THE ONE-EIGHTH RELATIONSHIP THAT CONSTRAINS DEGLACIAL SEISMICITy AND CAVE DEVELOPMENT IN CALEDONIDE MARBLES DEGLACIALNA SEIZMIč NOST IN RAZVOJ JAM V KALEDONSKIH MARMORJIH: RAZMERJE ENA PROTI OSEM Trevor f AULKNER 1 Izvleček UDK 550.34 :551.44(234.652) Trevor Faulkner: Deglacialna seizmičnost in razvoj jam v kaledonskih marmorjih: razmerje ena proti osem Razvoj jam v kaledonskih marmorjih na periodično poledene- lem 40000 km 2 velikem območju osrednje Skandinavije se je začel s tektonskim začetjem, procesom pri katerem so se prvotne prevodne razpoke ustvarile zaradi razbremenitev ob umikih ledenikov. Vzdolž teh prevodnih poti se v obdobju umikanja ledenikov in medledenih dobah razvijajo kraški kanali. Model tektonske incepcije, ki obravnava razvoj jam v “delno ločenem” vrhnjem delu skorje, temelji na opažanjih, da je največja globina jamskih rovov ( pri tem je mišljena razdalja med površino in rovom) manjša od osmine globine lokalne ledeniške doline. To nakazuje, da je nastanek prevodnih razpok povezan z izostatičnim dvigom in delno pogojen z velikostjo seizmičnosti, ki so jo povzročili diferencialni pritiski in dvigi, ki so nastali ob ledeniških dolinah med umikanjem ledeniških pokrovov v smeri od zahoda proti vzhodu. Razmerje ena proti osem je značilno tudi v drugih območjih kaledonskih marmo- rjev v Skandinaviji, na Škotskem in v Novi Angliji (ZDA), kar kaže, da so podobni procesi botrovali nastanku številnih jam na teh območjih. Ključne besede: Kaledonidi, deglacialna speleogeneza, podzemlje, rob ledu, incepcijske razpoke, marmor, neotekto- nika, razmerje ena proti osem, seizmičnost, kras v pasovih, razdalja površje-jama, tektonska incepcija, Weichselian, Skandi- navija. 1 Limestone Research Group, School of Geography, The University of Birmingham, Birmingham, B15 2TT, UK, e-mail: trevor@marblecaves.org.uk Received/Prejeto: 23.06.2006 COBISS: 1.01 ACTA CARSOLOGICA 36/2, 195-202, POSTOJNA 2007 Abstract UDC 550.34 :551.44(234.652) Trevor Faulkner: The one-eighth relationship that constrains deglacial seismicity and cave development in Caledonide marbles The formation of karst caves in Caledonide metamorphic limestones in a repeatedly-glaciated 40000km 2 region in cen- tral Scandinavia was initiated by tectonic inception, a process in which open fracture routes, primarily created by deglacial seismicity, provided the opportunity for subsequent dissolution and enlargement into cave passages in both deglacial and inter- glacial environments. The tectonic inception model built on re- ports of a ‘partially detached’ thin upper crustal layer in similar settings in Scotland and this paper shows that the present maxi- mum subsurface cave distance (i.e. the distance of a passage to the nearest land surface) is commonly less than one-eighth of the depth of the local glaciated valley. This suggests that frac- ture generation was related to the scale of isostatic uplift and was partly determined by the magnitude of seismicity caused by the differential pressure change and differential uplift that occurred along valley walls as the ice margin of each of the ma- jor Pleistocene icesheets receded from west to east. The maxi- mum one-eighth relationship is also commonly maintained in other Caledonide marble terranes in Scandinavia, Scotland and New England (USA), suggesting that many of the caves in these areas were formed by similar processes. Key words: Caledonide, deglacial speleogenesis, epigean, ice margin, inception fracture, marble, neotectonics, one-eighth relationship, seismicity, stripe karst, subsurface cave distance, tectonic inception, Weichselian, Scandinavia. ACTA CARSOLOGICA 36/2 – 2007 196 The repeatedly-glaciated 40000km 2 study area in central Scandinavia contains over 1000 individual metamor- phic limestone (marble) outcrops and has nearly 1000 recorded karst caves with a total passage length >72 km (f aulkner, 2005 and 2006a, f ig. 1; f ig. 1). f undamental differences between these caves and those formed in sed- imentary limestones derive from the metamorphic grade of the host bedrock with its very low primary porosity and from the fine-scale foliations and the consequent lack of partings guided by the initial bedding. Indeed, the foliation is commonly vertical in the western part of the study area, where sub-horizontal openings must be along joints or other fractures in the marble that has been metamorphosed up to amphibolite grade. The deepest cave is only 180 m deep, despite outcrop vertical rang- es reaching over 900 m. Caves tend to cluster together and are positioned