Petroglyphs at deeply patinated site in the Pilbara of Australia
The study of rock surface alterations in the service of dating rock art was the forerunner of scientific methods to address this issue. It was through the analytical and microscopic study of time-related changes to rock surfaces that 'direct dating' was developed in the 1970s (Bednarik 1979). Such changes may be reductive (resulting in loss of mass, e.g. erosion), additive (resulting in addition of mass, e.g. accretion) or transformational (resulting in chemical or physical changes of the substrate). The aging process of rock surfaces has been considered relevant to estimating the ages of petroglyphs for at least 180 years: Belzoni (1820) examined the numerous petroglyphs on Egyptian granite and noted the different stages of repatination, compared to the evenly dark-brown accretion on the unworked rock surface. The possibility of calculating the ages of the less patinated petroglyphs occurred to him:
"I beg my readers to pardon my thus speculating on a point which, in my humble capacity, can afford but little instruction; however, as the idea struck me, I lay it before the public (Belzoni 1820: 360-1)."
The form of patination Belzoni considered was almost certainly desert varnish, a ferromanganous deposit that has been intensively studied by geomorphologists for over a century, but its use in speculating about rock art remained sporadic and plagued by confusion. Among the researchers using this approach were Basedow (1914), Rhotert (1938, 1952), Mori (1965), Goodwin (1960) and Anati (1960, 1961, 1963, 1968). But statements about patination colour were often imprecise and sometimes misunderstood. For instance, Anati's key observation was misrepresented by two Australian writers. Anati's careful formulation,
"[I]n this region we know of no engraved surface from [Iron Age to recent] with a patination identical to that of the original rock surface. This seems to mean that in this area it took a minimum of 2500 years to reach ... the natural colour of patina (Anati 1963: 189)"
was rephrased thus:
"... no engravings have re-weathered to match the natural dark rock surface. As some of them are associated with the Iron Age, Anati believes it takes a minimum of 2500 years for a thin, initial surface patination to form in that region (Edwards 1971: 361)."
An almost identical error had earlier occurred in Mori (1965: 63), who corrected himself (Mori 1974: 89-90), substituting 'quasi scura quanto' for 'tanto scura quanto'.
There are other significant difficulties with using patination and weathering states in estimating ages of petroglyphs. To begin with, both weathering and patination processes are highly variable, depending on petrography, climate, topography, surface geometry, chemical environment and other factors. Secondly, there is no simplistic method of quantifying such changes, and attempts to do so (e.g. by measuring reflective properties of accretionary deposits such as rock varnish) have only resulted in most unconvincing results. Moreover, the role of engraved groove depth remains poorly understood, as does the influence of cation-scavenging micro-organisms and other processes of re-cycling accretionary matter. For instance, one question that was raised in a debate concerning the use of patina colour in estimating petroglyph age is the issue, to what degree does groove depth affect repatination rates? Does a shallow groove repatinate faster than a deep groove? If the process is endogenous this would be likely, but not in exogenous patinae such as rock varnish. The use of such phenomena to estimate rock surface ages, including those within a petroglyph, requires an intimate understanding of the processes active in repatination, and most comments one finds in the existing rock art literature fail to provide any informed analytical data about the patinae. They elicit no confidence in the conclusions drawn from such inadequate observations. Indeed, there are numerous instances of quite evidently false or misleading statements about the nature and significance of such features in the literature, in which surface deposits are incorrectly described, or used as evidence to prop up grotesquely implausible chronological assertions. Repatination can be affected by numerous factors besides the underlying lithology, such as water presence, climate, epilithic organisms, coarseness of surface texture, and the proximity of cation sources, such as sediments or nearby accretions.
Weathering rinds are zones of oxidation, hydration or solution forming parallel to clast surfaces and their thickness is a function of time (Carroll 1974; Colman 1981; Colman and Pierce 1981; Crook 1986; Gellatly 1984). The growth rate of weathering rinds can be quantified for a given rock type under given climatic conditions if it can be calibrated by another dating method, but it only yields imprecise results. Cernohouz and Solc (1966) developed a method for determining the ages of macro-wanes on sandstones, claiming an accuracy of +/-10-20%. Although their hypothesis was subsequently refuted (Bednarik 1992), they correctly recognised that weathering rind thickness is a function of surface geometry, and that this aspect is the cause of wane formation (see below).
Surface rinds often suffer from mass loss due to abrasion, erosion, frost action, Salzsprengung or exfoliation, which introduces a major error source. It may be preferable to measure subsurface rinds on submerged rock, as Colman and Pierce (1981) did, examining a large sample of clasts from B-horizons of deposits. They proposed a logarithmic function in the form of
d = log(a + bt) (1)
where a and b are constants, d is the rind thickness and time t can be determined. However, this is of limited use in estimating the age of petroglyphs. For instance, destructive sampling is usually out of the question, but in such cases one could consider the use of non-intrusive methods. In particular, the Schmidt hammer may be suitable for measuring degree of rock surface weathering (Birkeland et al. 1979; Burke and Birkeland 1979; McCarroll 1991). This instrument was originally designed to measure the surface hardness of concrete, but has been widely used on natural rock as well (Day and Goudie 1977).
The Schmidt hammer has almost never been used in rock art research (Campbell 1991; Sjoeberg 1994; for a preliminary but unsubstantiated and inconclusive attempt, see Pope 2000), nor have there been any serious attempts so far to employ weathering rinds in estimating rock art ages in any more than the most cursory fashion. Clearly there is more research required in this area.