A Thesis Submitted to the University of London for the Ph.D. Degree

BASALTIC VOLCANISM, VOLCANOES AND VOLCANIC ROCKS

MARINE AND LACUSTRINE ENVIRONMENTS

JOHN GILBERT JONES

A thesis submitted to the University of London for the Ph.D. degree

Department of Geology Imperial College March, 1968 2

ABSTRACT Intraglacial basaltic volcanoes near Laugarvatn, southwest Iceland, range from simple pillow lava piles to tuyas - volcanoes of tablemountain form in which a superstructure combining a carapace of sheet laTa on a pedestal of breccia envelops a tuff-mantled pillow lava pile. The pillow lava piles are inferred to be the product of effusion of lava into glacial meltwater, and the tuff tappings which most exhibit record a later phase of explosive emergence from the meltwater pond. The sheet lava and flow-foot breccia of the tuyas are believed to record an ultimate phase of lava effusion in air, following earlier phases of aqueous effusion and explosive emergence, in which flows advanced outwards into encircling meltwater on deltas of breccia formed at their fronts. The little that is known of the morphology and constitution of marine basaltic volcanoes suggests that they are basically similar to the Laugarvatn volcanoes and that concepts derived from the study of the latter have general application. Lxamination of the structural characteristics of Icelandic and Welsh pillow lavas suggests a mode of formation and propagation akin to the digital advance of pahoehoe: and a vesicularity study indicates the potential importance of pillow lavas as depth indicators. Review of the literature suggests that explosive activity, induced by extraneous water, is characteristic of eruption of basalt lava from a wet vent in shallow water and in air. Observations of the immersion of basalt lava are reviewed, and the significance of structural records of immersion is indicated. A re-examination of the type 'peperites' suggests that they are the product, not cf brecciation of lava intruded into wet sediment as is currently accepted, but of explosive projection of basaltic ejecta into steadily accumulating lime mud.

CONTENTS Page

Chapter 1. Introduction 5 Chapter 2. Icelandic intraglacial volcanoes Section 1. The Laugarvatn volcanoes - their 10 shape, gross structure, and pattern of development. Section 2. The Laugarvatn volcanoes - 35 detailed constitution and development. Addendum. Composition of the Laugarvatn 66 lavas. Chapter 3. Pillow lava. Section 1. Pillow lava and pahoehoe. 69 Section 2. Pillow lavas as depth 80 indicators. Addendum. Composition of pillow lava. 94 Chapter 4. Explosive basaltic volcanism in aqueous environments. Section 1. Basaltic eruptions from wet vents 97 in shallow water and in air. Section 2. A lacustrine volcano of central 109 France and the nature of peperites. Chapter 5. Immersion of basalt lava. Section 1. Observed immersion of basalt lava. 126

Section 2. Immersion of basalt lava - its 133. structural expression and stratigraphic significance. Chapter 6. Marine basaltic volcanoes. Section 1. Structural model of a marine 140 basaltic volcano. Section 2. Marine basaltic volcanoes 'high 141 and dry'. Section 3. Basaltic volcanoes of the present 147 ocean basins.

Chapter 7. Summary and conclusions. 155 4

Acknowledgements. 161

References cited. 162 Appendix. Clastic rocks of Espiritu Santo Island, New Hebrides.

Also bound in thesis; "Intraglacial volcanoes of southwest Iceland and their significance in the interpretation of the form of the marine basaltic volcanoes"; 'Nature', 212, 5062, p.586-588. 5

CHAPTER 1

INTRODUCTION 6

Basalt is the most voluminous of the Earth's volcanic products and is particularly widespread in the ocean basins. Yet the subject of basaltic volcanism in aqueous environments — its processes and products — has long remained an area of almost total ignorance. The reasons for this are clear enough, for the processes of basaltic volcanism in aqueous environments are largely inaccessible to observation as, until recently, have been most of the younger products. The incipient influx of data on the morphology, constituents and geophysical characteristics of totally and largely submerged oceanic volcanoes is creating an urgent need for a sound basis for interpretation. And an understanding of the processes and products of basaltic volcanism in water, based on studies of present or at least Cenozoic aqueous environments, is likewise essential for any substantial progress in the study of older basaltic rocks of aqueous origin in the stratigraphic column. This thesis attempts to present a coherent and comprehensive outline of the principal processes and products of basaltic volcanism in aqueous environments. It incorporates in comparable amount data from my own studies, principally of Icelandic intraglacial volcanoes, and data from the literature. With the objective of the thesis in mind, my fieldwork and literature survey have been directed principally to the elucidation of what I have judged to be the more important aspects of the topic. I have thus made no attempt to provide a complete geological account of the areas of investigation of the kind that might be expected in a local or regional study. Nor has it been my intention to provide a compendium of all published data, interpretation and 7

opinion related in any way to the topic. Concerning the products of basaltic volcanism in aqueous environments, this thesis confines itself almost entirely to description and discussion of macroscopic forms, structures and textures. This does not imply any denial of the relevance and importance of the microscopic, chemical and other aspects of the subject. However, I judged a comprehensive, integrated framework of basic macroscopic observation to be of more value to the subject in its present state than a more intensive examination of any limited aspect. A large number of thin sections has been examined in the course of the work and care has been taken to see that interpretations based on macroscopic observations do not conflict with microscopic characteristics. The fieldwork on which this thesis is based includes six months in Iceland, principally in the Laugarvatn area; three weeks in Auvergne, central France; and three weeks in the Strumble Head area of Pembrokeshire, Wales. In addition pillow lava localities on the Italian mainland and in Elba and Sicily were briefly visited. The field— work in France proved much less productive than I had anticipated, exposure of the peperites being very poor and their nature quite contrary to what current literature had let me to expect. With hindsight the time would have been devoted to a field of study more productive within the context of the thesis. The thesis has been written as a series of short papers, several of which are published or in press. Chapter 2, Section 1 and Chapter 3, Section 1 contain the substance of papers to be published in the Quarterly Journal of the Geological Society of London and in the Journal of Geology, respectively. And a considerable part 8

of Chapter 6, Section 1 has been published in 'Nature' under the title "Intraglacial volcanoes of southwest Iceland and their significance in the interpretation of the form of the marine basaltic volcanoes" (bound in the thesis). I have included in the thesis only such material as has been accepted for publication or that I consider to be in a condition suitable for publication. Appended is a paper entitled 'Clastic rocks of Espiritu Santo Island, New Hebrides", published in the Bulletin of the Geological Society of America. It is an outcome of fieldwork in the New Hebrides undertaken prior to my registration as a candidate for the Ph.D. degree in the University of London. 9

CHAPTER 2

ICELANDIC INTRAGLACIAL VOLCANOES 10

SECTION 1

THE LAUGARVATN VOLCANOES - THEIR SHAPE, GROSS STRUCTURE, AND PATTERN OF DEVELOPMENT

1. Introduction The intention of this section is to describe and interpret the shape, gross structure and interrelationship of a group of basaltic volcanoes in south-west Iceland. These volcanoes occur in the Ticinity of the village of Laugarvatn, about 30km E.N.E. of Reykjavik. They stand in a cluster immediately north of the road which runs from Thingvellir, through Laugarvatn to Geysir, and about 20km south-west of the Langjokull icecap. They are part of a Quaternary volcanic belt with a NE-S7; trend which runs from the Langjokull in the north-east to Cape Reykjanes in the south-west (Map 1, inset). All are part of the "Moberg Formation" of Icelandic geologists (Kjartansson, 1959). According to Kjartansson, the rocks of the Moberg Formation show uniform magnetic polarity of the present cycle and are generally considered to be of late Pleistocene age. The constituent rock types are briefly described and discussed at the beginning of the section to facilitate description of the volcanoes themselves. More detailed lithologic descriptions are given in Section 2.

2. Rock types and their significance All the rocks of which the volcanoes of the Laugarvatn region are built consist of a basalt which, when fully crystalline, contains essential olivine, clinopyroxene, plagioclase and opaques. 11

2.1. Lavas Sheet lava and pillow lava are important components of the Laugarvatn volcanoes. The term sheet lava denotes extruded lava which in section is seen to consist of tabular units with lateral extent many times their thickness. It includes both fragmental (aa) and non- fragmental (pahoehoe) varieties. The term pillow lava is restricted to lavas consisting of abundant units which in section display numerous spherical and elliptical outlines and which contain little or no fragmental material. The contrasting forms of sheet and pillow lava are believed to be a consequence of extrusion in air and water respectively Pillow lavas are generally accepted as a criterion of the aqueous environment and such an interpretation of pillow structure in the Laugarvatn area is in full accord with other facts and inferences.

2.2. Fragmental rocks In the Laugarvatn volcanoes fragmental basaltic rocks - palagonitic glassy breccias and vitric tuffs - are as abundant as the basalt lavas. The breccias are predominantly unsorted, glass-rich fragmental rocks with clasts ranging from coarse fragments of glass-encrusted lava, often of highly irregular outline, to glass fragments of sand grade. They occur most abundantly beneath, but in structural continuity with horizontal or gently inclined sheet lavas. The zone of transition between breccia and overlying sheet lava (subsequently referred to as the 'passage zone') is gradational and horizontal when undisturbed. Attenuated sheets and fingers of lava diverge from the base of the horizontal sheet lava with rapidly increasing inclination, giving the passage zone a stratification of variable inclination, but locally sheet lava passage zone

flow-toot breccia

FIGURE I . STRUCTURE OF IMMERSED FLOW 13

consistent direction, and pass down by fragmentation into commonly structureless flow-foot breccia (Fig.l). These flow-foot breccias thus have the same relationship to the sheet lava as the foreset have to the topset beds in a deltaic deposit. Structural relationships of this kind were first described by Fuller (1931) in the Columbia River basalts and were ascribed by him to the flow of basalt lava from air into water. "A fluid lava on entering a local body of water would tend to granulate like molten slag and would thus form a fine breccia which would accumulate to a depth approximately equal to that of the water. The fine breccia would settle until its surface attained an angle of repose which, owing to the roughness of the fragments, would be relatively steep. If the molten cascade continued to pour into the water, the accumulation of granulated glass would gradually advance like the foreset bedding of a delta. The inclined bedding would be preserved by the thin sheets and the ropy or ellipsoidal masses which failed to granulate. Except for the possible effects of rising steam, the flow would gradually advance on top of these foreset beds as if on dry land'. Other breccias occur within or as a veneer on pillow lava piles and clearly result from the fragmentation of pillow lava. These will be referred to as pillow breccia. The tuffs consist very largely of sand-size fragments of clear basalt glass palagonitised in varying degree.

3. The volcanoes; their shape and structure 3.1. Tindars (see Map 1) Those volcanoes of the Laugarvatn area which in this section are termed 'tindars' (tindar is Icelandic for peaks or pinnacles) form steep-sided linear ridges and linear 3

• o 0 E

C e 0 0 C. a Co

0 0

E CZ

.1, 0

. 0

E DC 15

groups of steep-sided mounds with profiles which are in some instances jagged, in others smooth. The tindars are aligned NE-SW, the volcanotectonic trend in the Quaternary volcanic belt of south-west Iceland. Kalfstindar, viewed from the south-east (P1.1), illustrates well the range in form, structure and lithology shown by the tindars. The southern end of the mass has a smooth profile and, with the exception of a patchy veneer of stratified pillow breccia and tuff low down on the flanks, consists entirely of pillow lava. The central part of the tindar has a more rugged profile. It rises 150m above the southern portion, the increased height resulting from a capping of stratified fragmental rocks on top of the pillow lava pile. These fragmental rocks are predominantly tuff. Sheets and lenses of pillow lava and pillow breccia, common near the base of the capping, become less abundant upwards and are absent from the upper part. The northern portion of Kalfstindar is marked by three distinct peaks with summits 300 m or more higher than the southern portion. On the southernmost of these sheet lava is interbedded with tuff within the uppermost 90m (P1.10A). Dykes, inconspicuous in their dimensions and inconsistent in their trends, are exposed along the length of Kalfstindar. One of the largest, a tuff-filled dyke about 3m wide, occurs close to the crest of the tindar just southwest of the 666m peak and parallel to its axis. Pillow lava, dipping strongly outwards as is common on the flanks of Kalfstindar, merges imperceptibly with the lava of the dyke wall. Others can be traced upwards into the sheet lavas of the 826m summit. PLATE 2

cliff section in sheet lava SKRIDA SKRIDUTINDAR KLUKKUTINDAR HLODUFELL 17

The other tindars in the area have been only cursorily examined but their similarity to Kalfstindar is obvious. They visibly consist predominantly of tuff. On Skridutindar, as on Kalfstindar, sheet lava is interstratified with tuff within the uppermost 70m of the main peak. Several dykes trending with the axis of the tindar are visible high on its western flanks.

3.2. Tuyas (see Map 1) The more equidimensional volcanoes of the Laugarvatn area - Hrafnabjorg, Laugarvatnsfjall, Middalsfjall, Efstadalsfjall, Skrida, Hlodufell, Raudafell and Hognhofdi - have the form of benched or terraced platforms surmounting steeply sloping bases and are herein termed 'tuyas'* (Mathews, 1947). The basis of tuya structure, expressed in the tablemountain form, is a capping of gently dipping or sub-horizontal sheet lava on a base of flow- foot breccia, but this is rendered more complex in detail by the close superposition of several such lava/breccia units, expressed in benched profiles. The tuya Skrida (P1.2) displays with clarity those features which are characteristic of the tuyas of the Laugarvatn area. The scree of its steep flanks mantles and is derived from crumbling flow-foot breccia which passes upwards, through the conspicuous foresets of the passage zone, into sheet lava at or a little below a conspicuous break in slope at about 800m (P1.3A). This sheet lava/flow-foot breccia unit is surmounted by another

*Volcanoes of this form are frequently termed 'tablemountains' in descriptions of Icelandic geology. • N* • • - • .1•.i•'

40,

tr. 71,n1111.111a4k, -4&110000!1 - •

B 19

less extensive lava/breccia unit. Unlike the other tuyas of the area Skrida displays a well preserved summit crater and a string of craterlets along an axial crest of NNE-SSW trend. The tuyas of the Laugarvatn area differ most notably in the extent to which the earlier products of the volcano are exposed beneath flow-foot breccia. The most extensive of the tuyas, Laugarvatnsfjall and Middalsfjall, show no sign of such early products, the flow-foot breccia of their periphery resting directly on a glaciated surface. Skrida, which is much smaller in extent than either of these, exhibits inconspicuous buttresses of tuff projecting from its north-east flanks. Efstadalsfjall, Raudafell and Hognhofdi, which are smaller still in terms of the extent of their capping lavas, exhibit conspicuous foundations of pillow lava, surmounted by a considerable volume of tuff in the case of the least extensive, Raudafell. A substantial pile of pillow lava projects from beneath flow-foot breccia at the southern end of Hlodufell.

3.3. A structural series The Laugarvatn volcanoes can be ranged in a series of increasing structural complexity. Some (e.g. much of Kalfstindar) consist simply of pillow lava and pillow breccia. Others consist visibly largely of tuff (e.g. Skefilfjoll, Klukkutindar, $kridutindar and Tindaskagi) but this commonly rests on and largely envelops pillow lava and is occasionally interbedded with sheet lava at the summits. Among the tuyas the least extensive are the most complex in terms of their visible structure. Thus the sheet lava and flow-foot breccia of Raudafell and Ho&nAofdi quite clearly rest on and partially envelop a common tindar- like foundation which extends beyond Hognhofdi as a typical 20

tindar, Kalfstindur (off eastern edge of Map 1). The more extensive tuyas visibly consist solely of sheet lava and flow-foot breccia. In the case of Laugarvatnsfjall and Middalsfjall, the internal structure can only be inferred. But in the case of Skrida the axial crest is exactly coincident with the trend of Klukkutindar which merges topographically with its south-west corner. One can readily envisage a northerly continuation of this tindar, coincident with the crest of Skrida, enveloped within and masked by sheet lava and flow-foot breccia. Indeed its presence may be revealed in the buttresses of tuff projecting from the breccias of the northeast face.

4. The genesis of the volcanoes 4.1. A growth sequence The structural series outlined in the previous subsection is believed to be the expression of a growth sequence involving three phases - an aquatic effusive phase, an emergent explosive phase and an aerial effusive phase. The growth sequence of the tuya, the end member of the series, is envisaged thus (Fig.2). An initial phase of effusion in water gives rise to a pile of pillow lava. As the pillow lava pile mounts towards the water surface effusion gives way to explosive activity which mantles the volcano with tuff. Some time after breaching the water surface the volcano resumes effusive activity and lava flows spread out from the vent, giving rise to breccia as they encounter water and advancing outwards over the deltaic accumulation of breccia thus formed.* *Saemundsson (1967A), working independently and indeed unknown to me in the area immediately to the southwest of the Laugarvatn area, has reached basically identical conclusions on the structure and growth sequence of a tuya (see especially his Fig.6). ///

3 nil Did 22

DESCRIPTION OF FIGURE 2. GROWTH OF A TUYA.

A. Aquatic effusive phase. Magmatic heat causes melting of ice sheet above eruptive fissure forming meltwater vault. Within this vault erupting lava builds a steep-sided pillow lava pile. B. As pile mounts roof of vault collapses forming intraglacial lake. Effusion gives way to explosive phase of emergence and resulting tuff accumulates between walls of ice on top of pillow lava pile. C. Emergent explosive phase gives way to aerial effusive phase. Lava issues from vent and pushes out into meltwater lake on deltas of flow-foot breccia. D. Advanced stage of aerial effusive phase. Products of earlier eruptive phases overwhelmed and buried by flow-foot breccia. E. Withdrawal of icesheet exposes eruptive pile in characteristic form of tuya. E'F.Alternatively, with temporary cessation of volcanism, progressively thickening icesheet overwhelms tuya. Subsequent eruptive phase results in partial mantling of earlier sheet lava/flow-foot breccia unit by another such unit. G. Withdrawal of icesheet exposes tuya of compound structure expressed in benched morphology. Symbols as for Plate 1 & 2. 23

4.2. Environment of volcanism The volcanoes of the Laugarvatn area give clear evidence of the existence during volcanism of bodies of water at levels of up to 1,000m above present sea level. Such heights are clearly well outside the range of eustatic fluctuations of sea level. Nor is there any evidence or likelihood of wholesale tectonic emergence and uplift. What then was the nature of these water bodies? I believe that the now generally accepted hypothesis of intraglacial growth of the volcanoes of the "Moberg Formation" is applicable to the volcanoes of the Laugarvatn region which, I envisage, grew within local bodies of water melted in an icesheet by volcanic heat. The evidence for the contemporaneity of glaciation and volcanism is strong. Wherever the substratum of the volcanoes is exposed, it can be seen to have been glaciated and the volcanoes themselves have all suffered the effects of glacial erosion. These effects are most conspicuous in the uniform N - S elongation of the larger tuyas which coincides not with the regional volcanotectonic trend but with the direction of ice movement recorded by glacial striae (see Map 1). This phenomenon is most strikingly apparent on Skrida where the crestal fissure runs NNE while the long axis of the tuya runs slightly west of north. One of the most striking characteristics of the tuff- capped tindars is the frequently low bedding dips exhibited by the tuffs despite their commonly elevated position. On Skefilfjall and on the southern crests of Skridutindar and Tindaskagi the tuffs even have the form of shallow troughs. Explanation of these features is rendered unsure by the uncertain degree to which these volcanic piles have been degraded by moving ice. However it is probable that banking against walls of ice is responsible for the shallow dips 24

and it is possible that the trough structure results when tuff, falling on an icesheet, is washed back into the meltwater lake in meltwater formed by its own residual heat. Sedimentation from the lake margins in this manner would form basins with axes coincident with the crest of the volcanic pile. Confinement by walls of ice may likewise be at least partially responsible for the encircling cliffs of flow- foot breccia which are so notable a feature of the tuya Hlodufell.

4.3. The explosive phase In the volcanoes of the Laugarvatn region pillow lava of the aquatic effusive phase is mantled by and separated from sheet lava of the aerial effusive phase by tuff. This tuff is the product of an explosive phase which coincides with the emergence of the volcano from the meltwater lake. Explosive activity is characteristic of the emergent phase of marine basaltic volcanoes as has been recently emphasised during the emergence of the new island volcanoes of Capelhinos in the Azores and Surtsey off the south coast of Iceland (see Chapter 4, Section 1). It has also characterised the historic eruptions of Iceland's active intraglacial basaltic volcanoes. The 14 eruptions of Katla, recorded since 894 AD, have all been explosive (Lacroix, 1923) as were the eruptions of Grimsvotn in 1922 and 1934 (Barth, 1937). In the case of both emerging marine and intraglacial volcanoes the product of the explosive activity is a tuff consisting very largely of clear basaltic glass which is very susceptible to palagonitic alteration. 25

4.4. Hypotheses of tuya growth In explanation of the Icelandic tuyas both Kjartansson, and van Bemmelen and Rutten have proposed hypotheses of intraglacial growth. In a recent paper Kjartansson (1964) summarises his concept of the growth of tuyas, first advanced (in Icelandic) in 1943. "The socle of moberg (or palagonite i.e. vitreous clastic rocks of basaltic composition) and pillows was formed by subglacial eruption piling up this material in water-filled vaults melted in the ice sheet from beneath. Later on, as the ice had melted through and the accumulated material reached above the water, the eruption became subaerial and changed its character, producing the normal lava flows which are now seen to cap the tablemountains and in some cases form veritable shield volcanoes on top of them". According to van Bemmelen and Rutten (1955) a developing tuya passes through three phases of growth (see their Fig.41); (i) subglacial eruption forming palagonite tuff and breccia; (ii) local melting of the ice cap with formation of irregular layers of "nodular" lava in a lake of meltwater; (iii) a complete regression of the icecap surrounding the eruption centre with subaerial extrusion of basalt lavas on the "top- plateau", some of which flow down into the surrounding plains. Neither of these hypotheses recognises the existence of flow-foot breccias and the entire foundation is thus assigned to a pre-emergent phase. The growth sequence which I have postulated to account for the tuyas of the Laugarvatn region is anticipated in part by Mathew's (1947) interpretation of the form and structure of a group of seven tuyas in northern British Columbia. The tuyas of the Tuya-Teslin area may thus be conceived as originating partly by subaqueous and partly by subaerial eruption of basic lava through intraglacial 26

lakes. The first products of the eruption, which would resemble the subglacial breccias and pillow lavas of Iceland, would accumulate in a cone or low mound at the scene of the activity The higher parts of Ash Mountain (consisting according to Mathews of a cone of granulated basalt containing a few isolated pillows) may be typical of the subglacial or subaqueous stage of the eruption. Provided a cone were built above the surface of the water, however, flat lying flows could have been extruded from the summit and these, on reaching the shores of the lake, would have granulated to form deltas of breccia which would then have been overridden by the advancing flows".

5. Regional relationships and the history of volcanism (see Map 1) 5.1. Vatnsheidi lava and breccia The Vatnsheidi lavas form the foundation on which rests Middalsfjall, and part at least of Laugarvatnsf jail, Efstadalsfjall and Raudafell-Hognhofdi. They outcrop principally in the valley between Laugarvatnsfjall and Middalsfjall and to the east of Middalsfjall where they constitute a prominent platform overlooking the river Bruara. They also appear ds an inlier in the deeply incised gorge of the Eruara where it runs between Raudafell and Hognhofdi. The Vatnsheidi lavas rest on flow-foot breccia, most extensively exposed along the scarp to the west of the Bruara where the passage zone stands at about 270m. 27

5.2. Laugarvatnfjall, Middalsfjall, Efstadalsfjall Hrafnabjorrr The conspicuous accordance of the flat top of Efstadalsfjall and the southern platforms of Laugarvatnsfjall and Middalsfjall when viewed from the south or east reflects a common structural element, a sheet lava/flow-foot breccia unit with a passage zone characteristically at 560m*. In detail the structure of Laugarvatnsfjall and Middalsfjall is rendered more complex by the presence of another lava/ breccia unit with a passage zone at about 650m. On these tuyas the relationship between units is nowhere exposed with clarity but appears to be one of simple superposition as on Skrida. On Laugarvatnsfjall the upper unit (the boundaries of which are difficult to delineate in their entirety owing to inadequate exposure) overruns and obscures the boundaries of the lower unit on the eastern flank of the tuya. On Middalsfjall the lower unit, which forms the southern promontory of the mass, disappears

*Related as it is to a water level at the time of formation, the passage zone of a single lava/ breccia unit may be expected to be horizontal if undisturbed. It may however be affected by regional tilting, faulting, or simply by gravitational collapse. No tilting has been discerned in the Laugarvatn area though the author has observed it on the tuya Fagradalsfjall near Grindavik, southwest Iceland. The effects of faulting are to be seen along the south-west face of Laugarvatnsfjall. Gravitational collapse is prevalent along the margins of the Laugarvatn tuyas and has occurred both during and after the cessation of volcanism. Thus on the south face of Laugarvatnsfjall a tilted, collapsed block of sheet lava at one locality is overridden by the flow-foot breccia of a later flow. At another locality a fracture along which collapse has occurred is draped by moraine, clearly deformed while in a still plastic condition. In this instance collapse may have occurred beneath superincumbent ice or during its withdrawal. 28

northwards beneath an extensive upper unit degraded by glaciation to a pile of breccia surmounted by sheet lava residuals. As has been noted these three tuyas rest, at least in part, on a surface of older (Vatnsheidi) lavas. This surface has been observed at the foot of mantling tuyas at four localities, in two of which it proved to be veneered with morainic material and in one instance to bear striations. The tuya Hrafnabjorg would seem best assigned to this group of volcanoes, for the rocks of a sheet lava/flow- foot breccia unit with a passage zone at or near 560m are preserved as small platforms on the northern and eastern flanks of the mountain.

5.3. The tindars9 Skrida and Raudafell-Hognhofdi The string of volcanoes stretching from Kalfstindlar to Skridutindar of which Skrida is a part, is clearly younger in age than Laugarvatnsfjall on whose surface it rests for much of its length. Similarly the line of volcanoes which includes Raudafell, Hognhofdi and Kalfstindur rests at its southern end on Middalsfjall. An 800m passage zone is common to the tuyas Skrida, Raudafell and Hognhofdi.

5.4. Hlodufell As an island in a sea of young lava, Hiodufell's structural relationships are obscured. This magnificent tuya, which rises nearly 700m above its surroundings, displays a conspicuously compound structure - a basal sheet lava/flow-foot breccia unit, expressed topographically as a prominent 9:0m bench, surmounted by another prominent lava/breccia unit with a break in slope at a height varying between 1060 and 1100m,(P1.3B). 29

5.5. Preglacial and postglacial lavas* The volcanic succession of the Laugarvatn region is not confined to intraglacial volcanoes. Volcanoes of extraglacial origin, both older and younger than intra— glacial volcanoes are contiguous. To the south—west of Laugarvatnsfjall stands the low, glaciated shield volcano Lyngdalsheidi. Its lavas are extensively veneered with glacial drift and alluvium and their full extent and structural relationship to Laugarvatnsfjall are not apparent on the ground. However it is evident on air photos that these lavas extend to the foot of Laugarvatnsfjall and that they are finely sliced by a swarm of fractures parallel to but much more closely spaced than the conspicuous fractures on Laugarvatnsfjall (the pattern employed in Map l. is intended to convey this aspect of their appearance). Bearing in mind that the Lyngdalsheidi lavas lie directly southwest and on trend with Laugarvatnsfjall, it is clear from their closely fractured condition that they are older than those of Laugarvatnsfjall and that they must thus extend beneath the southern portion of the tuya. Lapping up against the cluster of intraglacial volcanoes and encircling the more isolated are the very young, unglaciated lavas of the shield volcanoes Skjaldbreid and Lamba and the crater row Tindafjallaheidi.

