observed.Second generation olivine growth is commonamong Koolau extrusives (Wentworth andWinchell, 1947: 65) and Honolulu Series rocks(Winchell, 1947:24).

No new chemical analyses are presented here,but the petrographic evidence points to chemicalequivalence between these intrusives and theflows enclosing them. There is negligible al-

501

teration of the country rock adjacent to intru-sives. Only a slight baked band appears, whichis often difficult to distinguish. No metasomaticeffects were observed.

Paragenesjs

Olivine crystallized first in these rocks,usually reacted with the melt, and was slightlyaltered to iddingsite, probably during concentra-tion or upward migration of volatiles. Orthopy-roxene and feldspar followed olivine after aconsiderable interval. In several dikes olivinephenocrysts, alone in glass at dike margins, werewell developed. In some margins these pheno-crysts were sharply euhedral, indicating thatresorption had not yet begun. In other dikesresorption was already well advanced in thesesolitary phenocrysts.

The complicated interactions of feldspar andorthopyroxene cannot be generalized. A few in-trusives show unequivocal precedence of one tothe other, but contemporaneous growth is therule. Minor ore inclusions occasionally appearin orthopyroxenes, but they were not observedin the feldspars. This might suggest earliercrystallization of the feldspar; but the feldspar-magnetite antipathy persists even to the finematrix material. It apparently is not a reliablecriterion of crystallization sequence.

Orthorhombic pyroxene could not be identi-fied in crystals less than 0.2 mm across, and thelimit usually is nearer 0.3 mm. Hypersthenecrystallization apparently ceased before thegroundmass period. Resorption suggests an in-terval during which orthopyroxenes were not inequilibrium with the melt. Such resorbed crystalsoften present their rounded surfaces to in-timate contact with feldspar in glomerocrysts,interfering with growth of the latter sufficientlyto indicate that resorption began, at least, sub-stantially before the groundmass stage.

Feldspar phenocrysts grade imperceptibly intofine-grained microlites, showing continuousseparation to a sharp boundary with interstitialglass. Smaller feldspar laths commonly, thoughnot invariably, form subparallel flow patternsaround glomerocrysts and other phenocrysts(Fig. 3). This may roughly distinguish in-tratelluric from matrix feldspar crystallization.

A small quantity of ore crystallization mayactually precede olivine formation in most of

502

FIG. 3. Flow pattern of matrix laths near aglomerocryst composed predominantly of hypersthen~.Vesicle in upper center is 0.5 mm in diameter.

these intrusives, thus being the first solid phaseformed. Octahedrons of magnetite and platesof ilmenite occur scantily in the mafic pheno-crysts of dikes and sills. Most of the ore-mineral crystallization belongs to the ground-mass phase. In hyaloophitic chilled borders afew magnetite and ilmenite inclusions appear inolivine, but ores in the marginal glass werebeyond the resolving power of the microscope.With more gradual cooling, skeletal magnetitecrystals grew abundantly in and across matrixpyroxene and glass. The fine dust in tachyliteindicates separation of the ore minerals to thevery end of the liquid phase.

A gap is usually evident in the sizes ofpyroxenes between smallest hypersthene andlargest clinopyroxene. This probably representsa considerable interval of time, but it also reflectsthe change in conditions between magma cham-ber or feeder and the ultimate site of emplace-ment. Clinopyroxene shells on euhedral ortho-pyroxene phenocrysts rarely document a moreintimate succession.

PACIFIC SCIENCE, Vol. XXIII, October 1969

Optical homogeneity, and particularly the ab-sence of exsolution lamellae, suggest that ortho-pyroxene in these intrusions did not develop byinversion of earlier-formed clinopyroxene. Oc-casional olivine cores in orthopyroxene pheno-crysts indicate formation of the latter as reac-tion rims.

Volume abundances of clinopyroxene andorthopyroxene seem to be in crude inverse rela-tion (Fig. 4). Perhaps early formation oforthopyroxene phenocrysts reflects lower tem-perature of initial crystallization, shortening theinterval between beginning clinopyroxeneseparation and the ultimate freezing to glass.The substantial proportion of pigeonite ob-served in matrix material results from quenchingof the melt, which prevents its inversion toorthopyroxene.

The clinopyroxenes are characteristically an-hedral and interstitial to matrix plagioclase.Their intimate association with tachylite andthe arrangement of magnetite and ilmenite in-clusions in orderly arrays across glass-pyroxene

I I I I I

40

39 0op

38o

37

36o

o

o

owb

o

30

1234567Orthopyroxene volume %

FIG. 4. Modal volumes of pyroxenes in Kaau riftintrusives, p, Palolo Quarry pegmatite; wb, dense in-terior of Waialaenui bud.

