Magnetic Polarity Transition

The results obtained from the rock magnetic analysis indicate that the main fraction of the NRM at Weinan is carried by detrital magnetite. However, to interpret the data as records of the reversing geomagnetic field, it is necessary to establish that it was acquired at, or soon after deposition of the loess. There are no direct ways of precisely establishing the lock-in depth of sedimentary magnetisation and estimates vary, from virtually zero to a few times the duration of a polarity transition itself. For instance, it has been suggested that the transitional directions recorded in marine deposits may arise from a smoothing of non-antipodal stable directions before and after a geomagnetic reversal. To first test this extreme hypothesis, we have compared the intermediate VGPs obtained for the MB and UJ transitions at Weinan with the synthetic VGP paths generated at the same site assuming heavy smoothing of immediately pre- and post-transitional directions. For the UJ transition the synthetic VGP path lies over the Pacific (Fig. 8b), and differs from the VGP path obtained from the paleomagnetic data (this holds true also if we plot the south VGP for this transition). Therefore, smoothing of pre- and post-transitional directions, as suggested by Langereis et al., does not provide a realistic explanation for the position of the path. For the MB transition, the synthetic VGP path lies over eastern Asia, and thus coincides with the Asian part of the VGP path obtained from the paleomagnetic data. Smoothing, however, fails to account for the points situated over North America. Records of reversals displaying VGPs in both preferred longitudinal sectors (or antipodal clusters of VGPs over these sectors) have already been observed in both sedimentary and volcanic studies [15,38]. The presence of VGPs over the two preferred antipodal sectors at Weinan is thus a known characteristic of reversal records and reinforces the idea that heavy smoothing cannot account for the records recovered from this section.

Moreover, to explain the transitional records in terms of smoothing of pre- and post-transitional directions, the magnetisation must be acquired over a sufficiently long time before and after the actual reversal for virtually all traces of the intermediate field to be lost, say several tens of thousands of years [39]. For the Weinan section this would correspond to several metres of sediments. Smoothing to this amount would also lead to an observed duration of the transition of the order of a few times the real duration. This is not observed at Weinan: the estimates of 5000 and 3200 yr obtained for the MB and UJ transitions are quite consistent with the values reported from other studies of the same transitions.They are, in particular, consistent with the first estimate of the length of a polarity transition obtained more than 25 years ago [40], using resuits from sequences of volcanic rocks, which are devoid of any possible smoothing effects. It is worth noting that this value has been confirmed by all the following studies of reversals. The recent observation that the Blake geomagnetic event is recorded in the Xining section is additional evidence that loess may record fast geomagnetic signals on this time scale.

Finally, on a shorter time scale, the presence of short-lived directional fluctuations may also be considered as evidence that large smoothing did not occur, provided that they do not arise from spurious artefacts of the magnetisation or of the measurements a n d / o r incomplete magnetic cleaning in the laboratory. The directional fluctuations between successive samples observed at Weinan are significantly larger than possible inaccuracies in the measurement or the determination of the final paleomagnetic direction and of intercorrelation of the two parallel samplings. Incomplete cleaning of an overprint acquired after the transition in a stronger field can produce large spurious directions when vectorially added to a primary direction acquired in a relatively weak transition field and could thus be invoked as a possible cause for these fluctuations. If this were the case, however, the vector differences between the paleomagnetic directions of successive samples would be predominantly aligned with the direction of the dipole field at the site. Uncleaned secondary components do not appear to play any major role in the observed fluctuations.

As mentioned above, it has been suggested that geometrical artefacts in the construction of the VGPs might lead to a 90 ° site path dependence which, combined with a distribution of sites strongly biased towards Europe, might lead to the preponderance of the paths over the two preferential antipodal sectors. At Weinan, the two longitudinal sectors are clearly apparent, but this corresponds to a ~ 0 ° or - 180 ° site path distance. Therefore, this record is not consistent with the hypothesis of a geometrical artefact, or with other suggested mechanisms, such as an increased inclination error at low field, which would introduce a bias towards a 90 ° site path distance.

All these observations tend to suggest that the magnetisation lock-in depth at Weinan is of the same order of magnitude, if not shorter, than in the best marine or lacustrine sediments used so far in studies of polarity transitions. Therefore, we believe that the observed records more realistically reflect at least the main trends of the real geomagnetic signal related to the reversal process, rather than a rock magnetic artefact of the loess section at Weinan.

The Weinan section provides new experimental data on geomagnetic reversal records from a region which has been largely ignored so far. The most remarkable characteristic of these records is the noticeable preponderance of the transitional VGPs in the two preferential sectors defined by marine or continental records of reversals. The different considerations discussed above suggest that a real geomagnetic signal is recorded, at least in the main trends, and thus reinforce the idea that the preferential sectors may have a physical significance. The statistical approach of McFadden et al., used in connection with a data base considering only polarity reversals and not short events, sustains the hypothesis of genuine geomagnetic behaviour (hypothesis Exciting). Should this hypothesis be proven true, it would imply that the mantle exerts a significant control over the reversal process. Theoretical evidence does exist to support this. Notwithstanding all these points, the sensitivity of the results to the addition of just the two records from Weinan is a clear indication that it is crucial that additional records be obtained from sites outside the present preferred-site bands[PDF][PDF].