Rock Weathering on the Eastern Mountains of Southern Africa: Review and Insights from Case Studies

ARTICLE IN PRESS

Journal of African Earth Sciences xxx (2009) xxx–xxx

Contents lists available at ScienceDirect

Journal of African Earth Sciences

journal homepage: www.elsevier.com/locate/jafrearsci

Rock weathering on the eastern mountains of southern Africa: Review and insights from case studies

P.D. Sumner a,*, K.J. Hall a,b, J.L. van Rooy c, K.I. Meiklejohn a a Department of Geography, Geoinformatics and Meteorology, University of Pretoria, Pretoria 0002, South Africa b Geography Program, University of Northern British Columbia, 3333 University Way, Prince George, BC, Canada V2N 4Z9 c Department of Geology, University of Pretoria, Pretoria 0002, South Africa article info abstract

Article history: The mountains in the eastern region of southern Africa are of significant regional importance, providing Received 7 November 2008 for a diverse range of land use including conservation, tourism and subsistence agriculture. The higher Received in revised form 15 April 2009 regions are comprised of flood basalts and are immediately underlain by predominantly aeolian-origin Accepted 28 April 2009 sandstones. Our understanding of the weathering of these basalts and sandstones is reviewed here, with Available online xxxx particular focus on the insights gained from the Lesotho Highlands Water Project and an ongoing study into the deterioration of rock art. While the chemical weathering attributes of the basalts have been sub- Keywords: stantially investigated, it is evident that the environmental surface conditions of rock moisture and tem- Weathering perature, as affecting weathering processes, remain largely unknown. Within the sandstones, studies Lesotho Drakensberg pertaining to rock art deterioration present insights into the potential surface weathering processes Rock art and highlight the need for detailed field monitoring. Outside of these site-specific studies, however, little Lesotho Highlands Water Project is understood of how weathering impacts on landscape development; notably absent, are detail on weathering rates, and potential effects of biological weathering. Some palaeoenvironmental inferences have also been made from weathering products, both within the basalts and the sandstones, but aspects of these remain controversial and further detailed research can still be undertaken. Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction on the South African province of KwaZulu–Natal has recently been declared a Transfrontier National Park and is thus of considerable The eastern escarpment region of southern Africa is the highest environmental and cultural importance. While the significance of mountain range south of the central African Kilimanjaro massif. the mountains from a climatological, hydrological and ecological Called the Maloti mountains or Lesotho highlands in the west, perspective is generally appreciated, comparatively little is known and Ukhahlamba–Drakensberg in South Africa in the east, the of the area from a geomorphological viewpoint and few studies de- highest summit Thabana Ntlenyana reaches 3482 m a.s.l. (Fig. 1). scribe weathering processes within the region. Two focus areas, the The landlocked Kingdom of Lesotho is bordered entirely by South study of rock art deterioration and the engineering feats of the Africa, and Lesotho’s eastern international boundary is delineated Lesotho Highlands Water Project, have provided insights into by the watershed which approximates the position of the moun- weathering processes, while other studies draw palaeoenviron- tain range escarpment. Late Triassic to early Cretaceous flood bas- mental conclusions from weathering products. alts comprise the area above approximately 1900 a.s.l., underlain San rock art, painted onto sandstone, is a major tourism and by Karoo sediments the uppermost sequence of which is known as cultural heritage attraction. Interpretation of rock art has drawn the Clarens Formation (Fig. 2). Above 3000 m the climate supports much attention from archaeologists and anthropologists but geo- marginal periglacial conditions with a mean annual air tempera- morphological studies have also contributed through an ongoing ture in the region of 5 °C and precipitation less than 1000 mm analysis of rock art deterioration by weathering (Hall et al., p.a. (Nel and Sumner, 2008). Temperatures are warmer in the foot- 2007a). Recent studies utilize new techniques and have improved hills in the east where fluvially incised valleys expose the sand- our understanding of, and asked new questions pertaining to, the stones and precipitation increases to around 1200 mm p.a. weathering of sandstone. At higher altitudes the hydrological (Sumner and Nel, 2006). Much of the mountain area bordering importance of the mountains is best illustrated by the Lesotho Highlands Water Project (LHWP). The LHWP transfers water from the highlands across catchments to the central industrial region * Corresponding author. Tel.: +27 12 420 2881; fax: +27 86 699 2661. of South Africa and was considered one of the largest civil engi- E-mail address: [email protected] (P.D. Sumner). neering projects then undertaken in the world (Waites, 2000).

Please cite this article in press as: Sumner, P.D., et al. Rock weathering on the eastern mountains of southern Africa: Review and insights from case studies. J. Afr. Earth Sci. (2009), doi:10.1016/j.jafrearsci.2009.04.010 ARTICLE IN PRESS

2 P.D. Sumner et al. / Journal of African Earth Sciences xxx (2009) xxx–xxx

Fig. 1. Location of basalts and underlying sandstones in eastern southern Africa. See Waites (2000) for further detail on the transfer dam and tunnel locations of the LHWP.

Water transfer tunnels drilled through basalt, as well as the studies. Comments are also made on the palaeoenvironmental requirements for construction aggregate, prompted studies into inferences derived from weathering products. basalt weathering and durability and much was learned from problems encountered with the project. Both the rock art study and the 2. Weathering of the basalts water transfer project have thus contributed signiﬁcantly to a relatively small but important body of literature on rock weathering 2.1. Composition and alteration of the basalts processes in the region. This paper reviews our current understanding of basalt and sandstone weathering in the mountains Volcanic eruptions heralded the splitting up of the super and elaborates on the insights gained from the above two case continent Gondwana and created a lava pile with an original

