AMCS Bulletin 19 / SMES Boletín 7 — 2002 35

Small Subcrustal : Examples from , * Ken G. Grimes

RRN 795 Morgiana Rd., Hamilton, Victoria, 3300, Australia; [email protected] * Presented in 2002, revised in 2006.

Introduction Peterson & others (1994), Hon & oth- them pressure ridges). Unfortunately, ers (1994) and Kauahikaua & others on the relatively old (20-40,000 year) This paper discusses the formation of (1998) and which is illustrated in Figure flows in Victoria weathering and veg- a type of small lava that forms by 3. Recently, Halliday (1998a & b) has etation growth has reduced much of the “Subcrustal drainage” of lava flows. Ex- described two types of small lava cave: surface to a cracked and jumbled rubble. amples will be drawn from the Western His “sheet flow caves’ and ‘hollow vol- Thus, definitive axial clefts are difficult District Volcanic Province of western canic tumulus caves’ which he regards to identify and the new (genetically- Victoria, Australia (Figure 1). Caves as being distinct. I will argue that these from that province are numbered in are probably just two of several possible the Australian Index with a “3H” members of a continuum of forms which prefix, abbreviated here to just “H” e.g. have been referred to as “Subcrustal H-70 (Matthews, 1985). lava caves” (e.g. Stevenson, 1999). In a review/study of active volcanoes in Hawaii, Peterson & others (1994) pro- The terminology of surface lava flow posed two distinct models for the forma- features and their caves has become tion of lava tubes: either by the roofing rather complex and confusing in recent over of linear surface lava channels (Fig- years, so I will list here some terms - and ure 2); or by the draining of still molten my intended usage. material from beneath the solidified crust Surface lava features — what is a of pahoehoe flow lobes (Figure 3). The tumulus? The changing usage of “tumu- former process produces relatively large lus” affects the definition of a “tumulus” and simple lava tubes. However, this cave ! Walker (1991) gave the term paper will concentrate on the smaller, “tumulus” a genetic definition which but commonly complex, caves formed both expanded the term to incorporate by localisation of flow beneath the crust all lava rises, including elongated ridges, of thin flow lobes or sheet-flows, and and narrowed its usage to those rises subsequent partial draining - a process that show evidence of inflation, given that has been progressively recognised by opened axial clefts on the crest, but and described by Peterson & Swanson which have no evidence of lateral com- (1974), Wood (1977), Greeley (1987), pression (if there was, Walker would call

Figure 1. Location map of the Western District Volcanic Province, and its main lava Figure 2. Three ways of forming large cave areas. lava tubes by roofing a surface channel. 36 AMCS Bulletin 19 / SMES Boletín 7 — 2002 based) usage of the term “tumulus” has mix of his “tumuli”, “pressure ridges”, an example of the latter type. limited use. In Victoria the usage of “lava rises” and “lava rise pits”. In Victoria, speleologists have used “tumulus” has always been restricted For this paper I will use the descrip- the term “blister cave” for the small, to the distinctive, steep-sided, roughly- tive, non-genetic, term “lava mound” simple, isolated chambers found under circular mounds described by Ollier to describe all high areas within a lava the stony rises (Figure 4). However, (1964) in the Harman valley (many of field. Lava mounds are likely places care is needed to avoid confusion with which do have obvious large summit for the formation of small drained sub- another usage of that term for small clefts) and the more general term “stony crustal lava caves, whatever the process chambers formed by gas pressure (Gib- rises” is used for the chaotic complex of mound formation. son, 1974, and Larson, 1993). I suggest of broader hummocks and hollows that Cave types: In any discussion of lava usage of lava blister for those inflated by occur on many of Victoria’s younger caves and their genesis it is important to liquid lava (and later drained), and gas lava surfaces. This local usage of “stony distinguish between active (lava-filled) blister for those generated by gas pres- rises” would seem to correspond to the proto-caves and the drained tubes