*These terms are applied on a strictly local basis with reference to the phase of glaciation which the Laugarvatn volcanic rocks record. 3 0-

5.6. History of volcanism The oldest major intraglacial structure in the Laugarvatn region is the Vatnsheidi mass*. Its complete structure is unknown but one can state, knowing the altitude of the passage zone and the attitude of the foresets, that the Vatnsheidi lavas at the present perimeter of the mass advanced south-east and south into a meltwater lake with a surface altitude of 270m above present sea level. With the end of this phase of volcanism, the ice sheet surmounted the Vatnsheidi mass and glaciated its surface. A new eruptive phase saw the construction of Laugarvatnsfjall, Middalsfjall and Efstadalsfjall on and astride the margins of the Vatnsheidi mass. These tuyas exhibit a conspicuous passage zone at 560m and would appear to have grown simultaneously within a single meltwater lake or at least in several interconnected water bodies. Hrafnabjorg may be assigned to this phase of volcanism. In their turn these volcanoes were submerged beneath a cover of ice and Laugarvatnsfjall and Middalsfjall became the foundation for the still younger tindars and the tuyas Skrida and Raudafell-Hognhofdi. These tuyas evidently grew within water bodies with a common level at 800m. The intraglacial volcanoes of the Laugarvatn region bear clear witness of having developed in a sequence of meltwater lakes of successively higher surface altitude (a sequence which may be further extended by Hlodufell with passage zones at about 900 and 1050m). *A much degraded mass f intraglacial origin embedded in the south-east flank of Laugarvatnsfjall, is demonstrably enveloped by Vatnsheidi lavas at its northern end. 31

This phenomenon presumably reflects a progressive increase in the surface altitude of the ice sheet during the period of intraglacial volcanism, probably as a consequence of increasing thickness of ice*, though volcanic increments to the landsurface may have played a part. Such a reconstruction throws light on the compound structure of the Laugarvatn tuyas, typically consisting as they do, not of one but of two or more superimposed lava/ breccia units. This characteristic is readily explained in terms of recurrent eruption from these vents within an icesheet of continually increasing surface height.

6. Vent type and volcano form The Quaternary volcanic zone of Iceland is intensely fissured and these fissures, the youngest of which appear on the ground as gaping crevices (gja), clearly provide the channels by which much if not most lava reaches the surface (see also Walker, 1965). This has almost certainly occurred in the case of the strongly linear tindars which extend en echelon from Kalfstindar in the south-west to Skridutindar in the north-east and which can be seen to coincide with a colinear swarm of fissures conspicuous on the platform of Laugarvatnsfjall (Map 1). In the case of the tuyas no relationship to fissures is immediately apparent except in the case of Skrida which displays a row of craterlets along its crest in direct continuity with the elongate Klukkutindar. However observation of the course of the Surtsey eruption strongly suggests that the tuyas may also be fissure fed. For the first week of the eruption the

*One can infer local ice thicknesses in excess of 500m for Raudafell-Hognhofdi and Hlodufell. FIGURE 3 Contrasted profiles of tuya Skrida (right foreground) and shield volcano Skjaldbreid (left background). On Skrida vertical lines mark cliffs in sheet lava and oblique lines mark slopes of flow—foot breccia scree. 33

emerging volcano had the form of 'a relatively narrow ridge split lengthwise by a fissure on which there were during the first days of the island's existence no distinctly separated vents, the explosions occurring. along the entire fissure". On the eighth day "the form of the island was that of an ellipse, the long axis in the direction of the fissure. The main activity was in the southwest—most of three active vents Gradually the island became almost circular and as only one crater has been active most of the time since then, although sometimes with more than one vent, the eruption may be said to have changed from a linear to a central one" (Thorarinsson et al, 1964).

7. Intraglacial and extraglacial volcanoes Nothing is more striking in the Quaternary volcanic belt of Iceland than the contrasting forms of those volcanoes which have grown within ice sheets and those which have not. While the volcanoes of intraglacial origin appear as steep—sided piles of pillow lava and bedded tuff or as fortress—like tablemountains, with slopes of up to 35°, those of extraglacial origin appear as smooth cones and whale backs of lava with slopes never exceeding 10°. These contrasts are dramatically evident in the northern and western parts of the map area (Map 1, Fig.3). Here we have beautifully exemplified the effects of eruptive environment on basalt lava, for there are no significant differences in lava composition (see addendum to Chapter 2). 34

8. Synopsis The volcanoes described in this section grew up beneath or within an icesheet and this accounts for the peculiarities of their form and structure. The tuyas are the product of three growth phases: (a) Aquatic effusive phase. Magmatic heat causes melting of the ice sheet above the eruptive fissure forming a meltwater vault. Within this vault erupting lava builds up a steep-sided pillow lava pile. As the pile mounts toward the surface of the icesheet the roof of the vault collapses and an intraglacial lake is formed. (b) Emergent explosive phase. As the pillow lava pile approaches the lake surface, effusion gives way to explosive activity. The resulting tuff mantles the pillow lava pile and, banked against the walls of ice, fills the lake as the volcano rises above water level. (c) Aerial effusive phase. With the end of the explosive phase lava issues from the vent and flows advance outwards into a metlwater lake on deltas of flow-foot breccia. If eruption is sufficiently prolonged the products of earlier eruptive phases are overwhelmed and buried by advancing flow-foot breccia. Recurrent volcanism from single vents within a thickening icesheet has resulted in tuyas consisting of several superimposed sheet lava/flow-foot breccia units. Volcanism has not ceased with the withdrawal of the icesheet but, in their gross morphology, the volcanoes which have arisen since its withdrawal stand in marked contrast to their intraglacial counterparts. 3

SECTION 2

THE LAUGARVATN VOLCANOES - DETAILED CONSTITUTION AND DEVELOPMENT

1. Introduction This section describes in detail the constitution of two of the Laugarvatn volcanic piles: Kalfstindar and Raudafell-Hognhofdi. These were selected for detailed study because they display with remarkable clarity the constitution and structure of basaltic volcanoes which have commenced activity in water of moderate depth and have emerged in the course of their growth. Description of the constitution and structure of the volcanoes is preceded by definition and description of the basic rock types and assemblages.

2. Rock types In order to have a comprehensive integrated terminology for the whole suite of rocks based on genetically significant but observable properties, I have found it necessary to coin several new terms and to provide my own definitions of old ones. New terms have been coined by adding prefixes to the established basic terms pillow lava and breccia.

2.1. Lavas Sheet lava As defined in Section 1 the term sheet lava denotes extruded lava which in section is seen to consist of tabular and lenticular units with a lateral extent many times their thickness. In the Laugarvatn volcanoes these 36 units may be wholly non-fragmental (pahoehoe) or they may consist of an irregular sheet or lens of coherent lava enveloped in a breccia of scoriaceous varicoloured lava fragments (aa). Pillow lava The term pillow lava denotes lava consisting of abundant structural units which in section display numerous circular and elliptical outlines, and which incorporates little or no fragmental material. Seen in section the Laugarvatn pillow lavas have a textbook appearance (P1.4A,B). All pillows have a glassy crust up to lcm thick, and radial prismatic jointing is characteristic. All are vesicular, the vesicles (not including pipe vesicles) being present throughout the pillow though most abundant in the upper half where a concentric zonal arrangement is characteristic (P1.4B). Radially oriented pipe vesicles are a common feature of the lower half of pillows. The form and fabric of the Laugarvatn pillow lavas is the subject of Chapter 3, Section 1. Para-pillow lava The term para-pillow lava has been coined for lava consisting of structural units similar to pillows in their dimensions, fabric and internal structure (P1.8A,B), but with typically ragged outlines and separated by a breccia of glass. The surface of these units is remarkably jagged, and the breccia which encloses them commonly contains lava nodules, some no larger than a pea, with a hackly glass crust. Voluminous irregular masses of closely jointed lava (P1.8C), up to tens of meters across, are an integral part ,.)f- the total structure of para-pillow lava.

2.2. Fragmental rocks Tuff The term tuff denotes volcanic fragmental rocks with 37 clasts predominantly (more than 75%) smaller than lcm. The Laugarvatn tuffs consist largely or wholly of fragments of clear glass*. These clasts are characteristically though variably vesicular, but rarely pumiceous. Clasts with diameters many times that of the vesicles have scalloped margins (vesicular embayments) and tend to be roughly equant in form: clasts closer in size to vesicle diameters are often less equant and more shard-like in form with less intricate boundaries. Sorting is very variable, the coarser tuffs being generally poorly sorted and the finer tuffs moderately to well sorted. Breccia The term breccia denotes volcanic fragmental rocks consisting substantially (more than 25%) or predominantly of angular clasts larger than lcm. Two basic types are distinguidhed. Parabreccias characteristically display the texturally heterogeneous aspect of phenoclasts in a matrix. (P1.7D): orthobreccias have a texturally homogeneous aspect insofar as no visual phenoclast/matrix distinction is possible (P1.7A,B,C). The distinction is essentially similar to that between ortho- and para-conglomerates (Pettijohn, 1957, p.254). The term tuff-breccia is applied to parabreccia in which the matrix is tuff (P1.7D). The orthobreccias of the Laugarvatn volcanoes range in constitution from almost wholly lithic to almost wholly vitric. °lasts in the lithic breccias are polyhedral lava blocks. In the vitric breccias they range from small (less than 5cm) angular fragments of glass to large irregular lithic elements (not all can be accurately designated as fragments) encrusted with glass, even on what are clearly fracture faces. The characteristically bizarre shapes of large lithic elements in these vitric breccias, especially when seen on naturally polished *The clear basalt glass ('sideromelane') which is characteristic of the Laugarvatn tuffs and vitric breccias and which crusts the pillows is commonly palagonitised in varying degree. 38

surfaces, warrant a succinct term for the rock type. I call it chaobreccia (chaos-breccia). The breccias of the Laugarvatn volcanoes constitute two significant though not always readily separable assemblages. One assemblage consists of the breccias which are intimately associated with and flank the pillow lava piles. The other consists of the breccias which occur beneath and in structural continuity with sheet lavas. Characteristic of the former assemblage are the pillow breccias (P1.7A,B,C), orthobreccias which from their content of recognisable pillow fragments and their lithological homogeneity can reasonably be inferred to have resulted from the fragmentation of pillow lava. The pillow breccias range from the lithic Billow jointblock breccias (P1.7C) consisting almost entirely of cobblesize (5 to 25cm) pillow jointblocks with glass confined to remnant pillow surfaces, to the vitric pillow chaobreccias (P1.7B) in which most recognisable pillow fragments take the form of arcuate plates of glass-encrusted lava. 'Whole' pillows in the latter are typically void (P1.7B). The latter assemblage, embracing all the varieties of breccia described above, incorporates the flow-foot breccias, breccias which characteristically occur beneath and in structural continuity with overlying sheet lava. These are the products of the immediate or ultimate fragmentation of lava which flowed from air into water. While the structural relationship to sheet lava is their defining characteristic, the flow-foot breccias do show other group characteristics by which they differ from the breccias of the pillow lava piles. While lithic clasts in the flow-foot breccias, as in the breccias of the pillow lava piles, are predominantly of cobble grade, boulder size clasts (greater than 25cm) are far more common in the

40 former than in the latter: and the clasts in outcrops of flow-foot breccia frequently display a much greater range of crystallinity, vesicularity and colour (most notably in the presence of reddish clasts) than clasts in breccias of the pillow lava piles.

3. The Kalfstindar volcano The basic morphology and structure of Kalfstindar have been outlined in Section 1 (P1.1). In the following sub- section the constitution and structure of the pillow lava pile, the tuff pile of the middle peaks (i.e. the 666m peak and next peak to north - see P1.1), and the tuff pile of the north (826m) peak are described separately.

3.1. Pillow lava pile On Kalfstindar observation of the structure of the pillow lava pile is restricted to the relatively shallow gullies on its flanks. At least to the depth to which these are cut (50m at most), the pile consists almost entirely of pillow lava with minor irregular intercalations of pillow breccia. In general the plane of flattening of the pillows shows outward dips of up to 20° and is transected by the steep* scree-mantled slopes of the pile (P1.4D). Locally the inclination of the plane of flattening increases to as much as 40° in outcrops of elongate pillows with long axes parallel (P1.40). An important feature of the Kalfstindar pillow lava pile is an upward increase in pillow vesicularity (described in detail in Chapter 3, Section 3). It is clearly seen in

*Thirty readings of the inclination of pillow scree slopes on Kalfstindar and its northerly continuation gave values ranging from 22° to 31o with an average of 270. ' , : • '

+•:•?, 1•• •

Ct 0

9 SIV'T.d. 4 3 jUxtaposed photos of pillow lava from the bottom and top of the pile (P1.4A,B: there is a vertical separation of 250m). Occasional irregular impersistent dykes of variable orientation occur within the pillow lava pile (P1.4D) and may have fed flows of pillow lava. One unequivocal pillow lava feeder, of which mention has already been made, is exposed near the crest of Kalfstindar (P1.5A,B). This dyke is about 3m wide and runs parallel to the axis of Kalfstindar. Outward dipping pillow lava merges imperceptibly with the lava of the dyke wall (P1.5A). The center of the dyke is filled with tuff (P1.5B, top left), and where this tuff infilling has been eroded away the billowy walls of the dyke (P1.5B, right of field) show grooving, as is common on the walls of lava conduits.

3.2. Tuff pile - middle peaks The tuff pile, as the term indicates, consists predominantly though by no means wholly of tuff. In places the tuff forms homogeneous outcrops tens of meters high with a barely perceptible stratification (P1.9B, left of field). More commonly stratification is conspicuous (P1.6B4 O), though the strata, reflecting fluctuations in clast size and sorting, are typically diffuse. Stratification is most pronounced where the difference in grainsize of adjacent strata is greatest.. Small scale cross stratification is almost ubiquitous and a large scale structure akin to festoon cross bedding (Pettijohn, 1957) is apparent in some sections of the pile (P1.9A, top). Normal grading is apparent in some outcrops though in general graded units cannot be defined. However rare outcrops do show well defined graded units up to 5m thick. :

4if

Pillow breccia, Para-pillow lava and, in lesser amount, pillow lava occur at the lower levels of the tuff pile, intimately associated with tuff. Vertical and lateral gradations from orthobreccia to parabreccia to tuff and vice versa (P1.9A) are common. Pillow lava occurs sometimes as concordant and relatively regular bodies, sometimes as irregular clusters within which single pillows appear partially or wholly enveloped in tuff (P1.8D). Para-pillow lava is characteristically discordant in the details of its contacts with tuff and commonly in its gross distribution also. In the cliff face east of the 666m peak where it is most conspicuous, para-pillow lava forms bodies of extraordinarily irregular outline (P1.80). Its relationships with tuff can only be described as chaotic. In places the two lithclogies are crudely interstratified while more commonly the one appears as contorted stringers in the other. Generally the content of pillow breccia, pillow lava and Para-pillow lava within the tuff pile diminishes rapidly upwards and is nil in the upper part of the pile. Irregularly discordant boundaries are not restricted to para-pillow lava/tuff contacts. Such boundaries do occur between tuff and tuff, and also between tuff and breccia (P1.9B). At the base of the tuff pile one commonly encounters highly vesicular, fine vitric breccia (often a chaobreccia) with the aspect of a coarse tuff (P1.6A). It frequently occurs in bodies up to tens of meters thick, typically almost or completely unstratified. The pillow lava with which it is often associated and the fragments of which it often encloses (P1.6A) is typically highly vesicular with very strong concentric vesicle zoning. •

•• 48

The tuff pile rests saddle-like astride the pillow lava pile. The essence of the structural relationship is clearly conveyed in P1.9D. It can be seen that the tuff and intercalated breccia is draped or banked against the flanks of the pillow lava pile and thickens remarkably away from its axis. Dips of up to 32° at the head of the wedge immediately adjacent to the pillow lava pile decrease rapidly outwards to low dips of variable direction,

3.3. Tuff pile - north peak The tuff outcropping on the north peak above about 600m differs distinctly from the tuff of the middle peaks and from tuff similar to the latter outcropping lower on the east flanks. When tuffs of comparable grade are considered (compare P1.6B - middle peaks tuff, and P1.6D - north peak tuff), it can be seen that in the north peak tuff (above 600m) the strata are much thinner. And while even the coarsest clasts in the middle peaks tuff consist wholly of glass, many of the coarser clasts in the north peak tuff are crystalline, though very finely so. Furthermore tuff enclosing lava blocks on the north peak occasionally displays impact structure (bomb sags), a feature nowhere seen in the middle peaks tuff. The salient feature of the north peak is a sequence of ten or so as lava flows with tuff interbeds above 715m (P1.10A). The flows have an average thickness of 3m, with a coherent lava/scoria ratio of 1/2. A mass of coarsely fragmental lava which forms an isolated eminence at the southern end of the peak is the focus of several converging dykes (P1.103) which cut the tuffs on which it rests. At least one of these can be traced into the mass itself. Pq

0 50

Conspicuous on the east flank of the north peak beneath the level of the sheet lava are breccia masses containing many lava boulders, some a meter and more in diameter (P1.10C). In addition to their large size, the phenoclasts have an angularity which is quite foreign to the pillow breccias, and they are poorly or non-vesicular. Some display a peripheral, varyclose spaced prismatic joining normal to the bounding fracture faces. The matrix consists of a breccia of variably vesicular, more finely crystalline clasts and vary little glass.

3.4. Kalfstindar's water level Where the tuff pile stands highest on the north peak of Kalfstindar the summit at least was demonstrably emergent, for it contains a body of as lava flows. The lowest of these flows occurs at 715m, while pillow lava outcrops at 585m and breccias diagnostic of aqueous conditions to 600m on the east flank of the north peak. A water level thus lay between 600 and 715m, and is subsequently assumed to have stood at 700m. This value could be as much as 100m too high but this is thought unlikely.

4. The Raudafell-HoLmhofdi volcano The mountains Raudafell and Hognhofdi are part of a single crudely elliptical massif 9km I,ong and 3km wide, elongate NE-SW. (Map 1). Like Kalfstindar this massif has as its basic unit a pile of pillow lava on which rest the separate superstructures of the embryo tuyas Raudafell and Hognhofdi (Fig.4).

RAUDAFELL North South sheet lava stony pillow lava • and breccia 800m

„•,„„,,_,”1 •..-:.../..,0';,. , o,.4•• • tuff .•-/-;- ..., ?c tuff ..6. ....0.1.r..:',0,..4/ • .• •o.;;,‘"0/..:-5* pillow lava pile i;,.,„,4i, 0/, ..,,,,0// , /, •• • .....:::.1:-... • .-.c, -., /--/. ,.,,,p/f 4, _,./ Is, i:.i, .°,,,, -,,-; 5 . -;.1:.•:••••?/ 1--c'",.'y 7,,/,';',•.•.:.*.' ':' / --/.f, ..:.:..;• .....,--f-4------ge.-:-:,;;:,:?:•,?.., ... / ••••• •:.• • dly, ,/o / • Y r • -..- 1,.. Jo • /' c...!" 5-11. //% /C.'...• ' t• r •-/0 / , < !I ./ .. I / ' /

HOGNHOFDI sheet lava tuff sheet lava 800m 800m flow-foot breccia a\`. :41:63xe, ‘CA‘ . , 1 flow-foot • c C2C7 C c.c x C r-',14111 pillow lava pile C/ C .1 • C. g breccia 111141141i T Dig im '117n z2/'') . • I \ c c iffij c. % C , LII \ \.\\\, , ./- 7,c/ 6 ,‹) Cc c c2c? Ji ll`\\\\ -3 `j‘111 41: 111:17 c7v:i11141.1111iC)111: CP • 'PP.. • •. 2. /",c7.41, ... ass) 45 c, • . >cf a • c- • . lifYref : %.9 O Pah CN • • • Z cmil‘\ :" ":••••11t 11 4r1r.s: "‘P CO t a. • •P' e• 0. • jr•o• .."`jr.a c. 5,. SeOgrzr at. 'T.-4w.: e,

FIGURE 4. Raudafell-Hognhofdi from the east. A adjoins B. 52

4.1. Pillow lava pile Though largely mantled by the Raudafell and Hognhofdi superstructures, the pillow lava pile has an extent delimited by the perimeter of the massif as a whole. Its horizontal dimensions are thus 9 x 3 km and it has a visible maximum height (thickness) of about 300m. Its constituent pillow lava is similar to that of the Kalfstindar pillow lava pile and like the latter shows a distinct upward increase in vesicularity (Chapter 3,Section 3) . A remarkable section through the pillow lava pile in the south wall of Bruara gorge (';Tap 1) shows a nested series of slump scars filled with pillow breccia and pillow lava (P1.11A,B; the cliff face in B goes off at right angles from the right hand end of the cliff face in A). The lower slump scars have formed the loci of intrusion of basalt dykes (P1.11B); the uppermost is delineated by tuff (P1.11A).

4.2. Raudafell superstructure (Fig.4A) The roughly circular mass termed the Raudafell superstructure is founded on the southern end of the pillow lava pile. It consists of a pile of tuff, breccia, pillow and para-pillow lava, with a very incomplete capping of sheet lava. The northeast face and southwest slopes of Raudafell consist predominantly of tuff, resting visibly on the pillow lava pile. This tuff has the same range of character as the tuff of Kalfstindar and shows the same upward pattern of change. Tuff (and intercalated breccia) immediately overlying the pillow lava pile is similar to that of the middle peaks of Kalfstindar, while tuff at higher stratigraphic levels is like that of Kalfstindar's PLATE 11 54

north peak (P1.11C). On the southwest slopes tuff forms thick (as much as 100m) wedges, expressed topographically as benches (P1.110), which thin rapidly upslope. these wedges stratification decreases from 200 or more at the head of the wedge to 00 at its outer and upper margin. On the southeast slopes however tuff is less abundant than pillow and para-pillow lava and breccia. The voluminous pillow lava has characteristics which clearly distinguish it from pillow lava of the pillow lava pile (P1.120 is of the same lava type on Hognhofdi). Vesicles (exclusive of pipe vesicles) are typically few and small, or even absent altogether, and show no systematic variations with height in the pile (See Chapter 3, Section 3). Jointing is frequently irregular rather than prismatic. And the lava is a lighter grey than that of the pillow lava pile. It will subsequently be termed 'stony' pillow lava. Para-pillow lava and breccia (Pl. 12B, 13A,B) intimately associated with the stony pillow lava are in general similarly poorly vesicular. In a deeply incised gully the stony pillow lava appears as irregular pods spliced with steeply inclined wedges of breccia, with tabular and lenticular intercalations of tuff. (P1.12A). Elsewhere on the southeast slopes the relation ship of tuff and breccia can only be described as chaotic (P1.12B). The scree of the western and northern flanks, from which project small promontories of the same constitution as the southeast flanks, is derived from breccias which at about 800m pass up into horizontal sheet lava. This junction is marked by an abrupt positive break in slope which gives Raudafell its distinctive tablemountain profile when viewed from the north and west. ePP'

4it

E-I 56

The sheet lava of Raudafell is exactly similar to that of Kalfstindar's north peak, and boulder breccia like that of the north peak occurs in minor amount on the western and northern flanks. At the summit of Raudafell occur formless piles of red cinder. Tuff, visibly resting in places on sheet lava, outcrops at intervals along the length of the NE-SW crest.

4.3. Hognhofdi superstructure (Fig.4B) The Hognhofdi superstructure consists basically of a platform of sheet lava on an elongate pedestal with the characteristics of the northwest and southeast flanks of Raudafell. No body of tuff of dimensions comparable to those exposed on the northeast and southwest flanks of Raudafell is visible. The constituent rock types do not differ from those described for Raudafell. The principal outcrops on the largely scree covered slopes of the pedestal are crudely stratified, intercalated breccia, tuff-breccia and tuff with bedding varying in inclination from 0° to 30° (exclusive of slump blocks). The breccias pass upwards into the sheet lava of the platform at about 800m. Within the passage zone (P1.11D, below break in slope) occur intercalated sheets, lenses and more irregular bodies of crudely pillowy or hackly- jointed lava (P1.13D). Enveloped in and resting on the fragmental rocks of the pedestal are pods of stony pillow lava (P1.12C) which in a number of instances stand out from the pedestal slopes as small conical mounds with a rounded apex (P1.12D). A similar mound is conspicuous on the northern slopes of Raudafell. A subsidiary but ill-defined sheet lava/flow-foot breccia unit, culminating in the main peak, surmounts the

principal unit in the northern half of its extent. Mounds of tuff prominent on the soutwest corner of the platform probably rest on sheet lava, at least in part, though mantling scree obscures relationships.

4.4. Raudafell-Honhofdi's water level The principal passage zone on Hognhofdi, as on Raudafell, stands at 800m and records the surface level of the meltwater bodies within which these piles were constructed.

5. Affinities mode of formation and emplacement of rock types The textural and structural affinities of pillow lava and pahoehoe are such (Chapter 3, Section 1) that I believe pillow lava may justifiably be regarded as the subsqueous equivalent of pahoehoe; and the frequently observed mode of formation of pahoehoe toes, whereby a flow front advances by the successive protrusion of small bulbous flow units, may be accepted as a model of pillow lava formation. Horizons or clusters of pillows apparently isolated in or separated by tuff are common on Kalfstindar and Raudafell-Hognhofdi. However exposure in three dimensions makes it clear that these bodies are no more entities than are the pillows of pillow lava proper (see also Chapter 3, Section 1). Their apparent isolation reflects distension of the pillow lava fabric by tuff, a condition which can be readily explained in terms of advance of a pillow lava flow during simultaneous rapid deposition of tuff. Para-pillow lava I believe to be the subaqueous equivalent of as lava. It has the same gradational 59

textural relationship to pillow lava as does as to pahoehoe, the boundary between the types being an arbitrary one in each case. It seems probable that the discordance which characterises the contacts of para-pillow lava with tuff, in contrast to those of pillow lava, is to some degree a product of the erosive action of a hackly fragmental flow on a subsurface of unconsolidated or semi- consolidated tuff. This, together with concurrent rapid accumulation of tuff and repeated slumping for which there is ample evidence, is quite adequate to explain the gross form of bodies of para-pillow lava and the details of their relationship to tuff. Exposures on the flanks of Kalfstindar reveal with marvellous clarity the mode of formation of the pillow breccias. Plate 5D shows a buried cliff of pillow lava against which rests an accumulation of crudely stratified pillow orthobreccia, parabreccia (tuff-breccia) and tuff, with dips away from the cliff face as high as 400 immediately adjacent to the face. The story it tells is clear. Pillow lava debris, as it fell in greater or smaller amount from the cliff, came to rest on the steep slopes of mounting bank of tuff accumulating against the face. The orthobreccia is the pure scree, the parabreccia an admixture of pillow scree and tuff. A s:,mewhat similar exposure is seen in Plate 50 (the upper right portion of the field of 50 overlaps the lower left portion of the field of 5A). In this instance the buried pillow lava cliff appears as an obliquely discordant, outwardly inclined discontinuity (P1.50 - top right to bottom left), the lower portion of which is the locus f injection of a dyke. At the base of the exposed cliff section is a substantial accumulation of pillow chaobreccia, shown in close-up in Plate 7B. It is reasonable to infer that this chaobreccia, containing in abundance the shells and shell fragments of 60 voided pillows, has resulted from the collapse and crumbling of a pile of semi-consolidated pillow lava which was drained of its still-fluid lava on fragmentation. A subsequent eruptive episode mantled with tuff the chaobreccia and the pillow lava cliff froth which it fell, and new lava cascaded down the tuff-mantled face (P1.5A). In such an environment of steep slopes, instability and collapse, much of the pillow breccia must have attained its ultimate position of rest as a result of mass movement, as does much of the pillow scree on the flanks of the pile at present. Some of the discordant boundaries within the tuff pile appear to have resulted from gullying of the tuff substratum by such rubble flows (P1.9B). It is clear from their composition and stratigraphic position that the boulder breccias are the product of the total fragmentation of as lava flows, an effect which gravitational collapse provides a satisfactory explanation. The boulder-size clasts are fragments of the coherent centers of such flows, fragmented in some instances while still hot, for some of the boulders exhibit peripheral prismatic jointing normal to the fracture faces. One of these boulder breccias apparently moved down the east flank of the north peak of Kalfstindar us a rubble flow, cutting deeply into its tuff substratum (P1.10D). As indicated in Chapter 2., Section 1, the Laugarvatn vitric tuffs are believed to be the products of explosive eruptions of the kind which have marked the historic activity of Iceland's active intraglacial basaltic volcanopq, and which appear characteristic of basaltic eruptions from wet vents in shallow water and in air (Chapter 4, Section 1). The texture of the tuffs suggests that they are the product of explosive vesiculation and this finds strong support in stratigraphic-lithologic relationships. For the tuff piles of Kalfstindar and Raudafell rest on pillow lava piles which 61

show a conspicuous upward increase in vesicularity, the pillows at the top of the piles being highly inflated: and the tuff is linked to underlying highly vesicular pillow lava by a transitional rock type - the highly vesicular fine vitric breccia described previously and shown in Plate 6A. Though it is difficult to assess to what extent the tuffs were projected to the positions in which they now occur, the prevalence in the tuff piles of discontinuities thought to be slump scars (for instance see P1.9C - mid-field, below skyline) suggests that redistribution of tuff by mass movement following gravitational collapse was a common event.