P.D. Sumner et al. / Journal of African Earth Sciences xxx (2009) xxx–xxx 3

Fig. 2. Typical geological succession (after Eriksson, 1983) on an east–west transect through the escarpment, such as at Giant’s Castle in Fig. 1. thickness of approximately 1400 m (Duncan and Marsh, 2006). Mineralogical composition is important when considering Individual lava flows vary from less than 1 m thick to more than chemical weathering of the basalts. Primary minerals crystallized 50 m, with an average flow thickness of 6 m in the northern directly from the molten lava due to lowering in temperature parts of Lesotho. Flow contacts are generally welded and tight, and pressure either in the upper levels of a feeder channel or fis- indicating a relatively quick succession of lava outpourings that sure, or on extrusion at the surface. A distinct order of crystalliza- lack weathered flow contacts. The lower occurrences in the lava tion occurred, with olivine, where present, crystallizing first, pile contain some volcanic clastics and tuffs with volcanic brec- followed by plagioclase feldspar, then clinopyroxene. Solidification cias. These rock types are an indication of early explosive volca- took place in most of the lavas after the crystallization of the nic activity although the extent of the lava flows suggests that pyroxene, presumably as a result of extrusion. Residual liquid the lava poured out from fissures as low viscosity flood basalts. trapped in the interstices between crystalline phases underwent Typical of the basalts are the abundance of amygdales filled with rapid supercooling and solidified as glass, trapping within it incip- various secondary minerals. Thin flows are amygdaloidal ient crystals of opaque iron oxide magnetite or the iron–titanium throughout the flow thickness while thicker flows exhibit a dis- oxide ilmenite (van Rooy, 1992). Augite is the dominant pyroxene tinct zoning (from top to bottom) within the flow. The top layer with plagioclase the most abundant mineral of the rocks. Glass is is fine-grained, highly amygdaloidal and glassy as a result of the relatively abundant in most rocks and may reach 25% by volume sudden temperature drop and pressure decrease. Due to the ra- and much of the glass is altered to a variety of radiating micaceous pid outpouring of one flow on top of the other the upper parts of mineraloids. Zeolites are widespread in amygdales in areas of some flows also exhibit autobrecciation. The central part of the hydrothermal alteration and may exhibit a well-defined zonation thick flows is massive with only scattered amygdales while the (Hay, 1978). Clinoptilolite or mordenite characterizes the shallow- basal part is again fine-grained and contains numerous amyg- est and coolest zones, and progressively deeper zones commonly dales, some of which are typical pipe amygdales orientated per- contain analcime or heulandite, laumontite, and wairakite. The pendicular to the flow contacts. On both sides of the flow typical gas vug infill in Lesotho can be summarised as quartz contacts, thin zones occur where the basalt is usually very (agate) and calcite in the basal 170 m of the lava pile and zeolite fine-grained and more brittle than the adjacent rock further in the upper 1230 m. The zeolite zone is further defined by a lower away from the contact zone (Bell and Haskins, 1997). These brit- (250 m) laumontite zone and an upper mesolite zone. tle zones fracture preferentially under stress conditions. The early-formed olivine crystals are small and almost always Karoo volcanic rocks are crystalline and are therefore subjected completely replaced by minerals produced by deuteric action. Deu- to both mechanical disintegration and chemical decomposition (or teric alteration is characteristic of volcanic rocks which, as a result mineral transformation) depending on the environment (Weinert, of the sudden dramatic drop in containing pressure, release huge 1980). Weathering products will be typical of basic crystalline par- volumes of gases previously held in solution in the melt when ent rocks (text compiled by H.H. Weinert in Brink, 1983). In less extrusion at the surface takes place. Gases trapped within the rock mountainous areas, and outside the study area, such as the Spring- constitute a potent agency for the transformation of the host rock bok Flats, highly active residual soil from basalt can sometimes while solidifying. At least part of the deuteric alteration observed reach significant thickness (Brink, 1983) but the thin soils and in the Lesotho basalts took place virtually at the time of solidifica- the exposure of large areas of comparatively fresh basalt in Lesotho tion or immediately thereafter. This includes changes to the pri- are due to intensive erosion in the montane setting. While the mary minerals such as pseudomorphous replacement of olivines, environmental setting appears conducive to chemical alteration the localised but intensive attack on the plagioclase feldspars, as the dominant form of weathering of the basalts, several authors and partial or complete replacement of the original glass. Davidson (Irwin et al., 1980; Brink, 1983; OSC, 1986) also allude to mechan- (1972) postulated on three types of deuteric alteration: clay ical breakdown due to cool temperatures. mineral alteration, carbonate alteration and zeolite alteration. A