randomly in the vertical dimension, whilst commonly remaining within 50 m of the overlying surface. Additionally, there are no regional-scale caves. This is despite some narrow marble outcrops being sev- eral tens of kilometres in length, forming stripe karsts. Commonly, just a single streamway underlies upper-level relict phreatic passages with few vadose elements, creat- ing an upside-down, vadose-beneath-phreatic, morphol- ogy. Recharge to the karst is primarily allogenic and cave stream discharges commonly remain unsaturated with calcite (Lauritzen, 1981; Bakalowicz, 1984). Autogenic recharge is relatively insignificant, mainly occurring dur- ing the spring snowmelt. These caves have their own morphological style, recognisable right across the area, which differentiates them from caves formed in ‘classical’ karsts in sedimen- tary limestones. Because the caves are relatively short and epigean and there is a complete absence of long, hypo- gean, cave systems, speleogenesis by the (chemical) inception horizon hy- pothesis (Lowe 1992; Lowe & Gunn 1997) is unlikely. The tectonic inception model (f aulkner 2005; 2006a) proposed that it is only open fracture routes that could provide the opportunity for dis- solution and enlargement into cave passages in the Caledonide marbles. It was hypothesised that the dimen- sions of these fractures are related to the magnitude, and perhaps to the frequency, of local earthquakes and commonly-small tectonic movements that arose from the isostatic rebound that accompanied deglaciation at the end of each major Pleistocene glacial. f aulkner (2006a) described in detail the evidence for deglacially-induced earthquakes at the end of the Weich- selian glaciation in Scandinavia. An additional mechanism for fracture generation may arise from earth- quakes caused by the downhill sliding of a glacial ice mass (Ekström et al. 2003). The tectonic openings formed INTRODUCTION Fig. 1: Study area location, showing a) major caves and b) neotectonic earthquakes in Scandinavia from 1750–1999, from Dehls et al. (2000). TREVOR f AULKNER ACTA CARSOLOGICA 36/2 – 2007 197 along inception surfaces between the marble and adja- cent aquicludes and at inception fractures that are entirely within the marble and are commonly (though not uni- versally) parallel to, or orthogonal to, the foliation. The model built on reports of a ‘partially detached’ thin up- per crustal layer in similar settings in Scotland (Daven- port et al. 1989). It is supported by observations of later neotectonic movements in most relict cave passages and sporadically on the surface from evidence such as fault gouges, slickensides and sharp edges that produce long narrow shadows. This paper explores relationships be- tween cave dimensions and their local external geologi- cal and geomorphological attributes, providing more evi- dence in support of the model. NEOTECTONICS AND CAVES The areas of Scandinavia that are seismically the most ac- tive at present are the south Norway coast area, which has few carbonate outcrops, and that part of the county of Nordland in Norway that is north of the study area. That northern area, from Mo i Rana to Narvik, has caves up to 20 km in length and up to 580 m in vertical range (VR). In contrast, the study area may have a comparable density of karst caves, but with lengths and VRs only up to 5.6 km and 180 m (f aulkner 2005). The neotectonics map (f ig. 1) after Dehls et al. (2000) thus seems to suggest a rough relationship between the frequency and magni- tude of earthquakes, and the dimensions of karst caves in a region. It is therefore hypothesised that the depths of inception fractures below the surface, and hence the total lengths of potential proto-conduits, are related to the magnitude, and perhaps to the frequency, of local deglacial earthquakes, assuming some correlation with the neotectonic pattern. Because that seismicity was re- lated to the scale of isostatic uplift and to the differential pressure change that occurs along valley walls as a major icesheet recedes (Davenport et al. 1989; Ringrose et al. 1991), it follows that cave depth and length = function (fracture depth and extent) = function (strength of tec- tonic activity) = function (change of ice thickness during deglaciation). f rom the above conclusion, the caves with the great- est dimensions should lie along the Swedish border area, because the icesheet was thickest there during each gla- ciation (probably ~2 km thick at the Last Glacial Maxi- mum, covering all the local peaks) and because there