The methodology and equipment for palaeomagnetic studies in the 1960s and 1970s was often rather primitive by today's standards. For example, Doell and Cox (1967) imply that the standard procedure for getting a palaeomagnetic direction was simply to measure the natural remanent magnetization, and assume that it represented a reliable primary magnetization direction if it proved sufficiently resistant to one or more of the standard demagnetization techniques.
In contrast, in a modern palaeomagnetic study, I'd expect a far more elaborate methodology: ten or more AF or thermal demagnetization steps per sample (possibly with a pilot study to compare effectiveness of demagnetization methods), measuring remaining remanence at each step; inspection of data in various projections; palaeomagnetic directions obtained by fitting one or more demagnetization components using, for example, PCA (Kirschvink, 1980), Linefind (Kent et al., 1983), or some form of great-circle intersection analysis (e.g. McFadden & McElhinny, 1988); and rock magnetic work (e.g. coercivity spectrum analysis) to provide a better idea of what's carrying the magnetization, and when and how the magnetization components were formed. Stability tests such as the fold test or reversal test might also by employed to check the reliability of the remanence.
In this context, I remember once being told that some large proportion (tens of per cent) of palaeomagnetic directions determined during the early days of the field had subsequently been invalidated when sites and samples were re-analysed using modern methodologies. Occasionally I come across an individual study which explicitly invalidates an earlier one; for example, Turner et al. (1989) revised the findings of Kennett and Watkins (1974), citing ‘serious errors in the interpretation of the results’.
I've never come across a published study which attempts a quantitative survey of how often such invalidations occur. Does anyone know of any publications which provide such a review of early palaeomagnetic studies and their reproducibility -- or lack thereof?
Doell, R. R. and Cox, A. (1967). Analysis of palaeomagnetic data. In Collinson, D. W., Creer, K. M. and Runcorn, S. K., editors, Methods in Palaeomagnetism, number 3 in Developments in Solid Earth Geophysics, pages 340–346. Amsterdam: Elsevier.
Kennett, J. P., & Watkins, N. D. (1974). Late Miocene—Early Pliocene paleomagnetic stratigraphy, paleoclimatology, and biostratigraphy in New Zealand. Geological Society of America Bulletin, 85(9), 1385-1398.
Kent, J. T., Briden, J. C., & Mardia, K. V. (1983). Linear and planar structure in ordered multivariate data as applied to progressive demagnetization of palaeomagnetic remanence. Geophysical Journal International, 75(3), 593-621.
Kirschvink, J. L. (1980). The least-squares line and plane and the analysis of palaeomagnetic data. Geophysical Journal International, 62(3), 699-718.
McFadden, P. L., & McElhinny, M. W. (1988). The combined analysis of remagnetization circles and direct observations in palaeomagnetism. Earth and Planetary Science Letters, 87(1), 161-172.
Turner, G. M., Roberts, A. P., Laj, C., Kissel, C., Mazaud, A., Guitton, S., & Christoffel, D. A. (1989). New paleomagnetic results from Blind River: Revised magnetostratigraphy and tectonic rotation of the Marlborough region, South Island, New Zealand. New Zealand journal of geology and geophysics, 32(2), 191-196.