Evolution of the geomagnetic reversal rate since 160 Ma: Is the process continuous?

Two main interpretations have been advanced for the evolution of the geomag netic reversal rate lambda since about 160 Ma. These interpretations have q uite different implications for Earth history and associated events and pro cesses. Our own interpretation is that lambda gradually and continuously de creased from its value of about 4 1/2 per million years at about 160 Ma to zero at about 118 Ma, then at about 83 Ma lambda began to recover, again gr adually and continuously, until reaching its present value of about 4 1/2 p er million years. The Cretaceous Superchron occurred from about 118 to abou t 83 Ma, when the reversal rate was zero. Recently, Gallet and Hulot [1997] suggested an alternative interpretation where mu (=1/lambda) was constant from 160 Ma to about 130 Ma, at which time the process was rapidly and stro ngly perturbed. This perturbation led to a discontinuous increase in the me an length mu of intervals and caused the Cretaceous Superchron. The process then gradually recovered from the strong perturbation with the mean length mu gradually decreasing until the reversal process again became stationary at about 25 Ma. This interpretation is predicated on mu being constant fro m 160 to 130 Ma, whereas our interpretation requires a gradual increase in mu over this time interval. Consequently, a new statistic is suggested to d istinguish between the null hypothesis that the reversal process is station ary against the specific alternative that there is a trend in the mean leng th mu. Application of this statistic shows that the data require a trend to ward larger values of mu in the reversal process going from 160 to 130 Ma a nd that this trend continues strongly to the beginning of the superchron. T hus the addition of a discontinuity at 130 Ma is neither required nor suppo rted by the data. Furthermore, application of this statistic shows that the trend toward smaller values of mu from 83 to 25 Ma appears to continue fro m 83 Ma to at least 12 Ma. Certainly there is no requirement for a change i n the style of process at 25 Ma. Overall, the Gallet and Hulot interpretati on seems untenable with the available data. Despite this, the Gallet and Hu lot interpretation has great value in that it is inherently testable and ou r ability to reject it has increased our knowledge about the reversal proce ss.