Annual radiocarbon calibration re-dates the Thera eruption

Thursday, August 16, 2018

Charlotte Pearson leads a re-examination of controversial Thera eruption ¹⁴C dates.


Akrotiri, Thera: the excavated remains of the Minoan town destroyed by the eruption (photo: Gretchen Gibbs).
Gretchen Gibbs

Tree-Ring Lab faculty member Charlotte Pearson is the lead author on a paper re-evaluating the dating of the Thera eruption, a critical event for dating the archaeological chronologies of the Aegean, Near East, and Egypt. The huge eruption (which literally blew away most of an Aegean island) had far-reaching effects, but scholars have debated the precise dating of this event for many years, with some archaeologists and historians favoring a date no earlier than the mid 16th century BCE, possibly as late as the early 15th century BCE, but the radiocarbon evidence pointing to a date no later than the early 16th century BCE, and possibly in the late 17th century BCE. The new paper uses the nuances of how radiocarbon measurements get translated into calendar dates to suggest a reconciliation of these two different estimates.

Radiocarbon dating works because cosmic rays continuously produce a small amount of radioactive carbon in the Earth's atmosphere, ¹⁴C. By photosynthesis plants take in this ¹⁴C from the air along with the stable forms, and animals in turn pick it up from the plants or herbivores they eat, but no new ¹⁴C come in once the carbon is locked up in stable tissues such as wood, or an organism dies. Radioactive decay then gradually removes the amount of ¹⁴C relative to the stable forms of carbon, so by measuring the proportion of ¹⁴C in a sample from something that was once a living plant or animal you get an estimate of when it died, or (importantly in the case of wood) when it stopped exchanging carbon with the living parts of the organism. The physics of radioactive decay are unvarying: it removes the ¹⁴C at a constant rate; however the starting amount of ¹⁴C in air has fluctuated over time, so on their own ¹⁴C measurements can only provide approximate dates. Tree-ring dating provides a way to turn the raw ¹⁴C measurements into accurate dates (incidentally opening a window onto the ¹⁴C fluctuations in the atmosphere).

Tree-ring dates are based on matching the patterns in the annual growth rings of trees to give true dates (accurate to a single year), so they are independent of any of the assumptions of radiocarbon dating, and can act as a natural directory of ¹⁴C concentrations through time. If someone has measured the ¹⁴C in many samples of tree-ring dated wood, and you have a ¹⁴C measurement for a sample of unknown age, you can find the tree-ring dated sample with the same ¹⁴C concentration, and then ideally know the actual date for the unknown sample. In practice this process, called calibration, is generally presented with the ¹⁴C measurements from known age samples graphed as a line, called a calibration curve, and there are three obstacles to getting unique accurate dates. Firstly, the ¹⁴C measurement from the unknown sample will be subject to various statistical and laboratory errors, adding some uncertainty to the measurement, secondly the measurements from the known-age samples used to produce the calibration curve will have their own uncertainties, and thirdly there may be times when samples from different true dates contain the same amount of ¹⁴C: if there is a run of several dates where samples have nearly the same amount of ¹⁴C it is called a plateau in the calibration curve. To represent these limitations, radiocarbon dates generally get presented as ranges of years, or as graphs showing the probability of the date representing a particular true year.

The known-age samples used for calibration have generally been small blocks of wood containing about ten tree rings. Most radiocarbon dating experts used to think that any loss of accuracy caused by averaging together several adjacent rings was comparable to the laboratory measurement errors, but a recent discovery has shown that at times the ¹⁴C in the atmosphere suddenly increases by huge amounts within a single year, so there is an exciting move to make new calibrations based on annual measurements. The paper is based on one of these new annual radiocarbon calibration curves covering the controversial period of the Thera eruption, with the University of Arizona Accelerator Mass Spectrometry Lab measuring bristlecone pine wood from the White Mountains of California and oak wood from County Kildare in Ireland (using two different species from places thousands of miles apart increases confidence that the calibration is good for general use). At its ends the new calibration curve corresponded closely to the current internationally agreed curve, known as IntCal13, suggesting that there were no problems from laboratory-specific biases in the measurements.

However the middle part of the new calibration curve is distinctly different from IntCal13, and this difference has a large effect when you calibrate the existing radiocarbon dates for the Thera eruption using the new calibration rather than IntCal13: both curves have one of those plateaus, the ambiguous periods where a run of several different true dates will show the same amount of ¹⁴C, but the new curve positions the plateau so that more of the Thera radiocarbon dates fall onto it, pushing the range of true calendar dates into the 16th century BCE, and opening a possible reconciliation with the archaeological date ranges.