6. Glacial effects and influences Both Kalfstindar and Raudafell-Hognhofdi rest on glaciated surfaces, a fact best illustrated in the exposures of the Bruara gorge, the deep incision between Raudafell and Hognhofdi (Hap 1). Here the base of the pillow lava pile is extensively exposed (P1.11B - sinuous line at base of cliff face). It can be seen to be separated from a basement of older (Vatnsheidi) sheet lava by a variable thickness of tillite - a fine grey mudstone containing rounded and in some instances striated pebbles, cobbles and boulders of basalt. A similar tillite forms a prominent veneer on parts of the Raudafell-Hognhofdi pillow lava pile. It also occurs on both Raudafell-Hognhofdi and Kalfstindar as discontinuous wedges and lenses between the pillow lava pile and overlying tuff, para-pillow lava etc., a fact which indicates quite clearly that it was deposited in part duri:g the course of volcanism. The picture of close-pressing walls of ice suggested by this observation finds support in other structural 62

characteristics of these piles, most importantly the low bedding dips of tuff in elevated positions alluded to in the previous section. MoreovQr, on both Kalfstindar and Raudafell-Hognhofdi one finds blocks of tuff, more than 100m long in one instance, standing on end with stratification vertical. Several such blocks can be seen embedded in the precipitous southeast face of Hognhofdi, set in the flow-foot breccia which was once quite clearly banked against them. These are thought to be slump blocks of tuff initially deposited on the icesheet itself, which slid into the water-filled crevice between the flanks of the piles and the ice walls. They suggest that, in some instances at least, cliffed faces of tuff and breccia, like the southeast face of Hognhofdi, owe their present form to impounding ice. It seems highly probable that in any aqueous environment other than intraglacial a very large proportici. of clastic material would move under gravity to the foot of the pile and there accumulate as an apron of interstratified tuff and breccia. A problem which I am at present unable to resolve is the degree to which the pillow lava piles owe their form to their specifically intraglacial environment of origin. In the absence of evidence to the contrary, I am inclined to believe that their form would be the same in any aqueous environment.

7. Eruptive histories of Kalfstindar and Raudafell- Hoomhofdi. The eruption which built Kalfstindar commenced from a fissure which at its southern end stood more than 500m below the surface level of superincumbent ice. 63

Initially the lava, quietly welling into meltwater, accumulated along the eruptive fissure as a steep-sided pile of pillow lava. Structural analogy (see Chapter 3, Section 1) suggests that the mode of growth of the pillow lava pile was essentially similar to that of a pile of pahoehoe flows. As the pillow lava pile mounted and the hydrostatic pressure at the level of the vent became less, progressively more volatiles exsolved until the point was reached at which the frothing lava could no longer remain coherent. Unstable masses of frothy lava probably broke away as they were extruded and crumbled as they flowed, rolled and slid downslope, giving rise to the transition breccia. Ultimately, at a water depth of less than 200m, the mounting pile attained a level at which the pressure of the exsolving volatiles caused explosive fragmentation of the lava at the instant of eruption, and the formation and accumulation of tuff commenced. The early explosive episodes were accompanied or separated by the effusion of a few small flows of pillow lava, and, in greater amount, of para-pillow lava, envisaged moving down the steep flanks of the pile as relatively viscous flows of erosive fragmental lava. Continual wasting of consolidated and closely jointed pillow lava and collapse of semi-consolidated pillow lava contributed coarse fragmental material to the banks of tuff accumulating against the flanks of the pillow lava pile. Slumping, mass movement and gullying of finer substrata by flows of coarser fragmental material seem to have been the characteristic concomitants of eruption. At its northern end the pile ultimately emerged and explosive activity of dwindling intensity (suggested by the much reduced thickness of tuff strata as noted) permitted the extrusion of a number of as lava flows. The material of these flows, like the other constituents of the pile, 64 was involved in gravitational collapse and moved down the flanks of the pile as bouldery rubble flows. As with Kalfstindar, the construction of the Raudafell- Hognhofdi massif commenced with effusion of pillow lava and the building of an elongate steep-sided pillow lava pile, and was followed, at the southern end of the pile at least, by explosive activity which mantled .the pillow lava pile with tuff (the tuff of Raudafell). In all probability a similarly explosive episode capped the northern (Hognhofdi) portion of the pillow lava pile with tuff, but this is now obscured by the products of still later activity. Whereas on Kalfstindar the eruption apparently remained dominantly explosive to the end, explosive activity on Raudafell-Hognhofdi was succeeded, though never totally extinguished, by a phase of effusion in air. Lava, flowing from the summit vents of the now emergent tuff piles, almost immediately encountered the water of the encasing meltwater ponds whose level stood at 800m above present sea level. Shattered and granulated lava cascaded down the flanks of the tuff piles and underlying pillow lava, mantling these earlier products with banks of breccia. At the same time the backflow and flank eruption of lava degassed at the emergent summit vents is believed to have given rise to pods of stony pillow lava (see Chapter 3, Section 3). At the northern end of the massif where, judging by the constitution and dimensions of Hognhofdi, the phase of effusion in air persisted longest, the pillow lava pile and capping tuff were almost totally buried and obscured. A last feeble recurrence of activity in a still- thicker icesheet gave rise to the sheet lava/flow-foot breccia unit culminating in the present peak of Hognhofdi. Table 1. Chemical composition of lavas of the Laugarvatn area

Intraglacial pillow lavas Postglacial lavas 1 2 3 4 5 6 7 8 9 10

46.84 48.04 46.28 SiO2 48.15 47.96 16.24 14.05 13.65 12.26 14.64 A1203 16.4 16.2 15.8 15.2 '6.5 Fe 0 2.12 3.02 3.68 2.07 2.39 2 3 11.7 12.3 12.4 12.8 12.7 FeO 8.74 8.52 9.39 10.78 10.79 MgO 9.5 10.6 10.5 8.6 9.4 5.84 7.75 8.73 9.18 9.46 Ca0 13.1 12.5 12.3 11.7 2.4 13.66 13.38 12.46 12.04 11.89 2.8 2.20 1.90 1.60 1.82 1.19 Na20 2.8 2.5 2.5 2.7 0.3 0.33 0.29 0.35 0.38 0.37 K20 0.3 0.2 0.4 0.4 0.38 1.04 0.38 0.22 0.18 H20+ 0.08 0.42 0.40 0.08 0.12 H20- 1.5 2.00 1.40 2.01 3.00 2.00 TiO2 1.4 1.3 1.4 1.9 P 0 0.10 0.15 0.30 0.15 0.36 2 5 Mn0 0.20 0.20 0.20 0.31 0.33 100.04 100.08 99.99 100.33 100.00 6

ADDENDUM COMPOSITION OF THE LAUGARVATN LAVAS

Table 1. presents analyses of lavas of the Laugarvatn area (as delineated by Map 1.). Partial analyses of four pillow lavas and one postglacial lava were made by the method of atomic absorption by A.J.Thompson of Imperial College:* The five complete analyses of postglacial lavas are quoted from Tryggvason (1943) to whom reference may be made for petrographic details. The analysed specimens are as follows:- 1. Pillow lava; northern foot of Hrafnabjorg. 2. Pillow lava; southern end of Kalfstindar (Reydarbarmur). 3. Pillow lava; prominent conical mound with summit at 700m, north slope of Raudafell. 4. Pillow lava; 670m on southwest face of Hognhofdi. 5. Postglacial lava; Lamba lava, lkm southeast of southeast promontory of Hlodufell. 6. Postglacial lava; summit crater of Skjaldbreid. 7. Postglacial lava; Skjaldbreid lava lkm west of summit crater. 8. Postglacial lava; Skjaldbreid lava 9km west southwest of summit crater. 9. Postglacial lava; lava from Tindafjallaheidi crater row,lkm east of Hrafnabjorg. 10. Postglacial lava; lava from Tindafjallaheidi crater row, south tip of Kalfstindar (Reydarbarmur).

* I am grateful to Dr.J.R.Butler for permission to quote these analyses 67

CHAPTER 3

PILLOW LAVA PLATE 1 4

• f,„4 • rAlCr , 1. , •••41 *f." .14..;44. ' " • ••

C''tt 'e °a'

44•Y-''14.4• A 44141 • k)4N• .44 7444,1444 41,,,, •

-• s - ,P44611.9,0t. •

For details of plates 14 to 18 see p.78 69

SECTION 1 PILLOW LAVA AND PAHOEHOE

1. Introduction The resolution of the question of the mode of formation and propagation of pillow lava is a vitally important one for the worker in the field of submarine volcanism. For without the discipline of a concrete and sound hypothesis, coherent interpretation of the structure of volcanoes of the present oceans or of the remnants of marine basaltic volcanoes in the stratigraphic column is impossible. Lewis (1914) drew attention to what he believed to be a close structural similarity between pillow lava and pahoehoe and he inferred by analogy that pillow lava was formed and propagated in the manner of the observed digital advance of slow-moving pahoehoe flows. Snyder and Fraser (1963B) have objected that "The superficial similarity between pilloWs and pahoehoe has been un- necessarily confusing. Practically every modern textbook with a section on pillowed lavas describes an observation about the recent lava pillows actually seen by Anderson in the process of formation on Samoa. Actually Anderson's original observations and photographs make it plain that he was observing pahoehoe forming by a budding process which happened to take place at the shoreline''. I would agree that the lava depicted in Anderson's photograph of 'pillow lava chilled by seawater' (Anderson, 1910, Pl.LII) is indistinguishable from what is commonly termed pahoehoe But I strongly question Snyder and Fraser's conclusion that the similarity between pillow lava and pahoehoe is superficial. PLATE 15

A 71

2. Pillow lavas of Iceland and Wales The setting of the Icelandic pillow lavas has been described in the previous chapter. Significantly the pillow lava piles are completely undeformed and the pillow lavas can thus be seen in the positions in which they cooled and consolidated. The Welsh examples, from Strumble Head, Pembrokeshire, are Ordovician in age, marine in origin, basaltic (spilitic) in constitution, and form tabular units up to several hundred meters in thickness separated by units of pillow breccia and thin sedimentary partings. Though these pillow lavas have suffered faulting and rotation, they remain in general remarkably undeformed. Despite the differences in age and in environment of origin, the form and fabric of these pillow lavas is remarkably similar. Though it is the more regular forms which catch the eye, the range of two-dimensional form of the pillows is considerable, from spherical to amoeboid (P1.14A, 15B). Their only unifying formal characteristic when seen in two dimensions is their roundness (not to be confused with sphericity). A mutual conformity between the outlines of adjacent pillows is typical and curved surfaces show a predominant upward convexity (P1.14A, 15A). Where the fabric of the pillow lavas can be adequately seen in three dimensions (P1.14B, 16A,B), they have an entrail-like aspect, the pillows appearing as swollen protrusions, irregularly bulbous digitate cylinders, and vermicular appendices. They closely resemble the Alaskan example figured by Capps (1915) and most notably the pillow lavas of Hawaii's submarine slopes (Moore and Reed, 1963). In no locality where I have seen pillow lava adequately exposed in three dimensions have the pillows appeared to my eye to be discrete sacks, nor could I satisfy myself PLATE 16

A 73

that any I examined closely were so. Single rounded terminations are common (P1.14A), but the pillows showing them, when it is possible to trace them back, either coalesce with another or disappear into the tangled mass. In some outcrops on the flanks of the Icelandic pillow lava piles (as on Kalfstindar - Chapter 2, Section 2) the pillows are notably elongate, with long axes parallel and conspicuously inclined (P1.14B). On the Kalfstindar pile they can be seen dipping outwards from and in direct connection with an axial dyke feeder (Chapter 2, Section 2 and P1.5A,B). Pahoehoe lava is a prominent constituent of some intraglacial and extraglacial volcanoes in the Laugarvatn area and its fabric has also been examined near the source of the Askja (Iceland) 1961 lava flow. The outlines of pahoehoe toes* (P1.18), which are within the same size range as pillows, are indistinguishable from those of pillows. And pahoehoe units* display the same mutual conformity and upward convexity as do pillows, though their commonly larger dimensions make these features less apparent. Furthermore, though the shape and dimensions of most pahoehoe units in an outcrop are markedly different from those of pillows, the structural relationship between units is essentially the same. In two dimensional exposures of pahoehoe, as in two dimensional exposures of pillow lava, some units appear discrete while others merge laterally with what appear to be distinct yet connected units.

*I apply the term 'pahoehoe units' to all structural entities within a body of pahoehoe lava, and the term 'pahoehoe toes'to the smallest, typically bulbous pahoehoe units. PLATE 17

A 75

3. Mode of formation and propagation of pillow lava There is nothing in the form and fabric of the described pillow lavas that is not compatible with a process of formation and propagation analogous to the digital advance of slow-moving pahoehoe flows. Macdonald (1953) describes the process thus: "The entire front advances by successive protrusion of one small bulbous toe after another. Most toes advance only a few feet before they chill to immobility, after which the skin of the flow ruptures at some other point and another toe is sent out". Some toes "extend themselves by a process of inflation, the skin remaining unbroken, or if broken, quickly sealed again by lava squeezed from within. The front is gradually built up by the accumulation of a heap of these toes, generally 1 to 3 feet in diameter, lying alongside and on top of each other". In correspondence, J.G. Moore has raised the question of the propagation of pillow lava over long distances by such a process: "Off Hawaii, I have traced by ocean-bottom camera what appear to be single pillow flows for many miles, the top surfaces of which are made of nothing but pillows, many of which clearly bud from others The thought of a long chain of hundreds of mutually connected pillows no more than a foot or two in diameter is not reasonable, yet there does not seem to be a master channel or tube which can feed the pillows at the distal end". Structures which may well be master channels or tubes are quite common in the Icelandic pillow lavas. Mostly they appear in vertical section as roseate columnar-jointed masses, with glassy margins which mould to adjacent pillows, frequently with height greater than width, and from several meters to tens of meters in height (P1.17A). The more regular in form are very similar in aspect to the lava tube fillings in a subaerial olivine basalt flow described and figured

by Waters (1960). Less commonly these structures have the outlines of very large pillows and enclose flat-floored domical cavities underlain by an accumulation of very thin, horizontal lava sheets (P1.17B - I have seen similar bodies in pillow lava at Aci Castello at the foot of Etna). It seems plausible that flows of pillow lava of the kind described are fed by enclosed conduits at or beneath the surface of the flow, in much the same manner as many pahoehoe flows (Anderson, 1910; Macdonald, 1953).

4• Conclusion Of the rocks described in the literature as pillow lava, some (such as those described by Snyder and Fraser, 1963A) are clearly different in terms of fabric from the rocks described here, whatever the genetic relationship between these rock types may be. However many, if not most, appear identical in form and fabric to the Welsh and Icelandic examples, and I submit that digital advance in the manner of pahoehoe flows adequately accounts for the morphology of these pillow lavas. 78

DESCRIPTION OF PLATES 14 - 18

Plate 14 A, Vertical section in pillow lava. Initial orientation undisturbed. Note rounded termination near center of field. Laugarvatn area, southwest Iceland. B, Oblique section in pillow lava viewed from slightly above the field of the photo. Initial orientation undisturbed, Laugarvatn area southwest Iceland.

Plate 15 A, Oblique section in pillow lava. Initial orientation disturbed. Facing to top right corner of field. Strumble Head, Wales. B, Oblique planar section in pillow lava. Initial orientation disturbed. Facing to top right corner of field. Strumble Head, Wales.

Plate 16 A, Upper surface of pillow lava unit revealed by erosional stripping of thin sediment cover. Strumble Head, Wales. B, Truncated digitate cylinder in pillow lava. Initial orientat:'1 little disturbed. Strumble Head, Wales. Plate 17 A, Roseate prismatic-jointed lava masses in vertical section in pillow lava, Initial ._rientation undisturbed. Laugarvatn area, southwest Iceland. B, Large cavernous pillows in vertical section in pillow lava. Initial orientation undisturbed. Laugarvatn area, southwest Iceland. Plate 18 Pahoehos toes on 1961 lava flow of Askja, central Iceland. ON r4 80

SECTION 2 PILLOW LAVAS AS DEPTH INDICATORS

1. Introduction In an important recent paper Moore (1965) has described a suite of dredge samples, fragments of basalt pillow lava, from the submarine part of the east rift zone of Kilauea volcano, Hawaii. "The vesicularity and bulk density of the basalts show a systematic change with depth. Samples collected from progressively deeper water have a higher specific gravity and contain fewer and smaller vesicles". These findings give pillow lavas a new significance as environmental indicators and offer the hope that with further data it may prove possible accurately to infer depths of eruption of lava both in the present oceans and in the stratigraphic column. Icelandic intraglacial basaltic volcanoes (Chapter 2) offer a unique opportunity for the study of the products of basaltic volcanism in the shallow water environment, in particular pillow lava. These volcanoes are especially valuable in that it is frequently possible to determine with reasonable assurance the surface level of the meltwater bodies within which they grew. This section examines the vesicularity of the pillow lava of two of these volcanoes in relation to inferred depths of extrusion and uses the Icelandic and Hawaiian data to interpret the pattern of vesicularity in a sequence of Ordovician pillow lava in Wales. The methods employed in sampling pillow lava in the present study were to photograph sections of single pillows normal to the plane of flattening of the pillows at regular stratigraphic intervals and to collect oriented specimens from the top and bottom of such pillow sections incorporating the outer surfaces. PLATE 20 Sections of type 1 pillows — Raudafell—Hognhofdi

- M004111M A — depth 530m

D — depth 300m 87_ 2. Pillow lavas of Icelandic intraglacial volcanoes The pillow lavas of the volcanoes Kalfstindar and Raudafell-Hognhofdi (Chapter 2, Section 2) form the basis of this study. In these volcanoes two varieties of pillow lava, designated type 1 and type 2 in this section, are distinguished on lithologic and stratigraphic grounds. The type 1 pillow lavas, which in the field are a dull lustreless grey in comparison to the much lighter-appearing type 2 pillow lava, form the basal pillow lava piles of both volcanoes. Type 2 pillow lava ('stony' pillow lava of Chapter 2, Section 2) occurs as pods and lenses embedded in the tuff and breccia of the Raudafell-Hognhofdi superstructure, but is absent from Kalfstindar. Levels in the Kalfstindar and Raudafell-Hognhofdi piles are subsequently designated as depths relative to waterlevels at 700m and 800m a.s.l. respectively (Chapter 2, Section 2),

2.1. Depth trends in type 1 pillow lava Plates 19 and 20 show photos of sections of single pillows taken on approximately linear traverses up the flanks of the pillow lava piles of Kalfstindar and Raudafell-Hognhofdi. The vertical separation between the top and b,:,ttom pillow in each traverse is 280m and 220m respectively. These photos demonstrate a conspicuous increase in pillow vesicularity with decreasing depth, accompanied by the appearance and progressive intensification of concentric zonation of vesicles in the more vesicular upper halves of the pillows*.

*So called 'pipe vesicles' and 'vesicle cylinders' (see Waters, 1960) are common in the pillow lavas described in this section (see for instance P1.19C,D, 21B). They are very variably developed, even within pillows of single outcrops, and no systematic relationships with other pillow characters or with the pillow lava stratigraphy have been discerned. In this section these structures, termed 'pipes', are not included in the terms 'vesicle', 'vesicularity', 'vesiculation' etc. FIGURE 5

0.1 O - • 0

• rs • 0 te • • 0 0.2 • • lome ki • • • 0

• 0 r 0.3 • • • 0 te • • f wa S. h o 0.4 t • 0 • Dep NEM • • • • • 0.5 • MEM

50 40 30 20 10 0 Volume percent vesicles

Symbols: large circles — type 1 pillows: small circles — type 2 pillows 84

The trends shown in Plates 19 and 20 are characteristic of pillow lava of the basal piles. In the field it is the absence or the presence and intensity of vesicle zonation which is most readily observed. Conspicuously zoned pillows are predominant at depths of less than 350m and rare below 450m and conversely faintly zoned or unzoned pillows are predominant below 450m and rare above 350m. The inhomogeneity of vesicle distribution makes it impractical to measure the vesicularity of the whole pillow. However vesicle counts were made on specimens from the lower halves of the pillows where the inhomogeneity is less pronounced. A glass plate on which was engraved a 3mm grid was placed on the sawn surface of the specimen and centered 5cm from the lower margin of the pillow. On each half of the specimen approximately 100 points (grid intersections) were counted, exclusive of intersections on pipes, and an average vesicle percentage was obtained from the two counts. While the figure thus obtained is not the vesicularity of the whole pillow, it provides a crude measure of relative vesicularity. The graphical distribution (Fig.5), though diffuse, is consistent with the qualitative observations. In obtaining measurements of vesicle diameters only circular and elliptical vesicles were measured (the minor axis being measured in ellipses) since no meaningful 'diameter' can be measured for vesicles of less regular form. Using a steel rule graduated to 0.5mm, measurement was made on opposing surfaces of sawn specimens from the upper and lower halves of pillows. The five largest vesicles of regular form within 10cm of the pillow margin were measured on each surface and average vesicle diameters calculated for pillow tops and bottoms. Figure 6A shows the plot of vesicle diameters thus obtained. It is to be noted that vesicles are consistently smaller in the less vesicular lower halves of the pillows

g Dept h of water cn 0.5 0.1 0.4 0.2 0.3 0 Symbols: • Mi A small opencircle— topoftype2pillow large filledcircle— topoftypeIpillow N.. 11 ,

open 2.0 41111 41110-- - 10= •

•

• 1.

• le

1 0

bottom • INF 4 ik110

0 • ..•••

1.5

1 8 •=0 :

Diameter ofvesiclesinmm O

/ Cly • FIGURE 6 f

= 0 ° 0 0° 1.0 0 C;

• 0 0 0 0 0 0.5

0 2.5 2.01.51.00.50 I 0 1 86

than in the upper halves. From 550m to 300m the vesicle diameters show a clear increase with decreasing depth. A marked decrease in diameters of measured vesicles with decreasing depth above about 300m is expressed in a sharp inflection of the trend (most clearly defined by the plot of bottom diameters). Of the 10 pillow samples from depths less than 280m, 8 show prevalent vesicles of irregular form and/or a macroscopically perceptible but very fine vesicularity of the whole body of the pillow. Only 5 of the 20 samples below 280m show these characters to a pronounced degree and 2 of these are aberrant samples which plot outside the concave margin of the delineated field. In thin section the very fine vesicles are of highly irregular form and are intimately interwoven with the crystalline mesh in a fabric closely similar to Fuller's (1939) 'diktytaxitic' fabric. The -bends of increasing vesicularity and vesicle size with decreasing depth established by Moore (1965) for the pillow lavas of Hawaii reflect the response of a lava of relatively uniform composition (most importantly water content) to upwardly diminishing environmental pressure. Such presumably is the significance of the trends delineated by the Icelandic pillows, all of which are olivine basalts of very uniform character (see addendum to Chapter 2). The decrease in the diameters of measured vesicles above 300m reflects the irregularity of form and the concomitant impracticality of measurement of many or most of the larger vesicles in these shallowest pillows. This irregularity appears to be the consequence of distortion and incomplete coalescence. The 'diktytaxitic' fabric was attributed by Fuller (1939) to "the escape of the volatile constituents of the 87 final mesostasis", and this would seem an obvious interpretation of the interstitial microvosicularity briefly alluded to above. It may well be that the distortion and coalescence of the larger and presumably earlier formed vesicles in the shallowest pillows resulted in some degree from the frothing of their matrix as a consequence of the late exsolution of volatiles from the mesostasis, a process inhibited by greater external pressure at greater depths. The vesicle diameters as measured provide a strikingly close-fitting extension to Moore's (1965) curve (see Fig.6B), And the sharp bend of the composite curve at 500m parallels the sharp bend of Moore's specific gravity curve.

2.2. The character of type 2 pillow lava Unlike pillow lava of type 1, type 2 pillow lava shows no depth trends in vesicularity and vesicle size, but tends to contain fewer and smaller vesicles than type 1 pillow lava at the same depth (Figs.5, 6A); and neither vesicle zoning nor interstitial microvesicularity is encountered in type 2 pillows. Yet there is no indication of any significant compositional difference between type 2 and type 1 pillow lava (see addendum to Chapter 2: analysis 2 is of type 1 pillow lava, analyses 3 and 4 of type 2 pillow lava). The key to the differences is to be found, I believe, in the different stratigraphic environments of the two types. Type 1 pillow lava, as noted, constitutes the basal pillow lava piles of Kalfstindar and Raudafell-Hognhofdi and records a phase of effusion of lava in water which preceded emergence (Chapter 2). Type 2 pillow lava on the other hand is closely associated with the tuffs and flow- foot breccias produced during and after emergence of Raudafell-Hognhofdi from the meltwater lake. Some type 2 FIGURE 7

STRUMBLE HEAD

0.55

O"0 0 • 1.15 -0 o0 Non-pillowed lava 00 0 Pillowed and X 00 0 non-pillowed lava 0 0 C 0 o o 0 0 Pillow lava and aaa...... pillow breccia 11111 Basement

Stratification

/F Fault F / / F

0.65 0 500 m 89 pillow lava is almost certainly the product of immersion of lava flows extruded at the emergent summit vents. Most, I suspect, is the product of eruptions on the submerged flanks of the emerged volcano. This supposition is supported by the presence of a number of small conical mounds on the flanks of Raudafell—Hognhofdi which consist of type 2 pillow lava and are very suggestive of accumulations at parasitic vents (P1.12D). The characteristics of type 2 pillow lava can be readily explained in terms of partial degassing of the lava at the emergent summit, accompanied and followed by either extrusion and subsequent immersion of lava flows, or by backflow and re—emergence of lava on the submerged flanks (summit eruption followed by backflow and flank eruption is common in Hawaii; see for instance Richter and others, 1964). The relatively low volatile content of the lava at the time of pillow formation is reflected in the relative paucity and small size of vesicles and the absence of interstitial microvesicularity at even the shallowest depths. The apparent lack of any correlations with depth would be expected in terms of variable degassing.

