Zoomorpoh, apparently macropod, and other pictograms, Australia
Besides carbonates and oxalates there are many other forms of mineral accretionary deposits on rock surfaces that may conceal or underlie rock art, and while their essential components contain no radiocarbon, a variety of inclusions may be present in them. Among these residues are windborne particles, such as pollen and spores, micro-organisms and so forth. Others are derived from organisms occurring on the rock surface, such as algae or fungi. The most widely researched type of such mineral deposits are rock varnishes, dark-brown to almost black ferro-manganous coatings found in many environments, but best preserved in arid regions due to their high-pH and low precipitation regimes.
Although exceedingly thin (mostly <2 to 500 microns), rock varnishes often comprise complex stratigraphies which contain small quantities of organic matter, derived either from aeolian deposition or from locally derived micro-organic remains, microbial metabolic products and plant-derived detritus. The involvement of micro-organisms in the formation of some forms of rock varnish was conclusively demonstrated by Scheffer et al. (1963), although earlier observations had pointed in that direction (e.g. Francis 1920). Analyses in the late 19th century and the first half of the 20th century had frequently implicated non-organic sources (e.g. Loew 1876; Merrill 1898; Walther 1924), while Blake (1905) recognised that the accretions must be at least partially of exogeneous origin. The comprehensive analytical work of Engel and Sharp (1958) ushered in modern studies of rock varnishes (Allen 1978; Krumbein 1969; Krumbein and Jens 1981; Perry and Adams 1978; Potter and Rossman 1977, 1979).
The cation recycling of such deposits by microbes accounts probably for at least some of the stratigraphical complexities of the varnishes, and it may also effect the incorporation of introduced matter, including carbonaceous matter. White (1924) suspected pollen to be a varnish-forming factor, mistakenly believing them to be rich in iron and manganese. They do, nevertheless, occur in rock varnish, as do other airborne particles. Its main components are essentially clay minerals, commonly accounting for two-thirds of the deposit's bulk. It needs to be emphasised that the terms rock varnish and desert varnish probably refer to the stable products of a number of quite heterogeneous processes and sources, which merely lead to similar end effects. The term has often been misused, especially by archaeologists, to describe a variety of dunkle Rinden and other accretionary deposits that do not resemble rock varnish. Even weathering rinds with some iron patination have been so misidentified at times (e.g. Pineda et al. 1988).
Since the early 1980s, Ronald Dorn sought to secure radiocarbon dates from supposedly organic substances extracted from rock varnishes, taking advantage of the AMS method which had been introduced in the 1970s. Initially he did not use it to date any rock art directly, but to obtain numerical-age controls at nearby calibration sites for his cation-ratio method (i.e. to calibrate CR ratios; see below). When the CR method (Dorn 1983, 1986, 1990, 1992; Dorn and Whitley 1984; Dorn et al. 1992; Nobbs and Dorn 1988) became widely rejected, with the advent of the 1990s, Dorn resorted to applying AMS analysis directly to samples taken from petroglyphs. By that time Watchman, who had played a pivotal role in refuting the CR method (Watchman 1992), was developing the FLECS method for the same purpose (Watchman 1993), and both researchers then sought to improve their respective approaches to the analysis of carbonaceous substances. Whereas Dorn continued to work almost exclusively with rock varnishes, Watchman diversified his technique and, having earlier experimented with oxalates, included analyses of other accretionary deposits, notably silica skins which he had previously worked with for conserva-tion purposes.
Watchman's more rigorous approach is illustrated by a comparison of the two researchers' reports of their 1995 work at some of the Coa sites in Portugal, which they conducted under identical controlled conditions of a 'blind test' (Watchman 1995, 1996; Dorn 1997). Watchman sampled deposits postdating petroglyphs, but also collected two samples from older accretions that had been dissected by rock art. More importantly, he took control samples from a surface of known age, a railway cutting, thus realising that all the samples were contaminated. After locating the source of this distortion in graphite inclusions and correcting for it, he was able to provide minimum and maximum age estimates for a few petroglyphs. Dorn published his uncorrected results but at the same time rejected them, saying that he had lost confidence in the entire method of analysing carbonaceous inclusions in both varnishes and silica skins (Dorn 1996a, 1996b, 1997). In his 'change of perception' as he called it he admitted that he had noticed many 'anomalies' over the years, but this had apparently not prevented him from confidently publishing and defending his results. He admitted in 1996 that for over fifteen years he had made two critical mistakes which had 'blinded' him: he had falsely assumed homogeneity in his bulk samples even though it was clear that the organic matter was heterogeneous and of different ages; and he had assumed that the carbon was sealed into a closed system, until discovering that it was in fact an open system. 'To expect that no organic weathering occurred before a petroglyph panel was exposed by natural erosion was naive', he confessed (Dorn 1996a).
His mistakes were unnecessary, however, because in 1987 I had sent him my paper in which I had demonstrated that the carbon system of ferro-manganous accretions is open, and that there are substantial differences in the composition, pH and organic matter concentrations within the nano-stratigraphy of such deposits. If he had studied my paper (Bednarik 1979) he would have realised his mistake almost a decade earlier, and he would have understood that his approach had always been futile, indeed, ironically, that I had solved the problems he was wrestling with before he even learnt of their existence. This episode illustrates the need to conduct a thorough survey of the relevant literature before embarking on any scientific work. Hundreds of archaeologists and geomorphologists have completely relied on Dorn's methods and results, which he has now recanted and disowned (for further problems with Dorn's work, see Dalton 1998; Dayton 1997; Beck et al. 1998; Malakoff 1998).
In contrast to the controversial status of the dating of rock varnish, the analysis of carbon-bearing substances found in silica crusts has yielded considerably more convincing results, primarily because the silica seems to provide a comparatively closed carbon system. In addition, silica skins do not suffer from recycling of cations by micro-organisms, as is assumed to be widely the case with rock varnish. Watchman (1996) has produced excellent results from layered silica skins, by separating a layer preda-ting a petroglyph from a younger layer postdating the art, and securing organic carbon from both layers. Even though this procedure does not secure an actual dating for the rock art, it does provide tentative minimum and maximum ages for it.
Contrary to the tenor of Dorn's mea culpa, careful analysis of accretions remains a valid method of direct rock art dating (Watchman 2000; Campbell 2000). It does not provide numerical ages of rock art, but it does offer falsifiable data concerning the age of an entity that is physically related to the rock art in question. While there is a taphonomic problem in that the 'carbon age' of carbonaceous inclusions is probably not the same as the time of their deposition, the real qualifications that apply to such data are a great deal more complex than even Dorn has indicated lately. They are discussed below, under 'pitfalls'.