are deep glacial valleys with extensive stripe karst outcrops. This setting should cause the largest earthquakes at the end of each Pleistocene deglaciation, and therefore the most extensive and deepest set of fractures. This may ac- count, at least in part, for the presence near the border of four of the five longest caves of the study area: Korall- grottan (Sweden: 5.6 km), Labyrintgrottan (Sweden: 2.6 km), Stor Grubblandsgrotta (Norway: 1.9 km) and Sotsbäcksgrottan (Sweden: 1.9 km). The three deep- est caves also lie in this region: Ytterlihullet (Norway: 180m), Korallgrottan (144 m) and Sotsbäcksgrottan (110 m). One other long and deep cave, Toerfjellhola (Norway: 1.9 km and 101 m), lies along the flank of a major mountain ridge that runs parallel to the coast. SUBSURf ACE CAVE DISTANCE f aulkner (2006a) discussed the shallow nature of most cave systems in the study area, suggesting that caves in stripe karsts have formed entirely within an upper zone of fractured rock. It was hypothesised above that there is a relationship between cave VR (which is the vertical dif- ference between the highest and lowest explored points of a cave) and the local change of ice thickness during deglaciation. A more direct relationship is likely to be with the maximum distance of cave passages from the overlying surface. This subsurface cave distance is taken to be the length along a line orthogonal to the surface and the centre of any intersected passage (f ig. 2), which, in the extreme case of a cave passage behind a vertical cliff, could be a horizontal distance. In order to test the distance and relief relationship, the maximum subsurface cave distances of 39 of the deeper caves of the study area (obtained from cave sur- vey sections) were plotted against the local relief differ- ences (f ig. 3). The local relief differences were taken from 1:50000 topographical maps by measuring the height above the valley floor of the local ridge-shoulder, where a consistently steep slope profile becomes less steep. Caves can occur at any altitude along this profile, and the total lengths of the profiles were always less than a few kilome- THE ONE-EIGHTH RELATIONSHIP THAT CONSTRAINS DEGLACIAL SEISMICITy AND ... ACTA CARSOLOGICA 36/2 – 2007 198 tres. There are few caves in areas of local relief difference of less than 100m, and none of these have VR>10 m. The complete scatter diagram for all 884 recorded caves would show a poor correlation between subsurface cave distance and the local relief difference, because the mean VR of the caves is only 8.8 m and they occur in valleys of many different depths. However, f ig. 3 shows that the maximum distance of cave passages (and there- fore of dissolutionally-enlarged inception fractures) from the surface is commonly one-eighth, or less, of the extent of the change of local relief. This maximum envelope for the relationship of subsurface cave distance to local relief difference appears to be approximately linear, at least for a local vertical relief of up to 400 m, and perhaps up to Fig. 2: Subsurface cave distance and other terms. Fig. 3: Relationship between maximum subsurface cave distance and local relief difference. The straight line indicates the maximum one-eighth relationship that constrains nearly all the known caves in central Scandinavia, of which 39 of the deeper caves are indicated. The codes in parentheses give the geological zones used by Faulkner (2005). z1–z8, z A–z C and KL are located progressively eastward into Sweden. The wide geographical spread suggests that the maximum relationship applies throughout the study area. 800m. Most parts of the study area are represented by the caves shown in f ig. 3, showing that the one-eighth rela- tionship probably applies across the whole area. THE INf LUENCE Of Ex TERNAL ATTRIBUTES ON INCEPTION f RACTURES The relationship between seismicity and cave devel- opment is supported by the evidence summarised by f aulkner (2005) and in Table 1. The karst type, for which three ranges of foliation dip were defined, influences the mean VR for the study area caves, but in a manner that is perhaps paradoxical. Thus, caves in Low Angle Karst (LAK, foliation dip 0–30°) tend to be deeper than aver- age, caves in Angled Stripe Karst (ASK, 31–80° ) have a similar mean VR to that of all 884 recorded karst caves, and caves in Vertical Stripe Karst (VSK, 81–90°) have a smaller mean VR. However, other cave dimensions vary less consistently with karst type. Those caves in close proximity to a major thrust zone (designation T=1) have