and sure. “Blister” should only be used on “hummocky pahoehoe” of Hon et al chambers (i.e. caves) which appear at its own where the genesis is uncertain. (1994) but some have relatively flat the end of the eruption – as discussed surfaces that correspond to their “sheet by Halliday (2004). The basaltic Western District Volca- flows” and there are also transitional In an earlier paper (Grimes, 1995) I nic Province (previously known as the forms. In Walker’s (1991) terminology described complex, lateral, levee-breach Newer Volcanic Province) of western the Victorian “stony rises” represent a systems associated with lava channels at Victoria has over 400 identified eruptive Mount Eccles, and distinguished them points and it ranges in age mainly from from smaller, isolated, drained chambers Pliocene (about 5 Million years) up to in the surrounding “stony rises” but very recent times (5ka), though there are did not suggest a formal nomenclature. some volcanoes as old as 7 Ma (Joyce, The terminology of Halliday (1998a,b), 1988, Joyce & Webb, 2003, Price & which is based on the surface lava flow others, 2003). Lava caves are known character (after Walker, 1991), is difficult across the whole province (Figure 1), to apply to the Victorian subcrustal caves but are most common in the younger because of the problem in distinguish- flows associated with Mount Eccles ing “tumuli” (sensu Walker, 1991) from (20-33 ka, Head, & others, 1991, and P. other lava mounds. I also suspect that Kershaw, per comm, 2005) and Mount rather than two discrete genetic types Napier (about 32 ka, Stone & others, of small subcrustal cave as proposed by 1997). Recent summaries of both the Halliday (“sheet flow caves’ and ‘hollow surface landforms and the volcanic caves volcanic tumulus caves’), we have a of the province appear in Grimes (1995, broad continuum of forms with a number 1999); and Grimes & Watson (1995). of distinctive end-members. The earlier literature on lava caves of I suggest that the cave classification the region by Ollier, Joyce and others should not be tied to the surface termi- is reviewed in Webb & others (1982) nology until the processes of cave devel- and Grimes & Watson (1995) and only opment are better known. Also, basing some of those papers are referenced here. the cave nomenclature on the surface The younger lava flows have surfaces lava forms may be confusing cause and ranging from strongly undulating (“stony effect—rather we should be explaining rises”) to flat. some surface mounds and tumuli as a At Mount Eccles the main volcano result of localised subcrustal flow in is a deep steep-walled elongated cra- tubes, not the cause (see conclusion). ter which contains Lake Surprise. At The unifying factor in all these caves is the north-western end the crater wall that they form by drainage from beneath has been breached by a lava channel a broadly-crusted lava flow; hence I that flows west and then branches into will refer to them here collectively as two main channels (referred to locally subcrustal lava caves. as ‘lava canals’) running to the west- In this paper my discussion will northwest and to the south-southwest concentrate on the smaller subcrustal (Grimes, 1995, 1999). Extending to Figure 3. Formation of lava caves by sub- lava caves, those that form originally, the southeast from the main crater there crustal drainage of a series of advancing lava lobes. Step 3a is the situation if the rather than the larger more evolved forms is a line of smaller spatter and scoria source of lava ceases early in the develop- which can develop from them over time cones and craters. Several smaller lava ment; irregular caves form. Steps 3b and and which tend to become closer in channels run out from these. Lava caves 4 indicate the further evolution into more shape to the formed by roofing occur in a variety of settings. linear feeder tubes as lava continues to of surface channels. In Victoria, the Mt. Beyond this central area of explosive flow through the system. Hamilton lava cave (Figure 14) may be activity, basalt flows form a lava field AMCS Bulletin 19 / SMES Boletín 7 — 2002 37 about 16 km long and 8 km across. From The cone contains one group of complex the western end of this lava field a long lava tubes (Ollier, 1963). flow, the Tyrendarra Flow, runs 30 km In the late Quaternary