Journal Series 5 Number 1

Note that this is a re-print of the original publication, based on a scanned copy. During the process of converting the original paper copy to this electronic version, the original formatting, page layout and page numbers have been lost. All diagrams and surveys have been scanned from the original and are consequently of poor quality.



Combe Down Freestone Mines, Bath by AJ Butcher and RD Mehew

Goon's Hole, Burrington Combe by MT Mills

Book Review: The West Somerset Mineral Railway and the Story of the Brendon Hills Iron Mines

A Preliminary Investigation of Surtshellir, Iceland by MT Mills and C Wood


Published by the Shepton Mallet Caving Club

The Mineries, Wells Road, Priddy, Wells, Somerset, BA5 3AU



I must, first of all, apologise for the delay in producing this Journal and secondly, apologise for the poor quality of production.

Apart from a late addition to the content, two problems have beset this production. The first concerns the poor processing of electronic stencils, so much so that even after a second batch had been processed, I still consider them to be below the standard which I would like; and secondly the Clubs' duplicator has ceased to function efficiently and therefore it has been necessary to get some of the duplicating done elsewhere. I hope that the above difficulties will be appreciated.

However, in this edition there is the first of two articles on the Combe Down Freestone Mines at Bath. You may notice that the word 'Combe' is also spelt 'Coombe'. I am assured by the authors that both are correct and that the word has not been mis-spelt, although the former is that spelling which is in common usage. I gather that the calculations for this survey are to be worked out by computer!

The second article concerns a discovery made last June on Mendip. It is said that the intrepid discoverer was relieving himself at the time, though as you will discover a certain amount of 'bang' was still required to facilitate ease of passage!

Following, there is a book review, by our usual reviewer, concerning one of cavings' allied interests – mining. This is a second and revised edition concerning the mineral railway linking Watchet with the iron mines in the Brendon Hills.

Finally, a study and survey of Surtshellir, a lava tube in Iceland which was explored during the Clubs' Expedition to Iceland in 1970. This was carried out in one working day during a visit to the Hallmundarhraun lava field in West Central Iceland.

I hope that the contents of this issue will be considered worth the delay.


Combe Down Freestone Mines, Bath

An Introduction

Bath and the surrounding area has numerous workings, both opencast and underground, concerned with the production of Bath Stone. The extent of these workings just before the turn of the century can be seen in Figure 1, which is taken from "The Builder" (ref 1) published in 1895. Of these areas perhaps the best known is at Box where there are considerable underground workings. These have been surveyed and described in various Club publications (refs 2,3,4,5,6). Another area which, like Box, has underground workings is the Combe Down area of Bath. These have not been surveyed or documented to the extent of the Box area although AD Oldham has done some work in the area. Several visits are also reported in various club journals together with the method of working (refs 7,8,9,10,11,12). There are also workings at Odd Down but most of those are above ground.

We set out therefore, to produce a comprehensive survey of the known workings in the area not only for our own edification but to provide the local police with a map to assist them in searching for the children who occasionally become lost in the maze of passages.

The Mendip Cave Registry for the area has been compiled by Mike Baker, and gives the location of some twenty sites mainly situated within an area of about one square mile on Combe Down. Figure 3 shows these sites as well as the position of quarries in the area. Some of these quarries are now filled in. We have decided to use the Registry reference for each site as our survey reference. Where these are used in the text the last two figures only are quoted: (60)50. The workings are frequently linked underground which produces an extensive system of passages and when actually underground it is likely that one is not in the same system of workings as that of the entrance one went in by. Entrances to some mines have also been blocked in at some time in the past and access to these workings can only be gained, if at all, from other workings. For instance, in 22, although access is gained through the entrance noted in the Registry, it is probable that at least two other mines are connected to the workings behind 22 as there appears to be three other blocked entrances at various points in the workings. In fact, at two points the way on is via a very low passage and on emerging on the other side of these a roadway is gained leading to a (presumably) blocked entrance. When the survey is completed and compared with the relevant Ordnance Survey Sheet it should be possible to locate these blocked entrances on the surface.