larger-than-average mean dimensions, whereas those in close proximity to a ma- jor igneous pluton, and therefore in marble ‘remetamor- phosed’ by contact metamorphism (designation R=1) tend to be smaller. If the enhancing relationships are, in fact, directly controlled by the thrust attribute (rather than this just acting as a proxy for some other control- ling variable), then this implies that the reactivation of old thrusts by deglacial seismic shocks promotes fractur- ing in any adjacent marble outcrops, creating longer and deeper voids for cave inception. Two mechanisms are possible to explain the restricting case. f irstly, the pre- vious high-temperature contact metamorphism of met- alimestone may reduce its fracturing ability, by making the rock more homogeneous. Secondly, the presence of a large igneous pluton (such as occur in the Helgeland Nappe Complex in the west: f aulkner, 2006a, f ig. 1) may, of itself, reduce the magnitude of local earthquakes, and, therefore, their ability to create long and deep fractures. Of the six caves discussed previously, three occur dispro- portionately near a major thrust zone, and none lie in an outcrop subjected to contact metamorphism (the main country rock being mica schist), again lending support to the earthquake relationship hypothesis. The percentages of the 39 deep caves in f ig. 3 with external attributes R=1 or T=1 were compared with the same percentages in the total set of 884 recorded caves TREVOR f AULKNER ACTA CARSOLOGICA 36/2 – 2007 199 External attribute Mean VR <8.8 m Mean VR ~ c. 8.8 m Mean VR >8.8 m Notes Karst Type VSK ASK LAK Other cave dimensions (length, volume and cross- section) vary Contact metamorphism (R) R=1 R=0 Similar for all cave dimensions Thrust proximity (T) T=0 T=1 Similar for all cave dimensions Cave Location (as defined by f aulkner, 2005) Coastal, Valley floor and Paleic Surface Gently Sloping and Valley Wall Ridge and Valley Shoulder Locations in bold have similar influences for mean length and mean volume Tab. 1: External attribute influences on mean cave vertical range. VSK: Vertical Stripe Karst, ASK: Angled Stripe Karst, LAK: Low Angle Karst. in the study area. R=1 only occurs in 18% of the 39 caves compared with 23% overall and T=1 occurs rather more often (13%, compared with 11%). Although these per- centages may be statistically similar, the differences at least suggest that the individual attributes that promote the deeper caves towards the one-eighth limit of the dis- tance / relief relationship are the same as the attributes that influence vertical range, as shown in Table 1. f ur- thermore, caves lying below coastal, valley floor and paleic surface (i.e. one that has hardly been modified by glaciation) locations are under-represented by the 39 deep caves, caves beneath gently sloping and valley wall locations are similarly represented, and caves in ridge and valley shoulder locations have two to three times the representation. Thus, the steeper is the local topography at the cave location (as defined by f aulkner 2005), then the more likely a deeper cave will be formed. The proximity of the caves in f ig. 3 to the one- eighth ‘limit’ can be considered in terms of the compe- tition between their various external attributes. Only Øyåskjeleren and Svartdalgrotta exceed the normal maximum relationship, with subsurface cave distances that reach about one-seventh the local relief difference. Not only are these two caves situated in the valley shoul- der cave location, they also both lie behind large vertical cliffs, suggesting that the effect of seismic shock is mag- nified even more by very steep topography. In the case of Svartdalgrotta, this overcomes the restrictive effect of an adjacent, but small, intrusive outcrop. Korallgrottan is shown at the one-eighth limit, probably because of its proximity to a thrust, despite lying essentially in a val- ley floor location. (However, its maximum distance from the surface is only estimated approximately). JOBshullet also lies on the one-eighth line, despite being surrounded by an enormous granite outcrop, probably because it has its cave location in a narrow ridge of marble, which is seismically very favourable. f ractures in ridge and shoul- der locations are also more likely to open farther by ice wedging and by gravitational mass movement, explain- ing why caves in these locations have the largest num- bers of entrances per cave, creating many through caves (f aulkner 2005). It was suggested by f aulkner (2006a) that part of Yt- terlihullet achieves