lava flows of southwards to the present coast and Mount Eccles and Mount Napier, in continues offshore for a further 15 km the Western District Volcanic Province, (sea level was lower at the time of the we find both cave types described by eruption). This long flow must have had of Peterson & others (1994) and also a major feeder tube, but no drained sec- isolated “lava blister” caves - I will tions have been discovered to date. draw my examples from those areas. Mt Napier and the Harman Valley The complex lava cave system at Mount flow: Mt Napier, about 20 km northeast Hamilton appears to be a further-evolved of Mt. Eccles, is a steep cinder cone “feeder” system. Figure 5. Turtle Cave, H-90, at Byaduk, capping a broad lava shield 10 km in Most of the longer caves known at is an example of a simple “lava blister” diameter. Some lava caves occur on Mount Eccles are in or adjacent to the cave. The name derives from its resem- the lower slopes of the cone, and on lava channels, but there are a number blance to an empty turtle shell. the lava shield, but the main cluster is of small caves scattered throughout the at the Byaduk Caves, at the start of a area, and the known distribution may (Figure 6); and small, isolated, drained long lava flow that follows the Har- simply reflect the more intensive ex- chambers (“lava blisters”) within the man Valley for at least 20 km to the ploration along the main canals. There stony rises (e.g. H-78; Figure 4). west. Other lava caves occur further are several types of lava cave in the At Mount Napier, and in its long down the valley, as do an excellent set area. Roofed channels include Natural flow down the Harman Valley we find of sharply-defined tumuli (Ollier, 1964). Bridge (H-10; Grimes; 2002b), which both very large tubes (which might be It was at the Byaduk Caves that Ollier has the distinctive “gothic” ceiling of roofed channels, though the evidence is & Brown (1965) derived their ‘layered tubes formed by overgrowth of a levee ambiguous) and many small subcrustal lava’ model of tube formation - which bank (Figure 2c), and also possibly Tun- caves. Some of the small subcrustal is still invoked by some authors (e.g. nel Cave (H-9; Grimes, 1998). The caves are exposed, along with their con- Stephenson, 1999). remainder are shallow, low-roofed caves taining lava flows, at various levels in the Mount Hamilton is a broad lava cone that fall into two types: complex, levee- walls of collapse dolines formed above surrounded by “stony rise” lava flows. overflow systems on the sides of the the large tubes; for example, the upper There is a large lava crater at the summit. major lava channels, e.g. H-51 & H-70 of level of Fern cave (H-23, Figure 13) and H-74 and H-108 (Figure 12). Others are shallow isolated caves on the flow surface (H-31, 90, 91 and 106; Figures 4 & 12). One shallow cave has an open feeder from below that connects to a larger ‘feeder’ tube at depth (H-33, Figure 13).

The shallow lava caves involve a broad array of styles ranging from simple sin- gle chambers to multi-level, complexly- interconnecting systems of tubes and chambers. All gradations occur between these extremes, but the group has in common the dominance of shallow, low- roofed, irregular chambers and small- diameter tubes. They also grade (and possibly evolve over time) into larger and more-linear “feeder” tubes. Thus, while we can identify several distinc- tive types, there are many transitional forms that are hard to classify. Their genesis is discussed in more detail later in this paper. Figure 4. Examples of small, simple, subcrustal caves; mostly Simple drained lava mounds and associated with low lava mounds. H90, 91 and 107 would be “lava blister” caves: Scattered through called “lava blisters”; H-78 is a “peripheral remnant” left by the the stony rises there are small, shallow, collapse of the roof of shallow chamber; H-31 is approaching low-roofed chambers; typically only 1m a linear “tube” form; H-11, 33 and 106 are grading to the more high with a roof 1m or less thick. These complex forms. can be circular, elongate or irregular in 38 AMCS Bulletin 19 / SMES Boletín 7 — 2002 plan; up to 10m or more across but grad- Cave (H-31, Figure 4) described by Ol- can squeeze through the rubble into a ing down to small cavities only suitable lier & Joyce, 