Petroglyps with Thamudic inscriptions at Yatib, Saudi Arabia, about 3000 years BP
Petroglyps with Thamudic inscriptions at Yatib, Saudi Arabia, about 2500 years BP
Style and technique
Exacavation and proximity
Patination and weathering
Radiocarbon analysis of mineral accretions
Radiocarbon analysis of inclusions in accretions
Microerosion analysis
Luminescence dating
Dating of pictograms
Other methods
Pitfalls and unsuitable methods
The above articles provide a comprehensive introduction for the newcomer. The more experienced researcher may wish to proceed to the main body of this webpage, its specialist library:

Articles on rock art dating

For a cutting-edge current rock art dating project, see:

The EIP Project: dating the oldest known rock art in the world

It has long been apparent to philosophers of science that confusion concerning scientific matters is usually attributable to shortcomings of language. The term ‘to date’, for instance, has a variety of meanings, and confusion about the dating of rock art (or the dating of archaeological remains, for that matter) initially stems from this ambiguity. As a noun, ‘date’ may refer to the date shown on a coin, book or building, presumably representing the time of minting, printing or completion, i.e. an ‘absolute date’. But it may alternatively refer to a time period of some considerable duration (e.g. hundreds of millennia) from which an archaeological, palaeontological or geological phenomenon supposedly ‘dates’. The corruption imposed on the first meaning becomes apparent when the term is used in the second meaning but the precision implicit in the first meaning is often attributed to such usage.

Significant problems also arise when the scientific (i.e. refutable) dating information provided by archaeometrists is interpreted as if it referred to real ages. In nearly all cases, such data are subject to significant qualifications, which in archaeological use are not adequately taken into account. Even attempts to compensate for the routine misuses of dating results have been misguided. For instance, the introduction of once fashionable terms such as ‘absolute dates’ and ‘relative dates’, or the use of ‘calibrated’ radiocarbon dates have only provided cures worse than the disease; rather than correcting the problem they tried to conceal it. Apart from some rare exceptions, archaeometry does not provide results that could reasonably be defined as ‘absolute dates’, and radiocarbon assay results are no basis for calculating ‘absolute calibrated dates’. Thus the practice of distinguishing between B.P. and b.p. carbon isotope ‘dates’ is misleading, in that it elicits a false sense of security in practitioners. Reference to a calibration curve proposed for bristlecone pine in some part of California does not compensate for the numerous inherent qualifications of radiocarbon results, it merely compounds interpretational confusion.

The difficulties of estimating the age of rock art are considerably greater than those of ‘dating’ archaeological finds in most cases, and the methodologies developed for these two entities differ considerably. Nevertheless, it is most instructive to consider the difficulties of archaeological dating before focusing on the dating of rock art.

Very few methods are known of absolute dating, such as dendrochronology (analysing the annual growth rings of sectioned tree trunks) and counting of annual ice layers (particularly in Greenland and Antarctica). Another possibility might be the annual luminescence banding in carbonate speleothems (Baker et al. 1993). All other methods used in dating yield results that cannot be expressed in calendar years, sidereal time or absolute historical time, irrespective of any ‘calibration’ used.

The radiometric method most used in estimating archaeological time is the analysis of carbon nuclides. Especially in the Anglo-American version of archaeological practice, this method now forms the chronological backbone of the discipline. All radiometric techniques (as well as some non-radiometric dating methods, such as fission-track analysis) provide sets of statistical information thought to relate to the age of samples; they do not yield sidereal or calendar ages. The probability statements they offer can be expressed at one, two or more standard deviations, which means that the true age of the sample would be within stated tolerance values if all secondary qualifications were ignored.

In the case of radiocarbon, the remnant content of an unstable and thus radioactive isotope of carbon is determined to estimate the time when its decay to nitrogen commenced. For this to be accurate it is necessary to know the initial concentration of carbon-12, 13 and 14; the decay rate; and that the process had not been influenced by other factors. There are problems with all these conditions:

1. The initial atmospheric concentrations of 14C and d13C are not known to us. We can only make assumptions about past atmospheric carbon nuclide regimes, which introduces a significant unknown variable. Dendrochronology has documented major variations in past carbon regimes.
2. The decay rate is expressed in the half-life of radioactive substances. This is now thought to be about 5733 years for radiocarbon, but we continue to use Libby’s original estimate of 5568 years, for consistency.
3. The assumption that the ratio of carbon isotopes in the carbon reservoir of the Earth are the result of a complete and rapid equalisation is not necessarily valid.
4. The assumption that the ratios of carbon isotopes are only altered by radioactive decay is only approximately valid.
5. The levels of radiocarbon cannot be measured with real accuracy.

These issues are too complex to be examined in any detail here. They are well understood by radiocarbon scientists, but are not often reflected in the pronouncements of archaeologists about such results, as the excessive confidence in them indicates. Of greater concern still are the ways in which these results are then misused in the course of archaeological model building. For instance, it is commonly assumed that charcoal excavated in sediment strata provides secure dating for such deposits. All components of a geological sediment possess an ‘age’, most are significantly ‘older’ than the deposit and may have been redeposited many times by a variety of processes, while some are younger. Instead of referring to the ‘age’ of such a sediment or deposit it is more appropriate to refer to the time of its most recent deposition. Charcoal itself has no radiocarbon age, such an age denotes the time when the tree in question assimilated carbon dioxide from the atmosphere, not the time it stopped doing so, or the time the wood was incompletely burned (Schiffer 1986; Fetterman 1996). Moreover, due to its extremely low specific gravity relative to other sediment particles, charcoal tends to ‘float’ on top of a deposit through trampling or other disturbances, and it is easily transported by wind, water and biological agents (e.g. insects, burrowing animals). Its taphonomy ensures that few charcoal pieces are recovered from the locations of actual hearths, so there is no proof that they provide precise dates for the final deposition of the sediment they were recovered from (Bednarik 1989a). Even when the charcoal is extracted from a well-defined hearth, there is the possibility that charcoal or wood of various ages has been used. Moreover, the presence of charcoal in the vicinity of stone tools offers no proof that it is of anthropic origin, especially in a region that is prone to natural fires. These very brief comments should merely convey an inkling of the true complexity of dating issues, more details relating to carbon isotope analysis are given below.

Fundamentally, methods of estimating the age of, or ‘dating’, rock art, are divided into indirect methods and direct methods, as defined on these introductory pages.
For a detailed reading list providing up-to-date and in-depth scientific information please go to Articles, and select specific articles from the list.