View article

To identify age peaks and other features in an isotopic age distribution, it is common to perform a kernel density estimation or similar analysis. A key aspect of this estimation process is the choice of an age resolution bandwidth that best reflects the random variable and other assumptions on the data. Probabilistic kernel density analysis of large databases (up to nearly 40,000 samples) of U/Pb zircon ages suggests an optimum bandwidth of 25–30My for many key features, which yields approximately 40 peaks with confidence levels of c≥0.9. Because of natural redistribution processes, geographic sample bias may be minimized by jointly analyzing isotopic ages from both orogenic granitoids and from detrital zircons. We show that the relative heights of age peaks are commonly controlled by the local geographic distribution of samples and are not necessarily correlated with total geographic extent. Eight peaks with …