1968, p70. small ‘peripheral remnant’ cave. for rabbits. Some Victorian examples These caves generally are found be- More complex overflow caves as- are shown in Figure 4, and include Turtle neath low lava mounds (with or without sociated with the lava channels at Mt. Cave (which looks like an empty turtle the central clefts required to class them Eccles are generally shallow systems shell) illustrated in Figure 5. In section, as “tumuli”!), though in some cases the formed in the levee banks on each side the outer edges of the chambers may be surface relief may only rise half a metre! of the channels and would have fed smoothly rounded or form a sharp angle These small simple chambers have been small lateral lava lobes or sheets when with a flat lava floor. The ceiling may be locally called “blister caves” (see discus- the channel overflowed or breached arched or nearly flat, with lava drips, and sion in the Terminology section). through the levee (Grimes, 1995). Fig- sometimes has a central “soft” sag that A large cluster of -defined tumuli ure 6 shows the lateral caves associated would have formed while the crust was (sensu Walker, 1991) occur in the Har- with the South Canal at Mt. Eccles, and still plastic. Commonly, the thin central man Valley (Ollier, 1964). One of these Figure 12 shows a group of shallow part of the roof has collapsed and we is reported to be hollow by G. Christie caves adjacent to a large collapsed feeder find only a peripheral remnant hidden (pers comm) who entered it as a child, tube at Byaduk. behind rubble at the edge of a shallow but has not been able to relocate it. There Some of these lateral caves are simple collapse doline (e.g. H-78, Figure 4). is a ‘donut’ shaped tumulus which pre- linear tubes (e.g. H-48, 89, and the proxi- The more elongate versions grade into sumably has resulted from collapse of a mal part of H-53), but mostly they are small “tube caves”; for example, Shallow central hollow. Within its annulus, one branching systems with complexes of

Figure 6. Examples of more complex subcrustal caves formed in thin overflows from a lava channel at Mt. Eccles. See Fig. 7 for detail of H-70/72. A detailed map of H-51 is included in the supplementary material on the CD. AMCS Bulletin 19 / SMES Boletín 7 — 2002 39 Figure 7. Detailed map of Carmichael Cave (H-70) at Mt. Eccles. 40 AMCS Bulletin 19 / SMES Boletín 7 — 2002

exposed in the wall of a large collapse doline (H-74, 106 & 108; Figure 12). The elongated doline formed over a deeper large feeder tube (up to 25 m wide and 15 m high) and the thin flows may have been fed by overflows from the feeder tube, through roof windows. The three shallow caves comprise low- roofed branching passages and chambers very similar to those found beside the channel at Mount Eccles (Figure 11). In the lowest cave (H-74) there are intru- sive lava lobes that may have entered through roof holes from the overlying lava flow. Likewise, in the next high- est cave (H-108) a lava fall drops a metre to a short section of lower-level Figure 8. Two-level passage in H-70, looking south from section X22 (Fig. 7). Note the passage that might be in the same flow “window” on left which might be the remains of a partition between two lava lobes. as H-74. More complex stacked systems also low passages that bifurcate and rejoin, are late-stage additions; running over occur. These can be fed from below, or open out into broad low chambers. an earlier pahoehoe floor. through a skylight in a major feeder tube, The shape suggests draining from be- Where not disrupted by breakdown, or laterally from a remote source. The neath the thin solidified roof of a series the walls and roof typically have thin upper level of The Theatre (H-33) is a of coalesced flow lobes. Only a few of (2 - 20 cm) linings. These conceal the small subcrustal cave system obviously the passages are large enough to stand in, original wall, but in a few places fallen fed from below as the shallow branching typically (but not always) those nearest linings have exposed “layered lava” tubes occupy an isolated raised mound the proximal end - the channel entrance comprising thin sheets with ropy or and a drain-back tube allows access to (e.g. H-48, H-53, H-70). Most passages hackly surfaces (eg the proximal end of lower levels of low-roofed chambers and are