4 P.D. Sumner et al. / Journal of African Earth Sciences xxx (2009) xxx–xxx further type of deuteric action is where mineralizing fluids perco- lower altitude, corresponding with the sandstone-basalt contact late through the solid rock and deuterically alter the lava. The (1920 m), Sumner and Nel (2006) measured basalt temperatures products of such deuteric action are visible in veins, in the round and recorded a minimum of À6.7 °C and a maximum of 56 °C. Both vesicular cavities formed by bubbles of gas evolved under high studies note a poor correlation between air and rock surface tem- pressure in the still-fluid lava, and in angular interstitial voids that perature, and on potential for thermal stress. Notably absent are probably represent gas-filled cavities in parts of the flow in which data on rock moisture and moisture chemistry conditions. Thus solidification was complete at the time of the release and migra- the efficacy of processes such as wetting and drying, salt weather- tion of the gas. This deposition of the secondary minerals was ing and freeze-thaw is speculative, particularly in winter when likely a slow process starting at some indefinite time after consol- conditions are generally dry, while the role of biological activity idation of the rock and possibly continuing to the present day at and its potential integration with other weathering processes (Eti- depth in the volcanic pile. enne, 2002) remains unknown. Where gas cavities are prominent, alteration of the basalt may be pronounced. These cavities are usually filled with, quartz, agate, 2.2. Case study: the Lesotho Highlands Water Project (LHWP) chalcedony, and calcite, or with zeolites such as heulandite, stilbite, thomsonite, and scolecite (Walker and Poldervaart, 1949). The LHWP transfers water from storage dams in the mountains Cox and Hornung (1966) and the Lahmeyer MacDonald Olivier of central Lesotho to South Africa (Fig. 1) through a system of tun- Shand Consortium (OSC, 1986), however, found a considerable nels. Katse and Muela Dams were constructed first and linked by a variety to exist in the minerals thought to have been deposited 45 km long Transfer Tunnel, bored with a 5 m diameter through in cavities in the basalts from percolating mineralizing fluids, the basalt. From the Muela Dam, the 36 km long Delivery Tunnel exits most abundant of which is the zeolite stilbite. Chlorite, in finely- in a tributary of the Vaal River in South Africa. The tunnel passes aggregated form and in close association with montmorillonite, through the underlying sandstones and mudstones of the Karoo fills some vesicles completely or occurs as the rims of some amyg- Supergroup and was lined segmentally during boring with precast dales. Although plagioclase feldspar appears generally fresh, there concrete units. Mohale Dam supplies Katse Dam via a 31 km inter- are signs of alteration in most mineral grains. This is seen as a pen- connecting tunnel, also drilled through basalt. While initial plans etration of the mineral, along cleavage planes and irregular frac- were that the tunnels through basalt would not need to be lined tures, by thin films of a clear to slightly greenish clay mineral. as required in the sedimentary sequences, during a visual inspec- The main mineral produced in this type of replacement appears tion of the Transfer Tunnel in 1994 it was found that large sections to be montmorillonite (Deer et al., 1963) but the changes generally of the tunnel length showed severe weathering. The quality and exert little influence on the durability of the rock. A more serious durability of the exposed rock and imminent sidewall disintegra- form of deuteric alteration is the zeolitization of the feldspars. tion eventually led to the decision to line the tunnel. The tunnel Early-formed euhedral crystals of olivine occur in almost all the lining consisted of precast concrete segments with sound dolerite basalts, and their complete replacement by deuteric minerals is al- as coarse aggregate (LHC, 1994). This had a significant influence most universal. The dominant replacing mineral is again montmo- on the construction period as well as the final cost of the project, rillonite, which can be positively identified optically because of its and the weathering processes involved thus warranted further occurrence as unusually large crystals in some of the pseudomorph investigation. crystals. Chlorite and hematite are closely associated with mont- In early exposed rock core samples and from literature it was morillonite in the olivines. The latter forms rims around the origi- noted that fracturing of certain basalts, referred to as ‘‘crazing”, oc- nal crystals and preserve their characteristic original fracture curs. This was initially attributed to the action of swelling clays. system, thereby facilitating distinction of olivine crystals from Petrographic study showed, however, that the ‘‘crazing” is not other concentrations of deuteric montmorillonite. genetically related to swelling clays although the fractures exploit The rocks studied during the Lesotho Highlands Water Project patches of these clays as weak spots. ‘‘Crazing” was also confined feasibility investigations showed negligible weathering, but practi- to the coarsely amygdaloidal parts of basalt flows, generally in cally all would have been subjected to deuteric alteration during their upper zones. The fracture systems were found to originate the late stages of solidification or shortly thereafter. This alteration in amygdales filled mainly by zeolites (OSC, 1986). The conclusion manifests itself in the form of the transformation of original (devit- of the Lahmeyer MacDonald Olivier Shand Consortiums was that rified) glass and early-formed olivine crystals wholly or partially to the presence of shrinkage cracks in montmorillonite-rich patches swelling clays of the smectite group. The feldspars in the rock may, indicates that the basalts are in a dehydrated condition (OSC, to a lesser extent, be replaced by zeolites and clay minerals such as 1986). The zeolites are noted for their capacity of holding water montmorillonite. Montmorillonite was identified as the major molecules loosely and of giving up this water easily (LHC, 1994). mineral present through thin section staining, XRD analysis, At high temperatures all the water is driven off irreversibly and although minor amounts of nontronite also occurs (Van Rooy, there is evidence that structural changes take place. For some zeo- 1992). The main cause for degradation or rapid weathering in these lites, the decrease in volume can be as much as 22% that will pro- rocks is the presence of swelling clay minerals as demonstrated in duce shrinkage forces in the closed system such as an amygdale. the case study below. This may initiate the ‘‘crazing” and the failure of the rock in which While chemical alteration typifies basalt on exposure, relatively the zeolites occur. Rapid deterioration encountered in some basalt little is known of physical weathering aspects. Mechanical fractur- rock types was later ascribed to a combination of the mineralogy of ing, flaking and spalling (Grab, 2007a) and extensive block produc- the different rock types that respond to changes in moisture con- tion (e.g. Boelhouwers et al., 2002; see below), are reported for the tent on exposure, the changes in stress regime on excavation, basalts. Although, with the exception of dilatation (or unloading), and the intrinsic geotechnical characteristics of each rock type mechanical weathering is based primarily on rock temperature (LHC, 1994). and rock moisture conditions little is known of these environmen- It is now accepted that the main cause of basalt deterioration is tal conditions. Grab (2007a) reports on short-term rock tempera- the expansion of swelling clay minerals (e.g. montmorillonite) and ture oscillations that appear conducive to thermal stress at active zeolites (e.g. laumontite) occurring within the rock mass 3060 m near Sani Pass. Longer duration monitoring at 3300 m (LHC, 1994). Volume changes in these minerals occur as soon as highlights the thermal potential for frost activity, with minimum they are exposed to moisture changes. Microfissuring, jointing, temperatures at the rock surface of À13.6 °C(Grab, 2007b). At a fissuring or crazing will cause an incremental increase in rock