Figure 1 – Combe Down Bath Stone Workings, taken from "The Builder"

It is necessary to quarry or mine for Bath Stone as the best building stone is situated some way below the surface. On Combe Down this stone may be anything from 3 feet to 30 feet below the surface and found in beds about to 2-4 feet thick, making up from 4 to 20 feet of useful stone. Above this the overburden is made up of ragstones and the like which were sometimes used as road metal. "The Builder" (ref 1) contained a series of articles on the mines, giving location, method of workings and quality of the stone obtained, the workings described being shown in Figure 1. In the introduction to the article is mentioned "sweeping adverse criticisms with regard to the quality (of Bath Stone)" and it goes on to note the use of immense quantities of bad and worthless stone and the widespread use of the name Bath Stone in connection with stone raised in other areas. It criticises both architects and quarry masters alike, and, to quote, with regard to the former "We go so far as to say that barely one architect in fifty, having ordered a certain kind of Bath Stone ever knows whether he gets that, or some other kind", sweeping criticism indeed. The article goes on to mention that although there were quarry masters who were perfectly honest there were some agents who would sell under any convenient name. There is a humorous footnote in connection with this: "At one quarry in the Bath area, in answer to an inquiry as to what the stone raised there actually was called in the market, we were informed 'Anything; all Bath Stone are so much alike that architects cannot tell one from another. We have sometimes called it Portland when builders require that stone'".

Figure 2 – Information from "The Builder" redrawn to larger scale
(note original scan is almost illegible)

Figure 3 – Combe Down stone workings, from Ordnance Survey map

It has proved rather difficult to relate the map shown in Figure 1 to the Ordnance Survey map, Figure 3. In order to present a comparison, therefore, Figure 2 has been drawn in a similar manner to that of Figure 1, but to a larger scale. It would appear from a comparison of these plans that the railway is slightly displaced on Figure 1. However, to substantiate this completely requires more research and a thorough walk through the area and this, as yet, has not been done. The only quarry that can be positively located is No. 10 at Mount Pleasant. All the other quarry numbers are a matter of some conjecture at present.

The fact that "The Builder" describes quarry No. 11 as "the only stone mine on Coombe Down" would seem to indicate that large scale mining did not commence until after 1895. David, whose paper is reported (ref 13) includes Combe Down under his general heading but does not specifically mention the area in connection with actual mining. He notes the amount of stone raised in the whole area as 2,000,000 cubic feet during one year, an indication of the size of operations at about the turn of the century. Calvert (ref 14) however, seems to indicate that all the extensive underground workings were carried out before the 1880's as he dismisses the Combe Down area as being "exhausted as a commercial prospect" after this date. However, much more research is required in order to produce a detailed history of the area and it is hoped that this will be the subject of a future article.

The survey was commenced at site 22 which is situated in an overgrown quarry. The nearer parts of the mine have been used recently for mushroom growing and because of this it is sometimes known as "Mushroom Mine". This is a fairly typical Bath Stone working with a main "access roadway" from which headings have been worked leaving large areas with the roof supported at intervals by stone pillars. This is known as "Pillar and Stall" working. Some of these pillars have smooth finished areas on which the quantity of stone removed was recorded. At one point these records are still visible. We have referenced the whole system as 22 although as noted above it is likely that there is more a than one mine making up the complex maze of workings.

A method of surveying was formulated to provide a quick traverse of the system. The relevant CRG grading would be approximate to grade 5. The system adopted involves a line traverse through the workings with each station being marked and numbered. When the completed line is calculated and drawn up a further trip into the mine with the survey will enable the passage detail to be determined and drawn in on the spot. The actual edge of the working is at some points rather vague due to the stacking of rubble at the sides of the main "roadway" and it is hoped that by using the method described, this problem will be overcome.

Another problem with this type of mine survey is the infinite number of closed traverses which could result if the "pillar and stall" method of mining were fully surveyed. Collins (ref 15) found that the use of the plane table in this type of system could provide a simple solution to the problem. We have, however, decided that an accurate survey of these workings would be too time consuming and therefore, we intend only to make a partial survey. An initial line survey is to be made starting from a prime survey station and following the wall around returning (hopefully) to the prime station. Each station will be marked as before. Several lines will then be surveyed across the workings until the opposite wall is reached and each line will be related to appropriate stations. Permanent stations will also be marked at various points to provide position references in the workings, related to the survey. The stone pillars will be shown diagrammatically for clarity. At the present time the survey is continuing, following the wall around and from a preliminary plot of the readings, it can be seen that we have progressed approximately 150 metres and are moving away from our prime station.

Proposed Method of Calculation

In order to avoid spending a large amount of time calculating the survey data by hand it was decided to investigate the use of a computer for this task. An initial study of the sub-computations led to a simple program which is capable of calculating the position of each station from a survey using a "leap-frog" technique.

With reference to Figure 4, a program based on this basic layout will deal with the calculations required in a simple open traverse. Complications do arise if minor legs are taken from the main survey line. This can be overcome by the suitable addition of named storage space for each station position. One such case has arisen but this was of a single reading and was overcome by including in the data a reading which reversed the previous reading. However, the program will not deal with a closed traverse and the rounding off of errors.