Dikes in the Southeastern Koolaus-BIGELO\X'

boundaries establishes the separation of clino-pyroxenes to the very end of the groundmasscrystallization period. Apparently pyroxene,labradorite, and ores continued to form togetheruntil the remaining melt congealed to glass.Minute crystallites of pyroxene and ore dustcan be seen in this glass under high magnifica-tion.

Feldspar-Magnetite Relationships

The antipathy between feldspar and mag-netite in these rocks is worthy of further com-ment. Formation of magnetite in the intratel-luric stage is suggested by well-formed octa-hedrons occurring as inclusions in olivine and,rarely, in orthopyroxene phenocrysts. Most ofthe magnetite formed in the groundmass periodalongside plagioclase laths, but was not includedin them. In view of the greater crystallizingforce of magnetite as compared with plagio-clase, it seems that the latter could not effectivelyrestrain growth of the magnetite structure.Pyroxenes, which do host abundant matrixmagnetite in these rocks, have substantiallygreater crystallizing force than plagioclase.Evidently factors other than lattice strength aredecisive.

Possibly, formation of magnetite was not con-tinuous in these intrusives; but plagioclasecrystallization encompasses such a wide intervalthat contemporaneous separation must haveoccurred. Contemporaneity is not necessary tomake magnetite available for inclusion in feld-spar, however. Earlier-formed magnetite couldeasily be included in later-formed plagioclase.This is undoubtedly the manner of inclusion ofmagnetite in olivine phenocrysts in these rocks.If olivine formed by epitaxis on early magnetite,the latter might be thus denied to early plagio-clase phenocrysts; but when more than one suchinclusion appears in optically continuous olivinethis mechanism seems doubtful. The coalescenceof several olivine crystals into optically homo-geneous aggregation is unlikely.

If, as Vogt (1921: 321) suggests (in an al-together different chemical environment), themagnetites cluster themselves by synneusis, thisglomerocryst might serve as a crystallizationnucleus for early olivine. The magnetite in-clusions in olivine phenocrysts of these rocksare not typically contact clustered, but they are

503

distinctly localized. Such a mechanism can ac-count only for the absence of magnetite in thelarger phenocrysts of plagioclase. The inhospi-tality of matrix feldspars remains unexplained.Apparently some unique magnetite-plagioclaseantipathy is acting here. In other rocks feldsparcommonly includes ores, though they tend tofavor pyroxene hosts.

It is suggested that in some of these intrusivesa few grains of magnetite formed before olivinebegan to incorporate iron. Ore formation thenceased because the iron content of the melt wasslightly diminished and/or because the estab-lished centers of magnetite nucleation wereenclosed in the growing olivine phenocrysts.The long, very thin olivine phenocrysts hint thatconsiderably more olivine may have formed atan early stage than has survived emplacementand cooling.

As cooling proceeded, orthopyroxene pickedup the iron returned to the melt by olivine re-sorption. After an indeterminate interval, dur-ing which formation of orthopyroxene andplagioclase considerably diminished the liquidphase, the magma was intruded into the dikefissure and cooled rapidly. As other componentsof the melt were depleted, the iron concentra-tion increased. Much iron was accommodated byclinopyroxenes in the groundmass. Linear arraysof magnetite in these clinopyroxenes might sug-gest exsolution. But unless striking changes inoxidation potential are postulated in magma atleast several hundred feet beneath the shieldsurface, substantial quantities of ferric ironwould have to be admitted into pyroxene struc-tures on this hypothesis. Drickamer, Lewis, andFung (1969:885) have demonstrated a reversi-ble reduction of ferric iron to the ferrous statewith increasing pressure in a variety of sub-stances. They indicate that this transition istemperature-sensitive, but their data do notreach the temperature of a basaltic melt. Untilthis process has been verified with rock-formingsilicates at melt temperatures its applicabilityhere remains speculative. In any case, it isdoubtful that exsolution from pyroxenes couldaccount for alignments of magnetite octahedronsthat are commonly continuous across pyroxeneboundaries into glass. Lattice planes in pyroxenehosts do not seem to control magnetite align-ment.

504

Cmmtloporphyry

The phenomenon of cumuloporphyry (Jo-hannsen, 1931 :203), though widespread in agreat variety of rocks, has been neglected in thepetrographic literature. It is sometimes used as adiagnostic property of igneous rocks on theassumption that the aggregates so designatedcould form only with the freedom of movementavailable in a melt (Vance and Gilreath, 1967:529).