P.D. Sumner et al. / Journal of African Earth Sciences xxx (2009) xxx–xxx 5 porosity and this together with the strength of the rock material within shelters or caves, sandstone rock surfaces can experience and the quantity and distribution of active minerals will determine rapid temperature fluctuations, exceeding 2 °C/min, when exposed the rate of disintegration (LHC, 1994). Smectite clay minerals, de- under partly cloudy conditions (Meiklejohn, 1994). Minimum tem- rived from the deuteric alteration of olivine, glass and plagioclase, perature measured is À6.1 °C and maximum 50.1 °C(Sumner and and active zeolites, laumontite–leonhardite and scolecite are pres- Nel, 2006) while daily ranges regularly exceed 30 °C(Meiklejohn, ent in amygdales and as interstitial secondary minerals in all the 1994). Potential for thermal fatigue and shock is thus apparent basalts. Vertical zeolite zoning and dominant zeolites present were but no further field evidence is available. Detailed atmospheric determined through exhaustive XRD analysis and research by Dun- and rock surface measurements have also been taken within rock levey et al. (1993). The influence of the active minerals is also art shelters (Meiklejohn, 1994; Hoerlé, 2006). Meiklejohn (1994) dependent on the texture of the parent rock. If the texture results notes a greater activity of weathering in exposure-limited sites in low permeability rock material, the mineralogy may then not be underneath rock overhangs where the presence of moisture ap- sufficient on its own to cause deterioration of the rock within the pears to be the most important driver of sandstone weathering. design life of civil engineering structures. A low porosity, together with high microporosity, however, pre- cludes the movement of moisture through the sandstone massif, 3. Weathering of the sandstones although some seepage does occur along joint lines (Meiklejohn, 1994). Surface moisture fluctuations, which can also be driven by 3.1. Chemistry, weathering and durability of the sandstones air humidity, may cause salt crystallization, hydration and dehydration of minerals, precipitates and clays, and solution processes The Clarens Formation is the uppermost of sedimentary rocks (Meiklejohn, 1994). Chemical analysis shows that favourable mois- that comprise the Karoo Sequence; a large basin of sediments laid ture and temperature conditions alter the feldspars and break down between 200 and 160 million years ago. Overlain by the down calcite cement while chemical weathering products provide younger basalts, the Clarens is in turn underlain by the older Elliot precipitates at or near the surface that can cause mechanical ac- and Molteno Formations and shows a depositional palaeoenviron- tion. Meiklejohn (1994) does, however, note that much of the evi- ment that increases in aridity from the Molteno through the Elliot dence for chemical weathering points towards a wetter period in to the Clarens Formation (Eriksson, 1983). The sequences, includ- the past. ing the basalt layers, conformably overly one another but, except While the detailed study by Meiklejohn (1994) provided insight for the basalt transition, boundaries are often indistinct. In the cen- into the rock properties and environmental conditions within shel- tral Drakensberg region the Clarens Formation reaches a thickness ters, recent observations and access to new equipment has seen a of 300 m and is primarily of aeolian origin with predominantly resurgence in interest in weathering and rock art (see Hall et al., fine-grained sandstone, subordinate siltstones, and medium 2007a). Several advances have been made that provide further in- grained sandstone and occasional mudstone horizons towards sight into weathering of the Clarens Formation and are presented the base of the Formation. Eriksson (1983) identified four lithofa- below in the context of understanding rock art deterioration. cies, the dominant of which are massive cross-bedded structures that form the scenic cliffs commonly called the Little Berg. Based 3.2. Case study: sandstone weathering and rock art deterioration on petrographic work by Beukes (1970) and Eriksson (1981), the sandstones consist of 35–90% quartz, up to 20% feldspar with a ma- Historically, weathering of the Drakensberg sandstone has been trix of 10–60% (van Rooy and van Schalkwyk, 1993). Silica and cal- highly subjective and poorly understood. In the context of rock art, cite are present as cementing agents, but will weather at different available information has suggested radiation-induced thermal rates. Accessory minerals include zircon, hornblende, sphene, gar- stresses may play a role where paintings are exposed to direct sun- net and clay minerals kaolinite, illite, montmorillonite, chlorite, light (e.g. Loubser, 1991; Rudner, 1989; van Rijssen, 1987), the smectite with traces of mica (Beukes, 1970; Eriksson, 1981, more so if in association with moisture (e.g. Batchelor, 1990). 1983; Meiklejohn, 1994). Weathering and erosion of the structur- Moisture can also play a significant role (e.g. Avery, 1974) while ally weaker, closely jointed alluvial-origin sediments in the lower salts, and their expansion/contraction resulting from hydration/ sections of the Formation create overhanging shelters or ‘‘caves” dehydration cycles (e.g. Batchelor, 1990), may be a significant fac- within which numerous examples of San rock art are found. tor at some sites especially if the salts occur between the pigment Sandstone is more durable than the siltstones and mudstones, is and the rock (e.g. Loubser, 1991). Overall, although possibly sound less prone to slaking, and in the case of the Clarens sandstones, in theory, most assessments of weathering lack appropriate data. proved highly resistant to laboratory durability testing (van Rooy To exacerbate this situation, some recent studies have added a fur- and van Schalkwyk, 1993). Geotechnical investigations for the ther layer of obfuscation by misconstruing macro-cave climatic Lesotho Highlands Water Project found that the unconfined com- information for rock-surface micro-climatic data (e.g. Hoerlé and pressive strength of the Clarens Formation varied appreciably, Salomon, 2004) by (wrongly) equating non rock-surface, air condi- from moderately strong to extremely strong (Castro and Bell, tions within a shelter for those that occur at the (sub-)millimetre 1995). Where sandstone grains are loosely packed, with little or scale on the rock surface. Conversely, a well suited approach to no point contact, and where clay content is high, rapid weathering evaluating the degree of weathering (but clearly not the cause of is, however, expected due mainly to wetting and drying cycles. weathering) over recent (decadal) time spans is that of comparing When used as concrete aggregate, the sandstone absorbs water photographs of the same paintings at different times and evaluat- from the concrete mix, thus decreasing the workability and ing percentage loss (Hoerlé, 2005). As weathering rates are unlikely increasing the amount of water required, and subsequent drying to be linear (e.g. Egli et al., 2003; Viles, 2001) and environmental results in a high overall shrinkage. Together with later wetting conditions have probably changed, so this photographic approach and drying at the concrete surface, differential strains can result still has limitations. Thus, the weathering of San rock art is still lit- in cracking (Roper, 1959; van Rooy and van Schalkwyk, 1993). Lab- tle understood, and assumptions are often somewhat simplistic in oratory testing by Hall and Hall (1996) and Sumner and Loubser nature. It is therefore appropriate to review former comments and (2008) also confirm the susceptibility of the sandstone to wetting studies before outlining the state of current research. and drying cycles. The loss of San rock art, as with other sandstone cultural heri- As noted for the basalts, the current environmental rock tem- tage, is a result of ‘‘natural and anthropogenic factors, involving perature regime is poorly known. At exposed sites, as opposed to physical, chemical and biological phenomena. Deterioration