its exceptional (for this study area) 93m subsurface cave distance because it occurs in LAK with interlayered amphibolites that acted as inception horizons. This remains a valid factor, but the cave is also situated at the eastern shoulder of Bryggfjelldal, one of the largest and deepest glaciated valleys in central Scan- dinavia (f ig. 4). This is 5000m wide and 800m deep and lies below the Okstind mountain range that has the area’s largest remnant glacier. The cave is thus ideally situated to take advantage of deep fractures produced by high-mag- nitude seismic events that shook the area after each of its deglaciations. f rom f ig. 3, inception fractures formed along the observed amphibolite layers still lie within the limits of the one-eighth relationship. f rom the evidence in f ig. 3, fractures are created only rarely up to the one-eighth ‘limit’ . Additionally, their enlargement into cave passages at the depths reached must be constrained by the extent of the marble outcrop in that area, and by the geological and topographical inheritance: passages can only develop in size (even un- der deglacial conditions) if there is a suitable hydraulic pathway. Deep fractures that have no route back to the surface can only fill with static water, and not enlarge. Thus, some caves with subsurface distances that are well inside the one-eighth line can be explained by a lack of suitable marble outcrop. At the extreme, areas that do not exhibit cave systems, or that contain unexpectedly shal- low systems, despite containing extensive striped marble outcrops, such as Stordal near the coast (f ig. 1), may be areas of anomalously low seismicity. Indeed, the Stordal marble lies along the floor of a glacially-rounded valley THE ONE-EIGHTH RELATIONSHIP THAT CONSTRAINS DEGLACIAL SEISMICITy AND ... ACTA CARSOLOGICA 36/2 – 2007 200 that is surrounded by large plutons of quartz diorite and trondheimite, so that its lack of karstification can be as- cribed to the contact metamorphism restriction and to its location. The many short and shallow caves at Övre Ältsvatn in Sweden commonly lie in LAK, ‘remetamor- phosed’ in places by granitic intrusions, on a paleic sur- face plateau. Thus, their small dimensions probably de- rive from both contact metamorphism and their rather flat cave location. These reduce local seismic activity and, additionally, the paleic surface location restricts the op- portunities for deep hydrogeological drainage. The conclusion from the above commentary is that the largest positive influence on the production of long and deep inception fractures and therefore on cave di- mensions is the seismic magnification that can occur at ridge and shoulder cave locations, especially if near a re- activated fault or thrust. Inception fracture depths are restricted near igneous intrusions and at coastal, valley floor and paleic surface cave locations; foliation dip has a less consistent influence. Fig. 4: Bryggfjelldal from the entrance to Naeverskardhullet. The cave system has formed at the valley shoulder location, which is favourable for the seismic creation of fractures during deglaciation. Cave Karst Type Mean tier spacing (m) Mean shaft spacing (m) Spacing ratio shaft/ tier Klausmark System ASK 3 3–30 1–10 Two Bridges Cave ASK 4 20 5 Hornet Pot ASK 6 30 5 Lislvatngrotta ASK 3 25 8 Tourist Cave ASK 4 - 8 22 3–5 Svartdalgrotta LAK 10 10–20 1–2 Neptune’s Cave ASK 5 - 10 10–15 1–3 Balcony Cave ASK 2 4 2 Toerfjellhola VSK 5 12–22 2–4 Øyåskjeleren VSK 5 8 2 Eiterådalgrotta ASK 5 50 10 Sirijordgrotta VSK 8 16 2 Håpgrotta ASK 3 10 3 Green Valley Cave VSK 2 5 2.5 Jordhulefjellhullet VSK 4 20 5 Pustehola ASK 6 12 2 Brown Stains Cave ASK 4 10 2.5 Sarvenvårtoehullet ASK 5 20 4 Gevirgrotta ASK 5 15 3 Sarvejaellagrottene ASK 8 12 1.5 Jegerhullet ASK 3 10 3 Etasjegrotta VSK 2 7 3.5 Invasjonsgrotta VSK 13 40 3 Anastomosegrotta VSK 3 8 3 Møllebekkgrottene VSK 2 8 4 Geitklauvgrotta VSK 3 6 2 Kompassgrotta VSK 5 10 2 Blåfjellgrotta VSK 4 10 2.5 Høgligrotta ASK 4 10 2.5 Kvannlihola VSK 5 50? 