3. Pillow lavas of Strumble Head Wales Ordovician pillow lava is superbly exposed in the vicinity of Strumble Head, Pembrokeshire, and a four— kilometer stretch of coastline southwest of Strumble Head has been examined in detail. The rocks dip NNW — NNE at angles from 35° to 70° (see Fig.7). They are closely blockfaulted, the largest of the faults (see Fig.7) having a left—lateral component of about 75m. A number of small repetitions of sequence are a consequence of this faulting. The lower part of the sequence consists very largely of pillow lava with subsidiary pillow breccia. The upper part PLATE 21

Sections of pillows — Strumble Head

A — base of sequence (yes. diam. 0.4mm) B — upper middle of sequence (yes. diam. 2.15mm)

C — top of sequence (yes. diam. 1.7mm) 91 consists very largely of non-pillowed lava, mostly in units tens of meters thick, some of which are coarsely brecciated, and less commonly in tabular units up to several meters thick with abrupt, rounded ends, closely associated with pillow lava. Thin, well-bedded sedimentary partings occur at all levels in the sequence, though more prominently towards the top. In the pillow lavas of Strumble Head the variations in size and abundance of vesicles are far more extreme than in the Icelandic pillow lavas. This contrast is most apparent when pillows from the base of the sequence aro compared with pillows from an outcrop near the top of the sequence (P1.21A,C). Yet an indiscriminate record of vesicularity or vesicle size shows no clear trend (as would be true for thc. Icelandic pillows were no distinction made between pillows of type 1 and type 2). Figure 7 records the vesicle diameters of ten pillows in the Strumble sequence. Two of these pillows show characteristics which set them apart from the rest. The pillow with vesicle diameter of 1.4,comes from a unit of pillows which differ from those of the other sampled units in having an abnormally thick spherulitic (variolitic) skin. The pillow with a vesicle diameter of 0.55mm comes from a pillow lava unit which is unique in the succession in containing abundant feldspar phenocrysts. If these aberrant samples are excluded, a pattern emerges which, with the exception of a reversal at the base of the sequence, can be readily related to the curve in Fig.6B. Starting with values of 0.65mm and 0.4mm at the base of the sequence, the vesicle diameters rise to 2.15,about two thirds of the way up the sequence and then fall to 1.651.1m at the top. That the ultimate decrease in vesicle diameter is of the same character as that shown by the shallowest of 92

the Icelandic pillows of type 1 is strongly supported by the extreme vesicularity, conspicuously irregular vesicles, interstitial microvesicularity and strong zonation of the pillow lava of vesicle diameter 1.7mm (P1.210). Until much more depth-related data on vesicularity and vesicle size is available, the only confident assertion one can make on depths of extrusion in the Strumble Head sequence is that the pillow lavas at the base of the sequence were extruded in water appreciably deeper than for pillow lavas toward the top of the sequence. On the basis of the available data it is postulated that the strongly zoned pillow lava at the top of the sequence was extruded at a depth of less than 150m_ and the pillow lava at the base of the sequence at a depth in excess of 500m and possibly as much as 1000m.

...... • • . • . • • • • • • . .

Lu cc J (.1 LL

. . . . . •... . • .. . • • . ' ...... • . .. • • • • • • . • • . • . - . • .

......

•• .. • ..' : • ...... • •

1 i i I v• CI) 01 0:3),1 + 010N 94

ADDENDUM COMPOSITION OF PILLOW LAVA

A rigorous examination of this topic would involve a detailed review and discussion of the nature and genesis of spilitic rocks, something outside the scope of this thesis. However, since Mesozoic and older rocks have commonly provided the basis for discussion (see for instance Valiance, 1965), it may be useful to review what little information is available on the chemical composition of Cenozoic pillow lavas. As the terms pillow lava and pillow have been applied to diverse structures of presumably diverse origin, the only analyses considered here are of rocks which photographs or personal experience show to be essentially the same and to fall within the definition of pillow lava given previously (Chapter 2, Section 1). Moore (1965) gives 9 analyses of 6 dredge samples from the submarine portion of the east rift zone of Kilauea. It is not certain that all are samples of pillows but some certainly are. These very fresh lavas are estimated to be no more than a few hundred years old (Moore, 1966) Noe—Nygaard (1940, 1950) and Yagi (1965) give analyses of pillow lava from three Icelandic localities. All are of Quaternary age and completely unaltered. Vuagnat (quoted in Vallance 1965) gives 2 analyses of a pillow from Pliocene pillow lavas at the foot of Etna. Gage (1957) quotes 3 analyses of lower Oligocene pillow lava at Oamaru, New Zealand. Bailey et al (1924), give an analysis of pillow lava from Mull (Eocene?). The Mull pillow lavas have suffered "pneumatolytic changes" involving"a certain amount 95

of albitisation of the basic felspars" and complete alteration of olivine to serpentine, chlorite and magnetite. Figure 8 is an alkali; silica diagram on which is marked the Hawaiian tholeiite field (Macdonald and Katsura, 1964). All of the pillow lava analyses plot within this field. 96

CHAPTER 4

EXPLOSIVE BASALTIC VOLCANISM

IN AQUEOUS ENVIRONMENTS 97

SECTION 1 BASALTIC ERUPTIONS FROM WET VENTS IN SHALLOW WATER AND IN AIR

1. Introduction As the recent eruptions of Capelinhos (1957-58) in the Azores and Surtsey (1963-66) off the coast of Iceland have emphasised, the behaviour of erupting basalt lava is strongly influenced by the eruptive environment, and the products of eruption vary accordingly. This section (1) reviews the eruptive behaviour of basalt lava in the marine shallow-water environment as observed or inferred from its products; (2) discusses the basis of the environmental influence; (3) indicates the probable depth limits of the processes involved and (4) exemplifies the influence of a shallow water environment on erupting basalt lava by reference to the constitution of Icelandic intraglacial volcanoes.

2. Eruptive activity of Surtsey and Capelinhos The eruptions of Capelinhos and Surtsey, both of which commenced in water about 100m deep, have provided the first detailed accounts of basaltic eruptions from vents which for some of their eruptive duration were quite certainly 'wet'.

2.1. Principle types of activity and their products The Capelinhos eruption was dominantly explosive for the first seven months and effusive for the last five. The Surtsey eruption (considering only the principal phase which built the island of Surtsey) was explosive for the 98

first four months and effusive for the following fourteen. The explosive phases of these eruptions were characterised by two types of activity which, following Thorarinsson and others (1964), are termed intermittent explosion and continuous uprush. Intermittent explosions have the character of sporadic bursts of 'tephra fingers' with a sheaf—like or cockstail form in aggregate, while continuous uprush is marked by persistent jetting of tephra in the form of a cylindrical eruption column. The effusive phases were characterised by the appearance of a lava lake; lava fountaining, typically of a sporadic ("Strombolian") character in the case of Capelinhos; and outflow of lava. In their ultimate state, Capelinhos and Surtsey consisted of a crudely crescentic pile of vitric ash (simple in the case of Capelinhos, compound in the case of Surtsey), the product of the explosive phase, encircling a cone (Capelinhos) or rampart (Surtsey) of spatter and partially surrounded by an apron of lava, the products of the effusive phase. At both Capelinhos and Surtsey continuous uprush assumed the nature of paroxysms, from a few minutes to many hours in length, separating generally more extended periods of intermittent explosions. The relationship between the two types of activity is depicted with clarity in Thorarinsson's diagram of the behaviour of the Syrtlingur vent during a three hour period (Thorarinsson, 1965A, Fig.6). This diagram makes it clear that the overall behaviour is not one of alternation but is cyclic, each cycle consisting of a crescendo of intermittent explosions culminating in a paroxysm of continuous uprush. While the height of the tephra fingers of intermittent explosions rarely exceeds 500m, the tephra column can rise to 2km during continuous uprush. According to Thorarinsson and 99

others (1964), the eruption columns resulting from continuous uprush contained proportionately more tephra and less vapour than the eruption columns of intermittent explosionso continuous uprush was by far the more effective in building up crater walls and enlarging the island, and caused the main tephra falls. It seems evident to me that intermittent explosion characterises periods when the lava column is relatively static or when the eruptive rate is low, while continuous uprush reflects a high and relatively constant rate of eruption of lava, an inference consistent with their relative productivity.

2.2. Relationships between eruptive activity and eruptive environment During the Capelinhos eruption Tazieff (1959) saw on the same day the birth of two flank vents, one just above and one just below sea level; the former exhibited sporadic lava fountaining (OStrombolian" activity) while the latter gave rise to intermittent explosions ("pseudo- Vulcanian" activity). Tazieff also notes that on many occasions a collapse of part of the volcanic pile resulted in an abrupt change from "Strombolian" to "pseudo-Vulcaniar° activity.' During the Surtsey eruption Thorarinsson (Thorarinsson and others, 1964) observed two vents, "the activity of the westernmost one - near sea level - being either of the continuous uprush or of the tephra-finger type, whereas the inner one, at higher level displayed ...... lava fountaining " An excellent photograph taken by Thorarinsson (Thorarinsson, 1964, cover photo) shows these two vents in simultaneous eruption, the one nearer the sea exhibiting continuous uprush and the other firefountaining. 100

Continuous uprush, it seems to me, is the behaviour of basalt lava erupting from a vent which, if dry, would exhibit firefountaining (see also Richter and Eaton, 1960, quoted in sub-section 5 below). The aspect of continuous uprush differs from firefountaining in the much greater power of the jetting, reflected in the height to which the ejecta rise, and in the much less luminous character of the ejecta, presumably a consequence, at least in part, of their smaller size and lower capacity for heat retention.

3. Location of basaltic tuff rings on oceanic island volcanoes As we have seen, the explosive phases of Capelinhos and Surtsey have left a record in crescentic piles of vitric ash - ash rings breached by the sea and the outflow of lava. In the Azores the products of a number of prehistoric, presumably Capelinhos-like eruptions form littoral tuff rings or their remnants (Hadwen and Walker, in press). In the Galapagos Islands, as noted by Charles Darwin (1876), tuff rings, "from their structure, their size and number, present the most striking feature in the geology of this archipelago I saw and received accounts of twenty-eight of these craters; of these twelv- form separate islets" while "of the remaining sixteen, some form promontories and others stand at a little distance from the shore". All the tuff rings of Oahu (17 are listed by Wentworth, 1926) stand close to the sea, as noted by Stearns and Vaksvik (1935). And nine of ten islets just offshore of Uvea in the Wallis Islands (Stearns, 1945) are tuff rings or remnants. In a coastal area of recent basaltic volcanism on the island of Tutuila, Samoa (Daly, 1924; Stearns, 1944) four cones lie on a four mile fissure 1 0 1

which transects the coast. The two cones furthest from the sea are cinder cones and mark the source of a voluminous pahoehoe flow: the two cones nearest the sea are tuff cones. A similar relationship on Oahu is described by Stearns (Stearns and Vaksvik, 1935): "A crack opened diagonally across the east end of the Koolau Range from one shore to the other. The lava from this crack that erupted through the coral produced the Koko Crater tuff cone, but on the windward side, where it broke out in the talus about 250 feet above sea level, it produced a normal flow. Unfortunately it cannot be definitely established that these eruptions took place at exactly the same time, but there is no field evidence to the contrary". Thus the distribution of tuff rings on oceanic island volcanoes confirms what the observed activity of Capelinhos and Surtsey suggests - that the explosivity of basaltic eruptions may be strongly influenced by the proximity of the sea.

4. Location of basaltic tuff rings in a terrestrial setting Terrestrial environments rich in groundwater, like the littoral marine environment, appear to favour the explosive eruption of basalt lava. Thus Searle (1965) notes that "In the Auckland volcanic district the site of the eruption appears to have exerted a determining influence as to whether the eruption would be predominantly explosive or predominantly effusive in type. Where the terrain was high-standing eruptions were predominantly effusive and scoria cones and lava flows were the principal 1 0 2

accumulative landforms produced. Where the terrain was low—lying and particularly where the surface rocks were Pleistocene silts, eruptions were explosive and explosion craters with tuff cones were the main constructional products". 103

5. The influence of extraneous water on erupting basalt lava Most of the geologists who have described the Capelinhos and Surtsey eruptions, and the products of such eruptions - the basaltic tuff rings, have attributed the explosive activity to the action of extraneous water, an inference explicit in the terms 'phreatic' and 'phreatomagmatic' commonly applied to such activity. Likewise Moore and others (1966) believe that the major explosions which followed the initial lava fountaining during the Taal eruption resulted from lakewater gaining access to the volcanic conduit. And according to Richter and Eaton (1960) the 1960 flank eruption of Kilauea Mtarted as a dazzling incandescent curtain of fire Three hours after the initial outbreak, seawater gained access to the main vent area producing dense volumes of steam, salt and fine ash clouds that roared 2000 feet into the air". Extraneous water invading fractures in the quenched crust of a lava column and flashing into steam may well be the basis of intermittently explosive behaviour. It may be that minor surges in the magma column suddenly generate fissures in the lava crust into which water may enter and thus generate steam explosions (see Jaggar, 1908). However such a process seems unlikely as an explanation of continuous uprush which is in fact not strictly explosive but is a jetting more akin to lava fountaining though much more powerful. If continuous uprush is essentially the behaviour of basalt lava erupting from a vent which, if dry, would exhibit firefountaining, then the exsolution of juvenile volatiles is presumably an integral element of the behaviour. A mechanism which would account for its more 1 0 4 potent character is outlined by Saemundsson (1967 B) who suggests that lava rising in a volcanic conduit with the sort of velocities manifested in lava fountaining will cause low pressures at the conduit walls (Venturi effect) and, in a saturated environment, a consequent influx of water, transformed instantaneously into steam,

6. Depth limits of explosive basaltic eruptions in water The two basic processes likely to result in explosive disruption of basalt lava erupting in water are steam explosion, as for instance when water invades fractures in the cool crust of still—hot lava, and explosive vesiculation. According to McBirney (1963) "sudden explosive expansion of water can occur under two possible conditions: (1) when water at a pressure below its critical point is heated through a temperature interval that crosses the boiling point for the appropriate sub—critical pressure, or (2) when the pressure on water at temperatures below the critical point is reduced isothermally through an interval crossing the vapourisation curve. Even at shallow depths neither condition is likely to apply to water exsolved from vesiculating fluid lava which will already be far above its critical temperature". However the first condition will apply to extraneous water which comes in contact with hot lava. Steam explosions can thus result only at depths less than 2160m — corresponding to a pressure of 216 bars, the critical pressure of water. Considering only the process of explosive vesiculation, McBirney (1963) states that "the depth at which explosive eruptions can occur is primarily a function of the water content of the magma". Basalts from the 1 0 5

submarine portion of the east rift zone of Kilauea contain 0.45 ± 0.15 per cent water, and this Moore (1965) believes to be the original water content of the magma from depth. Macdonald (1963) estimates that for Hawaii the amount of gas in the rising magma, of which probably two thirds or more is water, is "between 1 and 2.5 per cent during the early gas-rich phase of the eruption and between 0.2 and 0.7 per cent during later phases. The former figure is close to the maximum of 2 - 3 per cent estimated by Daly and the latter is close to that of 0.5 per cent assumed as a likely proportion of volatiles in basaltic magma by Bowen. If the initial rush of gas during an eruption represents a gravitative concentration of gas from a considerable portion of the magma body, the average gas content of the magma at depth is probably much closer to 0.5 than to 2 per cent". Tazieff (1951) has estimated a gas content of 3.1 per cent for the eruption of Nyiragongo in 1948. Using McBirney's graph (McBirney, 1963, Fig.3) which relates gas-liquid volume ratios to water depth for basalt at 1100°C, given the water content, it can be predicted that basalt lava containing 0.5 per cent water will not undergo disruption due to vesiculation at a depth of more than about 50Cm. For basalt lava containing 2.5 par cent water, which would seem a reasonable maximum on the basis of the present scanty data, the depth limit of explosive vesiculation would be about 2000m.

7. Basaltic vitric tuffs of Icelandic intraglacial volcanoes Icelandic intraglacial basaltic volcanoes illustrate with remarkable clarity the influence of environment on 1 0 6

erupting basalt lava. The basaltic vitric tuffs which are a prominent constituent of these volcanoes have been shown to be the product of eruptions from shallowly submerged or slightly emerged and presumably saturated vents. Furthermore these tuffs may be equated with a phase of emergence, for they characteristically rest on and succeed pillow lavas recording a phase of effusion in water* and are enveloped by and largely precede as and pahoehoe lavas and associated breccias marking a phase of effusion in air. Moreover they are demonstrably the product of explosive vesiculation commencing at depths of no more than two or three hundred meters in response to decreasing hydrostatic pressure at the summits of mounting eruptive piles (Chapter 2, Section 2).

8. Conclusions Explosive activity, induced by extraneous water, appears to be characteristic of eruption of basalt lava from a wet vent in shallow water or in air. Explosive generation of steam when extraneous water comes in contact with hot lava provides a satisfactory explanation of this environmental influence. The depth limits of explosive basaltic eruptions will be about 2000m for steam explosions and may be a comparable depth for explosive exsolution of volatiles, the exact value in the latter instance depending primarily on the water content of the lava.

*For the Surtsey eruption Thorarinsson (1965A) and Kjartansson (1966) have postulated an initial effusive phase with the production of pillow lava, preceeding explosive emergence. It should be stressed that for Surtsey this eruptive sequence, however likely, is supposition, not fact as the English summaries of these papers might suggest.

contorted • bedding

. • poor outcrop „. ‹, d tor ' •4- .ti .,.... , , . \••••\... • • . a. ,,;./.* • ..k...,, agglomerate 4 * • * . .. . t', • ,:41. 0 Y -.A6Ci:A.1%.4-7-.11 ••4! 7 . 4 ' 8T•••••44, ,,v \ & i•I• A k.:•.1 • .:?c : 0 , 61 II • .. % 5 1 ..• 35. 7 . . ci . ... .• •••• .7 • •• a c:. • • •• -13 1c,"`t? . lava , . , 0.; ' • .. 4 • 0 si `7. A...0 •ft:421,,i.:: *, 32. • I.. • •17. ';7 •_,, ••'/, •••47. • - i -= • • 600 '-` • ' 1 ' •••: '• ' ... . • •17 • 4 6 i: • ": . N 1 • .1*P '• 77.• .*C7 • o ',k7; ' ; .7. ' • ci• • Y ' %‘ i. r.""111.11: .: ti-•:- • .• , /4 1 •••••••41 •cl• • • . • '• -4 • • , , v 4:3 ' V. *p. • V . 1/ .0%.r. . o 0 .. 60 :• VY . .17...!!•• • • • • • g1 • . V: • • 1% • 6. V• i.. of • 4 560 • • 17 • '' I.'. . * , •„•• • • • • t:: . • A• :. •i . v. 1.11. p9 .1..v. • b "•.:' * . . .43. .• 'A lo •', , ' 75 il ••••• • : /z.;/ • ..b..• •60 • . 0 .. • /• " • •4 4.. 4. „ re ...: U / , ../..* ”4, • .- • •,),- - - „ • . • a II o itot 'o II . • i.....- ...... v..4. s.‘ " " 32", rd.. • :C/.• Cl. ..1. .. °f .'-* :9 ,...... ,. • • •`' !••••to 5,5 '‘••••6: •7esama•-" 4-11. •• ....45AL... % /5: ' :-. ' .. 4, ta 0 / .4 it •. ti ," • el ...... • li ,•.4",1 u.... 7 i,...1, ‘‘ ,, 1; • • . ' ' : '• Cl . ' 99 . • : : iiii. 5: ;° . ..'4 . :"'''' --...-' •r• ... 4 ., • 0 4, LA.... . 0 s i ...... : .7..• ,, .1.* 1...... e .7 • . . .•V •., ..1 . . 7 'n •••••2 :1..-- ••6 5 ,- • i . v. .:. ,: •;i;L 3.. • •5 :0 . .• •••••n••. •• ' •• ' . . . .(.I .A • .c ••. ..64 . Il'' l' • 2 ., " . ' :'o ','•• . • • 411. 7• /. • . • NI' i''' • . • .• V • •• F". • • .• . 1-Iii F'.,7•. .• .. ". . . - . , . :' .P.: • ••4• '• V .• • •• ki x;,••• -, . . • • . . • . 4 i ... •z• • • ...... :.. , • ••?1* ** V• • • •0 *#.• • • ...... 1... , .1.%% '. • • • • T • • 4 ••• .:.•• . ,..,. • -. • . • (12, ••• ei • •• • • . . , •A -F4. • .i::::•57.• •- ;45 '. 4 •. .. r 9 • ,,,% tt.' t : ‘ I% lava • °t. .17" , : :41: • • • . • .. • .ity.., . , ...... ft.. „ :,...... z. * " • .• 77 • : ;47-,cr ... - ... .' . • ...... •• 47.c7 . . v 4'1 ,- tuff & agglomerate .:..:-:.•.“ . • .- ey• • . b., • ,,.....: ..,-• ... • e::t? ..7. : • -Au V, •C7. . 4 .— -.. •••• 4..

• ..‹? • • • .;1 1.V: - 11,.....:•-' .... •, ,.. _ ' • • • limestone ,,,...... - • • • ' , - t•.,..-.., -- • o• - , /,' /. i ' ' ....• 0 - z• e of platy * „le" •. • is• ..:, 4,... 0 jointing x4• : . ..•••/ _... , 0,, 5 ,..• . •T• AP . ..f. / a•••• :1.,-,D, ••••••( bedding ••••:*'' ;,•• . • li f• •• . • ° P. ....s 4. 4 ,.. 0? .,. —.. ; ... r....* r_.;4. * contours in meters ",•-• '"V‘'.;.- • ' ••••= /1 - .. . .1.6'• U• ..= 1. 'w0 '*....0 . •.% 41.4• 11 •-•.-7••• - 0 200 m Z:. /e •

FIGURE 9. MAP OF S-E SLOPES OF GERGOVIA GERGOVIA PUY MARDOU 1

median sheet 3b -660m— • . • •• . • •_. . • .. . . 2c,d 3a ck• : *.;:,7 • • •e): • - to • • •• • . • ; ••‘, • ••• •fi,-: 4.:v".••• • basal 3c.d G. • • •. A • • - • • •.:3 • : • 4 • .•;r *.• :t=7 • : : . • -c7 • • • • —620m— sheet \ • 43. 'et %VP... • •••51..;:/: ' ..t>.' "7•••••.0.- 6;"-"cr:•;5:. •'6", • • :till • • v • ..v.;' 41-: 6.•:", -7> -• • ".47 • • -i3; it • 9. / * et V. 0 „ =fly sx .0 be '6A.: 0: . . II iroir A ;5. 7= 6•46.- " 0 pi 1, • • e -;;\ rzyb ill. \\—is ° ‘0, • 00' " •-• No.••• •••• •.". - -:e? : . . . • • . . • - ••.• • • • . •

FIGURE 10. GERGOVIA SEEN FROM S-E . Symbols as in Fig 9 . Photo localities indicated. 109

SECTION 2

A LACUSTRINE VOLCANO OF CENTRAL IIRANCE AND THE NATURE. OF PEPERITES

1. Introduction The hill of Gergovia, 7 km south of Clermont-Ferrand, forms the head of a promontory extending into the basin of the Grande Limagne from its western boundary. It is an erosional residual cut in lacustrine strata of Oligocene age and capped by Miocene basalt lava (Michel, 1948). 'A small hill on its eastern flanks, Puy Mardou, marks the location of a volcanic vent whose stratified products constitute much of the eastern face of Gergovia and extend westwards with diminishing thickness beneath the basalt capping. Exposure of this volcanic edifice is largely restricted to the southeast flanks of Gergovia and is an investigation of this oblique section which forms the basis of this account. The vent and its products are designated the Gergovia volcano. Gergovia is the type locality fir a class of rocks known as 'peperites'. It is the intention of this section to re-examine the nature and origin of these rocks at the type locality.

2. Structural-lithologic elements of the Gorgovia volcano Figs. 9,10,11) The disposition of the volcanic rocks can best be seen in the view of Gergovia from the southeast (Fig.10). The volcanic pile rests conformably on near-horizontal limestone strata.which form the lower southeastern slopes. The volcanic sequence is: (1) a thick basal lava sheet, Z Z ailirld (2) a lower unit of bedded tuff*, (3) a thin median lava sheet, and (4) an upper unit of bedded tuff. Puy Mardou, standing out from the slopes of Gergovia and marking the northern boundary of the southeast flank, consists basically of an axial lava body with agglomerate at its margins.

2.1. Basal sheet median sheet and axial lava body Along most of its outcrop the basal sheet has a roughly uniform thickness of about 15m. It pinches out rather abruptly at its northern end. Its contacts with underlying and overlying strata are billowy (P1.22A) and generally concordant, though at its southernmost point of outcrop the upper contact is notably discordant. Platy jointing parallel to the contacts is conspicuous in places. The median sheet is expressed topographically as the lower of two prominent terraces high on the southeastern flank of Gergovia. It outcrops only at the northern and near the southern limits of its extent. At its northern end it is about 4m thick while at its southern and it is no more than lm thick. In its northerly outcrop the lower half of the sheet is highly vesicular, much altered and impregnated with carbonate while, apart from a relatively thin vesicular zone at its upper surface, its upper half is non-vesicular. Within these vesicular zones the vesicles are not regularly disposed in relation to the top and bottom surfaces of the sheet but appear to delineate structural units akin to flow units. *In this section the term tuff is applied to fraguental volcanic rocks with maximum clast size less than lcm: 'agglomerate' is applied to fragmental volcanic rocks with maximum clast size greater than lcm.

1 1 3

At its lowest point of exposure in two small quarries on the southeast face of Puy Mardou, the axial lava body has the form of an upwardly expanding dyke about 15 to 20m thick, in sharply discordant contact with horizontal limestone (P1.22B). In these quarries it showsa remarkable development of platy jointing parallel to the contacts and extending throughout almost its entire thickness (P1.22B, C). Above the level of the quarries outcrop is far from complete, but the attitude of jointing (Fig.9) gives one a fairly good picture of the structure of the mass. On the northeast flank of Puy Mardou this body is continuous with a basal sheet similar in all respects to the basal sheet on the southeast slopes of Gergovia. The rocks forming the basal and median sheets and the axial lava body are all very similar in handspecimen; dark grey and aphanitic but for phenocrysts of olivine. The similarity is likewise apparent in thin section. All display scattered olivine euhedra in a rock consisting principally of clinopyroxene, plagioclase and opaques. These latter occur as subhedral to euhedral crystals in a mesostasis which is a brown glass containing skeletal pyroxene in specimens from near contacts, and zeolite in specimens further from contacts. A little biotite is present in a sample from the center of the axial lava body near its lowest point of exposure.