6 P.D. Sumner et al. / Journal of African Earth Sciences xxx (2009) xxx–xxx processes often occur by means of phase transitions, most of which understand the potentially complex nature of sandstone weather- require months, years or centuries to take place” (Benedetti et al., ing if we are to preserve these cultural relics for future generations 2008, p. 155). Considering the possible age of some of the art (Pope et al., 2002). Equally, we need to understand the properties (>2000Y BP: Mazel and Watchman, 2003), it is perhaps more sur- of the paintings themselves (Odlyha et al., 1997), and how these di- prising it has survived so well. In some ways, the recognition of rectly impact their own weathering, as well as their interaction art durability is something not often identified (a ‘taphonomic’ with the rock on which they are painted. attribute? – Bednarik, 1994) and yet it is a significant factor. In- Antipathetical to the comment by Bradley et al. (2002) and deed, Bednarik (1994, p. 69) points out that survival of rock art Chalmin et al. (2003) note that while archaeologists have at- can be dependent upon (p. 70) ‘‘...certain techniques, pigment or tempted to interpret rock art, they have often disregarded the tech- paint types, locations, media, climates etc.” Some paintings having nical aspects of the paints; a sentiment echoed by Zoppi et al. been here for a few thousand years (Mazel and Watchman, 2003) (2002). The physicochemical attributes of the paint can offer infor- identifies that they have experienced substantial climatic change. mation on how the paints were created as well as on their proper- Thus, the present climatic variability is not unusual within the con- ties and hence how they may react to the environment. Indeed, text of changes having taken place during the Holocene (Alley Casellato et al. (2000, p. 217) argue that conservation intervention et al., 1997) and the apparent present degradation of the art re- can be more appropriate when ‘‘...quantitative physico-chemical quires that consideration also be given to other impacting factors. properties of the object are sufficiently known” as this helps eluci- Although used in a different context, Bradley et al. (2002) state date the modification the object has undergone as a result of that it may be just as important to study the rock as it is to inves- ‘‘...the action of time, weather and human beings.” Once we better tigate the art for (p. 109) in respect of the majority of (San) rock art understand the chemical composition so we may better under- studies, ‘‘the... paintings have come to dominate the discussion – stand the performance properties of the pigments (Hao and Iqbal, the rock itself appearing to be of secondary importance.” Bakkevig 1997). Further, our knowledge of paint composition, and its inter- (2004) points out that while such as an ancient ruin can produce action with the (light-transmissive) sandstone on which it is differences in microclimate and can change the (local) environ- placed, may be significant to an understanding of possible bacterial ment markedly, this is in stark contrast to rock art which has little colonization; given the nature of the rock and presence of the or- to no impact on the environment itself. Bakkevig (2004) also notes ganic paints, the absence of any bacterial colonization may be sci- that the vegetation of a site may not be the same as that which was entifically significant. Bacterial activity is a factor not widely present when the art was created. This is because at a number of investigated with respect to the San rock art, but colonization of sites vegetation has been changed in recent times with the goal iron oxide art in a sandstone rock shelter has been reported from of improving access and visibility; as argued by Hall et al. Spain (Gonzalez et al. 1999). Ultimately this may prove an impor- (2007a,b) and Arocena et al. (2008) in respect to such changes tant attribute, for Gonzalez et al. (1999) found a potential for affecting the microenvironment. Thus, there is a need to clearly microbial growth with hematite that could be significant from

Fig. 3. Examples of rock art on Clarens Formation sandstone: A shows an example of (ochre) pigment directly onto a naturally weathered rock surface, the pigment, by visual comparison, is quite thick; B shows what is considered an area of anthropogenically-smoothed rock on which paintings were located, the area to the top of the paintings appears quite different although the smoothed rock still retains the broad irregularities inherent within the rock; C and D are examples of areas that have been anthropogenically-smoothed and then covered by a clay-based ground; examples of both striae from the smoothing and sections where the ground has detached are evident.

P.D. Sumner et al. / Journal of African Earth Sciences xxx (2009) xxx–xxx 7 the perspective of conservation; the bacteria can also change the of some art, Deflorian et al. (2007) note that coating thickness colour of paintings. has a strong influence on weathering and that loss of adhesion of In broad terms, the San rock art appears to be on one of three the paint to the medium can also accelerate degradation. Loss of types of surface: (1) (seemingly) directly on the natural rock sur- adhesion is dealt with in detail by Perera (1996) where he found face, (2) on anthropogenically-smoothed rock surfaces, and (3) on most organic coatings under stress due to variation in temperature a clay ground placed on top of a smoothed surface (Fig. 3). Where and relative humidity. The result of these stresses (internal, ther- rock has been anthropogenically smoothed to create a surface for mal, hygroscopic) acts against adhesion and provokes delamina- the painting, this action has theoretically ‘‘zeroed” the weathering tion (Perera, 1996, p. 23); a high tensile stress is induced under clock (Pope et al., 2002, p. 213). However, the surface may not be dry, cold conditions and a compressive stress under humid, warm truly ‘‘fresh”, as it can possess a ‘memory effect’ insofar as conditions – conditions applicable to the Drakensberg. Thus, at the ‘‘...weathering often penetrates into the rock...”