Figure 4 – Simple Flow Chart of Basic Survey Program

In a spate of enthusiasm, the program was improved by adding a graph plotter package. The object of this was mainly to get an idea of where the survey was going and whilst it would be almost impossible to get a reasonable line drawn from the size available with the graph plotter, the small size survey obtained by the method was of use in checking major errors. The whole exercise was, in fact, purely academic and it is felt that no real justification can be made for its use. Another modification which has been made is the use of a "scaling factor" so that a direct plot can be made to a specific scale, viz. 25/334 inches to a mile.

It is hoped to add further modifications to the program, mainly to deal with closed traverses. The main problem involved here arises if several small closed traverses are made within a large closed traverse where they all inter-connect. It is hoped that it is justifiable to call surveying an art and the computer being an extremely logical beast of burden, which has little chance in competition. The use of this program has speeded up the basic calculations from approximately six hours in one case to fifteen minutes, the time required to produce a presentable form of readings.


The authors would like to thank Messrs R Mansfield and AD Oldham for their past, present and hopefully, continuing help with references and the like.


  1. Bath Stone, The Builder, LXVIII (2723), pp273-278 (April 13, 1895) and LXVIII (2724), pp291-295 (April 20, 1895)

  2. Tucker, R. and Reed, J. Mines in the Box Area, East Devon Caving Group Newsletter, 12-13 (1963)

  3. Baxter, R. and Mansfield, RW. Mendip Nature Research Committee Journal 1, (2), pp4-15, (1963)

  4. Baxter, RJ. Northern Cavern and Mine Research Society Memoira 51-53, (1966)

  5. Cotham Caving Group, (1967)

  6. Tucker, JH. Axbridge Caving Group Journal pp8-16, (1966), (Pub.1967)

  7. Mines in the Combe Down Area: Access to Caves and Mines. Mendip Caver 2(8), 63 (June 1966)

  8. Townsend Phil and "Flicka", Trip to Bath Stone Workings. Belfry Bull. (186), pp10-11 (August 1963)

  9. A (tkinson), D. Operation "Lunch Hour", Wessex Cave Club Journal (36), pp5-7, (October 1952)

  10. A (tkinson), D. Notes on Access, Wessex Cave Club Journal, (43), p3, (December 1953)

  11. Edwards, WT. Club Trip to Bath Stone Workings, Wessex Cave Club Journal, (103), p304, (October 1965)

  12. Harris, GF. Bath Stone, British Caver, 47, pp32-49 and pp80-86 (1967)

  13. David W. The Position of and Method of Working Bath Stone in the Quarries of the Bath Stone Firms. Trans. Institute Mining Engineers, 22, pp495-499, (1900-1)

  14. Galvert, M. Bath Stone Mines - Their History and Method of Operation. Belfry Bull. (185), pp2-5. (July 1963)

  15. Collins, S.J. Surveying in Redcliffe Caves, Bristol. Bristol Exploration Club Caving Report No. 1. (January 1956)

AJ Butcher and RD Mehew, April 1971


Goon's Hole, Burrington Combe

During a sojourn on Mendip last summer on Monday 8th June 1970, Alan Jeffreys and myself went over to Burrington and spent an hour or so exploring the newly opened Lionel's Hole. Upon returning to the afternoon sunlight Goon felt the call of nature and went into some roadside bushes to relieve himself and returned to report a hole at the base of a small rock outcrop which was emitting a cool draught.

The following Friday evening whilst Goon and I visited other caves in the Burrington area Phil Romford and Tim Large (BEC) visited Lionel's Hole and having returned to the car before ourselves they proceeded to the site of the draughting hole which had been mentioned to them. As they awaited our return they managed to dig out about 10 feet of loose stones gaining access to a rift about 6 inches wide about which Phil (1) at that time suggested might need the use of bang.

On the 20th June Alan Butcher, Tim Large and myself returned to the site and spent 2-3 hours digging. Tim had spent all day there alone the previous Tuesday and managed to extend the dig to about 20 feet, there now being about 12 feet beyond the first squeeze. Digging continued with two of us below the squeeze and one on the surface, the spoil being removed was almost entirely loose stones of various sizes. The passage was descending at about 33° and was open, but due to the loose stones was too narrow to negotiate without removing the spoil. At the end of this visit the cave could be entered for about 25 feet and seen to continue for another 10-15 feet, where it appeared to get larger (2). A cool draught was still emerging from below and it was thought that the cave might possibly connect with Lionel's Hole.

Our next visit was on 1st September when Tim Large, Bob Mehew, John Cooper and myself spent another 5½ hours digging. After about 2 hours Tim managed to loosen a large boulder but after an equivalent time spent pulling and pushing we came to the conclusion that this was larger than the squeeze and so had to remain below where it still forms an obstacle in the passage. The removal of this boulder made progress easier and Bob went through the second squeeze to where the passage opened out (relatively speaking) and this was followed on downwards past a couple of very small stal formations and over a boulder where the floor dropped and the passage was almost of crouching height. Here, the passage levelled out and ended in a tube about 18 inches in diameter with a gravel infill and only about 5-6 inches of air space over this. At this time one could see about 20 feet along the tube, the length of the cave including this was estimated to be 64 feet long and about 30 feet deep (3).