The relations between individual crystals inthese aggregates cannot be explained as epitaxis.Crystal boundaries may be intimate, but the in-dividuals, whether like or unlike species, showno tendency to grow around or envelop theirneighbors. The association is most commonlytangency or slight intergrowth (Fig. 5). Flowpatterns in matrix feldspar microlites show thatthe latter have piled against and streamedaround each glomerocryst. Relative movementof these phenocryst clots with respect to thesurrounding medium is unmistakable. They havebehaved as coherent units, withstanding what-ever forces attended this passage of the fluid.

Vance and Gilreath (1967: 553) attributethe formation of glomerocrysts to a process(synneusis) dependent upon ". . . episodes ofturbulence and their timing with respect to thecrystallization sequence...." Vogt (1921: 321)observed synneusis in cooling ore melts andmetallurgical slag, noting that magnetite andzincblende clusters formed within 30 minutesin an essentially static fluid. Spry (1953:255)commented that velocities of magma movementin dikes are probably too low to generate sub-stantial turbulence unless viscosity is very low

FIG. 5. Intergrowth of plagioclase and ortho-pyroxene in glomerocryst.

PACIFIC SCIENCE, Vol. XXIII, October 1969

due to superheating or unusually high volatilecontent. He held observable structures in bos-tonite dikes to indicate essentially laminar flow.

Flow properties in the intrusives of thisstudy are less definitive than those examinedby Spry, but the consistent parallelism ofmicrolites in some samples suggests laminarflow. This does not preclude earlier episodes ofturbulence at greater depth, however, especiallylocal vortices where the fluid passed an obstruc-tion or irregularity.

Vance and Gilreath also concluded fromcounting studies that phenocrysts have affinityfor their own species and antipathy for unlikespecies in synneusis. However, Rogers and Bogy(1958 :471) found by counting studies ingranites that like crystals were less likely to formin contact with each other. They contend thatthis is due to (1) growth of one crystal pre-venting nucleation of others in the neighbor-hood, and/or (2) early crystallization of likegrains causing them to be isolated from eachother by later-forming minerals.

The feldspar and orthopyroxene phenocrystsin this study show a marked tendency towardheterogeneous aggregates, although small single-mineral clusters do rarely appear.

Flow structures in igneous rocks may eventu-ally provide a basis from which to determinerate of magma movement during emplacement.The flow configurations around glomerocrystsobserved in this study may have resulted in partfrom gravity settling of these clusters aftermagma movement ceased. The mechanism ofsynneusis remains an open question.

SUMMARY

Exposed dikes and sills trending southwestand roughly perpendicular to the primary riftzone in the Waialae-Palolo area of Oahu arehypersthene-bearing tholeiitic basalts, intrusiveequivalents of the basalt lava flows of theKoolau Series. No Honolulu Series intrusiveswere found along a line joining Kaau Crater,Mauumae, Kaimuki, and Diamond Head, analleged secondary rift of the Koolau volcano.Bouguer anomaly maps show no indications ofsubstantial intrusive bodies along the southeastHonolulu Series trends, so simple dike feedersare inferred.

Dikes in the Southeastern Koolaus-BIGELow

The intrusives are typically porphyritic witha fine-grained, hypocrystalline matrix, commonlyintersertal in the coarser centers to hyaloophiticat chilled margins. Phenocrysts include olivine,hypersthene, and labradorite. Groundmassminerals are clinopyroxene and labradorite, theformer containing abundant ore inclusions. In-terstices contain ore-rich tachylite.

The crystallization sequence began witholivine, occasionally preceded by minor mag-netite. Orthopyroxene and plagioclase appearednext, in close association, with precedence un-certain. Orthopyroxene separation ceased beforegroundmass formation, plagioclase continuingto crystallize at increasingly numerous nucleiwith clinopyroxene and ores. The remainingfluid finally congealed to dark brown tachylitewith ore dust and incipient clinopyroxenecrystals.

Modal counts indicate an inverse relation inabundance of clinopyroxene to orthopyroxene,although further data are needed to confirmthis. Plagioclase and magnetite are antipatheticthroughout these rocks, despite the intimacy ofmagnetite with all other major species present.

Orthopyroxene and plagioclase are commonlyassociated in glomerocrysts, which may eventu-ally be useful as a basis to infer flow propertiesand crystallization sequences in magmas duringemplacement.

