Bayesian Chronological Modelling And Wiggle Matching

Origin and history

Bayesian chronological modelling is a statistical framework that originated in the field of radiocarbon dating, with its foundational principles developed in the late 20th century. The specific application of Bayesian analysis to archaeological chronology was pioneered in the United Kingdom during the 1980s and 1990s. Key methodological advances were made by researchers such as Caitlin Buck and colleagues, who integrated prior archaeological knowledge with radiocarbon likelihoods. Wiggle-match dating, a technique often used within Bayesian models, was developed to refine dates from tree-ring sequences and has its roots in dendrochronology from Europe and North America. The integration of these methods into user-friendly software, like OxCal, in the 1990s and early 2000s, facilitated widespread adoption in archaeological science. These techniques represent a major shift from simple calibrated date ranges to complex, model-based interpretations of time.

What it is for

This suite of methods is for constructing precise and robust chronological sequences for archaeological sites and events where standard radiocarbon calibration is insufficient. It is specifically used to resolve dating problems where materials are from short-lived contexts or where stratigraphic relationships are complex. The primary function is to incorporate "prior information," such as the known order of layers in a trench, into the statistical analysis of radiocarbon measurements. Wiggle-matching is for situations where a sequence of radiocarbon dates, such as from consecutive tree-rings, can be aligned against the known wiggles in the atmospheric radiocarbon calibration curve. This allows archaeologists to achieve sub-decadal precision in dating, far beyond the typical range of a single radiocarbon date. Ultimately, it is for testing and refining archaeological hypotheses about the timing and duration of human activity, settlement, and cultural change.

Overview

At its core, Bayesian chronological modelling uses Bayes' theorem to combine measured scientific data with structured archaeological knowledge. The process begins with the collection of radiocarbon dates from carefully selected, short-lived organic materials like seeds, twigs, or bone. The archaeologist then builds a formal statistical model that encodes the site's stratigraphy, such as stating that one context is earlier than another, or that a group of contexts are all contemporaneous. The software, such as OxCal or ChronoModel, runs simulations to find the chronological solutions that best satisfy both the radiocarbon evidence and the prior model. Wiggle-matching is a specific model type where a series of dates from a tree-ring sequence are fitted as a unit against the calibration curve's known fluctuations. The output is a revised, often much narrower, probability distribution for each dated event, presented as a modelled date range with a clear statistical confidence level.

What to know

Successful application is entirely dependent on the quality of the archaeological "prior" information; a flawed stratigraphic model will produce precise but inaccurate results. The materials dated must be critically assessed for their taphonomy and association with the human activity being studied; using long-lived or residual material invalidates the model. It is not a magic tool that creates precision from poor data, but a method for extracting maximum information from high-quality data within a clear archaeological framework. The technique requires close collaboration between the field archaeologist, who understands the site sequence, and the specialist who builds the statistical model. Outputs must be interpreted as probabilistic statements, not absolute truths, and the model's assumptions should be explicitly stated and published alongside the results. Understanding the "agreement index" provided by software is crucial, as low values indicate conflict between the radiocarbon dates and the prior model, necessitating re-examination of either.

Common questions

A common question is whether Bayesian modelling can date a single sample more precisely than a standard calibration; it cannot, as it relies on the constraints provided by multiple samples and prior information. Researchers often ask if they can use dates from old excavations, which is possible only if the stratigraphic relationships are exceptionally well-recorded and published. Many wonder about the cost, which is significantly higher than routine dating due to the need for multiple, targeted radiocarbon assays and specialist analysis time. A frequent point of confusion is the difference between a simple sequence model and a phase model, which groups events that are contemporaneous but not necessarily in a specific order. Users regularly question how to handle outliers, which are dates that do not fit the model and may indicate intrusive material, contamination, or errors in the stratigraphic premise. Finally, archaeologists often ask about the minimum number of dates required, which depends on the complexity of the model but generally starts at a minimum of five to ten for a meaningful analysis.

Pros and cons

A major pro is the method's power to achieve remarkable chronological precision, often turning century-spanning calibrated ranges into tightly defined decades, which can resolve long-standing historical debates. It forces a rigorous, explicit formalization of the site's stratigraphic narrative, which often reveals hidden ambiguities or inconsistencies in the initial interpretation. The primary con is that the results are only as reliable as the prior model; an incorrect stratigraphic assumption is powerfully reinforced, yielding a confidently wrong chronology. The complexity of the software and statistical concepts can lead to a "black box" problem, where archaeologists accept outputs without critically understanding the model's structure. A common mistake is the inclusion of dates on unsuitable, long-lived material like bulk charcoal or human bone collagen from adults, which introduces unknown temporal offsets. Researchers often regret choosing this approach when they have not budgeted for the necessary number of high-quality radiocarbon dates or the specialist consultation time, leading to underpowered or flawed models.

Who it suits

This approach best suits research-driven archaeological projects where constructing a high-resolution timeline is a central objective, such as investigating transitions between cultural periods. It is ideal for sites with excellent organic preservation and clear, well-understood stratigraphy, like waterlogged settlements, peat sequences, or deeply stratified tells. The method suits projects with sufficient funding to undertake a targeted dating program of 10-30 samples, rather than relying on a handful of dates acquired over many seasons. It is particularly valuable for dendrochronology projects where wiggle-matching can pin floating tree-ring sequences to absolute calendar years. It is less suited to small-scale rescue archaeology with limited budgets, sites with severely disturbed stratigraphy, or projects where the primary dating material is long-lived or of uncertain association. Ultimately, it suits interdisciplinary teams willing to invest the time in integrating field archaeology, scientific dating, and statistical analysis from the project's initial planning stages.