2.2. Tuff and agglomerate Bedded tuff forms the bulk of the Gergovia volcano. It occurs as two tabular units, separated by the median lava sheet. The upper unit is very poorly exposed and most of the following observations come from the lower unit. At more than 200m from the axial lava mass these it; it

• 4

41'

-4 1 1 5

units are broadly horizontal though within them bedding shows considerable undulation. At less than 200m bedding rises conspicuously towards the axial lava body and with increasing inclination up-section (15° at the base and 35° at the top of the upper unit, Fig.11). In the lower part of the lower unit however, bedding in the vicinity of Puy Mardou is highly contorted with dips of up to 90°. The most complete exposure of bedded elastic rocks is to be found on the east slope of Gergovia immediately opposite Puy Mardou where these rocks are at their coarsest - fine pumiceous agglomerates and tuffs (P1.23C,D, 24A) • Here bedding is diffuse and ill-defined. Elsewhere, both higher in the section and further from Puy Mardou, bedding is generally more clearly defined (P1.24B,C), though sharp boundaries between strata are rare. The tuffs consist of equant lava clasts which are typically largely cryptocrystalline or glassy. Though vesicles are almost invariably present, clast vesicularity varies greatly between and in some instances within samples. Clast outlines vary greatly according to the abundance and size of contained vesicles and may be ragged, scalloped, angular or crudely rounded. Forming a sort of aureole a few meters thick at all but the basal contacts of the axial lava body is an unbedded agglomerate consisting of scoriaceous lava clasts of typically ifregular outline set in a hard limestone (P1.23A;. This limestone is a matrix, not a cement, since in places the clast framework is clearly disrupted. 2_ prominent rock on the slopes of Gergovia immediately above Puy Mardou consists of a conspicuously laminated pumiceous agglomerate containing scattered rounded blocks of compact lava and of limestone. At the top of the outcrop the rock consists of ragged clasts of variably

A B 00.. .o...... -...... " -...... • ...... • ....., ...... o...... mIN , ••••••• ...OM •ff.m.• .••••. .... •Im.... a..... : 4 .. &f41::. A • :.ci .. c., ... ' . -- _.....--...... )-...... °. : 15:: V.• ....4P.':A.... .A. :Gk. • .:...t>: :. N.:.1tr. : c."7...?.'..s,...:4>. ; V, ...... •1::,'71• . ....7.0...... ,..r...„,.•.k>.. T 7 . • .(;, . . ui ...:u ..v m ii p ...; ,...P: • • • ..1;,p.- v j s v,.-. 1%, lf.... :011-,...... 0....11_ •. 4.- .9 = ..= VI • . • ..../..;.•• ..p, -57.. ...A. s„, „ it - m _ II - .:- II", •,".„.". ..- sl 4.., . -" .. • . ..(7. •...-'`-\'..;q: : ''.-:...'"*' --. .p. fp. • • 4 • % '11•717.-11 -k‘' U I/ 1... 4 • • • • • • • P. IA II \ ill ...= ..,.,! .% ...% 4. .0 q.,0„ 11:1 =. : I b .: 'Sit.*. --...... ;-;6: : '4Z.1 . ii...:11.,...‘",... ..2!"."': • ! : ::- .• . .••...... -•-• :2 -r."-. 0 1 200m ••...• :•.•'''' . • •. ..•11,..:-.• v \ •:—'7-----..••. .. . .• • •-• • . • • . • • . -

FIGURE II. SECTION ACROSS PUY MARDOU W— E. 1 1 7 compacted pumice impregnated with a carbonate cement (P1.23B) downwards it becomes more and more compact, passing imperceptibly into coherent lava at the foot of the outcrop. On Gergovia an unbroken series is shown from rocks consisting wholly of pyroclastic material (P1.24B), through rocks in which pyroclastic material is dispersed in a limestone matrix (P1.24D), to pure limestone. Despite a paucity of plagioclase in the glassy lava fragments of these fragmental rocks, there is no reason to suspect that their composition when unaltered differs significantly from that of the lava sheets and axial lava body*. However alteration is prevalent, the ferromagnesian minerals being largely replaced by carbonate, and glass phyllosilicate and carbonate.

2.3. Limest-.ne The limestone underlying, overlying, interbedded and intermixed with the volcanic material is very variable in its macroscopic characteristics; white, grey, pale buff and pale green in colour; soft and chalky in places, hard with a conchoidal fracture in others; unbedded or very indistinctly bedded in some outcrops, distinctly bedded in others. The limestone strata are tabular with the exception of minor nodular and lenticular masses of very hard flinty cream and dark grey limestone.

*Lichel (1953) places great emphasis on what he believes to be valid chemical differences between the lava of the peperites and of the associated lava sheets and dykes at Gergovia and elsewhere. I am dubious of the significance of these differences, particularly as all the peperites analysed are considerably altered; olivine is pseudomorphed by calcite, and vesicles contain calcite, chalcedony and zeolites. Michel himself notes that "the peperite outcrops are always rich in zeolites". 1 1 8

Those specimens which exhibit least diagenetic modification consist of finely granular carbonate containing minute rods 0.005 to 0,01mm in diameter and up to 0.25mm long, thought to be the tubular moulds of the thalli of filamentous algae. Several samples, notably of the limestone matrix of the agglomerate adjacent to the axial lava body and the nodular limestone, contain large numbers of hollow spherical bodies ranging from about 0.01 to 0.08mm in diameter, thought to be moulds of coccoid algal cells (also identified by Dangeard, 1931, in limestone from another locality at the same stratigraphic level in the Limagne Basin). The limestones are very similar in their microscopic characteristics to the algal tufa from Mono Lake described and figured by Scholl and Taft (1964). Golites and pisolites, commonly with a lava or mineral fragment as a nucleus, are present in these limestones. It is important to stress the complete stratigraphic continuity between the volcanic products and the limestone, both vertically and laterally. Though the products of volcanism are predominant in the vicinity of Puy Mardou, southwards they become progressively less abundant and more dispersed in a limestone matrix, and tuff-free interbeds of limestone become more numerous.

3. The character of the Gervia volcano It is quite clear that Puy Mardou marks the site of a volcanic vent. It consists basically of an upwardly expanding lava mass of funnel-like form in profile (Fig. 10) which at its lowest point of exposure has the form of a markedly discordant dyke. The fragmental volcanic rocks 1 1 9

are at their coarsest along the contacts of this mass and decrease in coarseness away from it. Bedding in the fragmental rocks becomes notably steeper and is at its most contorted in the vicinity of Puy Mardou. But for the common presence of a limestone matrix, the fragmental rocks are in no way different from the ejecta common to explosive basaltic volcanoes and, though exposure is inadequate to demonstrate whether or not a crater was present, the form of the axial lava mass is very suggestive of an upper vent and crater filling. High dips and general contortion of bedding in the vicinity of the vent,such as occur in the vicinity of Puy Mardou, are common characteristics of such volcanoes. The median sheet shows no intrusive characteristics and, as its structural inhomogeneity suggests, is almost certainly a flow. The nature of the basal sheet is less certain but its structural homogeneity is consistent with intrusive emplacement. However there is no reason to doubt its contemporaneity with volcanism and sedimentation. It is petrographically the same as the rest of the Gergovia volcanic rocks and its billowy margin is characteristic of a lava/wet sediment contact. If, as there seems no reason to doubt, the Gergovia volcano is basically a simple pile of basaltic ejecta, then the nature of its association with limestone can only be explained in terms of carbonate precipitation concurrent with volcanism. The exact nature of the lacustrine environment at Gergovia can be established from paleontological data. ILmong the abundant fossils recorded from the vicinity of Gergovia are nests of tortoise eggs which evoke for this stratigraphic position either low ground separating lacustrine depressions or conditions of temporary emergence during sedimentation (Michel, 1948). cn--) C C

.."•• • 4' • I '\`‘‘;`,. • 1,*')' • -1- \.\ • f•-.-

. .

FIGURE 12. INFERRED CHARACTER OF GERGOVIA VOLCANO 1 2 1

I envisage the Gergovia volcano as an ash ring rising a few tens of meters above the surface of an extremely shallow lake (Fig.12), and contributing to the steadily accumulating lime mud the products of sporadic explosive eruptions. At least once if not periodically during the period of activity the lakewaters must have receded completely from the area of the vent, for the lava flow which constitutes the median sheet gives no evidence of having flowed into or under water. From the Carson Desert of Nevada, Russell (1885) has described two well-preserved basaltic tuff rings with a history very similar to that postulated here for Gergovia. These tuff rings, the flanks of which merge imperceptibly with the desert surface, have 'flat' profiles of the kind which the Ger.govia section (Fig.10) indicates for that volcano. In the crater walls of the larger tuff ring, 50m at their highest, are "well exposed sections of stratified lapilli, mingled with an abundance of angular grains, kernels, and masses of basalt, some of which are two and three feet in diameter and scoriaceous, especia]ly in the interior Interstratified with the lapilli occur manly lake beds containing fresh-water shells and dendritic tufa Both the lapilli and the lake beds are evenly stratified, and exhibit diverse dips ....' A figured section of the crater wall contains about 15ra of "laminated marl containing lapilli" interbedded with 35m of "laminated lapilli". The Soda Lake craters are situated within the basin of the geologically recent but longer extant Lake Lahontan. Russell reasons that "from the presence of fossiliferous lacustral clays in the midst of lapilli, it seems evident that volcanic eruption was interrupted by periods during which the lake covered the craters". 1 2 2

4. Gergovia and peperites The history of investigation of Gergovia and most notably of the speculation of which it has been the source, goes back nearly one hundred and fifty years. Scrope visited the Auvergne in 1821 and agt.in in 1857. Of Gergovia he says (Scrope, 1858); an "alternation is to be seen of beds of limestone and marls with others containing numerous fragmentary volcmic matters, often in such abundance es to compose far the greater part of the rock and give it the character uf a peperino". In reference to a similar outcrop elsewhere, he remarks: "The strata interspersed with volcanic [-letter are perfectly parallel to those that are free from it, and present all the characters that we should expect in sediment slowly and tranquilly deposited in a body of water into which repeated showers of volcanic ashes and fragments were occasionally ejected from some neighbouring volcano in active eruption". According to Jung (1946), Scrape's ideas on the origin of the "calcareous peperino" or "peperites" as these rocks are termed by French geologists, were 'unanimously accepted by his successors and particularly by Pomel in 1844 with respect to the peperites of Gergovia. But in 1896, as a result of a meeting of the Geological SoCiety of France in Auvergne, Michel Levy ...... produced a new hypothesis. According to this the peperites were internal explosion breccias belonging to volcanism of Pliocene or Quaternary age. Opinions were thenceforth divided for a long time The question was, in fact, at its most confused when, in 1927, in a note relating to Indochina, A. Lacroix and F.Blondel announced new observations of foremost importance. These geologists were able to show that a peperite sioklar to those of the Auvergne had formed as a result of the intrusion uf 1 2 3

basaltic magma into diatomaceous ooze u (my translation). According to the most recent and comprehensive paper on the subject (Michel, 1953) in which Gergovia maintains the status of the type locality, the peperites of the Limagne are "rocks formed by intrusion, fragmentation and mixing of an andesitic lava in the still-plastic sediments of the lakes of the upper Staapian" (my translaticn). It will be apparent, even from this brief account, that peperites in general and the geology of Gergovia in particular have long been subjects of controversy. It would be neither practical nor profitable to attempt to evaluate here all that has been written on the geology of Gergovia. However I will briefly examine the views of Michel (1948, 1953) whose notions on peperites in general and Gergovia in particular, as far as I am aware, are current at present. In his paper on Gergovia, Michel (1948) rejects Scrope's hypothesis, reaffirmed in this study, on the grounds that it "explains with difficulty, at least at Gergovia, the formation of an enormous mass of peperites; it would be necessary to admit what seems improbable, that the ejectims must have gone on at the same time as sedimentation for a very long time. Besides, even in this case, sorting of the ejecta would take place, the heaviest sinking most deeply in the mud, something which is nowhere seen" (my translation). Michel's first intuitive objection can be countered by reference tD the history of the Soda Lake craters. And I can see no reason for accepting his arbitrary criterion of projection, based as it appears to be on unsupported assumptions. 124

In his subsequent paper Liichel (1953) advances as further evidence against the hypothesis of explosive projection "the vitreous state of the volcanic particles" and ''the absence of any surface flows associated with peperite outcrops". But vitreous clasts are a characteristic product of explosive basaltic volcanism, and at least one lava flow (assumed to be intrusive by Michel) is present at the type locality Gergovia.

5. Conclusio7 In proposing his hypothesis cf intrusion/brecciation, Michel-Levy (1890) concluded that "the question is still not definitely resolved" and that "it is prudent not to express too definite an opinion on the subject". Yet despite this hesitant advocacy, his hypothesis has shown signs of attaining the status of fact. In their terminological review, Wentworth and Williams (1932) wrote, concerning peperites: "These breccias were long ago recognised by Michel-Levy to result from the rapid chilling and disruption of basaltic magma intruded into wet Oligocene sediments " More recently Snyder and Fraser (1963A) have written: "Peperites have been universally ascribed to intrusion of magmas into incoherent, poorly consolidated or moist sediments". The fact that the geology of Gergovia has given rise and continues to give rise to diverse and contradictory interpretations and opinions should serve as a warning against the uncritical acceptance of an hypothesis of intrusion and brecciation of lava in wet sediment for rocks thought to be similar to the peperites of the Auvergne. 1 2 5

CHAPTER 5

IMMERSION OF BASALT LAVA 1 2 6

SECTION 1 OBSERVED IMMERSION OF BASALT LAVA

The immersion of basalt lava has been observed on many occasions and descriptions are perhaps more numerous than for any other aspect of basaltic volcanism in aqueous environments. Nevertheless the amount of geologically significant data contained in these accounts is to my mind very limited. The principal limitation of such accounts is the obvious inability of the observer, except in very rare circumstances, to view what happens to the lava once immersed. And in many instances the dangers of observation clearly must prevent close and careful observation even of processes occurring partially or wholly in air. Furthermore I suspect that the more striking explosive phenomena have caught the eye of most observers more than have less conspicuous but to my mind more significant phenomena. The general aspect of immersion varies enormously. The 1859 lava of Mauna Loa, which flowed into the sea for more than five monthp, was seen "quietly tumbling into the sea over a low ledge, perhaps six to eight feet high We experienced no discomfort whatever when we rowed within a few yards of the molten cataract, except from the heat; there were no gases but the rising steam, and nothing was thrown up' (Green, 1887). On the other hand the lava of the 1840 eruption of Kilauea "rolled down with restless energy to the sea, where, leaping a precipice of forty or fifty feet, it poured itself in one vast cataract of fire into the deep below, with loud detonations and fearful hissings, and a thousand unearthly and indescribable sounds The atmosphere in all directions was filled with ashes, spray, gases ...... " (Titus Coan, quctod in Brigham, 1909). 1 2 7

Immersion may result in instantaneous fragmentation as Einarsson (1966) observed during the Surtsey eruption. "On March 20th, 1965, I observed lava flowing into the sea and saw how it was completely split up into small fragments on contact with the water. A lm broad tongue flowed pretty rapidly down to the beach during retreat of the sea. The next wave overran the frontal part of the tongue and changed it completely into small fragments that ran down the beach with the backwash. In this way the lava was constantly cut off at the beach, being changed into glass fragments " On the other hand newly formed pahoehoe observed at sea level and awash by Anderson (1910) apparently remained intact. Referring to the flow of lava into the sea during the Surtsey eruption Thorarinsson (1965B) remarks; "From the very beginning of the lava eruption it has been striking how much of the lava becomes fragmented when coming in contact with the sea water. Explosive pseudoeruptions on a small scale could frequently be observed. A great amount of coarse, more or less glassy sand was formed because of fragmentation of the pahoehoe lava surface by rapid cooling, and also by the crushing effect of the breakers and the grinding of blocks by the surf. The sand thus formed builds up a collar of sand in front of the lava flow so that at low tide one can walk on a sandy beach in front of the advancing lava. The lava is thus constantly advancing over a layer of wet sand at sea level". Of the phenomena which may be associated with immersion, littoral explosion is the most commonly described. For instance the Keekee flow of the 1955 eruption of Kilauea (Macdonald and Eaton, 1964) entered the sea to the accompaniment of many small explosions which "were not 1 2 8

single puffs, but pulsating jets that continued for several seconds or minutes" and which "threw up columns of black ash—laden steam to heights of 50 to 100 feet". These pulsating jets presumably result when water trapped in quantity in and beneath the flow expands into steam and forces an exit (see Jensen, 1906). It is persistent jetting of this kind which builds littoral cones of glassy ash as much as 250 feet high. Nevertheless "only a small amount, probably never more than 5'j:), of the lava is thrown into the air and back on land by the steam explosions" (Moore, 1965), and, in terms of their ash yield, littoral explosions are probably of minor significance in the total context of immersion. Hawaiian geologists have emphasised a relationship between the occurrence of littoral explosions and the surface characteristics of flows. Thus Finch and Macdonald (1953) note that "Many aa flows entering the sea cause violent steam explosions which throw into the air large amounts of ash and cinder. This happened in 1840, 1868, 1919 and 1926. Many pahoehoe flows enter the water quietly with only moderate evolution of steam Macdonald (1944) ascribes the commonly explosive immersion of aa flows to the fragmental nature of their surface which "allows the water easy access to the highly heated, still fluid interior of the flow, and presents to the water a very large surface area of hot lava. Explosively violent generation of steam is the result". Steam explosions, presumably because they attract attention, are most frequently called on to account for fragmentation of lava during immersion. Yet it seems to me that thermal stress is likely to be more important, though to isolate it as a process is doubtless artificial in that its effects are almost certain to be accompanied 1 2 9

and enhanced by steam explosions. Significantly "chilling itself does not produce permanent strain in a glass body provided that the chilling occurs while the body isso hot and fluid that stress is relieved as quickly as it develops" (Carlisle, 1963). This provides a satisfactory explanation for seeming anomalies of behaviour in the granulation or continuing coherence of basalt lava on immersion, instances of which have already been described. Thorarinsson (1967) comments that "in Surtsey the relative amount of crumbling depended much on the rate of flow. When the flow was great, the surface lava flow did n..Dt stop at sea level, but continued down slope, but slowly flowing lava tongues were to a great extent crumbled to coarse sand when entering the surf zone". It may well be that the observed relationship between rate of flow and degree of granulation was not one of cause and effect but rather that both were governed by the varying fluidity of the lava. During the 1950 eruption of Mauda Loa it was observed that fast-flowing incandescent lava entered the sea with the generation of very little steam, while semi-solid blocks carried in it produced large bursts of steam. Finch and Macdonald (1953), who record the observation, inferred that the solid blocks disrupted the otherwise smooth, impervious skin of the lava and gave sea water access to the hot interior of the flow, resulting in a steam explosion. However it is equally possible that these steam bursts accompanied the shattering of the blocks which, unlike lava itself, were insufficiently fluid for stress relief by viscous flow to take place. 1 3 0

There is a common reference in geological literature to pillow lava observed in the process of formation by Anderson (1910). A recent discussion of the mode of submarine volcanism (Menard, 1964) contains the following account of Anderson's observations. "A small, slowly moving flow entered the sea with few steam explosions, and the water remained so calm that the flow could be seen progressing down the sea floor. Protrusions, or pipes of lava grew out from the front of the flow and shortly swelled into the typical sacklike forms of pillow lava". A brief reference to Anderson's description is sufficient to show that this is pure fantasy. Not only did the observed events take place largely in air, as Anderson makes quite clear, though the hot lava was periodically washed by the surf, but it is evident from Anderson's Plate LII (as pointed out by Snyder and Fraser, 1963B) that what he observed was simply the bulbous advance of pahoehoe, seen by chance at sea level. There is, to my knowledge, no unequivocal account of the observed formation of pillow lava resulting from the flow of basaltic lava from air into water. 1 3 1

SECTIoN 2 EMERSION OF BASALT LAVA - ITS STRUCTURAL EXPRESSION AND STRATIGRAPHIC SIGNIFICANCE

1. Introduction The consequences of the flow of basalt lava from air into water are such that immersion leaves a distinctive structural record of the water level of the time. Such records obviously have enormous significance for reconstructions of volcanic activity in aqueous environments and for stratigraphy generally. Yet, though such records must be common, not only in active or relatively young volcanic areas, but also in the stratigraphic column, they have received barely any attention since Fuller (1931) first realised their significance. It is the intention of this section to review the nature of immersed basaltic lava flows and to discuss the significance of extended records of immersion.

2. Constitution and structure of immersed basalt flows There are to my knowledge no published descriptions of the subaqueous constitution and structure of historic basalt flows which have entered water, and it is to instances of inferred immersion that one must turn for information. According to Waters (1960), the margins of the Columbia River basalt furnish hundreds of examples. "When flood basalts build an extensive lava plain there is sure to be much disruption of the former drainage. Streams are ponded against the margin of the lava flood and shallow lakes spread widely over the edges of the newly congealed lava. Flows from the new eruptions pour into these marginal lakes, filling them with a complex of pillow lavas 1 3 2

and granulated basalt glass". The most complete descriptions of these complexes are those of Fuller (1931, 1934) who first recognised their origin. Some magnificent examples of probably marine immersion have been recently described by Nelson (1966) from James Ross Island, Antarctica. And innumerable instances of basalt lavas which flowed into meltwater ponds are to be found in the Icelandic intraglacial volcanoes (Chapter 2, Section 1). In the Columbia River region both breccia and pillow lava*, sometimes separately, sometimes together, constitute the originally immersed portions of flows. In the Laugarvatn region and in the James Ross region, breccia is the characteristic product of immersion. Basaltic 'flow-foot' breccias, as the fragmental products of immersion are termed in this thesis, typically consist largely of fragments of basalt glass (sideromelane - or its alteration products, most notably palagonite). Crystalline elements (not all are fragmental) typically have a partial or complete glassy crust. The defining characteristic of these breccias is structural continuity with overlying sheet lava ('subaerial' lava). The zone of transition between breccia and overlying sheet lava, the 'passage zone' as it is termed in this thesis, is planar, and horizontal when undisturbed (Fig.l). It is characterised by attenuated sheets and fingers of closely jointed, glass-encrusted lava, often of 'pillowy' aspect, which diverge from the base of the sheet lava with rapidly increasing inclination and pass down into breccia with which they commonly merge by fragmentation. The more extended of these sheets and fingers, and the glass- encrusted, often 'pillowy' lava masses resulting from their

* 'Pillow lava' denoting lava of the kind figured in Chapter 3, Section 1. 1 3 3

fragmentation, characteristically define a steeply inclined (up to 40°) stratification within the breccias which Fuller (1931) likened to foreset bedding. This stratification is conspicuous in extensive outcrops cut normal to its strike, but is frequently inconspicuous in less extensive and otherwise oriented exposures. Fuller's fertile analogy with deltaic structures, which will be developed in this section, is the basis for the subsequent reference to the sheet lavas as topset lavas or topsets and to the flow- foot breccias as foreset breccias or foresets. Following Fuller (1931), the lithologies and structures described in the preceeding paragraph are attributed to the advance of lava at water level over a bank of granulated glass resulting from its immersion. The passage zone between sheet lava and breccia is not invariably horizontal. In an exposure I have examined in the Laugarvatn area, breccia foresets terminate upwards in a vertically transgressive zone of interdigitation of breccia and sheet lava, and similar relationships are described by Nelson (1966, p.40). Possible reasons for such a relationship are more appropriately discussed in the following sub-section. In its broad aspect the structural relationship between lava topsets and breccia foresets is strongly suggestive of angular unconformity (P1.3B: see Nelson, 1966, Pl.VII,c) and has probably been mistaken f._r such in the past. As Nelson (1966) emphasises, this relationship is essentially the outcome of constructional, not destructional processes. Yet it is clear that in some instances, especially in an open marine environment, destructional forces must play a part. In such instances the junction between lava topsets and breccia foresets will be locally, not a transition but an erosional discontinuity. 1 3 4

In a cliff section described and figured by Nelson (1966; p.39 and Fig.4) topset lavas have been extensively stripped from underlying breccia foresets at the level of the passage zone, presumably by marine erosion. In the intraglacial volcanoes of the Laugarvatn area the dip directions of breccia foresets in relation to the centers of the piles (which coincide with or approximate to the main eruptive centers) commonly vary between tangential and radially outward. Rarely they are inwardly directed. Mathews (1947) has made similar observations on intraglacial basaltic volcanoes in British Columbia. For the James Ross region Nelson (1966) records that "the dip direction of palagonite-breccias varies in an ordered and radial sense and was commonly followed through a change in direction of almost 180° of arc Such variations are, of course, typical of the foresets of sedimentary deltas and are equally to be expected in the foresets of lava/breccia deltas if one considers the process of immersion in plan view. Nelson suggests that "If the points of outflow of certain lava streams were maintained for some time, then palagonite breccia fans would probably be built out seaward in a manner comparable to deltaic sedimentation at the mouth of a river". A fan of this nature has almost certainly been constructed beneath the lava fan of Surtsey (see Thorarinsson, 1964, plate 46 and section V,p.27) as Nelson points out. Furthermore it has been observed (Thorarinsson, 1964, p.49 and plate 35) that basalt lava, on reaching water, is deflected along shore in its advance. Such a pattern of flow might be expected to result in a pronounced development of longshore or 'tangential' foresets. Waters (1960) has called attention to the potential utility of breccia foresets as an indicator of direction

FIGURE 13

A C. ,,a C C , C C ( , C ..) C . C 1 t C .) C C. C c ...) C C CCC pt >> C... C C c 7 C C. t C C C. C ', c 1 , 'b .1 c a c C , c 1 t c / C c 7 c., C c c C. (03 c e C. c C. C. c c_. C (.. /- 7 1 C c.. 7 CC .. t. C-70 -3C 7 C 7 c7C C-C '" C 3 L C e -7 ‘ C. le t c C '7 C e cn C ' C c '3 2 C.. t7 C il e" C7 -7 7 'ti." C C C°C... . 7 C ,. C 7 C.-2 C3CC 4... C., C,....3 C C. C. C-C 22 C7 3 C C CC - C -JO J C, cCc- ...-... 2 C. C

A. Rising waterlevel

B. Falling waterlevel

C. Fluctuating waterlevel

C C C 1/ C C c, c 3 C") .. C c to 3 Ccc, C 'e, CC `-' C. `-'e --) c c A C C e. D , C 7 3\c c > c- c. ..,` C. C C fi e, C c c \ c ,..) _ 1/4.. c ,„. --> t, c c C 1 C c_, -) c e, c 3 c. -3N-k., C C \ C- 3 1- -% C L . c C .3 C.. '' C c C - C.- '? c c 0 e, C. ") C. C' c -.. C.- c_., C - , C c 7 ? -. c c.. 1 C C C -- ) 'D c C , C C n C ? \ 1 C- c C. C- C ? c C. c

D. Static waterlevel and collapse

Symbols sheet lava — black: flow—foot breccia — lunate

passage zones — denticulate 136

of flow, such indicators being of particular importance for instance in locating otherwise unidentifiable lava sources. It will be apparent from the preceeding paragraph that, unless numerous observations can be made, the inferences that can be placed on the attitude of breccia foresets will be very limited (a qualification which is of course applicable to any indicator of flow direction).

