Statistical significance of magnetic fabric data in studies of paramagnetic granites
E. L. Pueyo, M. T. Román-Berdiel, J. L. Bouchez, A. M. Casas, J. C. Larrasoaña, 2004. "Statistical significance of magnetic fabric data in studies of paramagnetic granites", Magnetic Fabric: Methods and Applications, F. Martín-Hernández, C. M. Lüneburg, C. Aubourg, M. Jackson
Download citation file:
The low anisotropies of paramagnetic granites, due to magnetocrystalline anisotropy, require a statistical treatment of the anisotropy of magnetic susceptibility (AMS) data when systematic fabric studies are performed. Absence of statistical information on these data makes evaluation of their quality difficult. The statistical significance of magnetic fabric in granites is evaluated in this paper. Jelínek’s elliptical confidence angles for the three principal susceptibility axes (E13, E12, E23) of a specimen are used as markers of the quality of the AMS data. Comparing these markers at sample-, site- and massif-scale with the mean AMS axes that result from spherical statistical models helps clarify the reliability of the AMS data. This analysis is presented in detail for the plutons of Veiga and Trives (Spain). It is then applied to seven other massifs from the Pyrenees. We propose the following guides: (1) fabrics with E13 between 10° and 20° tend to isotropy; the directional data and the shape parameter should be considered with great care; (2) lineation is not reliable when E12 > 25°, i.e. when Kmax is almost the same as Kint; (3) similarly, foliation is considered as not reliable when E23 > 25°, i.e. Kmin does not easily differentiate from Kint. Errors attached to the mean Kmax and Kmin axes should always be produced, thus allowing further interpretation. In Trives and Veiga, ‘perfect’ triaxiality cannot be automatically assumed since foliation and lineation could be defined simultaneously in only 53% of the cases. Finally, a minimum of three cores (9 specimens) per site would considerably increase the proportion of reliable orientation data.