crawl-ways about a metre high with H-70). Most caves are at a single level, eventually to a large feeder tube at depth low arched roofs and flat lava floors (Fig- but some show evidence of several lev- (Figure 13). Lava would have welled ure 8). Some of the smallest passages els (only a metre or so apart vertically) up from this lower level and formed have smoothly-rounded cross-sections that either have coalesced vertically the surface rise in several stages (the (Figure 9). The ceiling is generally only into a single passage or chamber or are different “levels”), then drained back a metre or so below the present surface, joined by short lava falls (e.g. H-48, to leave the small tubes and chambers. and in places breakdown has exposed H-70 (Figure 7) and H-108). Fern Cave (H-23) comprises a large the base of overlying pahoehoe flows, At Byaduk, three caves occur in a ‘feeder’ tube at depth, but there is a indicating that the original roof was less stacked set of thin, 1-3m, lava flows higher level of low-ceilinged irregular than a metre thick. In some chambers the roof has sagged down in a smooth curve to reach the floor (Figure 10). Where not covered with introduced soil, the floors are generally pahoehoe, with smooth, platy or ropy surfaces; but sharp aa lava floors occur in several places (e.g. H-51 and H-70). Some of these

Figure 9. Small subcrustal tube, H-52 at Figure 10. Chamber in H-74, showing sagged parts of roof. Mt. Eccles. AMCS Bulletin 19 / SMES Boletín 7 — 2002 41 chambers and passages which appears to be in a younger flow that ran over the prior roofed tube (Figure 13). This flow would seem to have been fed from the large collapsed tube to the south, which might have been an open channel at that time. The present connections between the upper and lower levels of Fern cave are later accidents of collapse of the lower tube roof. The Mount Hamilton Cave (H-2) is a complex system of moderately large bifurcating tubes at several levels (Figure 14; Ollier 1963, Webb et al, 1982). It is dominated by linear tubes rather than the Figure 11. Low chamber in H-106, with lava drips and an intrusive lava tongue at left. broad low chambers typical of most other caves considered in this paper and may cave. Collapse of the roof can occur subcrustal lava tubes (e.g. Figure 12) but indicate a more evolved style of larger while the tube is active, as well as after those had not been mapped at the time of subcrustal lava cave (see below). it is drained. Ollier & Brown’s report. However, Ol- Genesis Ollier & Brown (1965) used the Vic- lier & Brown also referred to a still-finer torian lava caves, in particular those at layering within what are now recognised When discussing genesis one must Byaduk, to propose a “layered lava” as flow units - marked by sub-horizontal keep in mind the distinction between model of tube development. This is cracks, trains of vesicles, and small flat- active tubes (lava-filled) and drained similar to the more recent subcrustal tened cavities which may have stretch tubes (caves) – as discussed by Halliday models of Hon & others (1994) but their structures or small lava drips. They (2004). Only some active tubes will be concept of “layered lava” is confusing rejected the suggestion that separate drained and become accessible at the end as they seem to apply that term to two flow units were present, and believed of an eruption, most will remain filled distinct types of “layer”. The ex- that all the layers were “formed by dif- and solidify. As long as a tube or cavity posed in the collapse dolines at Byaduk ferential movement within one thick remains active, its form can evolve by, have flow units from 0.5 to 5m thick that lava flow” (not within thinner flow units) firstly, mechanical and thermal erosion are distinguished by lobate ropy surfaces and that they were “possibly shearing of its edges; secondly, solidification of at top and bottom, with small gaps and planes formed during flow just before its stagnant parts including linings, and partings between them and local areas solidification”. They recognised that the thirdly, partial drainage to form an open of rubble. These flow units host small flow somehow become differentiated

Figure 12. A set of three small subcrustal caves formed in separate stacked lava flows at Byaduk. Detailed reports and maps on H-106 and H-108 are in the supplementary material on the CD. 42 AMCS Bulletin 19 / SMES Boletín 7 — 2002