(Pope et al., scale of the art weathering there is a need to evaluate both the art/ 2002, p. 213) and future weathering-related changes to that rock rock physico-chemical properties and the environmental condi- may well reflect non-visible attributes created under the previous tions (and how these have changed with time). weathering regime. However, where a ground is applied to the In the ongoing study, research directions are: (1) further deter- smoothed rock (Hall et al., 2007a) and the paint put on top of that mination of the paint composition, coupled with (2) evaluation of then the situation is somewhat different, but some of the memory paint pigment properties, (3) on-going monitoring of environmen- attributes can still play a role at some point in the weathering pro- tal determinants at scales appropriate to determining pigment-to- cess. Pope et al. (2002, p. 220) note, cultural stone may not be that pigment, pigment-to-ground, ground-to-rock, and rock-to-pig- representative of the natural world in that ‘‘The acts of preparing ment stresses, (4) investigating environmental changes at art sites, cultural stone create structural stresses that are different from (5) studies of anthropogenic preparation of rock surface impacts on those found in nature.” Further, the changes to atmospheric chem- weathering, (6) measurement of pigment thickness as a determi- istry due to economic development, a notable factor in the past nant in painting/pigment longevity, and (7) using all of the forgo- 100 years of southern Africa, have a significant impact because ing to model the stress fields associated with the three identified present weathering activity may differ significantly from that of methods of paint applications. Indirectly, this will also help better pre-human-impact weathering (Pope et al., 2002, p. 221). Thus understand the weathering of the sandstone per se, and thus have the age of paintings (and, possibly, the impact of previous vegeta- applicability to the preservation of the many buildings composed tion cover) may well be critical to our full understanding of the of this material found throughout southern Africa. current weathering. In addition, the art age may be a useful tool for evaluating the rates of rock weathering, as proposed by King 4. Weathering and palaeoenvironmental inferences (1945), but the scarcity of accurate dates (see Mazel and Watch- man, 2003), as well as the potential degrading effects of pre- Within the Clarens Formation, many rock surfaces are covered weathered surfaces, militates against further detail in this regard. by a gypsum precipitate, originating from chemical alteration of Findings to date in the ongoing rock art weathering study were the sandstone. Given that rock art is painted on the surface of recently detailed in Hall et al. (2007a) and so it would not serve to the gypsum (an excellent example is the Battle Scene at Battle repeat them in detail here. The key points germane to weathering Cave), it is clear that some of these precipitates formed during pre- of the sandstone are, first, that the paint acts as surface modifier vious wetter climates (Meiklejohn, 1997). Accurate dating of the (Bullet and Prosser, 1983), significantly changing albedo, light art, which has thus far remained largely elusive, may thus provide transmissivity, thermal, and moisture properties of the rock, and a useful tool for estimating minimum ages for the wetter periods of second, where the paints are applied on top of a clay ground so the late Holocene. Within slope deposits, clast angularity, or the both the ground and the paint change surface properties. Thermal absence thereof, has been central to interpreting Late Pleistocene infra-red data have shown that white and red pigments, as well as palaeo-weathering processes. Sandstone, however, appears to be the rock, all have quite different responses to solar radiation that predisposed to angularity. In contrast under wet conditions chem- can lead to pigment-to-pigment as well as pigment-to-rock stres- ical weathering of the basalts produces rounded products, ses (Hall et al., 2007a,b). Thermal and moisture fluctuations can although angular blocks have also been observed (see below). Sev- lead to fracturing of the clay ground allowing the entrance of eral studies interpret angular sandstone clastic deposits as evi- microbial activity (Arocena et al., 2008). In many instances it is dence for former frost action, notably conditions expected the detachment of the clay ground rather than the weathering of around the Late Pleistocene Last Glacial (max. 18,000 B.P.) but the rock that is the destructive factor. It has also been suggested the presumptions have been strongly criticized (see Hall, 1992). (Hall et al., 2007a,b; Arocena et al., 2008) that recent changes to With specific reference to southern African conditions, Hall surrounding vegetation may have increased exposure to solar (1991, 1992) notes that angularity can be produced through sev- and moisture impacts leading to accelerated deterioration. eral processes, including thermal stress fatigue, dilatation and wet- At the local scale, the sandstone is reasonably uniform, with re- ting and drying. The presumption of oscillations around 0 °Casa gard to colour, thermal properties, albedo, moisture transmission, critical value for frost action in rock is also questioned and thus and light transmissivity, such that the nature of weathering is the palaeoenvironmental interpretation for the deposits, namely fairly constant in character. However, with paints and/or ground, a cold and relatively moist, frost active environment, is problem- so the properties change significantly within the area of applica- atic (Hall, 1992, 2007). Clastic form is thus considered as a poor tion. Deflorian et al. (2007, p. 244) note that, in order to evaluate indicator of palaeo-weathering process; a problem typically associ- the art weathering there is a need to measure ‘‘...a few different ated with interpreting landforms or weathering features from pro- environmental parameters affecting the organic coating proper- cesses (Turkington and Paradise, 2005). Within the basalts, ties: total amount of energy coming from UV radiation, time with however, more recently proposed palaeoenvironmental interpreta- relative humidity higher than a defined threshold (causing surface tions are drawn mainly from the depositional (or slope) environ- water condensation), environmental temperature higher than the ment rather than the block attributes alone. polymer Tg, temperature range, etc.” The key is the paints them- Several studies have focused on the nature of relict block accu- selves may vary in respect to their responses to these parameters, mulations at altitudes above 3000 m in Lesotho. Boelhouwers and in regard to the responses of the subsurface and surrounding (1994, 1999a), Boelhouwers et al. (2002), Boelhouwers and Sumner rock (e.g. Hradil et al., 2003). Of possible importance to longevity (2003) and Sumner (2004) report on basalt openwork block accu-

