The Calibration Problem
When Willard Libby invented radiocarbon dating in 1949, he assumed atmospheric 14C was constant over time. But tree-ring measurements in the 1960s revealed that it was not — 14C levels have fluctuated by up to 10% over the past 50,000 years. A radiocarbon age of 3000 BP does not mean 3000 calendar years ago; it could be off by centuries. Calibration using tree-ring dated samples corrects this fundamental discrepancy.
Building the Calibration Curve
Scientists measure 14C in tree rings of known calendar age (established by crossdating). Plotting radiocarbon age versus calendar age creates the calibration curve. The IntCal20 curve uses thousands of tree-ring measurements from oaks, bristlecone pines, and kauri trees spanning 14,000 years, extended to 55,000 years using other archives. This curve is the Rosetta Stone of chronological science.
Wiggles, Plateaus & Precision
The calibration curve is not a straight line — it wiggles. Where the curve is steep, calibration is precise: a small 14C range maps to a small calendar range. Where the curve plateaus or reverses, calibration is poor: a single radiocarbon date maps to multiple possible calendar ages spanning centuries. The Hallstatt plateau (800-400 BC) is infamous for making Iron Age dating nearly impossible with radiocarbon alone.
Bayesian Calibration
Modern calibration uses Bayesian statistics. The radiocarbon measurement (with its Gaussian error) is projected onto the calibration curve to produce a probability distribution on the calendar axis. Software like OxCal and CALIB compute these distributions, often revealing multimodal age ranges that reflect the curve's wiggles. Additional constraints (stratigraphy, sequence models) can narrow the calibrated range dramatically.