Figure 13. Complex, multi-level lava caves at Byaduk. Shallow subcrustal systems overlie large feeder tubes at depth. into solid parts and liquid-filled tubes, flow. The figure shows the case where continuing flow of hot lava through but a detailed understanding of how this a channel has overflowed, as along the a small feeder tube can enlarge it by happened had to wait on the observations Southern Canal at Mount Eccles, but erosion of the walls or floor (Peterson of active tube-fed lavas by later worker similar effects occur at the front of an & Swanson, 1974; Greeley, 1987). De- (e.g. Peterson & others,1994 and Hon advancing pahoehoe lava flow where the struction of the crust above the active & others,1994). Ollier & Brown did, lava is delivered by a channel or major tube can form skylights or local surface however, recognise that the tubes, once feeder tube, but then spreads out into a channels, and overflow from these can formed, could enlarge by eroding the series of lobes. These lobes grow by a form secondary flow lobes. Thus, pa- surrounding (layered) lava rock. process of ‘budding’ in which a small hoehoe lobes can be stacked vertically For a more detailed description of lobe develops a skin, and is inflated by as well as advance forwards so that the observed processes seen in active the lava pressure until the skin ruptures a complex three-dimensional pattern lava flows, and models deduced from in one or more places. Lava escaping of branching tubes and chambers can these, see Peterson & Swanson (1974), through the rupture develops new lobes form. Wood (1977), Greeley (1987), Peterson and so on (Figure C-1, 2, 3). If the sup- H-53 could be regarded as showing & others (1994), Hon & others (1994) ply of fresh lava is cut off, the still-liquid a transition from the low branching and and Kauahikaua & others (1998). The parts of a lobe may be drained to form a chambered systems at the (younger) dis- model used here is essentially that de- broad but low-roofed chamber (Figure tal end, to the more linear unbranching scribed in the last three of those papers C-3a). However, if fresh hot lava con- tube systems at the proximal end that (Figures 2 & 3). tinues to be delivered from the volcano it would develop in time as flow becomes Isolated small “lava blister” caves may become progressively concentrated more localised and organised to feed an found beneath low mounds in the “stony into linear tubes that feed the advancing extensive overflow sheet. The proximal rises” would form by the irregular drain- lobes, while the remaining stagnant areas end of this cave approaches the character ing of cavities beneath the thin solidified solidify (Figure C-3b, 4, 5). of a ‘roofed channel’ tube and determin- crust of a broad lava flow. The process Tubes formed by draining of lava ing the origin of simple large lava tubes is similar to that which forms other lobes and flows are generally smaller can be difficult as much of the evidence subcrustal drainage tubes (see below), than those formed by the roofing of a may have been removed by erosional but less organised so that only isolated channel – although inflation of the flow enlargement of the original tube, or be low-roofed chambers appear to result can provide a thickness of ten metres or hidden behind wall linings. beneath the high points of the undulating more in which larger subcrustal drain- The Mount Hamilton Cave (H-2, Fig- surface. Commonly the chamber roof age tubes can form. However, if flow ure 14) may be a further-evolved system sags (while hot) or later collapses so that continues after they are formed, several in which the original irregular chambers only a crescentic ‘peripheral remnant’ small tubes within a lobe complex may and small passages of subcrustal drain- survives, as at H-78 (Figure 4). coalesce by breakdown of their thin walls age caves in several stacked flows have Figure 3 illustrates the formation of or floors (the “partitions” or “septa” of combined and evolved into a more lin- more complex tubes and cavities by sub- Hon et al, 1994, and Halliday, 1998b) ear system of larger “feeder” tubes as crustal draining from beneath a crusted to form a larger feeder tube. Also, a lava flow continued through