10 Grønndalsgrotta LAK 8 16 2 Gielasvaratjgrottan LAK 2 3 1.5 Sotsbäcksgrottan LAK 10 20 2 Korallgrottan ASK 4 18 4 SUMMARy All 2–13 3–50 1–10 MEANS 5 16 4.6 Tab. 2: Spacing between passage tiers and between shafts. For karst types, see Table 1. Caves are Listed W-E. f RACTURE SPACING The surveys of 34 of the more complex caves in the au- thor’s cave databases reveal (Table 2) that the mean verti- cal spacing between sub-horizontal phreatic passage tiers varies from 2–13 m (overall mean c. 5 m) and the mean horizontal spacing between near-vertical shafts and joints varies from 3–50 m (overall mean c. 16 m). The ratio of mean shaft spacing to mean tier spacing for each cave varies from 1–10 (overall mean c. 4.6). All these ranges appear to be independent of karst type, as previously de- fined. Because Marrett et al. (1999) provided evidence that fracture apertures in limestone follow a power-law scaling, it might be assumed that, at any one time and place, fracture apertures commonly decrease with depth, so that the horizontal and vertical separations between TREVOR f AULKNER ACTA CARSOLOGICA 36/2 – 2007 201 tectonic fractures of a particular aperture size increase with depth, i.e. they become less frequent. However, from the survey sections of the two caves with the most pas- sage tiers in the study area (8 in Toerfjellhola and c. 20 in Etasjegrotta), there is little evidence of an increase in fracture spacing with subsurface cave distance, which in these cases approaches 50 m. It is assumed therefore that within the “partially detached thin upper crustal layer” of Davenport et al. (1989), fractures occur at essentially ran- dom intervals, but that the distance of this detachment from the contemporary surface equals the maximum subsurface cave distance. This random arrangement within an upper crustal layer contrasts with the finding of Milanović (1981, p48) that the “depth of karstification” found in boreholes in sedimentary limestone obeys an exponential law. CONCLUSIONS This paper has shown that the present maximum “sub- surface cave distance” is commonly less than one-eighth of the depth of the local glaciated valley, suggesting that fracture generation was related to the scale of isostatic uplift and was partly determined by the magnitude of seismicity caused by the differential pressure change and differential uplift that occurred along valley walls (which are typically aligned N-S) as the Weichselian and earlier icesheets receded from W-E. Some stripe karst outcrops in central Scandinavia also support permanent bodies or flows of water (lakes, tarns and streams: f aulkner 2005), which suggests a sporadic lack of speleogenesis. Thus, it is concluded that tectonic activities and fractures occur in clusters along the various outcrops. As each successive glaciation deep- ened glacial valleys and fjords further, the ice thickness variation, and therefore the intensity of seismic shocks in some earthquake zones, must have increased during the time of the Mio–Plio–Pleistocene glaciations. However, because the valley geography remained roughly constant, each cluster of large seismic shocks remained approxi- mately concentrated on the same position. Hence, each successive deglaciation commonly re-activated previ- ous fracture sets, and extended them farther along, and farther below, the contemporary surface than the pre- vious one. Because the present maximum subsurface cave distance is almost universally one-eighth the range of local relief, it seems likely that both the depth of the partially detached crustal layer and the maximum sub- surface distance of cave passages also increased at one- eighth the rate of glacial valley deepening. However, act- ing synchronously with this deepening, there is also the probability that previous palaeo passages were removed by the erosional lowering of the surface by glacial strip- ping. The competition between these two processes was explored further by f aulkner (2005), to create a general Caledonide model of cave development. In this model, fractures were created by a pulse of deglacial seismicity that accompanied the recession of each ice margin, af- ter which the presently relict phreatic passages enlarged by deglacial speleogenesis in cold water with little CO 2 (f aulkner 2006b) and the mainly vadose passages en- larged during interglacial speleogenesis. The general model was also found to be valid for karst caves in the other non-Arctic metamorphic Cale- donide terranes of northern Scandinavia, southern Scandinavia, New England (USA) and in the Dalradian Supergroup outcrops in Scotland and Ireland (f aulkner 2005), where the maximum one-eighth relationship is also commonly maintained. Only in some deep caves in northern Norway (e.g. Tjoarvekrajgge, the Okshola / Kristihola system and the Greftkjelen / Greftsprekka system) is the one-eighth relationship (dramatically) ex- ceeded. The likely explanation is that tectonic inception at such caves was promoted by longer-timescale, possibly aseismic, processes such as the long-term uplift of the Scandinavian