While moisture does cause a variety of physical rock deterioration processes, such as flaking or exfoliation, other such processes involve no chemical changes wrought by moisture. They have been somewhat neglected in the rock art conservation literature, but they can have devastating effects on rock art and are often much easier to alleviate. While Rosenfeld (1985) lists here hydration, ‘salt decay’ and the removal of cementing medium, I limit the term physical rock weathering to purely physical processes.Insolation or solar radiation is capable of breaking up some rock types through thermal expansion and contraction, particularly in arid or semi-arid regions experiencing significant diurnal temperature variations. Once controversial, the effectiveness of this process has been shown through the buckling of marble slabs. The ubiquitous sub-spherical boulders of some deserts are largely, though not entirely, the result of temperature stresses over great time spans. The rounding of boulders, most especially of granite, is the result of the geometry of temperature transference within the rock, which heats or cools protruding aspects or corners more readily. Over time the repetitiveness of the process along a stress zone leads to eventual material fatigue. Ideally the process would lead to a spherical shape (or more precisely, a body with the smallest possible surface area relative to mass), and the spalls are always of distinctly convex fracture surfaces. The process is assisted by fire damage which has the same effect but yields quite different spalls. In my estimate, based on the repatination of successive fracture facets and limited dating information, a typical gabbro or dolerite boulder in an arid and hot region might experience one insolation spalling event every one to four millennia. A lesser rate applies to granite, even boulders engraved in the Pleistocene may not bear any subsequent facets. It is sometimes suggested that moisture within the rock increases susceptibility to insolation fracture. This is quite possible in some rock types, such as sandstones, but is probably less effective in granites, for instance. Thermal stress fatigue is believed to be responsible for a variety of fracturing patterns other then the described spherical reduction, which have also affected rock art sites, particularly in combination with other factors (Hall and Hall 1991; Meiklejohn 1995).Lightning strikes result in often massive fracturing of boulders of any size, and their traces are easy to recognise. This is a major cause of rock reduction, especially in flat areas where rocky outcrops tend to be the only elevated features. Boulders near the peak or along upper ridges are favoured targets of lightning, which usually strikes the uppermost surface of boulders. A boulder may be broken in half, or a large chunk flaked off. At the point of impact the rock may be shattered, its surface is frequently glazed where surface minerals became fused and new ones formed instantaneously. If there are iron-rich rinds present, the minerals formed under very high temperatures of the electrical discharges tend to be of a light greenish to bluish colour. The fracture surfaces created by lightening resemble closely the fractures found on stone implements: point of percussion, positive and negative bulbs of percussion, and radial stress lines found on small stone flakes have their here much larger counterparts, on fractures sometimes measuring many square metres. Quite likely users of stone tools would have noticed such features in the landscape, and marvelled at the impact required to fracture such huge boulders.

Brushfires are a highly active agent of physical weathering, especially on rock types that are less susceptible to other forms of weathering. The heat developed by the burning of often very resinous vegetation in arid regions can be quite in-tense (temperatures exceeding 800ºC have been reported), resulting in rapid heating of a thin surface zone, and fatigue between this and the core mass of the boulder which is unaffected by this sudden temperature change. Due to the geometry of heat transference in a solid body, which obeys the same fundamental laws as wane formation (see microerosion analysis), protruding aspects can heat up rapidly and the flakes detached are always convex on the inside. In general these flakes are under 20 mm maximum thickness, although thicker specimens have been observed.

On particularly compact, unweathered granite these heat spalls are as thin as one millimetre, and they tend to have sharp edges all around. Typically these spalls are of 15-30 cm diameter on weathered granite. A single fire can remove as much as a square metre of surface from a medium-sized boulder, and if there is any rock art present on this surface, the effect on it is devastating. Selkirk and Adamson (1981) report losses of 2-6 kg of rock per square metre of sandstone in a single intense forest fire. Another effect of fire is a reddening of many types of rock, where the reducing parts of flames has converted any goethite present, be it in the rock fabric or on the surface, to haematite.
Kernsprung is a phenomenon of boulder fracture whose results resemble those of lightning strikes, but there is no impact apparent, nor any of the other features described above. The cause is not clearly established, but reports of extremely loud, explosion-like sounds in arid areas suggest that cumulative temperature-induced stresses could build up to the point of a spontaneous release, causing an apparently structurally sound boulder to split literally in half.

Tectonic changes in rock facies or rock structures can also affect rock art sites in a number of ways. These are structural responses, usually due to gravitational disequilibrium of some form. Not all such changes are attributable to natural causes, for instance subsidence due to underground extraction of coal is obviously an anthropogenic threat. Natural tectonic causes that have destroyed rock art are seismic activity, relocation of major sediment deposits which render rock structures unstable, and the effects of aquifer level fluctuations. The first two are fairly straightforward processes that destabilise rock masses bearing rock art. The third is less readily appreciated. Limestone cave systems often transect the upper limit of aquifers, and these water bodies can experience major oscillations. As the sea level has fluctuated by about 150 m during the Pleistocene, so did coastal aquifers. In the case of partially water-filled cave systems, this means that when the water table falls, rock masses that were tectonically stable under water become unstable once they emerge, because their gravity increases by one ton per cubic metre. Gravitational equilibrium needs to be re-established, which means that formerly stable structures collapse, including false floors, load-bearing members and lintels, and sometimes cave roofs. Where cave art occurs its survival can be affected by such tectonic adjustments.

Petroglyphs often occur near rivers, even on boulders presently located in rivers (in the Amazon region, for instance), and other decorated boulders occur in shallow sea water (e.g. in the general region of Vancouver, Canada). Such boulders are exposed to yet another form of physical damage, due to the kinetic energy of water. They may be moved by the water, or sand moved by the water may act as an abrasive on them. The impact of such fluvial damage is illustrated with the example of the Siega Verde site in Spain, whose numerous petroglyphs archaeologists have assigned to the Upper Palaeolithic. All these petroglyphs occur on soft schist rock and boulders, all are submerged annually under high-energy flood waters. The sharp-edged coarse quartz sand then being rafted past gauges the schist so much that I have measured a retreat of up to 30 mm per century. Needless to say, in such an environment the petroglyphs will not survive more than a few centuries at most, and those of Siega Verde are certainly doomed to destruction. Sand-blasting is not, however, limited to water-propelled quartz sand, it also effects aeolian erosion. While this is barely effective on hard and structurally sound rock, aeolian erosion is highly effective on softer rock types, most especially on calcarenite, other soft sandstones, even limestone, especially in coastal and arid environments of high wind velocity. Sandblasting can destroy pictograms and even deep petroglyphs in a matter of a few decades. These weathering processes are often facilitated by weakening of the rock fabric through moisture or salt action.


REFERENCES Bibliography of Rock Art Conservation