3. Extended records of immersion There can be few volcanic environments in which water levels are static for long. The basaltic volcanoes of the ocean basins show a marked propensity to subside, while in the island arc environment the relative positions of land and sea are constantly shifting. It is to be expected that extended records of immersion will be correspondingly complex. One can readily envisage a variety of structural configurations which may result from relative movements of water level and a volcanic pile or terrain during the course of volcanism in an aqueous environment. Intermittent eruptions during a period of rising water level (Fig.13A) may be recorded in superposed flow-foot breccia units and a rising sequence of passage zones. Overlap, and associated abrupt changes in thickness of units of flow-foot breccia (see unit 3, Fig.13A) may be anticipated. Intermittent eruptions during a period of falling water level (Fig.13B) may be recorded in contiguous, onlapping flow-foot breccia units and a falling sequence of passage zones. A fluctuating water level (Fig.13C) may result in a combination of overlap and onlap and a sequence of passage zones of fluctuating altitude. The same structural relationships may of course record a static water level and a subsiding, rising or oscillating volcanic pile or terrain. 1 3 7

Structural relationships within and between the intraglacial volcanoes of the Laugarvatn area are of the kind illustrated in Fig.13A. (see P1.3B) and are believed to record periodic volcanic activity within an ice sheet of increasing thickness (Chapter 2, Section 1). A fine example of structural relationships of the kind illustrated in Fig.13C. is described by Nelson (1966, p.38 and Fig.4) and attributed to a temporary reversal in an overall trend of subsidence during volcanism in a marine environment. The interpretations which can be placed on vertically transgressive passage zones of the kind briefly referred to in the previous sub-section will depend on their lateral extent. Where such a relationship between sheet lava and flow-foot breccia is regional rather than local one might reasonably infer a rapidly rising water level during the period of immersion, especially in an environment such as the lacustrine where waterlevels are frequently unstable and subject to rapid variation. Where such a relationship is local in extent, a probable explanation to my mind is repeated local marginal subsidence of topsets and foresets (Fig.13D), perhaps under the weight of constant increments of lava (see Waters, 1960, p.362). Though it cannot be demonstrated, this is thought the most likely explanation of the Laugarvatn example previously alluded to, since collapse of topset lavas, often involving rotation, is prevalent in the Laugarvatn area. Cne such block was seen to have been overrun by subsequent lava and breccia (as shown diagramatically in Fig.13D). 1 3 8

CHAPTER 6

MARINE BASALTIC VOLCANOES FIGURE 14

product of successive

vitric tuff of emergent explosive phase 1 4 0

SECTION 1 STRUCTURAL MODEL OF A MARINE BASALTIC VOLCANO

The structure and constitution of the Laugarvatn volcanoes (Chapter 2) provides the basis for a structural model of a basaltic volcano in an aqueous environment. In this model a basaltic volcano which has grown in water whose depth precludes explosive eruption of the magma in question will consist simply of a pile of pillow lava. In 'the case of a very shallowly submerged or slightly emerged volcano the pillow lava pile will be capped by tuff. A volcano which has established itself above water level will consist of a pillow lava pile surmounted by a shield of sheet lava on a pedestal of flow—foot breccia with a core of tuff (Fig.14B). In the subsequent sections of this chapter this model is evaluated in terms of what is known of marine basaltic volcanoes. 1 4 1

SECTION 2 MARINE BASALTIC VOLCANOES 'HIGH AND DRY'

1. Introduction Judging from the common occurrence of basaltic volcanoes within present mobile belts, one might expect to find numerous examples of basaltic volcanoes which have grown partly or entirely beneath sea level and which, as a result of subsequent tectonic events, now stand 'high and dry'. Basaltic volcanic rocks which have been erupted or deposited below sea level are certainly prominent in mobile belts, both active and no longer active. But recognisable volcanoes (in the sense of volcanic piles whose form and/or structure are largely preserved and exposed) are either very rare or, more likely, remain unrecognised and undescribed. The following section describes five examples uncovered by an extensive literature review.

2. The Troodos massif, Cyprus The Troodos massif (Cyprus Geological Survey, Memoirs 1-7; Gass and Masson-Smith, 1963) is a basic igneous complex of Cretaceous age. In plan it is an elongate, roughly rectangular body 50 miles long and 20 miles wide. In its present much eroded condition it rises from sea level at its lowest point of exposure to over 6000 feet at its centre. Pillow lavas characterise the massif at its perimeter where dykes account for less than 5 percent of the outcrop. Inwards the dykes increase in concentration to the point where the pillow lavas are reduced to narrow screens and ultimately vanish altogether. 142

The resulting sheeted complex is the most extensive facies of the Troodos massif and it in turn envelops a core of coarse grained basic and ultrabasic rocks. Intercalated in the pillow lavas are laterally impersistent, crudely tabular sills and more irregular lava bodies and "agglomerates" - unsorted rocks which commonly contain pillows in varying states of disaggregation and which grade into adjacent pillow lava. Alteration increases with depth within the massif. Sediments rich in iron, manganese (sometimes as nodules), and silica ("umberiferous shales, radiolarites, jaspers and manganiferous chorts") occur sporadically both within the pillow lavas and, more conspicuously, at their perimeter. Geologists of the Cyprus Geological Survey recognise and map a number of units within the Troodos massif, the principal divisions being (from center outwards) the plutonic complex, the sheeted intrusive complex or "diabase", and the pillow lava series. Unfortunately, it seems to me, the creation of these divisions, and the tendency to view them as ordinary stratigraphic units, has tended to obscure the evident broad structural unity and simplicity of the massif. Gass and Masson-Smith (1963) go so far as to place a "strong unconformity" between the sheeted complex and the pillow lava series though there is no unequivocal or even plausible evidence in any of the Survey memoirs for more than very local unconformities. The Troodos massif appears to be the remains of a single large marine basaltic volcano, a pileof pillow lava comparable in its exposed extent to the island of Hawaii. The nature and relationships of its various rock types as described in the Survey memoirs are consistent with such a view; a carapace of extrusive rocks (pillow lava) merging inwards through a zone of complex interrelationships to a core of coarse grained 1 4 3

rocks and a progressive inward increase in dyke concentration and alteration.

3. South Guam Island The southernmost portion of the island of Guam (Tracey and others, 1964), one of the Mariana group, consists almost entirely of basaltic rocks which outcrop in an oval area of about 40 square miles. They form a dissected cuesta that slopes 5 to 15° northeast. Pillow lavas with an exposed thickness of about 1400 feet form the lower part of the sequence. They outcrop on the steep western slopes of the cuesta. Intercalated in the pillow lavas, especially in the upper 500 feet, are tuffaceous beds and lenses of limestone. Overlying the pillow lavas and underlying the extensive eastern slopes of the cuesta is a blanket of pyroclastic rocks up to 700 feet thick. Exposures of this unit along the steep upper western slopes of the cuesta are almost wholly of "tuff breccia" consisting of unsorted, angular to subrounded lava fragments, similar in composition to the underlying pillow lava, in a predominant "sandy tuffaceous matrix'. "Bedded tuffaceous sandstone (vitric tuff according to the authors) and lenses .oFvolcanic conglomerate'" are Common in the east. Small erosional residuals- of basaltic lava flows overlying the pyroclastics are scattered over the gentle eastern slopes of the cuesta from the center of which .is /recorded a deeply weathered volcanic plug (Corwin and Tracey, 1965). The authors of the Guam memoir do not attach any particular significance to the sequence and other structural relationships of the Miocene volcanic rocks of South Guam. However their description of the nature and disposition of the various rock types is consistent with a single volcano 144

consisting of a pillow lava pile, with interbedded tuff near the top, surmounted by a flat-topped pile of flow-foot breccia(?) enveloping tuff, and in turn capped by sheet lava residuals (see Fig.14B). It is interesting to note that the limestone lenses containing reef material appear at the same level as the tuff interbeds in the pillow lava pile, presumably recording the establishment of reefs on the shoal flanks of the emerging volcano. If this interpretation is correct, subsequent events have included tilting of the entire volcano giving the present cuesta form, and erosional stripping of most of the capping lavas.

4. North Auckland Remarkably similar to the description of the South Guam massif is Brothers' (1965) brief description of the elongate mass 8 miles long which forms the Maungaru Range of North Auckland, New Zealand. "The range is tilted to the west, as can be judged from topographic form and westerly dips on intercalated sediments. Thus the deeper parts of the sequence are exposed along the eastern margin where fairly structureless basaltic pillow lavas have been intruded by a later suite of gabbro, and pyroxene and hornblende dolerites. The middle part of the sequence is characterised more by pillow lavas and massive basalts or dolerites that are exposed in the steep headwaters of creeks draining the eastern face. The more gentle westerly backslope of the Maungaru Range is built mainly of volcanic breccia with abundant glassy fragments and intercalated sediments. The same pattern appears in the Tangihua Range

The Maungaru and Tangihua ranges are two of a number of such massifs consisting of pillow lava, breccia and tuff 1 4 5

which occur within an area of widespread Cretaceous rocks, including greywackes and red and green siliceous mudstones and cherts. These massifs are generally irregular in plan and are as much as 25 miles in maximum extent. They are several thousand feet in height or thickness, and their flanks have slopes of up to 40° (Quennel and Hay, 1964). Quennel and Hay have suggested that these massifs may be of the same nature as seamounts, a suggestion elaborated by Farquhar (1966).

5. James Ross Island, Antarctica Reference has already been made to James Ross Island, Antarctica (Nelson, 1966), a Miocene basaltic volcanic pile comparable in area to the island of Hawaii. An extensive icecap obscures much of the 5000 ft high pile but exposure round the perimeter of the ice is sufficient to reveal its structure. It consists of a series of sheet lava/flow-foot bz.eccia units, attributed to the immersion of flows during intermittent phases of volcanism in an area of subsidence (see volcano superstructure in Fig.14D). Subsequent uplift has raised once submerged rocks to heights of 3000 ft and possibly 5000 ft above present sea level.

6. Santa Maria Island, Azores Though it can scarcely be classed as 'high and dry', the Azores volcano of which the summit forms the island of Santa Maria is appropriately mentioned here. The emergent portion of the volcano (Hadwen and Walker, in press) consists of an apron of basalt sheet lava and flow- foot breccia encircling and resting on an eroded founda,„Luli 1 4 6 of older subaerial basalt lavas. Miocene limestones rest on this erosion surface beneath the mantling breccias.

7. Conclusions On the basis of such scanty data no firm conclusions can be drawn on the general validity of the model proposed in Section 1. However the very little that is known is not inconsistent with the model and in several instances can be illuminated by it. The little we know of the Troodos massif demonstrates the existence of pillow lava piles at the upper end of the volcano size range. The meagre descriptions of the South Guam and North Auckland massifs are illuminated by the model in that it strongly suggests significance in relationships which otherwise appear random and insignificant. And James Ross Island clearly displays a constitution and structure of the kind envisaged in the superstructure of a marine basaltic volcano which as a result of intermittent volcanic activity during subsidence has periodically emerged (Fig. 14D) or remained emergent. 147

SECTION 3 BASALTIC VOLCANOES OF THE PRESENT OCEAN BASINS

1. Introduction The basaltic volcanoes of the present ocean basins include not only the numerous basaltic volcanic islands but the far more numerous seamounts (including guyots). According to Menard (1964) "about 2000 seamounts have been found in the ocean basins; several hundred have been surveyed enough to establish their shape, and about 50 have been dredged of bedrock. There remains no doubt that almost all, if not all seamounts are submarine volcanoes, because the bedrock is always basalt and the shapes and slopes are characteristic of no other landform However, it should be emphasised that the shape and location are the only things known about most submarine volcanoes". Menard estimates that 10,000 volcanoes with a relief of more than lkm exist in the Pacific Basin. This section attempts to present the principal morphological characteristics and to indicate the common constituents of submerged oceanic basaltic volcanoes, drawing on very scanty published data; and to interpret some of this information on the basis of data and concepts drawn from the study of the Largarvatn volcanoes (Chapter 2). Only some of the more detailed sources of relevant data are quoted.

2. Shape and constituents of oceanic basaltic volcanoes Seamounts show considerable variation in size and shape. In plan some are equidimensional, others are highly 148

elongate. Some have an unmistakeably volcanic profile, others are conspicuously flat-topped.

2.1. Seamounts without flat tops In comparison with oceanic subaerial basaltic volcanoes, seamounts without flat tops have notably steeper and less regularly inclined flanks as inspection of true scale profiles will show. For instance five seamounts southeast of Hawaii have "a relief that is about three times that of the subaerial portions of the various islands in the Hawaiian chain when expressed as a ratio of height to diameter. The greater relief is, of course, also indicated by steeper slope: a mean of 17° for the seamounts as compared to 11° for the submarine slope of Hawaii and 7° for the subaerial slope of Mauna Loa ...... "(Emery, 1955). The uneven slopes of seamounts without flat tops are well illustrated by the profiles collated by Menard (1964, Figs. 4.4. and 4.8.). Nine samples have been dredged from the two largest of the seamounts described by Emery and other seamounts southwest of Hawaii (Moore, 1965) and all are of altered basalt. Submarine photographs from one of these seamounts show pillow lava, and "tuff-breccia' was dredged from another. Photographs and dredge samples from seamounts of the Mid-Atlantic Ridge (Muir and Tilley, 1964) and the Carlsberg Ridge (Gann and Vine, 1966) are of basalt pillow lava.

2.2. Seamounts with flat tops (guyots) "The most characteristic feature of profiles of guyots is an abrupt change in slope of 15 to 40° between the steep sides and the almost flat top. Some guyots have tops that are level for as much as 10km, others have 149

slopes of 2 to 3° from the edge of the centre of the top, and some combine a marginal gentle slope with a central level top" (Menard, 1964). At their simplest the flanks of the flat-topped seamounts fall smoothly to the surrounding seafloor. However many guyots have profiles in which the upper and lower flanks are separated by a distinct negative break in slope (see for instance the profiles of some of the Mid-Pacific mountains, Hamilton, 1952). Almost all samples gathered on the summit platform of the well sampled Vema seamount (Simpson and Heydorn, 1965) consisted of "medium to dark brown tuffaceous agglomerate or volcanic ash of basaltic composition". Palagonitic basaltic "tuff-breccia" and "tuff" have been dredged from the break in slope on Sylvania guyot (Emery and others, 1954) and similar rocks are dominant in samples from the upper part of Cobb seamount (Hayudu,1962). On the other hand dredging from the platform of Fieberling Guyot (Carsola and Dietz, 1952) yielded a considerable quantity of angular to subrounded pebbles and cobbles of basalt.

2.3. Submerged portion of island volcanoes The morphology of the southeastern slopes of the island of Hawaii has been studied by Emery (1955). "The steepness of the lower submarine slopes was measured along 11 of the sounding traverses, using for each measurement the gradient of the steepest 6,000 foot depth zone and avoiding as much as possible the influence of secondary features. The mean of these values is 11°. In comparison the maximum, mean and minimum slopes of 6000 ft height zones of the subaerial part of Mauna Loa were found to be 11°, 7°, and 2° respectively. It is evident from these 1 5 0 measurements as well as from the inspection of the contour lines ...... and of the profiles that the submarine slopes of this part of Hawaii are steeper than the subaerial slopes". The submarine portion of the east rift zone of Kilauea has recently been photographed and dredged (Moore and Reed, 1963; Moore, 1965). Photographs were taken at depths of 10,800 feet and 14,100 feet. Of the 45 photos taken at these sites, the majority show pillow lava and the remainder rubbly surfaces. Fragments of basalt pillow lava were dredged along the axis of the rift in depths ranging from 500 to 5,000m.

2.4. Archipelagic aprons According to Menard (1964) archipelagic aprons are areas of characteristically smooth and very gently inclined seafloor encircling groups of modern and ancient islands (flat-topped seamounts). Most aprons form a smooth curve grading into the insular slopes. Upper parts of aprons have slopes of 1 to 2° which decrease outwards to merge with the flat abyssal plains. The estimated minimum volume of the Marquesas apron is 1 - 3 times larger than the total volume of the submarine part of the volcanoes (Menard, 1955).

3. Discussion Judging from published work there is at present insufficient data to establish with any confidence the constitution and structure of any single seamount, let alone of seamounts in general. Yet it seems to me that nothing that is known of the morphology and constitution of seamounts at present is inconsistent with the model outlined in Section 1 and much can be illuminated by it. 1 5 1

What we know of seamounts without flat tops suggests that they may be simply piles of pillow lava. Certainly their morphology indicates a constitution significantly different from the subaerial portions of emergent oceanic basaltic volcanoes and makes untenable Menard's (1964) assumption that submerged marine basaltic volcanoes are simple shields consisting of pahoehoe flows. What we know of flat-topped seamounts suggests that in some instances at least they may have or may initially have had a tuya-like structure. That most if not all were at some time at or near sea level is indicated by the recovery of rounded pebbles and cobbles, strongly vesicular lava and reef faunas from those guyots that have been dredged. In several instances, as on Cobb and Sylvania seamounts, palagonitic fragmental rocks have been recovered at or below the principal break in slope as might be anticipated in terms of a tuya-like structure. And their prevalence on the platform of Vema seamount might be explained in terms of the erosional stripping of a lava shield from its breccia pedestal (Fig.14C). The negative break in slope of some guyot profiles may well mark the junction between mantling flow-foot breccia and an underlying pillow lava pile for this is the topographic expression of just such a junction in several Laugarvatn tuyas. It has naturally been postulated and is now generally assumed that the guyot profile results from erosional truncation at sea level of an emerged volcano. I would not deny the importance of erosive processes at sea level in modifying the form of emerged volcanoes.But, if a tuya- like structure is characteristic of emerged marine basaltic volcanoes, then the effects of erosion in producing such profiles will in all such instances have been much less 1 5 2

than is generally assumed and may in some instances have been negligible. While it seems erosional truncation must be responsible in substantial degree for the final form of the truly flat-topped seamounts, seamounts with slopes of 2 to 3 degrees from the edge to the center of the top may be in some instances simply submerged, essentially unmodified volcanoes. The profiles of some flat-topped seamounts or guyots, particularly some in the northeast Pacific (Nayudu, 1962; Murray 1941), show marked shoulders on the flanks. These shoulders might be taken to be profiles of encircling terraces, but, according to Menard and Ladd (1963), "Detailed surveys show that terraces that can be traced around a seamount or guyot are very rare at best". Nayudu (1962) has suggested that the terraces on Cobb Seamount and Bowie Bank are primary volcanic features rather than wave planation surfaces; however the reasons he gives for this are not compelling and the suggested mode of origin is difficult to envisage. Ho infers that the presence of "primary" palagonitised basaltic fragmental rocks on these terraces, as revealed by dredging, makes wave planation an unlikely explanation and envisages the formation of terraces to occur below and independent of sea level by accumulation of Ipalagonite tuff erupted on a sloping surface". In fact marine erosion will operate regardless of the primary or secondary nature of the exposed material. A possible explanation for some of these imperfect terraces is that they formed at sea level as a result of the immersion of aerially extruded flows, and were carried below sea level by subsidence during quiescent phases of intermittent volcanic activity. Early constructional terraces thus formed might be overridden in places by the flows and flow-foot breccias of subsequent eruptions 1 5 3

(Fig.14D) and this would account for their lateral impersistence. The presence of palagonitised breccias on the surfaces of such terraces, as on Cobb and Bowie seamounts, could readily be explained by the erosional stripping of topset lavas from flow-foot breccias at sea level before the constructional terrace subsided below the surf zone (see Chapter 5, Section 2),- According to Menard (1964)9 "gravity observations suggest that submarine volcanoes have a relatively low average density", and "seismic refraction measurements show that the gross physical properties of near-surface layers of Pacific Basin volcanoes are quite different from hand specimens of the basalt of which they are constructed". In terms of his pahoehoe shield model of submarine basaltic volcanoes Menard suggests that draining of pahoehoe flow units may explain these anomalies. I suggest that a volcano constructed of pillow lavas, containing, as they may, considerable interpillow space, and of breccias, might be expected to show gross physical properties quite different from a handspecimen of the basalt of which it was constructed. The significance of archipelagic aprons lies in the volume of material they contain which may equal or exceed that of the encircled volcanoes. Menard (1964) suggests that archipelagic aprons are plains of 'fluid fissure flows" fed by fissures at the base of high volcanoes, though he notes that "the surface of the aprons is smoother than any known lava plain I; The form and seismic properties of these aprons are consistent with sedimentary accumulations emplaced by turbidity currents and allied processes. Menard (1964) has however rejected this explanation because "considerations of volumes of aprons and of the islands available as sources of debris rule out 154 any major contribution from this source". The assumption that erosion of the emergent portions of insular volcanoes provides the only potentially significant source of clastic material is extremely unlikely (see appendix). Breccias which result from the gravitational collapse of pillow lavas, either during or after cooling and consolidation, are prominent associates of pillow lavas in the Laugarvatn volcanoes. Gravitational collapse would most likely be particularly prevalent on the steep flanks of tall submarine basaltic volcanoes and the avalanches of pillow debris thus formed would come to rest at the foot of the volcano, building an apron. On such volcanoes as reach sea level explosive activity builds piles of ash which, as observation of Capelinhos (Zybszewski, 1960) has shown, are particularly susceptible to collapse and will further contribute to the building of an apron. Moreover, on those volcanoes which succeed in establishing themselves above sea level, continually accumulating unstable banks of flow-foot breccia will provide a ready source of clastic material. 1 5 5

CHAPT.L.1-i. 7

SUMMARY AND CONCLUSIONS 1 5 6

The principal obstacles to the elucidation of the subject of basaltic volcanism in aqueous environments are the present impossibility of direct observation of most of the processes and the general inaccessibility of the more recent products. In this situation Icelandic intraglacial basaltic volcanoes offer probably unique opportunities. Intraglacial basaltic volcanoes near Laugarvatn, southwest Iceland, range from simple steep—sided piles of pillow lava to fortress—like tablemountains (tuyas) in which a pillow lava pile with tuff capping is enveloped within a superstructure combining a carapace of sheet lava with gentle slopes on a pedestal of glassy breccia with steep slopes. The boundary between carapace and pedestal, which is planar and horizontal when undisturbed, is characterised by a downward passage from sheet lava to breccia and is believed to record the waterlevel of the meltwater pond within which the volcano grew. The variations in form, constituents and structure exhibited by the Laugarvatn volcanoes can be related to a sequence of development which, if it runs to completion, results in the formation of a tuya. It is inferred that in some instances volcanic activity was limited to effusion in water, resulting in steep—sided piles of pillow lava. With piles which mounted close to water level activity culminated in explosions which mantled the pillow lava pile with tuff. In other instances explosive emergence was superseded by quiet effusion of lava in air. Flows advanced outwards into encircling meltwater on deltas of breccia formed at their fronts, mantling and obscuring the products of the earlier phases of aqueous effusion and explosive emergence. 1 5 7

This interpretation of the constitution and structure of the Laugarvatn volcanoes provides the basis for a structural model of a basaltic volcano in an aqueous environment. In this model a basaltic volcano which has grown in water whose depth precludes explosive eruption of the magma in question will consist simply of a pile of pillow lava. In the case of a very shallowly submerged or slightly emerged volcano the pillow lava pile will be capped by tuff. A volcano which has established itself above waterlevel will consist of a pillow lava pile surmounted by a shield of sheet lava on a pedestal of flow-foot breccia with a core of tuff. Review of the small amount of published data suggests that nothing that is known of the morphology and constitution of oceanic seamounts and the submerged_ portions of oceanic island volcanoes is inconsistent with this structural model and much can be illuminated by it. What is known of seamounts without flat tops suggests that they may be simply piles of pillow lava, while what is known of flat-topped seamounts suggests that in some instances at least they may have or may initially have had a tuya-like structure. The aprons which encircle oceanic basaltic volcanoes are volumetrically as important as the encircled volcanic piles. Their constitution is unknown but, in the light of the constitution of the Laugarvatn volcanoes, they may well incorporate substantial volumes of pillow breccia, flow-foot breccia and tuff, emplaced by rubble flows and turbidity currents. Basaltic volcanoes which have grown partly or entirely beneath sea level and which now stand 'high and dry' are either very rare or, more likely, remain unrecognised and undescribed. Two definite examples, and three probable 1 5 8

examples, one of which has not previously been recognised as such, appear to have a constitution and structure consistent with the Laugarvatn model, though data is very limited. Pillow lava, as one of the principal constituents of basaltic volcanoes in aqueous environments, has been a particular object of study. The forms and fabric of excellently exposed pillow lava in Iceland and Wales are in many respects closely similar to the forms and fabric of pahoehoe. These similarities reinforce the long— standing though by no means generally—accepted hypothesis that the mode of formation and propagation of pillow lava is akin to the digital advance of pahoehoe. The typically basaltic and more specifically tholeiitic character of analysed Cenozoic pillow lavas indicates the probable basis of this structural similarity. Not only are pillow lavas invaluable indicators of an aqueous environment, but they also give promise of being invaluable indicators of depth of water at the time of eruption. The pattern of decreasing vesicularity and vesicle size with increasing depth recently established by Moore (1965) for Hawaiian pillow lava is also exhibited by pillow lava of the basal pillow lava piles of the Laugarvatn volcanoes, and by a thick sequence of pillow lava in Wales. However pillow lavas associated with the products of emergence in the Laugarvatn volcanoes do not exhibit these trends, probably because partial degassing of the lava at an emergent summit vent preceded their formation. As is generally the case, breccias containing recognisable pillow fragments or pillowy lava bodies are intimately associated with pillow lava in the Laugarvatn volcanoes. Their structural—lithologic relationships on 1 5 9

undeformed, little eroded and excellently exposed pillow lava piles strongly suggest that they are the product of the simple gravitational collapse of pillow lava, their variable texture and glass content reflecting the state of consolidation of the pillow lava at the instant of collapse. Para-pillow lava, a variety of fragmental lava encountered in the Laugarvatn volcanoes with a fabric very like that of pillow lava, is believed to be the subaqueous equivalent of as lava, as pillow lava is believed to be the subaqueous equivalent of pahoehoe. Peperite, a variety of volcanic fragmental rock now generally attributed to brecciation of lava as a consequence of intrusion into wet sediment, has been examined .at the type locality. On the basis of this investigation I have concluded that the type peperites are the product of explosive projection of basaltic ejecta into steadily accumulating lime mud, as suggested by Scrope over a century ago. Review of the character of the recent Capelinhos and Surtsey eruptions and of the geographic distribution of the products of such eruptions suggests that explosive activity, induced by extraneous water, is characteristic of eruption of basalt lava from wet vents in shallow water and in air. Explosive generation of steam when extraneous water comes in contact with hot lava provides a satisfactory explanation of this environmental influence. On the basis of McBirney's (1963) theoretical analysis of submarine ash formation, it can be predicted that the depth limits of explosive basaltic eruptions will be about 2000m for steam explosions and may be a comparable depth for explosive exsolution of volatiles, the exact value in the latter instance depending primarily on the water content of the lava. 1 6 0

Recorded observations of the flow of basalt lava from air into water show that immersion may be quiet or explosive in its general aspect and may or may not result in instantaneous fragmentation of the lava. There are no unequivocal records of the observed formation of pillow lava. Littoral explosions, though a conspicuous accompaniment of immersion in many instances, are probably of minor significance in terms of their yield of fragmental material. Explosive evolution of steam is probably of less importance than thermal stress in the fragmentation of submerging lava. Immersion of basalt lava leaves a distinctive structural record of the water level at the time of immersion and of short or long term movements of water level relative to the active volcanic pile or terrain. Such records obviously have enormous significance for reconstructions of volcanic activity in aqueous environments and for stratigraphy generally. 1 6 1