landmass or the spreading of the Atlantic Ocean, rather than mainly by deglacial seismicity caused by rapid uplift. In at least the first two of the given exam- ples, their relatively large subsurface cave distances may also be facilitated by the LAK nature of their (medium grade) marble outcrops, providing more opportunities for ‘conventional’ inception horizons to operate (c.f. Yt- terlihullet, as discussed above). The island of Shetland, also within the Dalradian Supergroup, was found to have no caves within its long marble stripe karst outcrops and only small exokarst features. Hence, Shetland provides the (null) end member of a Caledonide tectonic incep- tion series, because the Weichselian ice sheet was much thinner there, the relief is modest, and there was little de- glacial seismicity. Indeed, the island was uplifted during glaciation and the present land surface is actually falling during the Holocene, so that some valley floors are now drowned by the sea to create inland waterways that are locally called voes. If the conclusions in this paper are correct, the exis- tence of cave passages in Caledonide marbles can be used as a proxy for the formation of tectonic fractures, and the evidence provided should be important in the field of seismology: it implies that most fracture creation arises THE ONE-EIGHTH RELATIONSHIP THAT CONSTRAINS DEGLACIAL SEISMICITy AND ... ACTA CARSOLOGICA 36/2 – 2007 202 from local earthquakes caused by adjustment to local- scale differential ice load, and arises only exceptionally from earthquakes or slow tectonic movements caused by Scandinavian-scale isostatic uplift or by the present mid-Atlantic ridge-push that is widening the ocean. Ad- ditionally, the presence and structure of the cave passages themselves and any internal neotectonic displacements may provide a method to deduce the strength and nature of the deglacial earthquakes. ACKNOWLEDGMENTS This paper reports part of a wider project to study spe- leogenesis in Caledonide metacarbonate rocks (f aulkner 2005), for which Professor John Gunn and Dr. David Lowe were helpful and patient supervisors. Dr. Rod Gay- er generously invited me to attend his lectures on Caledo- nian–Appalachian Tectonics at the University of Cardiff, and a field trip with Dr. Colin Davenport and his stu- dents at the University of East Anglia to study neotecton- ics in the Scottish Caledonides was extremely beneficial. Philippe Audra and Art and Peggy Palmer are thanked for their supportive and constructive review comments. REf ERENCES Bakalowicz, M., 1984: Water chemistry of some karst en- vironments in Norway.- Norsk Geografisk Tidsskrift, 38, 3–4, 209–214. Davenport, C. A. &P . S. Ringrose & A. Becker & P . Han- cock & C. f enton., 1989: Geological investigations of late and post glacial earthquake activity in Scot- land.- In: S. Gregerson & P .W . Basham (Eds), Earth- quakes at North-Atlantic Passive Margins: Neotec- tonics and Postglacial Rebound, 175–194. Dehls, J. f . & O. Olesen & H. Bungum & E. C. Hicks & C. D. Lindholm & f . Riis., 2000: 1:3000000 Neotec- tonic map: Norway and adjacent areas.- Geological Survey of Norway. Ekström, G. & M. Nettles & G. A. Abers., 2003: Glacial earthquakes. - Science, 302, 622–624. f aulkner, T., 2005: Cave inception and development in Caledonide metacarbonate rocks.- PhD thesis, Uni- versity of Huddersfield. f aulkner, T., 2006a: Tectonic inception in Caledonide marbles.- Acta Carsologica, 35, 1, 7-21, Ljubljana. f aulkner, T., 2006b: Limestone dissolution in phreatic conditions at maximum rates and in pure, cold, wa- ter. - Cave and Karst Science, 33, 1, 11–20. Lauritzen, S-E., 1981: A study of some karst waters in Norway.- Norsk Geografisk Tidsskrift, 35, 1, 1–19. Lowe, D. J., 1992: The origin of limestone caverns: an in- ception horizon hypothesis.- PhD Thesis, Manches- ter Metropolitan University. Lowe, D. J. & J. Gunn., 1997: Carbonate speleogenesis: An Inception Horizon Hypothesis.- Acta Carsolog- ica, 26, 2, 38, 457–488. Marrett, R., O. J. Ortega & C.M. Kelsey., 1999: Extent of power–law scaling for natural fractures in rock.- Geology, 27, 9, 799–802. Milanović, P . T., 1981: Karst Hydrogeology.- Water Re- sources Publications, 434pp, Littleton, Colorado. Ringrose, P . S. & P . Hancock & C. f enton & C. A. Dav- enport., 1991: Quaternary tectonic activity in Scot- land.- In: A. f orster & M. G. Culshaw & J. C. Cripps & J. A. Little & C. f . Moon (Eds.), Quaternary En- gineering Geology, Geological Society Engineering Special Publication No. 7, 679–686. TREVOR f AULKNER