8 P.D. Sumner et al. / Journal of African Earth Sciences xxx (2009) xxx–xxx mulations and extensive blocky deposits that often converge into Avery, G., 1974. The preservation of rock-art with special reference to South blockfields and blockstreams similar to those found elsewhere, African problems and conditions. South African Archaeological Bulletin 30, 139–142. such as in Tasmania (Caine, 1983) and Norway (Rea et al., 1996). Bakkevig, S., 2004. Rock art preservation: improved and ecology-based methods can Block accumulation in Lesotho is superimposed over, and thus old- give weathered sites prolonged life. Norwegian Archaeological Review 37, 65– er than, slope colluvium and such deposits predominate on south- 81. Batchelor, A., 1990. Preservation of South African rock art. Unpublished facing slopes (Boelhouwers et al., 2002; Sumner, 2004). The origin Summarized Report. Human Sciences Research Council, Pretoria. of these block deposits is attributed to former colder periods, nota- Bednarik, R.G., 1994. A taphonomy of palaeoart. Antiquity 68, 68–74. bly within the Late Pleistocene, when mechanical weathering is Bell, F.G., Haskins, D.R., 1997. A geotechnical overview of Katse Dam and Transfer Tunnel, Lesotho, with a note on basalt durability. Engineering Geology 46, 175– considered to have predominated. Such an argument concurs with 198. findings for openwork accumulations within the dolerites of the Benedetti, D., Bontempi, E., Pedrazzani, R., Zacco, A., Depero, L.E., 2008. Eastern Cape (Sumner and de Villiers, 2002) and the quartzites of Transformation of calcium carbonate stones: some examples. Phase Transitions 81, 155–178. the Western Cape (Boelhouwers, 1999b) mountains. Significantly, Beukes, N., 1970. Stratigraphy and sedimentology of the cave sandstone stage, the presence and persistence of such openwork accumulations also Karoo system. In: Proceedings of the Second International Gondwana militate against arguments for any glaciation of the Highlands (see Symposium, South Africa, pp. 321–328. Boelhouwers and Meiklejohn, 2002, for a review on the glacial vs Boelhouwers, J.C., 1994. Periglacial landforms at Giant’s Castle, Natal Drakensberg, South Africa. Permafrost and Periglacial Processes 5, 129–136. non-glacial debate), supporting rather a former periglacial-type Boelhouwers, J.C., 1999a. A discussion on block weathering in southern Africa and environment with little permanent snow cover. In contrast to a its significance to periglacial autochthonous blockfield development. Polar cold-climate origin, similar block accumulations can also be asso- Geography 23, 12–22. Boelhouwers, J.C., 1999b. Relict periglacial slope deposits in the Hex River ciated with warmer environments (see Boelhouwers and Sumner, Mountains, South Africa: observations and palaeoenvironmental implications. 2003; André et al., 2008). Several problems thus remain in the Geomorphology 30, 245–258. interpretation of such landforms as entirely cold-climate limited Boelhouwers, J.C., Meiklejohn, K.I., 2002. Quaternary periglacial and glacial geomorphology of southern Africa: a review and synthesis. South African (Boelhouwers, 1999a; Sumner and Meiklejohn, 2004) and the high- Journal of Science 98, 47–55. land areas of Lesotho, where weathering horizons may date back to Boelhouwers, J.C., Sumner, P.D., 2003. The palaeoenvironmental significance of the Tertiary (Boelhouwers, 1999a), still provides a largely unex- southern African blockfields and blockstreams. In: Phillips, M. (Ed.), Permafrost – ICOP 2003. Swets and Zeitlinger, Lisse, pp. 73–78. plored avenue for further research. Boelhouwers, J.C., Holness, S., Meiklejohn, K.I., Sumner, P.D., 2002. Observations on a blockstream in the vicinity of Sani Pass, Lesotho Highlands. Permafrost and 5. Conclusions Periglacial Processes 13, 251–257. Bradley, R., Jones, A., Myhre, L.N., Sackett, H., 2002. Sailing through stone: carved ships and the rock face at Revheim, southwest Norway. Norwegian A small but significant body of literature has provided insights Archaeological Review 35, 109–118. into the weathering processes associated with the basalts and Brink, A.B.A., 1983. Engineering Geology of Southern Africa. The Karoo Sequence, vol. 3. Building Publications, Silverton, Pretoria. sandstones of the eastern escarpment region of southern Africa. Bullet, T.R., Prosser, J.L., 1983. Paint: a surface modifier. Physics Technology 14, While the case studies into rock art and for the Lesotho Highlands 491–500. Water Project provide detail on current state of weathering, and Caine, N., 1983. The Mountains of Northeastern Tasmania. Balkema, Rotterdam. Casellato, U., Vigato, P.A., Russo, U., Matteini, M., 2000. A Mössbauer approach to the potential processes, it is evident that a substantial amount of re- physico-chemical characterization of iron-containing pigments for historical search remains to be undertaken. Although some field data are wall paintings. Journal of Cultural Heritage 1, 217–232. available, the environmental conditions controlling weathering Castro, D.J., Bell, F.G., 1995. An engineering geological appraisal of the sandstones of the Clarens Formation, Lesotho, in relation to tunneling. Geotechnical and processes and rates remain largely unknown given the diversity Geological Engineering 13, 117–142. of the landscape including aspect and altitudinal conditions. Inves- Chalmin, E., Menu, M., Vignaud, C., 2003. Analysis of rock art painting and tigations using high resolution instrumentation and novel ap- technology of Palaeolithic painters. Measurement Science and Technology 14, proaches, such as with moisture and light penetration (see Hall 1590–1597. Cox, K.G., Hornung, G., 1966. The petrology of the Karroo basalts of Basutoland. et al., 2007a) are ongoing and illustrate the intricacies required American Mineralogist 51, 1414–1432. for field studies on weathering. Direct measurement of weathering Davidson, W.H., 1972. The influence of constitution of the engineering properties of rate is notably absent, however, although some inferences could be crushed volcanic breccias. Proceedings of the 6th Australian Road Research Board Conference 6 (5), 7–90. made, and how this impacts on landscape development is a key Deer, W.A., Howie, R.A., Zussman, J., 1963. Rock forming minerals. Chain silicates, question for the future. Any form of biological weathering has also vol. 2. Framework silicates, vol. 4. Longmans, Green and Co. Ltd., London. pp. not been reported on in detail. Lastly, interpreting palaeo-weather- 127–129. Deflorian, F., Rossi, S., Fedrizzi, L., Zanella, C., 2007. Comparison of organic coating ing processes and climatic reconstructions from weathering prod- accelerated tests and natural weathering considering meteorological data. ucts remains problematic, but the highlands area may provide new Progress in Organic Coatings 59, 244–250. insights into cross-climate linkages given long geological history Duncan, A.R., Marsh, J.S., 2006. The Karoo igneous province. In: Johnson, M.R., Anhaeusser, C.R., Thomas, R.J. (Eds.), The Geology of South Africa. Geological associated with weathering of the basalts. Society of South Africa. Johannesburg/Council for Geoscience, Pretoria, pp. 501– 520. Dunlevey, J.N., Ramluckan, V.R., Mithcell, A.A., 1993. Secondary mineral zonation in Acknowledgements the Drakensberg Basalt Formation, South Africa. South African Journal of Geology 96, 215–220. Magda Geringer assisted with the drawing of Fig. 1. Two anon- Egli, M., Mirabella, A., Sartori, G., Fitze, P., 2003. Weathering rates as a function of climate: results from a climosequence of the Val Genova (Trentino, Italian Alps). ymous referees are thanked for provided helpful comments on an Geoderma 111, 99–121. earlier draft of the paper. Eriksson, P.G., 1981. A palaeoenvironmental analysis of the Clarens Formation in the Natal Drakensberg. Transactions of the Geological Society of South Africa 84, 7– 17. References Eriksson, P.G., 1983. A palaeoenvironmental study of the Molteno, Elliot and Clarens Formations in the Natal Drakensberg and northeastern Orange Free State. Alley, R.B., Mayewski, P.A., Sowers, T., Stuiver, M., Taylor, K.C., Clark, P.U., 1997. Unpublished Ph.D. Thesis, University of Natal, Pietermaritzburg. Holocene climatic instability: a prominent widespread event 8200 yr ago. Etienne, S., 2002. The role of biological weathering in periglacial areas: a study of Geology 25, 483–486. weathering rinds in Iceland. Geomorphology 47, 75–86. André, M.F., Hall, K.J., Bertran, P., Arocena, J., 2008. Stone runs in the Falkland Grab, S.W., 2007a. Near-surface rockwall temperatures in the high Drakensberg islands: periglacial or tropical? Geomorphology 95, 524–543. basalt: spatio-temporal differences and possible implications for weathering. Arocena, J.M., Hall, K.J., Meiklejohn, K.I., 2008. Minerals provide tints and possible Zeitschrift für Geomorphologie 51 (Suppl. 1), 103–113. binder/extender in pigments in San rock paintings (South Africa). Grab, S.W., 2007b. Rock surface temperatures of basalt in the Drakensberg alpine Geoarchaeology 23, 293–304. environment, Lesotho. Geografiska Annaler 89A, 185–193.