the conduit AMCS Bulletin 19 / SMES Boletín 7 — 2002 43 system on its way to the lava field be- low. This suggestion is supported by the presence of small ‘proto-tubes’, 20-60 cm in diameter, that are exposed by breakdown in the walls and ceiling of the larger tubes in several parts of the cave (Figure 15). Conclusion Small subcrustal lava caves form by drainage of lava from beneath a thin crust developed on a lava surface. In its simplest form, drainage of lava from beneath high areas on the crusted sur- face will form simple isolated cham- bers - “lava blisters”. Complex nests of advancing lava lobes create equally complex patterns of active tubes and chambers which can later drain to form open caves. As lava continues to flow Figure 14. Mt Hamilton lava cave is an evolved system of larger, linear, bifurcating, through these complex systems they will subcrustal tubes. evolve by erosion and solidification to form larger, more streamlined, linear pressure in these local areas. A linear Acknowledgements tube systems that act as “feeder tubes” could thus produce a linear to carry hot lava to the advancing lava ‘tumulus’ or a chain of rounded ones. With acknowledgements to my pre- front. If sufficiently evolved, these linear Wider chambers along the line of the decessors who conceived most of the tubes can converge on the form of the, tube would have weaker roofs and hence ideas expressed here: In particular Don generally larger, linear tubes formed by explain the localised nature of the sensu Peterson, Ken Hon, Bill Halliday, and roofing of surface lava channels. Thus stricto tumuli. many other speleo-geologists. This the genesis of many large lava caves The unifying factor in all these caves report draws on the exploration and remains difficult to deduce. is that they form by shallow drainage mapping efforts of numerous speleolo- The “drained tumulus caves” de- from beneath a crusted lava flow - hence gists from the Victorian Speleological scribed by Halliday (1998a) & Walker they can be referred to collectively as Association and other groups over the (1991) would be a special case of the subcrustal lava caves. last 50 years. small subcrustal type in which the crust was pushed up into a tumulus (sensu lato) before it drained. Halliday’s (1998b) “sheet flow caves” are also a special case tied to a particular surface form. I would expect all gradations between these features and the more extensive systems which can form under both flat-topped “sheet-flows” and undulating “stony rises”. I suggest that the cave classification should not be tied to the surface ter- minology until the processes of cave development are better known. Also, basing the cave nomenclature on the surface lava forms may be confusing cause and effect—rather than argue that some types of caves form beneath/in tumuli and others beneath “sheet flows”, it might be better to say that tumuli tend to form above active localised flows within a sheet (i.e. above lava tubes). The hot flowing lava would inhibit thick- ening of the crust above the tube or chamber so that it would be weaker and Figure 15. A pair of small ‘proto-tubes’, with 10 cm thick linings, exposed in the wall of more likely to be uplifted by hydraulic a larger, more-evolved, in the Mt Hamilton lava cave. Scale bar is 10 cm. 44 AMCS Bulletin 19 / SMES Boletín 7 — 2002

References Fitzroy Dearborn, NY., 760-764. OLLIER, C.D. & JOYCE, E.B., 1968: HEAD, L., D’COSTA, D., & EDNEY, P., Further descriptions of Victorian (Note: Nargun is the journal of the Vic- 1991: Pleistocene dates for volcanic lava caves. Victorian Naturalist. 85: torian Speleological Association) activity in Western Victoria and im- 70-75. plications for Aboriginal occupation. PETERSON, D.W., & SWANSON, GIBSON, L.L., 1974: Blister caves as- in WILLIAMS, M.A., DE DEKKER, D.A., 1974: Observed formation of sociated with an Ethiopian Volcanic P., & KERSHAW, A.P. [eds] The lava tubes during 1970-71 at Kilauea ash-flow tuff.Studies in , Cainozoic in Australia, a re-appraisal Volcano, Hawaii. Studies in Speleol- 2(6): 225-232. of the Evidence. Geological Society ogy, 2(6): 209-222. GREELEY, R., 1987: The Role of lava of Australia, Special Publication, 18: PETERSON, D.W., HOLCOMB, R.T., tubes in Hawaiian volcanoes. US Geo- 302-308. TILLING, R.I., & CHRISTIANSEN, logical Survey, Professional