A tale of clusters: No resolvable periodicity in the terrestrial impact cratering record

Rampino & Caldeira carry out a circular spectral analysis (CSA) of the terrestrial impact cratering record over the past 260 million years (Ma), and suggest a ∼26 Ma periodicity of impact events. For some of the impacts in that analysis, new accurate and high-precision ('robust'; 2SE < 2 per cent) ⁴⁰Ar-³⁹Ar ages have recently been published, resulting in significant age shifts. In a CSA of the updated impact age list, the periodicity is strongly reduced. In a CSA of a list containing only impacts with robust ages, we find no significant periodicity for the last 500 Ma. We show that if we relax the assumption of a fully periodic impact record, assuming it to be a mix of a periodic and a random component instead, we should have found a periodic component if it contributes more than ∼80 per cent of the impacts in the last 260 Ma. The difference between our CSA and the one by Rampino & Caldeira originates in a subset of 'clustered' impacts (i.e. with overlapping ages). The ∼26 Ma periodicity seemingly carried by these clusters alone is strongly significant if tested against a random distribution of ages, but this significance disappears if it is tested against a distribution containing (randomly spaced) clusters. The presence of a few impact age clusters (e.g. from asteroid break-up events) in an otherwise random impact record can thus give rise to false periodicity peaks in a CSA. There is currently no evidence for periodicity in the impact record.