ACKNOWLEDGMENTS

I wish to thank Dr. G. A. Macdonald forcritically reading the manuscript. Field workwas supported by a National Science FoundationResearch Participation grant during the summerof 1967.

LITERATURE CITED

ADAMS, W. M., and A. S. FURUMOTO. 1965. Aseismic refraction study of the Koolau vol-canic plug. Pacific Science, vol. 19, pp. 296-305.

BLOSS, F. B. 1961. An introduction to themethods of optical crystallography. NewYork, Holt, Rinehart and Winston. 294 pp.

CROSS, W. 1915. Lavas of Hawaii and theirrelations. U. S. Geological Survey, Profes-

505

sional Paper 88, U. S. Government PrintingOffice, 97 pp.

DRICKAMER, H. G., G. K. LEWIS, Jr., and S. C.FUNG. 1969. The oxidation state of iron athigh pressure. Science, vol. 163, pp. 885-890.

GEORGE, W. O. 1924. Composition of glassfrom refractive index and density. Journal ofGeology, vol. 32, pp. 353-372.

HEINRICH, E. W. 1965. Microscopic identifica-tion of minerals. New York, McGraw-HillBook Co. 414 pp.

HITCHCOCK, C. H. 1900. Geology of Oahu.Bulletin of the Geological Society of Ameri-ca, vol. 11, pp. 15-60.

JOHANNSEN, A. 1931. A descriptive petrog-raphy of igneous rocks. Vol. I. Chicago,University of Chicago Press. 267 pp.

KUNO, H., and K. NAGASHIMA. 1952. Chemi-cal compositions of hypersthene and pigeon-ite in equilibrium in magma. American Min-eralogist, vol. 37, pp. 1000-1006.

KUNO, H., K. YAMASAKI, C. IIDA, and K.NAGASHIMA. 1957. Differentiation of Hawai-ian magmas. Japanese Journal of Geologyand Geography, vol. 28, pp. 179-218.

MACDONALD, G. A., and T. KATSURA. 1964.Chemical composition of Hawaiian lavas.Journal of Petrology, vol. 5, pp. 82-133.

ROGERS, J. J. W., and D. B. BOGY. 1958. Astudy of grain contacts in igneous rocks.Science, vol. 127, pp. 470-471.

SPRY, A. 1953. Flow structure and laminar flowin bostonitedykes at Armidale, New SouthWales. Geological Magazine, vol. 90, pp.248-256.

STEARNS, H. T. 1939. Geologic map and guideof Oahu, Hawaii. Territory of Hawaii, Divi-sion of Hydrography Bulletin 2, PI. 2, 75 pp.

STEARNS, H. T., and K. N. VAKSVIK. 1935.Geology and ground-water resources of theisland of Oahu, Hawaii. Territory of Hawaii,Division of Hydrography Bulletin 1, 479 pp.

STRANGE, W. E., 1. F. MACHESKY and G. P.WOOLLARD. 1965. A gravity survey of theisland of Oahu, Hawaii. Pacific Science, vol.19, pp. 350-353.

VANCE, J. A., and J. P. GILREATH. 1967. Theeffect of synneusis on phenocryst distributionpatterns in some porphyritic igneous rocks.American Mineralogist, vol. 52, pp. 529-536.

506

VOGT, J. H. 1. 1921. The physical chemistry ofthe crystallization and magmatic differentia-tion of igneous rocks. Journal of Geology,vol. 29, pp. 318-350.

WENTWORTH, C. K. 1926. Pyroclastic geologyof Oahu. B. P. Bishop Museum Bulletin,vol. 30, pp. 1-121.

WENTWORTH, C. K., and A. E. JONES. 1940.Intrusive rocks of the leeward slope of theKoolau Range, Oahu. Journal of Geology,vol. 48, pp. 975-1006.

WENTWORTH, C. K., and H. WINCHELL.1947. Koolau basalt series, Oahu, Hawaii.

PACIFIC SCIENCE, Vol. XXIII, October 1969

Bulletin of the Geological Society of Amer-ica, vol. 58, pp. 49-78.

WILLIAMS, H., F. J. TURNER, and C. M. GIL·BERT. 1954. Petrography: An introduction tothe study of rocks in thin sections. San Fran-cisco, W. H. Freeman. 406 pp.

WINCHELL, H. 1947. The Honolulu Series,Oahu, Hawaii. Bulletin of the GeologicalSociety of America, vol. 58, pp. 1-48.

WINCHELL, A. N., and H. WINCHELL. 1937.Elements of optical mineralogy. Part 2. NewYork, John Wiley & Sons. 551 pp.