We have obtained new magnetostratigraphic results from two Precambrian sedimentary sections from eastern Siberia and the southern Urals dated between ∼1100 Ma and ∼800 Ma. Sample magnetizations from the uppermost Mesoproterozoic Talakh-Khaya section in the Siberian Uchur-Maya region appear to be mostly carried by a mixture of magnetite and hematite. A sequence of 33 magnetic polarity intervals is recorded within the section. All reversals occur in the first ∼24 m, while the upper ∼160 m are characterized by a single magnetic interval of normal polarity assuming a Northern Hemisphere position of Siberia around 1000 Ma. The Lower Neoproterozoic Uralian Minyar section also possesses an ancient magnetization carried by magnetite and hematite. The high-temperature magnetization component obtained from the Minyar section establishes a sequence of 43 magnetic polarity intervals. Positive reversal tests obtained from the data sets presented here and from other previously analyzed Proterozoic sections provide no convincing evidence for a long-standing asymmetric geomagnetic field during the Proterozoic that would make the Precambrian field markedly less dipolar than during the Phanerozoic. The new data further reveal the occurrence of sharp transitions and alternations between long periods without any reversal (one superchron is observed from the Talakh-Khaya section) and periods with high reversal frequencies, possibly larger than 5–10 reversals per Myr. This characteristic may well be an important property of the Precambrian field, although rather sudden transitions between reversing and nonreversing states of the geodynamo may not be unique to this period.
 Reversals in the Earth's magnetic field over the past 150 Myr are well known due in large part to the analysis of the oceanic magnetic anomalies and to many magnetostratigraphic studies carried out over the last 40 years. For older periods of the Phanerozoic, magnetostratigraphic data obtained from biostratigraphically well-dated sedimentary sections are less abundant and have provided so far only fragments of the entire geomagnetic polarity reversal history. Nevertheless, the available data show that the geomagnetic reversal frequency has significantly varied over time, from a zero value during long periods without any reversal (the so-called superchrons) to periods characterized by frequent reversals, such as during the Middle Cambrian, the Middle Jurassic and during the Miocene (see, e.g., Pavlov and Gallet [2005, Figure 9] or Algeo  and Opdyke and Channell ). At present, three superchrons, each one lasting several tens of Myr, were detected during the Phanerozoic, one during the Early Paleozoic [e.g., Gallet and Pavlov, 1996; Pavlov and Gallet, 1998], another at the end of the Paleozoic and the most recent during the Cretaceous [e.g., Opdyke and Channell, 1996]. These geomagnetic features are often considered as exceptional phenomena in Earth history and are likely related to major volcanic and biological events [e.g., Courtillot and Olson, 2007, and references therein]. The dominant theory on their origin, which is supported by 3-D geodynamo simulations [e.g., Glatzmaier et al., 1999], is that they result from long- or medium-term (on the 10 Myr time scale) changes in core-mantle boundary conditions [e.g., McFadden and Merrill, 1984, 2000]. Hulot and Gallet , however, underlined the sharp transition between the reversing and nonreversing states of the geodynamo during the Lower Cretaceous. They suggest that this transition might principally reflect the nonlinear nature of the magnetohydrodynamical processes acting in Earth's core. If confirmed, such a possibility would challenge the assumption of some dominant mantle-induced effects on the reversal rate variations. Gathering a more precise description of the magnetic polarity time scale at the limits of the other superchrons would thus be of particular interest. We will show below that studying the geomagnetic field behavior during the Proterozoic, i.e., during the “Earth's middle age” as termed by A. Knoll (personal communication, 2008), may also provide valuable constraints on this topic.
 Only very few continuous magnetostratigraphic sequences can presently be considered to constrain the geomagnetic field behavior during the Precambrian. However, several differences between the Precambrian and Phanerozoic field have been tentatively proposed, including a higher contribution of nondipole components [Kent and Smethurst, 1998], the occurrence of asymmetric magnetic reversals (perhaps a consequence of a less dipolar field) [Pesonen and Nevanlinna, 1981; Nevanlinna and Pesonen, 1983] or, on average, less frequent geomagnetic polarity reversals [e.g., Roberts and Piper, 1989]. On the other hand, Smirnov and Tarduno  and Biggin et al.  have argued for a dominantly dipolar field in the early Earth's history from an analysis of the geomagnetic secular variation at the Late Archean and Early Proterozoic. But, each of the characteristics above clearly requires further investigation and close scrutiny of any potential effect that might be related to the late growth of the inner core [e.g., Coe and Glatzmaier, 2006; Lay et al., 2008; Biggin et al., 2009]. In this respect, the Proterozoic appears as a vast and promising domain of research. We report magnetostratigraphic results obtained from two Proterozoic sedimentary sections from Siberia and southern Urals dated between ∼1100 Ma and ∼800 Ma.