Later that month on the 26th September Bob Craig, Derek Harding and myself visited the cave with the intention of attempting to dig in the final tube. It was apparent that since our last visit members of the BEC had burrowed along the tube and in removing and stacking the spoil, had used up most of the spare space at the bottom of the cave. Thus we decided to remove our own spoil to the surface but found extreme difficulty in removing only three bags of spoil and concluded (4) that the squeeze would need removing if this cave is to be dug easily.

Thus from that chance observation this then is the present state of the dig, except that on 14th March 1971 Bob Mehew and myself ran a traverse line from the permanent survey point at the entrance of Lionel's Hole, down the road to the entrance of Goon's Hole, and then down the cave to the present end. The instruments used in the traverse and cave survey were a tripod mounted survey head with compass and clinometer read to ½°, 20 metre fibron tape read to 0.01m, and passage detail was taken at every station in the cave. In the lower and tighter sections of the cave, tripod mounting had to be dispensed with and the survey head was hand held. The survey is, therefore, to CRG Grade 5C-6C.

Figure 5 – Plan survey of Goon's Hole

Figure 6 – Elevation survey of Goon's Hole

The plan and extended section of the cave as shown on the accompanying survey are shown in relation to part of Lionel's Hole (shown by dashed lines on the survey), the detail of which survey has been taken from that by WI Stanton in the WCC Jnl 132, pp160-162, December 1970 and this source hereby acknowledged.

The altitude of the entrance of Goon's Hole is about 414 feet above sea level, the length of passage 75 feet and the depth 32 feet. From the accompanying survey it is apparent whereabouts in Lionel's Hole any connecting passage should be sought, but it remains for an aural connection to be made to confirm the conjuncture.


Book Review: The West Somerset Mineral Railway
and the Story of the Brendon Hills Iron Mines

By Roger Sellick

pub. David & Charles (Newton Abbot, 1970) @ £1.75

This is a revised edition of the above work first published in 1962 which has for some years been out of print and has been described as a model work on such a subject.

It has been said that the earliest mining on the Brendon Hills was in Roman times but this has yet to be proved, however, there was undoubtedly sporadic mining in the 16th, 17th, and early 18th centuries. In 1839 Sir Thomas Lethbridge and the Earl of Egremont had trial pits dug on their estates, but despite the resultant mines being offered for lease the venture failed because at that time the spathic ore was virtually unknown in this country and not liked because of the carbon dioxide given off during smelting. About 1851/2 Samuel Blackwell and Ebenezer Rogers took a comprehensive mining lease including the existing Lothbrook and Gupworthy mines. Their interest was taken over the following year by Thomas Brown, the Brendon Hills Iron Ore Co. was formed and from then onwards the mining flourished.

The existence of good iron ore proven, the principal difficulty was getting it to the nearest port at Watchet. Horse and cart was unsatisfactory for the quantity involved and so in 1854 the route for a railway line was surveyed. Following the West Somerset Mineral Railway Act 1855, work commenced in May 1856 and the line reached Roadwater in 1857, Raleigh's Cross in 1861 and Gupworthy in 1864 the total construction cost to that date being £98,000.

In 1867 the iron trade was entering a periodic recession, but Robert Mushet found that excellent steel could be produced by the addition of Spiegeleisen and for which the Brendon Hill ore was suitable. Thus from 1874-8 production of ore was over 40,000 tons per annum. Peace in Spain in 1876 led to the output of Spanish ore and the less productive Brendon Hills mines had ceased work by 1877. The price of ore in 1871 was 20/- per ton and had dropped to 14/- in 1876. In May 1879 the Ebbw Vale Co., which had taken over the interests of the Brendon Hill Iron Ore Co. in 1864, closed all the mines following a loss of £42,500 the previous year and 250 miners were out of work. In the Autumn three mines were re-opened and others followed and the output rose to 31,354 tons by 1882. The company paid no dividend for the years 1875-82, and the fall in price of Spanish ore and the invention of the Thomas Gilchrist basic process made the mines uneconomic. They were all closed by September 1887, the entrances closed and all equipment taken out in readiness for auction.

The railway survived until 1898 when receipts were down to £300 and annual expenditure was over £7000 per annum. The mining villages were largely in decline as early as 1890. In 1907 the Somerset Mineral Syndicate Ltd. and Withypool Mining Co. re-opened and worked certain mines, the railway was reopened by the former which survived the longer of the two until 1910. The rail track was taken up in 1920 and the WSMR Co. wound up in 1925.

