The Orbital Cycle Factory: Sixty cyclostratigraphic spectra in need of re-evaluation

Abstract

Conventional practice in cyclostratigraphy relies on invalid statistical tests to identify candidate orbital cycles in stratigraphic data-series. The practice is embodied in a presentation of the power spectrum that also displays statistical confidence limits (CLs). An investigation of sixty recent examples of this spectrum-with-CLs demonstrates the dependence of large numbers of cycle identifications on this procedure. The CLs, which are generated by default, do not provide criteria for the statistical test that they appear to represent, and cannot be used to measure significance at multiple frequencies. Continuing adherence to this practice is driven by precedent, by its ease of use, by its exclusively positive results, and by reluctance to engage either with the prior probabilities of orbital cycle preservation, or with the technical constraints of statistical testing. Readily available corrections to the procedure tend greatly to reduce or even eliminate cycle detections, but their promotion has generally resulted in vigorous defence of the status quo. Challenging such a highly productive paradigm may be unpopular, but it is time to confront two possibilities: that mathematically true cyclicity is less prevalent in stratigraphic successions than published results suggest; and that the conventional spectrum-with-confidence limits creates more candidate orbital cycles than it discovers: it is in fact an Orbital Cycle Factory.