ACKNOWLEDGEd EN T S

I would like to express my sincere thanks to my supervisor, Dr. G.P.L. Walker, for unflagging support and interest, provocative and stimulating discussion, and thorough and constructive reading of the text of the thesis. I also wish to thank Professor J. Sutton for the freedom and support which I have enjoyed as a student in his department. I gratefully acknowledge the financial support cf a Beit Fellowship and of an 1851 Research Fellowship. 162

REFERENCES CITED

Anderson, T., 1910, The volcano of Matavanu in Savaii: Geol.Soc. London Quart.Jour., 66,p.621-639. Bailey, E.B., Thomas, H.H., and others, 1924, The Tertiary and post-Tertiary geology of Mull: Scotland Geol. Survey Memoir. Barth, T.F.W., 1937, Volcanic ash from Vatnajokull: Norsk. Geol. Tidsskr., 17, p.31-38. Bemmelen, R.W. van, and Rutten, M.G., 1955, Tablemountains of Northern Iceland: Leiden (E.J.Brill). Brigham, W.T., 1909, The volcanoes of Kilauea and Mauna Loa on the Island of Hawaii: Bishop Museum Press, Honolulu, 222pp. Brothers, R.N., 1965, Tangiteroria district - Tangihua Volcanics; in New Zealand Volcanology - Northland, Coromandel, Auckland; New Zealand Dept. Sci. Ind. Res., Information Ser., 49. Cann, J.R., and Vine, F.J., 1966, An area on the crest of the Carlsberg Ridge: petrology and magnetic survey: Phil. Trans. R. Soc., A 259, 1099, p.198-217. Capps, S.R., 1915, Some ellipsoidal lavas on Prince William Sound, Alaska: Jour. Geology, 23, 1, p.45-51. Carlisle, D., 1963, Pillow breccias and their aquagene tuffs, Quadra Island, British Columbia: Jour. Geology, 71, 1, p.48-71. Carsola, A.J., and Dietz, R.S., 1952, Submarine geology of two flat-topped northeast Pacific seamounts: Am.Jour. Sci., 250, P.481-497. Corwin, G., and Tracey, J.I., 1965, in U.S.Geol. Survey Prof. Paper 403A, p.130. 163

Darwin, Charles, 1876, Geological observations on the volcanic islands and parts of South America visited during the voyage of H.M.S. 'Beagle': Smith, Elder and Co., London, (2nd edition). Daly, R.A., 1924, The geology of American Samoa: Carnegie Inst. of Washington Publication, 340, p.145-178. Dangeard, L., 1931, Algues inferieures dans le calcaire concretionne de la Limagne: Acad. Sci. Paris, Comptes- rendus, 192, p.172. Einarsson, T., 1966, Studies of temperature, viscosity, density and some types of material produced in the Surtsey eruption: The Surtsey Research Society, Reykjavik, Iceland; Progress Report 2, p.163-180. Emery, K.O., 1955, Submarine topography south of Hawaii: Pacific Science 9, p.286-291. Emery, K.O., Tracey, J.I., and Ladd, H.S., 1954, Geology of Bikini and nearby atolls: U.S. Geol. Survey Prof. Paper 260-A. Farquhar, 0.C., 1966, Seamounts ancient and modern: New Scientist, 29, 487, p.698-699. Finch, R.H., and Macdonald G.A., 1953, Hawaiian volcanoes during 1950: U.S. Geol. Survey Bulletin 996-B. Fuller, R.E., 1931, The aqueous chilling of basaltic lava on the Columbia River Plateau: Am. Jour. Sci., 21, p.281-300. Fuller, R.E., 1934, Structural features in the Columbia River Lavas of Central Washington - a criticism: Jour. Geology, 42, 3, p.311-320. Fuller, R.E., 1939, Gravitational accumulation of olivine during the advance of basaltic flows: Jour. Geology, 47, 3, p.303-313. Gage, M., 1957, The geology of Waitaki Subdivision: New Zealand Geol. Survey Bull., new ser., 55, 135pp. 164

Gass, I.G., and Masson-Smith, D., The geology and gravity anomalies of the Troodos Massif, Cyprus: Phil.Trans. Royal Soc. London, ser.A, 255, 1060, p.417-467. Green, W.L., 1887, Vestiges of the molten globe; part 2, The Earth's surface features and volcanic phenomena: Hawaiian Gazette Publishing Co., Honolulu. Hadwen, P., and Walker, G.P.L., A geological review of the Azores: Proc. Geologists Assoc. London, in press. Hamilton, E.L., 1956, Sunken Islands of the Mid-Pacific Mountains: Geol. Soc. Am. Memoir 64. Jensen, H.I., 1906, The geology of Samoa and the eruptions in Savaii: Proc. Linn. Soc. New South Wales, 31, 4, p.641-672. Jung, J., 1946, Geologie de l'Auvergne et de ses confins Bourbonnais et Limousins: Serv. Carte Geol. France, Memoire 11. Kjartansson, G., 1959, The Moberg Formation; in Thorarinsson, S., Einarsson, T., and Kjartansson, G., On the geology and geomorphology of Iceland: Geogr. Annlr. , 135-169. Kjartansson, G., 1964, Isaldarlok og eldfjoll a Kili. The retreat of the last glacial ice-sheet and some volcanoes in the Kjolur area, central Iceland. Museum of Natural History, Reykjavik, Iceland; Miscellaneous Papers, 38. Kjartansson, G., 1966, Stapakenningin og Surtsey. A comparison of tablemountains in Iceland and the volcanic island of Surtsey off the south coast of Iceland: Museum of Natural History, Reykjavik, Iceland; Miscellaneous Papers, 49. Lacroix, A., 1923, Comparaison de la composition chimique de deux laves d'Islande: C.r. Acad. Sci., Paris, 177, p.369-373. 165

Lewis, J.V., 1914, Origin of pillow lavas; Geol. Soc. America Bull., 25, p.591-654. Macdonald, G.A., 1944, The 1840 eruption and crystal differentiation in the Kilauean magma column: Am. Jour. Sci., 242, p.177-189. Macdonald, G.A., 1953, Fahoehoe, aa, and block lava: Am. Jour.Sci., 251, p.169-191. Macdonald, G.A., 1963, Physical properties of erupting Hawaiian magmas; Geol. Soc. America Bull., 74, p.1071-1078. Macdonald, G.A., and Eaton, J.P., 1964, Hawaiian volcanoes during 1955; U.S. Geol. Survey Bull. 1171. Macdonald, G.A., and Katsura, T., 1964, Chemical composition of Hawaiian lavas; Jour. Petrol., 5, 1, p.82-133. McBirney, A.R., 1963, Factors governing the nature of submarine volcanism Bull. volcanol., 26, p.455-469. Mathews, W.H., 1947, "Tuyas". Flat-topped volcanoes in northern British Columbia: Am. Jour. Sci., 245, p.560-570. Menard, H.W., 1956, Archipelagic aprons; Am. Assoc. Petroleum Geologists Bull., 40, p.2195-2210. Menard, H.W., 1964, Marine geology of the Pacific; McGraw- Hill, New York. Menard, H.W., and Ladd, H.S., 1963, Oceanic islands, seamounts, guyots and atolls: in The Sea, 3, Interscience, New York. p.365-385. Michel, R., 1948, Etude geologique du plateau de Gorgovia; Rev. Sci. Nat. Auvergne, 14, 1-68. Michel, R., 1953, Contribution a l'etude petrographique des peperites et du volcanisme Tertiare de la Grande Limagne: Soc. Hist. Nat. Auvergne, Memoire 5. 1 6 6

Michel-Levy, A., 1890, Compte rendu de l'excursion du 16 septembre a Gergovie et Veyre-Monton. Soc. Geol. France, Bull. 18, p.891-897. Moore, J.G., 1965, Petrology of deep sea basalt near Hawaii: Am. Jour. Sci., 263, p.4C-52. Moore, J.G., 1966, Rate of palagonitisation of submarine basalt adjacent to Hawaii: in U.S. Geol. Survey Prof. Paper 550D, p.163-171. Moore, J.G., and Ault, W.U., 1965, Historic littoral cones in Hawaii: Pacific Sci., 19, p.3-11. Moore, J.G., Nakamura, K., and Alcaraz, A., 1966, The 1965 eruption of Taal volcano: Science 151, 3713, p.955-960. Moore, J.G., and Reed, R.K., 1963, Pillow structures of submarine basalts east of Hawaii: U.S. Geol. Survey Prof. Paper 475-B., p.153-157. Muir, I.D., and Tilley, C.E., 1964, Basalts from the northern part of the rift zone of the Mid-Atlantic Ridge; Jour. Petrol., 5, 3, p.409-434. Murray, H.W., 1941, Submarine mountains in the Gulf of Alaska: Geol. Soc. America Bull., 52, p.333-362. Nayudu, Y.R., 1962, A new hypothesis for the origin of guyots and seamount terraces; Am. Geophys. Union Monograph 6, p.171-180. Nelson, P.H.H., 1966, The James Ross Island Volcanic Group of N.E. Graham Land: Br. Antarctic Survey, Sci. Rept.54.

Noe Nygaard, A., 1940, Sub-glacial volcanic activity in ancient and recent times: Folia Geographica Danica, 1, 2, p.1-68. Noe Nygaard, A., 1950, Hagongur, a partially subglacial

volcano in S.W. Vatnajokull: Medd. Dansk. Geol. For. 9 11, 5. 1 6 7

Pettijohn, F.J., 1957, Sedimentary rocks: Harper, New York. Quennel, A.M., and Hay, R.F., 1964, Origin of the Tangihua Group of North Auckland: New Zealand Jour. Geol. Geophys., 7, 3, p.638-639. Richter, D.H., and Eaton, J.P., 1960, The 1959-60 eruption of Kilauea volcano: New Scientist, 7, p.994-997. Richter, D.H., Ault, W.U., Eaton, J.P., and Moore, J.G., 1964, The 1961 eruption of Kilauea volcano, Hawaii; U.S. Geol. Survey Prof. Paper 474-D. Russell, I., 1885, Geological history of Lake Lahontan, a Quaternary lake of northwestern Nevada; U.S. Geol. Survey Monograph 11. Saemundsson, K., 1967 A, Vulkanismus and tektonic des Hengill-Gebietes in sudwest-Island: Acta Naturalia Islandica, 2, 7, 105 pp. (with English summary). Saemundsson, K., 1967 B, in Iceland and Mid-Ocean Ridges, S. Bjornsson (ed): Societas Scientarum Islandica, 38, p.66 Scholl, D.W., and Taft, W.H., 1964, Algae, contributors to the formation of calcareous tufa, Mono Lake, California: Jour. Sed. Petrol., 34, 2, p.309-319. Scrope, G.P., 1858, The geology and extinct volcanoes of central France; John Murray, London. Searle, E.J., 1965, Auckland volcanic district: in New Zealand Volcanology - Northland, Coromandel, Auckland: New Zealand Dep. Sci. Ind. Res. Information Ser. 49. Simpson, E.S.W., and Heydorn, A.E.F., Vema Seamount: Nature, 207, p.249-251. Snyder, G.L., and Fraser, G.D., 1963 A, Pillowed lavas I: intrusive layered lava pods and pillowed lavas, Unalaska Island, Alaska: U.S. Geol. Survey Prof. Paper 454-B. 1 6 8

Snyder, G.L., and Fraser, G.D., 1963 B, Pillowed lavas a review of selected recent literature; U.S. Geol. Survey Prof. Paper 454-1J. Stearns, H.T., 1944, Geology of the Samoan Islands: Geol. Soc. America Bull., 55, p.1279-1355. Stearns, H.T., 1945, Geology of the Wallis Islands: Geol. Soc. America Bull., 56, p.849-872. Stearns, H.T., and Vaksvik, K.N., 1935, Geology and groundwater resources of the island of Gahu, Hawaii: Hawaii Div. Hydrography Bull.l. Tazieff, H., 1951, Lieruption du volcan Gituro: Geol. Serv. Congo Belge, Mem.l. Tazieff, H., 1959, L'eruption 1957-1958 et la tectonique de Faial, Acores: Serv. Geol. Portugal, Mem.4.

Thorarinsson, S., 1964, Surtsey - the new island in the North Atlantic: Almenna bokafelagid, Reykjavik.

Thorarinsson, S., 1965 A., Sitt of hverju um Surtsey- jargosid. Some facts about the Surtsey eruption: Museum of Natural History, Reykjavik, Iceland; Miscellaneous Papers 46. Thorarinsson, S., 1965 B., The Surtsey eruption - course of events and the development of a new island: in Surtsey Res. Committee, Reykjavik, Iceland, Progress Report I, p.51-55. Thorarinsson, S., 1967, in Iceland and Mid-Ocean Ridges, S.Bjornsson (ed.): Societas Scientarum Islandica, 38, p.66. Thorarinsson, S., Zinarsson, Th., Sigvaldc.;.son, G., and Elisson, G., 1964, The submarine eruption off the Vestmann Islands, 1963-64: Bull. volcanol., 27, p.1-11. 1 6 9

Tracey, J.I., Schlanger, S.O., Stark, J.T., Doan, D.B., and May, H.G., 1964, General geology of Guam: U.S. Geol. Survey Prof. Paper 403A. Tryggvason, T., 1943, Das Skjaldbreid-Gebiet auf Island: Geol. Inst. Univ. Upsala, 30, p.273-320. Vallance, T.G.V., 1965, On the chemistry of pillow lavas and the origin of spilitesz Min. Mag., 34, 268, p.471-481. Walker, G.P.L., 1965, Some aspects of Quaternary volcanism in Iceland; Trans. Leicester Lit. Phil. Soc., 59, p.25-40. Waters, A.C., 1960, Determining direction of flow in basalts; Am. Jour. Sci., 258-A., Bradley Vol., p.35C-366. Wentworth, O.K., Pyroclastic geology of Oahu; B.P.Bishop Museum Bull. 30. Wentworth, C.K., and Williams, H., 1932, The Classification and terminology of the Pyroclastic rocks; Natl. Res Council, Bull. 89, p.19-53. Yagi, K., 1964, Pillow lavas of Keflavik, Iceland and their genetic significance; Jour. Fac. Sci. Hokkaido University, Ser.IV., 12, 2, p.171-183. Zbyszewski, G., 1960, l'eruption du volcan de Capelinhos; Bull volcanol.,23, p.77-101.

APPENDIX

J. G. JONES Department of Geology, Imperial College, London, England

Clastic Rocks of Espiritu Santo Island, New Hebrides

Abstract: The Miocene clastic rocks of southwest were their source. The salient characteristics of the Espiritu Santo, among which rudites are predomi- marine andesite volcano and its environs as a sedi- nant, consist of lava-derived and organic calcareous mentary system are reviewed, and the genesis of the constituents in all proportions. Their composition Santo clastic rocks is discussed in the light of this and texture indicate that marine andesitic volcanoes review.

CONTENTS

Introduction 1281 Environment of deposition 1286 Acknowledgments 1281 Detailed provenance 1286 Miocene clastic rocks of southwest Espiritu Santo 1281 References cited 1287 Petrography 1282 Lava-derived elastic rocks 1282 Calcareous elastic rocks 1283 Figure Mixed elastic rocks 1284 I. Regional setting of Espiritu Santo 1282 Provenance 1284 The marine andesite volcano and its environs as a sedimentary system 1284 Plate Following Modes of initial fragmentation 1284 I. Lava rudite, Espiritu Santo Island, New Modes of dispersal 1285 Hebrides New The ultimate accumulation 1285 2. Lava rudite, Espiritu Santo Island, 1282 Synthesis 1286 Hebrides The Santo elastic rocks in terms of the volcanic 3. Lava rudite, Espiritu Santo Island, New archipelagic environment 1286 Hebrides

colleagues at Imperial College and to my wife INTRODUCTION for critical reading of the manuscript. This Clastic rocks of direct or indirect volcanic paper was written during my tenure of a Beit provenance, often associated with calcareous Fellowship at Imperial College. sediments, are basic elements of island arc MIOCENE CLASTIC ROCKS OF geology, and for their correct interpretation an SOUTHWEST ESPIRITU SANTO understanding of the special characteristics of the volcanic archipelago as an environment of Espiritu Santo, the largest island of the New sedimentation is required. It is the aim of this Hebrides group in the southwest Pacific Ocean, paper to elucidate some of these characteristics rises above sea level in the middle of the New in the course of describing and interpreting the Hebrides Ridge (Fig. 1). It stands to the west characteristics of the Miocene clastic rocks of of the present volcanic axis of the ridge, which southwest Espiritu Santo. is marked by a chain of Quaternary volcanoes. The southwest quadrant of the island is the ACKNOWLEDGMENTS most elevated (nearly 2000 in maximum) and This study is based on a geological investiga- deeply dissected area in the New Hebrides. The tion of southwest Espiritu Santo carried out by oldest rocks in southwest Espiritu Santo 'are pre- the author in 1963. I am indebted to C. V. G. dominantly clastic rocks of Lower Miocene Phipps and A. J. Warden for initiating the proj- age, and these constitute almost the entire bulk ect; to the Geological Survey of the New of this part of the island. Fragmented by a fine Hebrides for financing the fieldwork; to my network of fractures and faults within which

Geological Society of America Bulletin, v. 78, p. 1281-1288, 1 fig., 3 pls., October 1967 1281 1282 J. G. JONES—CLASTIC ROCKS OF ESPIRITU SANTO ISLAND, NEW HEBRIDES

the strike and dip of stratification often show cobble- and boulder-size clasts often show radical variations, these rocks have yielded no rounding, though well-rounded' clasts are al- coherent structural picture on the scale of the ways a minor component. In some rocks the investigation. The thickness of the clastic pile component clasts show an unbroken size range probably exceeds 2000 m. (P1. 1, fig. 1). In others a real division into Among the Miocene clastic rocks rudites phenoclasts and matrix is apparent, the latter (breccias and conglomerates) are predominant commonly arenaceous (Pl. 1, fig. 2). Some in the region as a whole, although arenites and rudites are homogeneous in terms of the corn-

‘...._ —, sl

.._...... _ 1 ! ! --t , I 0 iD 1 1011". I Rib i"ir Ay C -: • I lr, er-rsilli I - L. - — a• ---"-- rii - - Too .•.-. 4 I. -"". • ' 1 1„e. poimor,,---‘...: r,, . . 1 1 J_e. , .. _ .______— — ; • NEW N

Z----- b - Figure 1. Regional setting of Espiritu Santo. Contours in fathoms

siltstones are locally preponderant. The clastic position and roundness of their clasts; others rocks consist almost entirely of one or more of are heterogeneous. three classes of components. The predominant In sequences consisting entirely of rudites, class includes lithic and crystal clasts derived the form and thickness of single beds is not from the fragmentation of lava. A subsidiary readily apparent. However, where rudites are but important class is calcareous organic ma- interbedded with arenites they clearly form terial, principally the whole and broken tests tabular beds, typically many meters in thick- of largely benthonic foraminifera and fragments ness (beds of over 30 m were noted), with a of reef-forming algae and corals. Fragments con- planar concordant base on arenite (P1. 2, figs. sisting of aggregates of clasts from these two 1, 2). In some rudites a crude layering is defined classes constitute a third and minor class. The by fluctuations in the maximum size of clasts, combination of these components in all proportions gives rise to the lithological spectrum of Rocks in which the largest clasts have a dimension no the Santo clastic rocks. greater than 10 times that of the smallest clasts are classed as well-sorted. Rocks with clast size range be- Petrography tween 10:1 and 20:1 are classed as moderately sorted. Rocks with clast size range greater than 20:1 are classed Lava-derived clastic rocks. The lava-derived as poorly sorted or unsorted. rudites (Pl. 1) are unsorted and poorly sorted' 2 The Wentworth grade scale is used in this paper. rocks consisting of lava clasts which range in 3 The terms "rounded" and "well-rounded" are used size from boulders 2 m in diameter to silt.2 The in the sense of Pettijohn, 1957.

Figure 1. Lava rudite consisting of clasts with a continuous Figure 2. Lava rudite showing division into phenoclasts and size range. Note variable rounding of clasts. White clasts are matrix limestone. Width of field 0.6 m

Figure 3. Lava rudite. Large rectangular clast 0.6 m across Figure 4. Lava rudite LAVA RUDITE, ESPIRITU SANTO ISLAND, NEW HEBRIDES Figure 1. Lava rudite with planar, concordant base on well- Figure 2. Lava rudite with planar, concordant base on moder- bedded arenite-siltstone. Note contorted clasts of stratified ately bedded arenite arenite-siltstone in rudite (largest of these nearly 2 in long).

Figure 3. Calcirudite•calcarenite turbidite showing grading Figure 4. Lava rudite and rubbly arenite. Height of rock and lamination face 5m

LAVA RUDITE, ESPIRITU SANTO ISLAND, NEW HEBRIDES Figure 1. Mixed rudite. Basically lava rudite with enormous Figure 2. Mixed rudite. Basically calcirudite (light gray) with load of calcareous clasts (white). Hammer rests on limestone large load of lava clasts. Note the highly variable rounding of boulder. the latter. Boulder 0.7 m in diameter

Figure 3. Mixed rudite. Basically calcirudite (white and dark Figure 4. Mixed rudite. Basalt clasts in calcilutite matrix gray) with enormous load of lava clasts (black) LAVA RUDITE, ESPIRITU SANTO ISLAND, NEW HEBRIDES MIOCENE CLASTIC ROCKS OF SOUTHWEST ESPIRITU SANTO 1283