Apart from a war effort investigation of the mines and a trial adit driven near Comberrow in 1936, this is then a summary of the history of the mines. They numbered some 30 shafts, 20 levels and many surface working sites, total output between 1855 and 1909 had been 750,000 tons and Raleigh's Cross and Langham Hill mines both reached depths of over 690 feet.

In addition this book which totals 128 pages and is well illustrated with maps and photographs gives details of the geology of the area, pumping and winding engines and other mine buildings, the mode of working the mines, mining accidents, the mining village communities, mining returns and details of the railway route, timetables, traffic, engines and rolling stock etc.

The only minor observations which I can make on this excellent work is that a geological map of the area would have assisted the geological description in the text, and a diagram would have helped explain the process of working the mines by stoping and heading development.


  1. SMCC Hut Log 7, p36, entry 12 June 1970

  2. SMCC Hut Log 7, pp44-45, entry 20 June 1970

  3. SMCC Hut Log 7, p65, entry 1 September 1970

  4. SMCC Hut Log 7, pp70-71, entry 26 September 1970

MT Mills, March 1971


A Preliminary Investigation of Surtshellir, West Central Iceland

This paper is the result of a one day visit to Surtshellir made by members of the Shepton Mallet Caving Club during August 1970. It has two aims:

  1. to make available for the first time in Britain a survey and description of this lava tube, the most famous in Iceland, and to record its forms and features which are of general and scientific interest

  2. to indicate lines along which future research in the lava tube may be carried out


The lava tube is found at latitude 64°47'N and longitude 20°43'25''W, and at an altitude of approximately 350m (1150ft), in the northerly arm of the Hallmundarhraun, a post-glacial basaltic lava flow in the county of Mýrasysla. The flow here occupies the lower part of the valley of the Nordlingafljót, a tributary of the river Hvitá. The approach to the lava tube is made from Kalmanstunga, where a well signposted track leads some 8 km (5 miles) to the area of roof collapse which marks its lower entrance. Two other lava tubes are found nearby, these being Viðgelmir and the more important (for our purposes here) Stefánshellir. An approximate location for both Surtshellir and Stefánshellir is marked on the 1/250,000 map of West Central Iceland (Midvesturland) published by the Geodetic Institute of Copenhagen in 1952.

The Lava Flow

The Hallmundarhraun has its source in the volcanic area immediately north-west of Baldjökull, which is part of the Langjökull ice cap, from where it flowed in a westerly direction around the Eiriksjökull and into the Hvitá valley. In its lower part the flow splits into two arms separated by the mountains of Strútur and Tunga. The northerly arm in which the principal lava tubes are found is some 15 km (9¼ miles) long, and occupies the relatively constricted Nordlingafljót valley, which here is never more than 1.5 km (1 mile) wide.

The lava is typically pahoehoe (Icelandic - helluhraun basalt lava which congeals with a smooth skin that wrinkles into ropey and corded forms), and lacks surface vegetation. In the vicinity of the lava tubes it is monotonously undulating with surface forms being represented almost solely by tumuli (mounds formed where localised pressure from the underlying liquid lava forces up the cooled lava crust) and pressure ridges (elongated ridges of lava crust pushed up by the movement of the underlying liquid lava), both of which are usually cracked open at their summit. Less important surface features are contraction fissures and collapse depressions (the latter formed due to the collapse of the roof of an underlying cavity). Downslope, to the north of Kalmanstunga, the surface topography of the flow is made up of hummocks and depressions, channels and ridges, which form an area of stony rises.

Neither the composition nor the age of the basalt is known. However, due to the small amount of breakdown in the tubes, to a lack of surface vegetation, to the little chemical, alteration of the basalt, and to still well preserved flow features, a very young age for the flow is suggested, possibly 2000 years or less.

The Hallmundarhraun has a definite multiple structure in the vicinity of the lava tube which is similar to that described by one of the authors at the lava tube Raufarholshellir (Wood, 1971). Good exposure of the lava may be observed at all the major collapse holes and in some of the deeper fissures. It is suggested, therefore, that this part of the flow advanced by the protrusion of successive lava tongues, one above the other, each tongue being called a flow unit.

The Survey

The survey took about 5-6 hours to complete and also involved an element of exploration as none had the benefit of previous knowledge of the cave. The instruments were only those readily available at the time – a climbing rope (subsequently measured) and a Silva Scout No. 5 compass read to the nearest two degrees. Length measurements were to the rope's full or half length, shorter distances being estimated as were wall distances and roof height at every station. It will be appreciated that the accuracy of the survey is no more than the equivalent of CRG Grade 2, and for this reason the survey was not plotted by means of co-ordinates as this was considered unjustified.