P.D. Sumner et al. / Journal of African Earth Sciences xxx (2009) xxx–xxx 9

Gonzalez, I., Laiz, L., Hermosin, B., Caballero, B., Incerti, C., Saiz-Jimenez, C., 1999. Nel, W., Sumner, P.D., 2008. Rainfall and temperature attributes on the Lesotho- Bacteria isolated from rock art paintings: the case of Atlanterra shelter (south Drakensberg escarpment edge, southern Africa. Geografiska Annaler 90A, 97– Spain). Journal of Microbiological Methods 36, 123–127. 108. Hall, K.J., 1991. The allocation of the freeze–thaw weathering mechanism in Odlyha, M., Boon, J.J., van den Brink, O., Baci, M., 1997. Environmental research for geocryological studies: a critical comment. South African Geographical Journal art conservation (ERA). Journal of Thermal Analysis 49, 1571–1584. 73, 10–13. Olivier Shand and Lahmeyer MacDonald Consortium (OSC), 1986. Lesotho Hall, K.J., 1992. A discussion on the need for greater rigour in southern African Highlands Water Project. Feasibility Study, April 1986. Unpublished Report. cryogenic studies. South African Geographical Journal 74, 69–71. Perera, D.Y., 1996. On adhesion and stress in organic coatings. Progress in Organic Hall, K.J., 2007. Evidence for freeze–thaw events and their implications for rock Coatings 28, 21–23. weathering in northern Canada: II. The temperature at which water freezes in Pope, G.A., Meierding, T.C., Paradise, T.R., 2002. Geomorphology’s role in the study rock. Earth Surface Processes and Landforms 32, 249–259. of weathering of cultural stone. Geomorphology 47, 211–225. Hall, K.J., Hall, A., 1996. Weathering by wetting and drying: some experimental Roper, H., 1959. Study of shrinking aggregates in concrete. Special Technical Report results. Earth Surface Processes and Landforms 21, 365–376. No. 502. CSIR, December 1959. Hall, K.J., Meiklejohn, K.I., Arocena, J., Prinsloo, L., Sumner, P.D., Hall, L., 2007a. Rea, B.R., Whalley, W.B., Rainey, M.M., Gordon, J.E., 1996. Blockfields, old or new? Deterioration of San rock art: new findings, new challenges. South African Evidence and implications from some plateaus in northern Norway. Journal of Science 103, 361–362. Geomorphology 15, 109–121. Hall, K.J., Meiklejohn, I., Arocena, J., 2007b. The thermal responses of rock art Rudner, I., 1989. The Conservation of Rock Art in South Africa. National Monuments pigments: implications for rock art weathering in southern Africa. Council, Cape Town. Geomorphology 91, 132–145. Sumner, P.D., 2004. Geomorphic and climatic implications of relict openwork block Hao, Z., Iqbal, A., 1997. Some aspects of organic pigments. Chemical Society Reviews accumulations near Thabana-Ntlenyana, Lesotho. Geografiska Annaler 86A, 26, 203–213. 181–194. Hay, R.L., 1978. Geological occurrence of zeolites. In: Sand, L.B., Mumpton, F.A. (Eds.), Sumner, P.D., de Villiers, S., 2002. On the palaeoenvironmental implications of the Natural Zeolites: Occurrence, Properties, Use. Pergamon Press, pp. 135–143. Amatola screes. South African Journal of Science 98, 598–601. Hoerlé, S., 2005. A preliminary study of the weathering activity at the rock art site of Sumner, P.D., Meiklejohn, K.I., 2004. On the development of autochthonous Game Pass Shelter (KwaZulu–Natal, South Africa) in relation to its conservation. blockfields in the grey basalts of sub-Antarctic Marion Island. Polar South African Journal of Geology 108, 297–308. Geography 28, 120–132. Hoerlé, S., 2006. Rock temperatures as an indicator of weathering processes Sumner, P.D., Loubser, M.J., 2008. Experimental sandstone weathering using affecting rock art. Earth Surface Processes and Landforms 31, 383–389. different wetting and drying moisture amplitudes. Earth Surface Processes Hoerlé, S., Salomon, A., 2004. Microclimatic data and rock art conservation at Game and Landforms 33, 385–390. Pass Shelter in the Kamberg Nature Reserve, KwaZulu–Natal. South African Sumner, P.D., Nel, W., 2006. Surface-climate attributes at Injisuthi Outpost, Journal of Science 100, 340–341. Drakensberg, and possible ramifications for weathering. Earth Surface Hradil, D., Grygar, T., Hradilova, J., Bezdicˇka, P., 2003. Clay and iron oxide pigments Processes and Landforms 31, 1445–1451. in the history of painting. Applied Clay Science 22, 223–236. Turkington, A.V., Paradise, T.R., 2005. Sandstone weathering: a century of research Irwin, P., Akhurst, J., Irwin, D., 1980. A Field guide to the Natal Drakensberg. Natal and innovation. Geomorphology 67, 229–253. Branch of the Wildlife Society of Southern Africa. Van Rijssen, W.J.J., 1987. Paintings in peril. South African Archaeological Bulletin 42, King, L.C., 1945. South African Scenery. Oliver & Boyd, Edinburgh. 5–9. Lesotho Highlands Consultants (LHC), 1994. LHWP Mineralogy report for review by Van Rooy, J.L., 1992. Some rock durability aspects of Drakensberg basalts for civil the project manager LHC. Report 164/448, May 1994. engineering construction. Unpublished PhD thesis. University of Pretoria. Loubser, J.H.N., 1991. The conservation of rock paintings in Australia and its Van Rooy, J.L., van Schalkwyk, A., 1993. The geology of the Lesotho Highlands Water applicability to South Africa. Navorsing van die Nasionale Museum Project with special reference to the durability of construction materials. Bloemfontein 7, 113–143. Journal of African Earth Sciences 16, 181–192. Mazel, A.D., Watchman, A., 2003. Dating Rock Paintings in the uKhahlamba– Viles, H.A., 2001. Scale issues in weathering studies. Geomorphology 41, 63–72. Drakensberg and the Biggarsberg, KwaZulu–Natal, South Africa. Southern Africa Walker, F., Poldervaart, A., 1949. Karroo dolerites of the Union of South Africa. Humanities 15, 59–73. Bulletin of the Geological Society of America 60, 591–706. Meiklejohn, K.I., 1994. Aspects of the weathering of the Clarens Formation in the Waites, B., 2000. The Lesotho Highlands Water Project. Geography 85, 369–374. KwaZulu/Natal Drakensberg: Implications for the preservation of indigenous Weinert, H.H., 1980. The Natural Road Construction Materials of Southern Africa. rock art. Unpublished Ph.D., Geography Depart, University of Natal Academica, Pretoria. (Pietermaritzburg). Zoppi, A., Signorini, G.F., Lucarelli, F., Bachechi, L., 2002. Characterization of painting Meiklejohn, K.I., 1997. The role of moisture in the weathering of the Clarens materials from Eritrea rock art sites with non-destructive spectroscopic Formation in the KwaZulu/Natal Drakensberg: Implications for the techniques. Journal of Cultural Heritage 3, 299–308. deterioration and preservation of indigenous rock art. South African Geographical Journal 79, 199–206.