Radiocarbon assays of submerged peat and shell beds in the northern Gulf of Mexico are plotted here to reconstruct sea level change over time. The vertical axis shows depth below modern sea level; the horizontal axis shows calibrated years before 1950 AD. The red curve is a best-fit line for rising sea level based on radiocarbon assays. Graphic by Ken Brown. See McFaddin Beach to learn more.
The use of radiocarbon dating to understand the chronological sequence of events and behaviors preserved in the archeological record requires a nuanced and critical interpretive approach. This section discusses the importance of vetting radiocarbon data and thinking critically about how a date relates to the archeological event or question of interest, whether new or legacy data.
In 1949/1950 when radiocarbon dating was first published, archeologists were not formally trained in chemistry or physics (and most still are not). To many, radiocarbon dating seemed like a magic black box which yielded objectively scientific results. The widespread belief in the objective power of radiocarbon science persisted for several decades, eventually waning with the rejection of the culture-historical approach to archeology. Advances such as accelerator mass spectrometry (AMS) to radiocarbon dating in the 1980s made clear that some radiocarbon assays were simply better than others. In recent decades, the popularity of Bayesian statistics for analyzing large radiocarbon datasets has refocused attention on the importance of rigorous radiocarbon reporting and data vetting.
Radiocarbon dates show that the Honey Creek midden formed over a 600-year period between about cal AD 1100-1700. This burned rock midden (BRM) is well dated by a sequence of 14 radiocarbon assays, including one predating the BRM and one postdating it. Shown here are probability distributions of each date. The inferred period of BRM accumulation is based on the overall probability patterns. This graphic shows why it takes more than one radiocarbon date to understand when a site was used. From Black et al. 1997, Figure 62. See Honey Creek to learn more.
Archeologists today have access to staggering amounts of radiocarbon data from large crowd-sourced databases like the Canadian Archaeological Radiocarbon Database. By employing data hygiene methods, as the vetting of large datasets is often referred, the archeologist endeavors to cull potentially unreliable and irrelevant assays from a data set, to build more accurate and precise chronologies. There is no one-size-fits all method for vetting radiocarbon data—each research question, archeological region, and statistical method has unique considerations that guide the selection and exclusion of radiocarbon data.
Critical Evaluation
How well was an assay reported? What is the relationship between the sample, its context, and the targeted behavior or event of interest to the archeologist? These questions are at the heart of radiocarbon data evaluation.
To critically evaluate radiocarbon data, the archeologist first collects information about the assay or assays of interest. This can be found in the reporting publication or other documents such as fieldnotes and laboratory records. Essential data includes the conventional radiocarbon age, sample material type, and laboratory number; see Reporting Results. Once the data about an assay is gathered, the critical archeologist considers whether problems with the assay or archeological context exist. Problems to consider include complexities inherent to the sample material like the old wood effect or uncorrected reservoir effects. Contextual problems might include mixed or disturbed archeological deposits. If assaying an artifact, consider whether it may have been reused or curated in antiquity, and whether modern or historical conservation treatments contaminated it. Assessing potential problems is a nuanced task that is contingent on the archeological question being addressed; an assay that may be unfit for one question may be fine for another.
Profile drawings showing the stratigraphic layers documented at the Rush site. The radiocarbon assays shown here include several that seem too old to make chronological sense. There are many reasons that radiocarbon age estimates can be or can seem incorrect, which is why multiple assays are needed to evaluate the age of a site deposit. See Rush site to learn more.
Perspectives vary on how best to select and apply data hygiene criteria to a large radiocarbon dataset. One approach is to score or rank the dates using criteria such as how it was measured (AMS versus conventional methods), how long ago the assay was made, material types, and size of the standard deviation. Another approach rejects scoring methods, instead emphasizing the importance of a holistic consideration of the sample, context, and date. Strict data hygiene criteria, while ideal in theory, can result in the exclusion of most of a radiocarbon data set. Upwards of 80% of assays may be rejected using the most-stringent data hygiene! Less-restrictive data hygiene criteria, which still meet minimum standards, are sometimes needed to maximize the quantity of dates used in analysis. For example, a large data set is required for some statistical methods such as summed probability distributions. In the end, there is no one-size-fits all method to data hygiene—the evaluation should be tailored to the research question and the statistical method chosen by the archeologist.
When interpreting radiocarbon data or analyses, the archeologist should rigorously defend their conclusions by publishing their reasons for including or excluding assays. This information not only lends credibility to the archeologist’s case, but it is also valuable information for future researchers who seek to use and reinterpret the interpreted radiocarbon data.
Replication A common way to increase confidence in dates is to replicate assays—that is, date the event of interest more than once to check the accuracy of dates. Replication can be done by splitting samples and sending them to different labs, using different pretreatment methods on a split sample, or by dating several different material types that are associated with the same event and context. Replication can help identify mixed deposits, reservoir offsets, effects of different pretreatment methods, interlaboratory error, and curation or reuse of materials. In addition, having multiple dates from the same single-event feature may allow a more precise date to be generated using statistical combination tools, such as R_Combine in OxCal. How many assays should be replicated? Common sense holds that the answer will depend on the project, but replication of about 10% of the assays in a typical radiocarbon dating program has been recommended.
This animation shows that multiple radiocarbon assays from the same sample will often yield different but overlapping date ranges. The calibrated radiocarbon dates are symbolized by the blue line above the horizontal axis. The blue dashed lines represent the date range at 1σ (68%). The radiocarbon age in RCYBP is the black normally distributed curve on the vertical axis. The green lines are the calibration curve. A benefit of making multiple assays from one sample is that the radiocarbon ages can be combined using statistical tools such as the R_Combine tool in OxCal, thereby making the age range more precise. Animation by Maarten Blaauw, Department of Archaeology and Palaeoecology, School of Natural and Built Environment, Queen's University Belfast.
The next section, Case Study: Radiocarbon Dating the Lower Pecos Canyonlands, shows the application of critical evaluation criteria to a radiocarbon dataset of almost 500 assays from the Lower Pecos archeological region of Texas. Using Bayesian methods and a summed probability distribution, the timing of earth oven plant baking, plant fiber artifact manufacture, and human population fluctuations are compared to data on environmental conditions through time and the intermittent presence of bison.
University of Texas at Austin
College of Liberal Arts