An attempt was made to ascertain the general longitudinal section of the tube by using a Thommen (Swiss made) pocket altimeter graduated in metres. Readings were taken at 16 stations during the course of the survey, but when this was checked again at the end of the day it was found that the reading obtained at the point of the first reading (roof collapse 1) had decreased by 33 m (100 feet) due to a rise in barometric pressure. Whilst these readings would give the approximate change in elevation between any two consecutive stations rather than throughout the whole tube, they are not given as it is felt they would be both misleading and inaccurate. The tube generally falls from roof collapse 4 towards roof collapse 1 and beyond, the total difference in level throughout the whole tube being in the region of 37 m (122 feet). The continuous fall of the tube is interrupted by the piles of breakdown, particularly beneath entrances, which are roof collapses. The direction of magnetic north as shown by the compass is indicated at the first survey station, but due to the magnetic anomalies of unknown magnitude this approximate direction is not necessarily valid for any other position in the cave.

Figure 7Survey of Surtshellir
(note original scan is almost illegible)

Comparison with Earlier Quoted Lengths and Previous Survey

Olafsen and Pavelsen (1806) contains the earliest recorded length for Surtshellir of 839 Klaftern (ie. 1534m or 5034 feet). This then was the generally quoted length except for those who used indefinite phrases such as 'runs for miles'. Bout et al (1955) state that Surtshellir is in fact a series of caverns of which one attains 1580 m. However, Corbel (1955) claims to have measured the cave with the greatest of care and found it to be 1250m and not 1580m as had been published in error in certain ancient works; ie. 739 Icelandic fathoms instead of 839, a slip of the pen by Olafsen. Kjartansson (1968) states the length as approximately 1700m, and Montoriol (1968) claims 2200m as a result of a survey (yet to be published) and measurement of the cave.

Thus, prior to the visit by the SMCC, lengths of 1250, 1580, 1700 and 2200 metres for the cave had been claimed. In fact, the authors found the cave to measure approximately 1970m (6450 feet) in length.

The only other known published survey of Surtshellir is by Corbel (1955), and this comprises a thumbnail size plan, longitudinal section and three cross-sections, all to the same scale. A comparison of the survey with the survey drawn up by the authors is given; Corbel’s survey having been enlarged three times for the purposes of the comparison. It will be apparent that this survey is even more of a sketch survey than the authors’.

The General Form of Surtshellir

Although a total of nine collapse holes were found in the area, all of which were aligned roughly in a north-east / south-west direction, only four of these are due to the collapse of the roof of Surtshellir. Indeed, the upper termination of the lava tube is the back wall of roof collapse 4, beyond which, as will be described later, the lava tube probably continues into the complex Stefánshellir system.

Surtshellir for much of its length in plan is of a unitary (ie. simple, sinuous and elongate) forms, though a smaller lava tube which intersects the unitary tube is of an anastomosing (ie. braided) nature. In total length, as has already been mentioned, the lava tube in 1970m. (6450 feet) and the main tube is seldom more than 9m (30 feet) in height.

Figure 8Comparison of Surveys of Surtshellir
(note original scan is almost illegible)

The main entrance is the most south-westerly roof collapse (roof collapse 1 on survey) and on descent a double tube system is encountered. A short climb down of 2m (6 feet) from the flow surface brings the explorer to the top of a breakdown pile which here rests upon a false floor. On descent of a further 4m (13 feet) the true lava floor of the main tube is reached which on the north side of the breakdown pile disappears beneath the false floor. The lower level cavity is generally 1.2-1.5m (4-5 feet) in height. In places it is partially blocked by breakdown or also its floor is covered by deep water. A clear way leads on beneath the false floor in a south to south-west direction though this was not pursued.

To the south-west of roof collapse 1, the tube may be explored for 520m (1700 feet) until the floor and roof meet at a lava seal. On descent of the breakdown pile beneath the entrance collapse, the way on into this part of Surtshellir is across sheet ice in a lava tube of arcuate section (cross-section C) some 6.6m (22 feet) in height. En route to the lava seal the explorer passes the altar, a cairn of rocks on which it has been recorded the early venturers left coins. Here also the cross-sectional form of the tube changes and is typified by cross-section A. In this part of the tube, the tube floor is concave and a rough clinkery surface is covered in its central part of a deposit of clay. On either side of the tube the floor rises 1.2m (4 feet) and is bent over to form narrow lateral gutters. The average height of the tube in this lower section is 3.6m (12 feet) and there is little or no breakdown. The walls are glazed and of a grey colour, as they are throughout the whole of the tube, and apart from a few red stalactites this part of Surtshellir is relatively featureless. The origin of the form of the lower tube is clear. Where during the draining of the liquid lava from the main conduit the level for a time remained stationary, a crust formed over its surface. On further lowering of the liquid lava the crust sagged beneath its own weight to form the distinctive cross-section we see today. At the first roof collapse also, as we have already seen, a crust similarly formed but did not sag, so that a fairly large cavity is now found beneath it. This cavity may therefore extend beneath much of the floor of the lower (south-westerly) part of Surtshellir, though it is nowhere to be seen.