certain layers being defined by an abundance andesite and basalt. Pyroxene andesite clasts of large clasts. Quite commonly the distribution show a considerable variety of color: gray, of clasts, in terms of size, is remarkably inhomo- black, pink, purple, yellow-brown, and, in al- geneous, with the largest clasts grouping in tered rocks, various shades of green. clusters. The shape and boundary form of lithic clasts The arenites and siltstones are similar to the unmodified by abrasive rounding or diagenetic rudites in most respects, but differ principally compaction is governed by crystallinity, vesic- in that they are typically moderately to well- ularity, and jointing. The larger clasts, par- sorted. They generally contain a higher propor- ticularly those of cobble and boulder grade tend tion of crystal to lithic clasts than do the to be equant, polyhedral blocks bounded by rudites. The lithic (including vitric) clasts show flat joint faces. Among the smaller clasts, all degrees of rounding, being predominantly strongly vesicular clasts tend to have irregular unrounded in many rocks, rounded in some, outlines marked in thin section by vesicular and even well-rounded in a few. Crystal clasts embayments, while nonvesicular vitric clasts show little or no rounding. Some arenites are have angular outlines. Arcuate, cuspate shard homogeneous in terms of the composition and forms are rare. roundness of their clasts; most are heterogene- The form of crystal clasts varies with size ous. grade. Crystal clasts of silt and fine-sand grade The arenites and siltstones form tabular beds typically display broken, angular outlines, varying in thickness from several meters to less while clasts of medium- and coarse-sand grade than a centimeter. In some sequences stratifica- frequently exhibit outlines unmodified by attrition is sharply defined (Pl. 2, fig. 1), the strata tion, indicated by the preservation of crystal are constant in thickness on outcrop scale, and faces or zoning. Not infrequently these clasts graded units are common. In other sequences carry partial rims of glassy groundmass. stratification is more lensoid and less sharply de- In rocks little affected by diagenesis, glass fined (Pl. 2, fig. 2), and no graded units are may be altered to palagonite. In those rocks in apparent. Sole structures were not noted, which diagenesis is more advanced, glass is though this may only reflect the general lack of commonly altered to a ferromagnesian phyl- exposure of bedding planes. Large-scale cross losilicate. Plagioclase often shows zeolitic stratification is completely absent. alteration, but pyroxene and hornblende are un- Rocks which are not adequately represented altered in most rocks. Zeolite, ferromagnesian by the simple grade terms "rudite" and phyllosilicate and carbonate are common as "arenite" are those consisting of scattered lava cements. pebbles, cobbles, and boulders in a predom- Calcareous material of the kind which con- inant, crudely laminated arenite base. These stitutes the calcareous clastic rocks is very com- might be termed rubbly arenites. By increase monly present, though in variable amount, in of coarse clasts they pass into rudites, by de- the lava-derived clastic rocks. crease into arenites, and gradations of this sort Calcareous elastic rocks. The calcirudites are can be seen in outcrop (Pl. 2, fig. 4). unsorted rocks consisting principally of angular Among the component lithic clasts of the fragments of reef-forming corals and algae, the lava-derived clastic rocks the most prevalent whole and broken tests of benthonic foraminif- type is a basaltic andesite, typically with phe- era, and typically ragged secondary clasts con- nocrysts of plagioclase and pyroxene. Basalt and sisting of these components in a calcilutite hornblende andesite are prominent, though less matrix. The boundaries of clasts in these rocks abundant than pyroxene andesite. The ground- are poorly defined on weathered surfaces, and mass of most of the smaller (<1 cm) lithic clasts the dimensions of the larger clasts are thus not is glassy, though holocrystalline clasts of these readily apparent. However, limestone boulders dimensions are common. It is common for the up to several meters across are commonly found smaller lithic clasts in a rock to show a complete in lava rudites, and there is no reason to believe range of groundmass crystallinity and a cor- that clasts of similar dimensions are not present responding textural range. Vesicularity, too, is in the calcirudites. The smallest clasts are silt highly variable, and the clasts in a thin section size. Lava-derived clasts, both crystal and lithic, frequently range from nonvesicular to pumi- are common in these rocks, the lithic clasts often ceous. However, pumice rarely constitutes more well-rounded. than a small fraction of the lava clasts. Tach- The calcarenites and calc-siltstones differ ylytic glass is prominent in clasts of pyroxene from the calcirudites only in showing size sort- 1284 J. G. JONES-CLASTIC ROCKS OF ESPIRITU SANTO ISLAND, NEW HEBRIDES ing, ranging as they do from poorly sorted to them. There is thus good reason to suppose that well-sorted. the Santo area was as remote from continental The bedding characteristics of the calci- influences in the Miocene as it is today. rudites and calcarenites do not differ from those Since a continental source can be eliminated, of their lava-derived counterparts, with which it is logical to investigate the origins of the they are commonly interbedded. Miocene clastic rocks in terms of the character Mixed elastic rocks. This heading groups of the New Hebrides region as it is today. The those rocks consisting of both lava-derived and islands of the New Hebrides chain represent calcareous material but showing a marked pre- peaks on an elevated strip of sea floor, the New dominance of neither, which thus lie at the Hebrides Ridge. These eminences are the center of the lithological spectrum. The nature product either of accumulation alone, the es- of the constituents and the textural and struc- sentially volcanic landforms of Quaternary age, tural characteristics of these rocks do not differ or of tectonic elevation, the latter including the from those of the rocks already described. Par- larger islands of the group such as Santo. ticularly spectacular are the mixed rudites. On Fringing reefs are a common feature of both the one hand are those rocks which are basically island types. Were the source or sources of the lava rudites but which carry an enormous load Miocene clastic rocks "highs" of tectonic origin of calcareous clasts (P1. 3, fig. 1). On the other one would expect a significant content of second hand are those rocks which are basically calci- cycle material in the clastic rocks, with some rudites but which carry an enormous load of clastic constituents measurably older than the lava clasts (Pl. 3, figs. 2, 3). rocks in which they occurred. Such rocks must A particularly interesting group of mixed assuredly be forming off the Santo coastline rocks are those rudites and arenites consisting of today. Yet these characteristics are notably lava clasts of uniform composition (typically absent from the Miocene rocks. Thus marine basaltic), showing uniformly slight or no round- volcanoes remain as the only potential source ing, set in a matrix consisting principally of consistent with the nature of the Santo clastic carbonate (Pl. 3, fig. 4). This matrix may con- rocks. tain foraminiferal, echinoid, molluscan, and other organic constituents and a variable ad- THE MARINE ANDESITE VOLCANO mixture of fine-sand and silt-size vitric and AND ITS ENVIRONS AS A crystal clasts of the same composition as the SEDIMENTARY SYSTEM large lava clasts. A characteristic feature of If, as it seems, the characteristics of the Santo these rudites is the presence of masses of frac- clastic rocks should be examined and evaluated tured lava which apparently have been arrested in terms of the environment of the marine in the process of drifting apart. andesite volcano, then it is important that we should have a clear concept of the salient char- Provenance acteristics of this environment. This section The Miocene clastic rocks of Espiritu Santo examines the marine andesite volcano and its give no indication of derivation from a landmass environs in the light of the concepts of stratig- of continental dimensions or constitution, and raphy and sedimentation. the basic simplicity of the assemblage argues against such a source. The rocks are texturally Modes of Initial Fragmentation and mineralogically immature. Their constit- The processes of fragmentation whereby an uents are essentially supracrustal in origin, andesite volcano may serve as a source of clastic specifically lava-derived and organic calcareous; material are many and varied. They include they are the product of a single sedimentary both pyroclastic and epiclastic processes. I use cycle and are of the same age. Diorite and gab- the term "pyroclasis" to include any process bro clasts are a minor component of some associated only with volcanism which results in rudites, but significantly they are commonly the fragmentation of juvenile lava and "epi- present as inclusions in lava clasts of the same clasis" for all other supracrustal processes of rudites. And clasts which might be taken as in- fragmentation. dicating more than one cycle of sedimentation Among the common pyroclastic events on commonly are clearly the product of the dis- insular andesite volcanoes are the following: (1) ruption of partially lithified sediment and are explosive vesiculation and disruption of erupt- of the same age as the material which encloses ing lava; (2) explosive disruption and disinte- MARINE ANDESITE VOLCANO AND ITS ENVIRONS 1285 gration of tholoids; (3) flow fragmentation of avalanches, and ash and mud flows. Rubble extruded lava; and (4) explosive disruption of avalanches are probably one of the most im- lava flows on reaching the sea. portant agents of dispersal of coarse clastic Little is known of the nature and importance material on tall andesite volcanoes as Fenner of pyroclastic processes on submarine andesite (1937) has suggested. volcanoes. However, Fiske and Matsuda (1964) A vigorously active volcano forms a structure in have demonstrated that explosive vesiculation which conditions of instability of large masses of occurs in the aqueous as well as in the aerial en- rock are likely to occur repeatedly. . . . Whether a vironment, and it seems probable that in minor explosion shatters a protruded mass and sets shallow water the explosive evolution of steam it in motion, or whether there is simply the breaking is an effective and important agent of pyroclasis away of a somewhat watersoaked cliff, perhaps ac- (Jaggar, 1908). companied by loose ejecta that have accumulated Of the epiclastic processes of fragmentation on slopes, the masses set in motion have an ability to affecting emergent andesite volcanoes probably move in their lower courses for long distances over gentle grades in a manner that is difficult to con- the most important are marine erosion and ceive but that is attested by abundant evidence. gravitational collapse. An insular andesite volcano is open to attack by marine erosion on all Some of this evidence has been recently re- flanks, and previously unfragmented lava in the viewed by Kent (1966), who notes that Iranian volcanic edifice will break un under the impact "rockfalls" or rubble avalanches have in com- of the surf (see also Richards, 1960). At the mon "lack of sorting, absence of attrition of same time, the characteristically steep slopes of constituent blocks, and a horizontal distance of andesite volcanoes, both above and below sea travel five to ten times the vertical height of the level (Kuenen, 1950, p. 526), are prone to rockfall." gravitational collapse or slumping (see also For information on the characteristics of rub- Kuenen, 1935, p. 294-297; Fairbridge, 1950) ble avalanches in water one can turn to the ex- which may be precipitated in some instances by cellently documented "slides" of the virtually explosive eruption or by the seismic shocks so undeformed Permian Delaware basin of Texas prevalent in the island arc environment. Such (Newell and others, 1953) which began as collapse may partially or wholly involve co- slumps of reef talus on the steeply sloping front herent lavas which are thus fragmented. Insular of the Capitan reef and came to rest at depths of volcanoes form suitable and ubiquitous founda- 900 to 1600 feet at distances of up to 10 miles tions for organic reefs which are also susceptible from the reef front, having moved for a number to fragmentation by marine erosion and gravi- of miles in some cases across an essentially hori- tational collapse. zontal sea floor. As a group these Permian rock Below the surf zone the effectiveness of avalanches show "general lack of sorting and marine erosion as an agent of initial fragmenta- stratification, coarse texture, angularity of tion is negligible. There is no reason to suspect, composing fragments and tabular form" and however, that gravitational collapse is not an "like great modern landslides produce hardly important epiclastic process on the flanks of any visible effect on the substratum." The most submarine andesite volcanoes and on the sub- extensive of these avalanches carried blocks up merged flanks of emergent volcanoes (see also to 14 feet in greatest dimension. Moore, 1964). Flows of sand- and silt-grade material are Gravitational collapse will, of course, involve doubtless prevalent on submarine volcanic both coherent lava and previously fragmented slopes where they fall within the class of tur- pyroclastic material, sometimes separately, at bidity currents and slurries (see also Fiske and other times jointly. On emergent volcanoes it Matsuda, 1964). may also involve talus banks of sand and rubble temporarily accumulated at sea level from the The Ultimate Accumulation products of aerial mass wasting. A broad encircling fan or apron consisting of rubble avalanche deposits, together with lahar Modes of Dispersal or mudflow deposits and ash flows, appears to be Among the common modes of dispersal of prevalent at the foot of terrestrial andesite vol- elastic material on insular andesite volcanoes are canoes (see also Swanson, 1966) and one might the following: (1) explosive projection, (2) flow- anticipate similar accumulations at the foot of age of fragmental material, including rubble marine andesite volcanoes. That they have not, 1286 J. G. JONES-CLASTIC ROCKS OF ESPIRITU SANTO ISLAND, NEW HEBRIDES

to my knowledge, been identified is presumably predominance of lava-derived clastic material a consequence of the relatively complex bathym- might suggest explosive projection as a possible etry of the "island arcs." Aprons are in fact mode of transport and emplacement, there is no common around basaltic volcanoes which rise evidence of this, such as impact structures from the floor of the Pacific Basin (Menard, might provide. It appears most likely that the 1964), though their constitution is not yet es- rudites were emplaced as rubble avalanches or, tablished. Nevertheless, their morphology and as they will be termed henceforth, rubble flows, geophysical characteristics are consistent with for their sorting and structural characteristics an accumulation deposited by submarine flows are consistent with such a process. of clastic debris. It is noteworthy that the The sorting and structural characteristics of volume of these "archipelagic aprons" in some the arenites and siltstones are closely compa- instances exceeds that of the encircled volcanic rable to those of the deep-sea sands (Kuenen, edifices by a factor of two or three. 1964). If Kuenen's conclusion that the latter are turbidites is accepted, then it seems likely Synthesis that most of the Santo arenites and siltstones Thus the marine andesite volcano and its have been deposited by turbidity currents and environs may operate as a self-sufficient, cen- associated processes. trifugal, sedimentary system within which the Thus the depositional environment of the central volcanic edifice serves as the clastic Santo clastic rocks appears to have been fed by source and provides environmental conditions rubble flows and turbidity currents, a picture which favor the movement of clastic debris which is consistent with deposition at the foot under the influence of gravity to an outer zone of marine andesite volcanoes. of sedimentation. In this system commonly as- Some of the Santo rubble flows apparently sumed relationships between the competence stripped and incorporated the floor over which of a clastic source and its extent and relief are they moved, for partially detached masses of not valid. For as long as lava continues to reach underlying sediment are occasionally to be seen the surface, the source will undergo continuous at the base of rudites bearing often abundant, regeneration. Nor, presumably, does the com- contorted, and frayed clasts of stratified arenite petence of the source depend on its degree of and siltstone. emergence, for a volcano may be totally sub- That the slopes of the depositional environ- merged and yet yield voluminous clastic ma- ment itself, as well as of the source area, oc- terial by pyroclasis or gravitational collapse. casionally became unstable (possibly as a con- When sedimentation in an andesitic volcanic sequence of syndepositional tectonic move- archipelago outside continental influence is ments) is indicated by occasional "intra-forma- viewed as a whole, the source and depositional tional" rudites in successions of well-stratified environments may be regarded as coextensive, rocks, consisting of clasts of stratified sediment, and all elastic material may be regarded as in- akin to that occurring above and below the digenous. unit, in a sandy or silty matrix.

THE SANTO CLASTIC ROCKS IN Detailed Provenance TERMS OF THE VOLCANIC Where were the rubble flows and turbidity ARCHIPELAGIC ENVIRONMENT currents initiated and what sort of materials were involved? The extremely common, inti- Environment of Deposition mate mixture of lava-derived elastic material What do the sorting and structure of the and shallow, neritic organic material suggests Santo clastic rocks tell us about their mode of that many slumps were initiated at or near sea transport and emplacement and, consequently, level, and this conclusion is reinforced by the about the environment of deposition? The un- commonly incorporated, moderately to well- sorted and poorly sorted nature of the rudites, rounded clasts, presumably rounded by wave the commonly tabular nature of the beds, and action. Homogeneous lava rudites consisting of the complete absence of shallow-water sedi- lava clasts of uniform composition showing uni- mentary structures excludes transport and formly slight or no rounding are presumably deposition by traction currents, while the size the product of fragmentation of a single lava of the larger clasts in most rudites exceeds the body. Lava rudites consisting of clasts of varied probable competence of turbidity currents (10 composition showing uniformly slight or no cm, according to Kuenen, 1956). While the rounding may be in many instances the product REFERENCES CITED 1287 of the wholesale collapse of a segment of the in some instances as a consequence of sudden volcanic edifice. The calcirudites are believed to loading of a reef by lava. Arenites consisting of be the products of the collapse of a reef mass or clasts of varied composition showing a varied the slumping of reef talus. Some of the mixed degree of rounding are probably the product of rudites presumably result from the joint col- slumping of littoral deposits and of unstable lapse of lava-derived and reef material, perhaps talus accumulated just below sea level.

REFERENCES CITED Fairbridge, R. W., 1950, Landslide patterns on oceanic volcanoes and atolls: Geog. Jour., v. 115, p. 84-88 Fenner, C. M., 1937, Tuffs and other volcanic deposits of Katmai and Yellowstone Park: Trans. Am. Geophys. Union, v. 18, p. 236-239 Fiske, R. S., and Matsuda, T., 1964, Submarine equivalents of ash flows in the Tokiwa Formation, Japan: Am. Jour. Sci., v. 262, p. 76-106 Jaggar, T. A., 1908, The evolution of Bogoslof Volcano: Am. Geogr. Soc. Bull., v. 40, p. 1-16 Kent, P. E., 1966, The transport mechanism in catastrophic rockfalls: Jour. Geology, v. 74, p. 79-83 Kuenen, P. H., 1935, Contributions to the geology of the East Indies from the Snellius expedition: Leidsche. Geol. Med., v. 7, p. 273-331 — 1950, Marine geology: New York, John Wiley and Sons, Inc., 568 p. — 1956, The difference between sliding and turbidity flow: Deep-Sea Research, v. 3, p. 134-139 — 1964, Deep sea sands and ancient turbidites in Bouma, A. H., and Brouwa, A., Editors, Turbidites: Developments in Sedimentology, v. 3, Amsterdam, Elsevier Menard, H. W., 1964, Marine geology of the Pacific: New York, McGraw-Hill Book Co., 271 p. Moore, J. G., 1964, Giant submarine landslides on the Hawaiian Ridge: U. S. Geol. Survey Prof. Paper 501-D, p. 95-98 Newell, N. D., Rigby, J. K., Fischer, A. G., Whiteman, A. J., Hickox, J. E., and Bradley, J. S., 1953, The Permian reef complex of the Guadeloupe Mountains region, Texas and New Mexico: San Francisco, W. H. Freeman and Co., 236 p. Pettijohn, F. J., 1957, Sedimentary rocks: New York, Harper and Bros., 718 p. Richards, A. F., 1960, Rates of marine erosion of tephra and lava at Isla San Benedicto, Mexico: 21st Int. Geol. Cong. (Copenhagen), Rept., pt. 10, p. 59-64 Swanson, D. A., 1966, Tieton Volcano, a Miocene eruptive center in the southern Cascade Mountains, Washington: Geol. Soc. America Bull., v. 77, p. 1293-1314

MANUSCRIPT RECEIVED BY THE SOCIETY JUNE 17, 1966 (Reprinted from Nature, Vol. 212, No. 5062, pp. 586-588, November 5, 1966)

INTRAGLACIAL VOLCANOES OF SOUTH-WEST ICELAND AND THEIR SIGNIFICANCE IN THE INTERPRETATION OF THE FORM OF THE MARINE BASALTIC VOLCANOES*

By .124. J. G. JONES Department of Geology, Imperial College of Science and Technology, London, S.W.7

Mosr of the basaltic volcanoes on the Earth's surface stand largely or completely submerged beneath the oceans and have grown largely or wholly in water. Oceanography has provided information on the gross morphology of submerged basaltic volcanoes and the submerged portions of insular basaltic volcanoes. Yet interpretation of this information is seriously hampered by the meagre data, from dredging and photography, on the detailed constitution of the volcanoes and by the present impossibility of direct observation of submarine basaltic volcanism except in extremely shallow water. Quaternary basaltic volcanoes in Iceland1,2 and in British Columbia' have been described which are believed to have grown within bodies of water melted by volcanism in icesheets which have since withdrawn or vanished. In the hope that volcanoes of this kind would throw light on the processes and products of sub-aqueous and particularly submarine basaltic volcanism, I carried out investigations of the morphology, structure and genesis of a group of such volcanoes in the Laugarvatn area of south-west Iceland. Of the volcanoes studied some grew entirely within water, their activity limited to quiet effusion, and these consist entirely of pillyw.ilvtLand breccias—the last .8g3 a result of fragmentatioicb3i'grAtational collapse. Other volcanoes, which apparently rose to or just above a water surface, show evidence of explosive activity in bedded vitric tuffs which typically mantle pillow lavas of an earlier, purely effusive phase. Still other volcanoes • It should be emphasized that all data on marine basaltic volcanoes are drawn from volcanoes in an oceanic setting, mostly in the Pacific Basin, and that data from island aro volcanoes have been excluded. The Icelandic volcanoes which I believe hold the key to the interpretation of these data consist wholly of olivine basalt. breached a water surface and continued their activity for a considerable time in air and, in these, explosive activity was quickly superseded by the quiet effusion of aa and pahoehoe flows. The constitution of the Laugarvatn volcanoes suggests that the typically quiet effusive character of aerial basaltic volcanism is also characteristic of aqueous basaltic volcanism (except in shallow water) and that pillow lava is - the' typical product of the latter. The explosive character of the phase of emergence in the Laugarvatn volcanoes would also seem typical of the phase of emergence of marine basaltic volcanoes as recently exemplified in the development of Capelinhos in the Azores and Surtsey off the coast of Iceland. Basaltic volcanism in shallow water is typically explosive, as wit- nessed by the tuffaceous character of parasitic vents at the shores and on the shallowly submerged flanks of insular basaltic volcanoes. One of the most striking features of the Quaternary volcanic belt of south-west Iceland is the contrasting forms of those volcanoes which have grown within icesheets and those which have not. While basaltic fissure eruptions in air have built long, low- (slopes less than 10°), wide mounds of shield-like profile consisting of aa and pahoehoe flows, fissure eruptions within an icesheet in which basaltic lava .has been poured into a body of meltwater have built long, steep-sided (slopes up to 300), narrow piles of pillow lavas (commonly mantled by vitric tuffs) which now stand above the surrounding country as prominent ridges. If pillow lava is typical of aqueous basaltic volcanism except in very shallow water, then one might anticipate - that marine basaltic volcanoes the growth of which has been wholly submarine should have slopes notably steeper than those which chaiacterize the emerged portions of island volcanoes. An inspection of true-scale profiles of submerged marine basaltic volcanoes for which there is no evidence of previous emergence (that is, excluding volcanoes terraced or truncated at sea. level) shows_ that this is so; however; there is unfortunately little published information on slope values. Two of the three conical seamounts described by Northrop and Frosch4, which are believed to be basaltic volcanoes, have summit slopes of 23° and 25° and the minimum slope value quoted for the flanks is 13°. According to Wentworth and Macdonalds the aerial slopes of basaltic volcanoes of the island of Hawaii do not generally exceed 10° and there seems little to support Menard's° assumption that submerged marine basaltic volcanoes are simple shields consisting of pahoehoe flows. Stearns7 has suggested, presumably on the basis of the observed explosive character of basaltic volcanism in shallow water, that -aqueous basaltic volcanism at any 2 depth is explosive. This, however,, seems unlikely on theoretical groundss and is certainly contrary to observation in the Laugarvatn region. Nayudu9 has suggested an endogenous mode of growth for submerged marine basaltic volcanoes akin to the growth of a tholoid. I have not found, however, any indications of a coherent mass of lava at the core of the Laugarvatn volcanoes. A dyke exposed at the crest of one elongate pillow lava rifle in the Laugarvatn area can be seen in direct connexion with outward dipping pillow lavas. This dyke is clearly the feeder and suggests a mode of growth of pillow lava piles similar to that of piles of pahoehoe flows. In the Laugarvatn region those volcanoes which breached a water surface and continued their activity for some time in air have a distinctive "tablemountain" form with steep flanks (up to and sometimes exceeding 35°) bounding an upper surface of comparatively gentle slopes (less than 10°). This morphology reflects a structure in which a capping of as and pahoehoe flows rests on a foundation of palagonitized glassy breccias often containing lava masses of crudely "pillowy" form. The most significant aspect of the structure of these volcanoes lies in the relationship between the overlying gently inclined or horizontal capping lavas and the underlying breccias which close to the base of the lavas-typically exhibited crude stratification of variable inclination. The passage between the breccias and lavas is gradational—the breccias have the same relationship to the lavas as do the foreset to the topset beds in a delta. Relationships of this kind result when lavas extruded in air advance into water" and the horizontal plane defined by the transition from lava to breccia marks the water level of the time. Where stratification can be detected these "flow-foot" breccias show outward dips. If the Laugarvatn "tablemountains" are typical of emerged basaltic volcanoes, then the submarine foundations of the lava shields of insular basaltic volcanoes may be expected to consist of flow-foot breccias and vitric tuffs of the emergent explosive phase, resting in turn on a pile of pillow lavas (Fig. lb). High volcanoes rising from the fleets of the oceans, however, tend to subside so that the submarine foundations of insular basaltic volcanoes are presumably rarely exposed to view and, to my knowledge, remain undescribed. Such a structure, if it stood without significant subsidence for a geologically short period of time, would be readily and rapidly decapitated by marine erosion, which would result in a flat topped marine volcano—a guyot (Fig. lc). Like the marine basaltic volcanoes which give no evidence of emergence at any stage, the slopes of guyots and of the submerged foundations of insular basaltic volcanoes seem to be significantly steeper than those of the aerial portions of 3 product 'of • tuCCetiave eruptive • phOses during tubsidenci

vitric tuit of emergent explosive p/10s. e.

(a) submerged

• FIT. 1. Postulated structure of marine basaltic volcanoes (vertical scale exaggerated). 4 emerged volcanoes as would be expected in terms of the proposed structure. (For submarine slopes of the Hawaiian Ridge and volcanoes in the Gulf of Alaska see refs. 11 and 12; for Bikini Atoll see ref. 13.) Tho flanks of the Laugarvatn volcanoes are typically mantled by the scree to which pillow lavas and breccias give rise, and it is the angle of repose in air which imposes an upper limit of 30° on the slopes of pillow lava scree and a slightly higher limit on breccia scree. Such scree must be prevalent on -the submarine flanks of marine basaltic volcanoes. Nearly all values for the inclination of sub= marine basaltic volcanic slopes so far reported fall below 25° and the values for the steepest slope are quoted by many authors to be between 22° and 25°. It seems probable that the value of 25° is the angle of repose in water of basaltic scree. The profiles of many guyots (for example, some of those of the Mid-Pacific mountains") show sudden sharp downward decreases in the ,angle of slope of the upper flanks. A similar break in slope has been noted on several of the "tablemountains" of the Laugarvatn region where "flow- foot" breccias lap on to underlying pillow lavas in the manner shown in Fig. lb, and such a relationship may well account in some instances for this detail of guyot morphology. Tho profiles of some guyots, particularly some in the north-east Pacific9.'5, show marked shoulders on the flanks. These shoulders might be taken to be profiles of encircling terraces but, according to Menard and Ladd": "Detailed surveys show that terraces that can be traced around a seamount or. guyot are very rare at best." Nayuche has suggested that the terraces of Cobb Seamount and Bowie Bank are primary volcanic features rather than wave planation surfaces; however, the reasons he gives for this are not compelling and the suggested mode of origin is difficult to envisage. Ho infers that the presence of "primary" palagonitized basaltic fragmental rocks on these terraces, as revealed by dredging, makes wave planation an unlikely explanation and envisages the formation of terraces_airto occur below and independent of A96, level by accumulation of "palagonite tuff erupted on a sloping surface". In fact, marine erosion will operate regardless of the primary or secondary nature of the exposed material. A possible explanation for some of these imperfect terraces is that they formed at sea level by the advance of aerially extruded flows into water, and were carried below sea level by subsidence during quiescent phases of intermittent volcanic activity. Early constructional terraces thus formed might be overridden in places by the flows and flow-foot breccias of subsequent- erup- tions and this would account for their lateral impersistence (Fig. 1d). The presence of palagonitized breccias on the surfaces of such terraces, as on the Cobb and Bowie 5 volcanoes, could readily lie explained by the erosional - .stripping of the • topset lavas from foreset, flow-foot breccias at sea level before the constructional terrace subsided below the and zone. _ According to Menard°: "gravity observations suggeit that submarine volcanoes have a relatively low average density", and "seismic refraction measurements show that 'the, gross physical properties of near-surface layers of Pacific Basin volcanoes are quite different from hand specimens of the basalt of which they are constiacted". - In terms of his pahoehoe shield model of submarine basaltic volcanoes Menard suggests that draining of pahoehoe flow units may explain these anomalies. I suggest That a volcano constructed of pillow lavas, containing, as they may, considerable interpillow space, and of breccias, might be expected to show greis physical properties quite different from a hand specimen of the basalt of which it was constructed. The volcanoes of the Pacific Basin are 'surrounded by ' aprons characterized by a smooth slope of less than 2° which grades into the volcano flanks° The outer parts of these "archipelagic aprons" are as level as abyssal plains and their profiles are much the same as the lowermost slopes of large terrestrial volcanoes. The significance of these aprons lies in, the volume of material they contain which is many times that of the encircled volcanoes. The form and seismic properties of these aprons are consistent with sedimentary accumulations emplaced by turbidity currents and allied processes. Menard' has, however, rejected this explanation because: "considerations of volume of aprons and of the islands available as sources of debris rule out any major contribution from this source". The assumption that erosion of the emergent portions of insular volcanoes provides the only potentially significant source of elastic material, however, is extremely unlikely. Breccias which result from the gravitational collapse of pillow lavas either during or after cooling and consolidation are prominent associates of pillow lavas in the Laugarvatn.volcanoes. Gravitational collapse would most likely be particularly prevalent on the steep flanks of tall _ submarine basaltic volcanoes and the avalanches of pillow debris thus formed would come to rest at the foot of the volcano, building an apron. On such volcanoes as reach sea level explosive activity builds piles of ash which, as observation of Capelinhos" has shown, are particularly susceptible to collapse and will further contribute to the building of an apron. Moreover, on those volcanoes which succeed in establishing themselves above sea level, continually accumulating unstable banks of flow-foot breccia will provide a ready source of elastic material. Menard' suggests that the archipelagic aprons are plains of "fluid fissure flows" fed by fissures at the base of high volcanoes; however, he notes that "the surface of the aprons is

6 - smoother than any known lava plain . . .". In fact, it seems probable that pillow lava is the form typically assumed by fluid basalt lava in an aqueous environment and„t shows no tendency to spread out in the manner of pahoehoe flows. In conclusion I would suggest that the morphological similarities between the volcanoes of the Laugarvatn region and marine basaltic volcanoes are sufficiently strong to warrant the further evaluation of the following ros tulates: (1) That pillow lavas and their fragmental derivatives are the typical constituents of submarine basaltic volcanoes, except in very shallow water. (2) That pillow lava piles grow upward in a manner akin to piles of pahoehoe flows. (3) That the lava shields of emerged marine basaltic volcanoes stand on a pedestal of flow-foot breccias. (4) That some guyots result from the erosional stripping of a lava shield from its breccia pedestal. (5) That the flanks of marine basaltic volcanoes are typically mantled with scree, the presence of which is reflected in maximum slope values commonly greater than 20° but less than 25°. (6) That some terraces on guyots are volcanic constructional forms built at sea level. (7) That archipelagic aprons are accumulations of elastic material consisting largely of basalt breccia and vitric tuff. Kjartanason, G., Mus. Nat. Hid., Reykjavik, Iceland, Misc. Papers, 38 (1964). Bemmeien, R. W. van, and Button, M. C., Tablemountains of Northern Iceland (E. J. Brill, Leiden, Netherlands, 1955). ' Mathews, W. H., Amer. J. Sci., 245, 560 (1947). Northrop, J., and Froseh, R. A., Deep Sea Res.,1, 252 (1951). Wentworth, C. K., and Macdonald, G. A., U.S. Geol. Survey Bull., 994 (1953). 'Menard, H. W., Marine Geology of the Pacific (McGraw-Hill, New York, 1984). Stearns, H. T., Hawaii Div. Hydrog. Bull., 8, 1 (1946). McBirney, A. R., Bull. Vokanol., 26, 455 (1963). • ▪ Nayudu, Y. R., Amer. Geophys. Union Monograph, 6, 171 (1982). "Fuller, R. E., Amer. J. Sri., 21, 281 (1931). Moore, J. G., U.S. Geol. Survey Profess. Paper, 501-D (1984). "Dietz, R. S., and Menard, H. W., J. Geol., 61, 99 (1953). " Emery, K. 0., et al.. U.S. Geol. Survey Profess. Paper 260-A (1954). 14 Hamilton, E. L., Geol. Soc. Amer. Mem., 84 (1956). "Murray, H. W., Bull. Geoff. Soc. Amer., 52, 333 (1941). " Menard, H. W., and Ladd, H. S., in The Sea, 3 (Interscience, New York , 1963). " Zbyszewski, G., Bull. Vokanol., 23, 77 (1960).

Printed in Great Britain by Fisher, Knight & Co., Ltd., St. Albans.