Paper HON, K., KAUAHIKAUA, J., DEN- R.L., 1994: Development of lava tubes 1350, 1589-1602. LINGER, R., & MACKAY K., 1994: in the light of observations at Mauna *GRIMES, K.G., 1995: Lava caves and Emplacement and inflation of pahoe- Ulu, Kilauea Volcano, Hawaii. Bul- channels at Mount Eccles, Victoria. hoe sheet flows: Observations and letin of Volcanology 56 343-360. in BADDELEY, G., [Ed] Vulcon measurements of active lava flows PRICE, R.C., NICHOLLS, I.A., & Preceedings 1995. Australian Spe- on Kilauea Volcano, Hawaii. Geo- GREY, C.M., 2003: Cainozoic Igne- leological Federation, . logical Society of America Bulletin. ous Activity (section 12.4.6, Western pp 15-22. 106 351-370. District province), in BIRCH, W.D., GRIMES, K.G., 1998a: Tunnel Cave, JOYCE, E.B., 1988: Newer volcanic (editor) Geology of Victoria. Geo- Mount Eccles. Nargun, 30(10): landforms. in DOUGLAS, J.G., & logical Society of Australia, Special 172-173. FERGUSON, J.A., [eds] Geology of Publication 23: 366-370. GRIMES, K.G., 1999: Volcanic caves Victoria, Geological Society of Aus- STONE, J., PETERSON, J.A., FIFIELD, and related features in western Vic- tralia, Victorian division. Melbourne. L.K., & CRESSWELL, R.G., 1997: toria. in HENDERSON, K., [ed] pp. 419-426. Cosmogenic chlorine-36 exposure Cave Management in Australasia JOYCE, E.B. & WEBB, J.A. (co-ordi- ages for two basalt flows in the Newer 13. Proceedings of the thirteenth nators), 2003: Geomorphology, the Volcanics Province, western Victoria. Australasian Conference on cave and evolution of Victorian landscapes Proceedings of the Royal Society of Karst Management, Mt Gambier, (section 18.10.1, Volcanic Plains). Victoria. 109(2): 121-131. South Australia. Australasian Cave in BIRCH, W.D., (editor) Geology STEPHENSEN, J., 1999: Emplacement and Karst Management Association. of Victoria. Geological Society of and tube development in long tube-fed Carlton South. pp 148-151. Australia, Special Publication 23: lavas in N. Queensland, Australia. *GRIMES, K.G., 2002a: Carmichael 553-554. Proceedings of the 9th International Cave (3H–70): A complex, shallow, KAUAHIKAUA, J., CASHMAN, K.V., Symposium of Vulcanospeleology of “sub-crustal” lava cave at Mount MATTOX, T.N., HELIKER, C.C., the UIS. Centro Speleologico Etneo, Eccles, Victoria. Nargun, 35(2): HON, K.A., MANGAN, M.T., & Catania. pp. 134-145. 13-17. THORNBER, C.R., 1998: Obser- WALKER, G.P.L., 1991: Structure and *GRIMES, K.G., 2002b: Natural Bridge vations on basaltic lava streams in origin by injection of lava under sur- (3H–10), Mount Eccles: a special type tubes from Kilauea Volcano, island face crust, of tumuli, “lava rises”, of lava tube. Nargun, 35(2): 18-21. of Hawai’i. Journal of Geophysical “lava-rise pits”, and “lava-inflation GRIMES, K.G., & WATSON, A., 1995: Research, 103: 27303-27323. clefts” in Hawaii. Bulletin of Volca- Volcanic caves of western Victoria. in LARSON, C.V., 1993: An illustrated nology, 53, 546-558. BADDELEY, G., [Ed] Vulcon Guide- glossary of lava tube features. West- WEBB, J.A., JOYCE, E.B., & STE- book 1995. Australian Speleological ern Speleological Survey Bulletin, VENS, N.C., 1982: Lava caves of Federation. Melbourne, pp 39-68. 87. 56 pp. Australia. Proceedings, Third Interna- HALLIDAY, W.R., 1998a: Hollow MATTHEWS, P.J., 1985: Australian tional Symposium on Vulcanospeleol- volcanic tumulus caves of Kilauea Karst Index, 1985. Australian Spe- ogy, Oregon, USA. pp 74-85. Caldera, Hawaii County, Hawaii. In- leological Federation, Melbourne. WOOD, C., 1977: The origin and ternational Journal of Speleology., 481 pp. morphological diversity of lava tube 27B (1/4), 95-105. OLLIER, C.D. 1963: The Mount Ham- caves. Proceedings, 7th International HALLIDAY, W.R., 1998b: Sheet flow ilton lava caves. Victorian Naturalist. Speleological Congress, Sheffield, caves of Kilauea Caldera, Hawaii 79, 331-336. England. 440-444. County, Hawaii. International OLLIER, C.D. 1964: Tumuli and lava Journal of Speleology, 27B (1/4), blisters of Victoria. Nature. 202, *Papers Grimes 1995, Grimes 2002a, 107-112. 1284-1286. and Grimes 2002b are included in the HALLIDAY, W.R., 2004: Volcanic OLLIER, C.D., & BROWN, M.C., 1965: supplementary material on the CD. Caves, in Gunn, J. (Editor) Ency- Lava caves of Victoria. Bulletin Vol- clopaedia of Caves and Karst Science. canologique. 28: 215-30.