To the north-east, between roof collapse 1 and 2, remnants of a false floor may be observed as a cornice-like feature bordering either wall. It is 1.2m (4 feet) in height and is bent over at the top in a direction towards the centre of the tube, so that deep gutters are found where it leans away from the wall. Its surface is coarse in contrast to the normal grey glazed walls of the tube. It is evident that this feature represents the remains of a false floor which was continuous with the false floor beneath roof collapse 1 and whose central part has collapsed due to contraction on cooling.

Breakdown is much more extensive between roof collapse 1 and 2, possibly due to the more 'open' nature of the tube. In addition, breakdown also increases towards each roof collapse. In cross-section the tube here is arcuate, has a flat floor and an average height of 8m (26 feet). Large expanses of the floor are not covered by breakdown, so that its coarse ropey nature may be observed, 'frozen' into which are many flow features.

Roof collapse 2 is extremely impressive. Collapse has taken place at the junction of the main tube and two smaller side tubes, resulting in an enormous hole some 40 m (130 feet) in length, 15m (50 feet) in width and more than 6m (20 feet) deep, with sheer and often overhanging walls. The four large tube entrances may well be imagined as railway tunnels. Climbing over breakdown up into this collapse hole from the main tube a pile of loose snow is encountered. Leading off immediately to the east is a tube whose floor is at a higher level than the main tube and whose dimensions are smaller. This tube rapidly closes down from a height of 3m (10 feet) to 1m (3 feet) and shortly leads to an area of breakdown. The way on is a crawl through the breakdown either into a tube to the south which soon becomes impenetrable, or into a tube to the north which leads back to the main tube between roof collapse 2 and 3.

Both in the southerly and northerly branch tubes the cross-section is low and wide (cross-sections E and F), often not high enough to stand in. There is very little breakdown and generally the floor is covered by large expanses of sheet ice, which has often covered itself by melt-water, making the surface extremely treacherous. These smaller tubes are also of a relatively featureless nature, relieved only by breakdown or ice formations. One feature of interest was found on the walls of the northern tube, where the glaze was covered to a height of 1.2m (4 feet) from the ice surface by a coarse lining, about 1.5 cm (½ inch) thick, and whose upper edge represented a remarkably level line. It can only be suggested that this layer was deposited during the draining of the conduit, when the lava level remained stationary for long enough for the lava to cool against the side walls, but not long enough for a crust to form over the flow.

The northerly tube opens out as the main tube is approached once more and eventually splits into two, the entrances of which are found some 4.5m (14 feet) above the floor of the main tube. At the point of diffluence is found the "Robber's Lair", an area 3m (10 feet) wide and 6m (20 feet) long which is enclosed by a dry stone wall 1m (3 feet) high with a doorway on the north side. Around the area are found many splinters of bone and it is said that outlaws hid in this part of the tube as early as the tenth century and about which there is much folk-lore. The two tubes on the south side of the main tube have direct continuations in the opposite (north) wall, at the same height above the main tube floor. The smaller loop tubes on the northern side of the main tube continue in a northerly direction for 60m (197 feet) to end in a lava seal. There is one point about these smaller tubes which is of interest. It appears that although they drained completely of liquid lava at an early stage, they were invaded once more by liquid lava from the main tube. This later lava did not penetrate as far as the seal, for in the northern terminal chamber the toe of the flow can be seen and the lava is here 0.6-1m (2-3 feet) thick. On re-draining, a crust formed over the lava that occupied one of the small tubes bordering the main tube, so that a false floor was left some 0.6m (2 feet) from the roof.

In the main tube, between roof collapse 2 and 3, the fall in the lava level was interrupted on fourteen separate occasions and on each occasion a thin crust some 2.5-5cm (1-2 inches) thick formed over the surface, for their remnants are seen along each wall. The smaller tubes drained when the lava level in the main tube was stationary some 0.6m (2 feet) below their lip. A 'frozen' cascade of lava is now found at the lip extending down only as far as the first and highest lava level in the main tube.

The other smaller tube entrance in the north wall of roof collapse 2, is just a loop tube, also of a high level nature. Breakdown between roof collapse 2 and 3 is extensive, though where the true lava floor is seen flow features are obvious. The floor is generally ropey and apart from the fourteen levels mentioned earlier the only other important feature in this part of the main tube is a small lateral bench, about 0.3m (1 foot) in height, bordering either wall of the tube. Some flow grooves are also evident in places on the walls.

