ABSTRACTSBoth the growing body of literature on the tree‐ring calibration of radiocarbon dates, and the variety of statistical approaches used, make it appropriate to review the basic principles involved in the construction of a calibration curve. While its ultimate validity is in part a geophysical problem, the appropriate form for the calibration function, given a specific body of data, is a statistical question centering upon the analysis of the measurement errors involved. The relationship between the calibration function and the inverse calibration function is examined, and the shortcomings of the available published calibration curves considered in the light of this examination. It is concluded that there is no reason to doubt the general principles on which the calibration is based, but a need exists for a more critical handling of the measurement data with a statistical treatment of the errors involved. The quality of the data could itself be improved by the use of replicate samples for inter‐laboratory projects specifically designed to investigate the magnitude of measurement er