Low paleosecular variation at the equator: a paleomagnetic pilgrimage from Galapagos to Esterel with Allan Cox and Hans Zijderveld

McFadden and Merrill (1995) suggested that the paleosecular variation (PSV) measured by the angular scatter of the virtual geomagnetic pole is minimal at the equator and should be smaller during a superchron than during the last 5 Myr. We revisited a key site of the 0–5 Ma database, the Galapagos archipelago, studied by Allan Cox in the early sixties. We obtained 79 sites with reliable mean directions on four islands (San Cristobal, Floreana, Santa Cruz and Pinzon), showing a larger proportion of transitional data than Cox (16 instead of 6%), because the sampling was concentrated on the Brunhes-Matuyama transition as delimited by Cox. This dataset allowed us to test the statistical method of Vandamme (1994) to separate PSV from transitional data. We obtained an angular scatter value of 11.2° (9.9–12.9°), instead of 16.8° for an a-priori rejection angle of 40°, compared with the 12.7° predicted from the global compilation (McFadden et al. 1991). Studies of sequences of lava flows are quite scarce in the Permian Kiaman Superchron, and the Esterel volcanics with their subequatorial paleolatitude are a good candidate to test the above prediction. We confirm the quality of the original data of Zijderveld (1975) and we improved the mean direction from one site. We also used new geological and geochronological data: Ar/Ar ages point to the period 264–278 Ma for a totally reversed volcanic sequence, in agreement with an ending of the Kiaman Superchron at 262–268 Ma. The extremely low angular scatter obtained (4 to 8°, depending on data selection) confirms the prediction, but an alternative interpretation invoking a post-volcanic Permian remagnetization is discussed.