Roof collapse 3 is also large, some 52m (170 feet) long, 15m (50 feet) wide and 5.5m (18 feet) deep. As an entrance to the cave it is not readily climbable. To the north-east a low dry stone wall has been built across the main tube, beyond which breakdown of the tube walls and roof is very extensive. Indeed, breakdown of the roof has led to the formation of two small collapse holes and very little of the true tube floor is seen here. The lava tube terminates at the most north-easterly roof collapse (roof collapse 4). The back wall of the hole so formed is of breakdown and no way was found through it, though a small tube is found near the collapse and leads off in an easterly direction.

The Relationship of Surtshellir and Stefánshellir

It has already been mentioned that altogether a total of nine collapse holes were discovered during a short investigation of the flow surface around Surtshellir. The five not already mentioned are due to the collapse of the roof of the lava tube Stefánshellir and are all found to the north-east of roof collapse 4 of Surtshellir. One of the authors explored part of the Stefánshellir system with AJ Butcher and RD Craig and found that it consisted of many interconnecting tubes whose total length must be equal to, if not greater than, Surtshellir. Although nearly two hours were spent exploring Stefánshellir, there was still a large number of side tubes left to investigate. One particular tube which was followed for a considerable length, averaging 3.6m (12 feet) in height, led in a south-westerly direction, downslope towards the upper termination of Surtshellir. Unfortunately due to a lack of time and deepening water this passage was not pursued to any termination, though it is certain that it must represent a continuation of the main tube of Surtshellir. Surtshellir and Stefánshellir are, therefore, most probably parts of an extremely large and complex tube system whose total length, it is estimated, must be about 4km, though this cannot be fully established until the complete exploration and survey of the two lava tubes is carried out.

Recommended Future Research in Surtshellir / Stefánshellir

Due to insufficient evidence no discussion on the origin of Surtshellir / Stefánshellir can be given here. What can be done, however, is to outline the main lines along which future research in the lava tubes may be carried out. These are listed below:

  1. Priority must go to an accurate survey of both Surtshellir and Stefánshellir to a standard of at least that reached by the SMCC survey of Raufarholshellir (Ellis, 1971). In this way their true form may be discovered and their relationship with one another established.

  2. A careful study of flow features both in the main tube and in the minor tributary tubes must be made so that the direction of flow may be found (this is particularly important in understanding the nature of the anastomosing form of Stefánshellir), and the history of the draining of the lava tubes ascertained.

  3. The age, composition and structure of the lava flow needs to be known.

  4. Once the method by which the flow advanced has been established, the origin of the tubes may be discovered by paying particular attention to their relationship with the flow structure. Here cross-sectional forms are important, though their interpretation in Surtshellir / Stefánshellir may prove to be more difficult than the interpretation of cross-sectional forms made by one of the authors in Raufarholshellir.

  5. The position of the tubes in relation to the lava will be of interest. The lava tubes are found, for instance, in an area of low gradient, while immediately down slope the area of stony rises appears to have formed where the gradient steepens. It would be of benefit to survey the long profile of the lava flow, therefore, besides several cross-profiles, to discover the true relationship of the lava tubes and the gradients of the lava flow.

It is only through many hours of tedious surveying and recording that the origin of Surtshellir / Stefánshellir may be discovered. It is hoped, therefore, that this paper is of help to any future visitor to Surtshellir and Stefánshellir whose intention it is either to explore or undertake research in the lava tubes, by briefly outlining their true form and discussing the nature of the problem involved in their interpretation.


The authors would like to thank P Bowler and BM Ellis for the assistance they gave in carrying out the survey which involved foregoing more attractive explorations.


  1. Bout, B., Corbel, J., Derruau, M., Garavel, L.; and Peguy, Ch.P. (1955), Geomorphologie et Glaciologie en Islande Centralo, Norois (8), p472 (Oct-Dec 1955).

  2. Corbel J. (1955), Les Karsts du Nord Oiest de L'Europe, Lyons, pp.260-261.

  3. Ellis, BM. (1971), The Survey of Raufarholshellir; to be published in Trans of CRG.

  4. Kjartansson, G. (1968), Lava Caves of Iceland; Cascade Caver 7(4), p15, (April).

  5. (Montoriol, J,) (1968), Club Montanes Barcelones Circular (Abril Mayo, 1968), p490.

  6. Olafsen and Pavelsen. (1806), Travels in Iceland, in a collection of Modern and Contemporary Voyages and Travel, 2, p69, Pub. Richard Philips, London.

  7. Wood, C. (1971), The Nature and Origin of Raufarholshellir; to be published in Trans, of CRG.

MT Mills and C Wood, May 1971


 Journal Series 05 Number 1