2. Geological Setting, Lithology, and Age of the Studied Sections
2.1. Malgina and Tsipanda Formations From the Uchur-Maya Region
 The Uchur-Maya region is located to the southeast of the Siberian platform (Figure 1). It comprises two adjacent tectonic structures separated by the north/northeast oriented Nelkan-Kyllakh thrust system. Situated to the west of this thrust system, the first structure, referred to as the Uchur-Maya undeformed zone, belongs to the Siberian platform in a strict sense. In this zone, multiple sections of nonmetamorphosed, flat-lying Riphean rocks are exposed. To the eastern side, the second structure, the so-called Judoma-Maya fold-thrust belt, represents the disturbed margin of the Siberian platform.
 We sampled the Malgina and Tsipanda formations in the Uchur-Maya undeformed zone. According to paleontological evidence [e.g., Semikhatov and Serebrjakov, 1983; Bartley et al., 2001], these two upper members of the Kerpyl group date to the uppermost Mesoproterozoic (Figure 2). These formations were deposited in shallow water environments in rift-related intracraton basins [Khudoley et al., 2001]. They are well exposed over several tens of kilometers along the middle course of the Maya river. The Malgina Formation is mainly constituted by red, green and pale argillaceous limestones which are gradually replaced by pale greenish, pinkish, yellowish and grayish dolostones of the Tsipanda Formation. In some sections, principally located to the western part of the region, the upper beds of the Malgina Formation are composed of dark, sometimes bituminous limestones with thicknesses reaching up to 30 m. Within the limits of the Uchur-Maya underformed zone, the total thickness of both the Malgina and Tsipanda formations is fairly uniform (∼400–500 m). Considered separately, however, the individual thickness of these formations can vary significantly. The Malgina Formation, for instance, ranges from 35 to 120 m thick.
 We collected 680 oriented hand samples from the Talakh-Khaya section exposed along the right bank of the Maya river, some tens of kilometers below the mouth of the Ingili river (Figure 1). The choice of this section was guided by the excellent conditions of exposure, by the relative absence of any tectonic or metamorphic event in this area, by the small proportion of dark bituminous beds and by the thickness of the Malgina Formation (∼110 m) in this location relative to elsewhere in the undeformed zone [Semikhatov and Serebrjakov, 1983]. The thickness of the Tsipanda Formation reaches ∼290 m here. The entire Malgina Formation was sampled with a narrow spacing between samples of approximately 10–15 cm in the lower ∼50 m and then with a spacing of ∼1–1.5 m in the upper part of the section including the lower one third (∼100 m) of the Tsipanda Formation. Note that this sampling was carried out across two subsections A and B ∼300 m apart with a clear ∼1 m thick stratigraphic overlap.
 Recent geochronological data indicate that the age of the Malgina and Tsipanda formations are younger than 1100 Ma, which is a U-Pb age obtained from detrital zircons collected from the underlying Totta Formation [Khudoley et al., 2006], and older than 1005 ± 4 Ma, which is a U-Pb age on baddeleyite extracted from mafic sills intruding the overlying Upper Riphean Lakhanda group and the Kandyk Formation [Rainbird et al., 1998]. Interestingly, pebbles from these sills have been found in rocks from the upper part of the Kandyk Formation, which is the lower member of the Uy group (Figure 2) [Semikhatov and Serebrjakov, 1983]. These sills were therefore emplaced during the deposition of the Kandyk Formation. Pb-Pb isochrons yielding two ages at 1043 ± 14 Ma and 1025 ± 40 Ma were also recently obtained on limestones from the upper and middle parts of the Malgina Formation and the lower part of the Lakhanda group (Milkon subformation–middle part of the Neryuen Formation), respectively [Ovchinnikova et al., 2001; Semikhatov et al., 2000]. These ages are considered to be closely related to the time of early diagenesis in the sediments. On the other hand, an age of 1035 ± 60 Ma was obtained for carbonates from the Sukhaya Tunguska Formation (Turukhansk region, northwestern Siberian platform) assumed to be coincident in time with the Tsipanda Formation [Ovchinnikova et al., 1995]. At present, all the available geochronological data therefore indicate that the Malgina and the Tsipanda formations were emplaced between ∼1060 Ma and ∼1000 Ma.
2.2. Katav Formation From Southern Urals
 The section of the Katav Formation presented here is located on the territory of the Bashkirian anticlinorium (southern Urals; Figure 1), which represents a thrust and fold zone at the southeastern edge of the east European platform. The carbonate Katav Formation is the second oldest formation of the sedimentary carbonate terrigenous Karatau group of Upper Riphean age, lying above the Zilmerdak Formation and below the Inzer, Minyar, Uk and Krivaya Luka formations (Figure 2). These Riphean strata were deposited in a shallow water marine basin along the platform margin [Maslov et al., 1997]. The main stages of tectonic deformations (folding and thrusting), which probably induced local rotations and relative displacements within the Bashkirian anticlinorium, have been dated to the Late Paleozoic [Giese et al., 1999]. However, there is also strong evidence for older Vendian to Ordovician deformations, most probably related to the Cadomian or the Baikalian orogen [Maslov et al., 1997].
 We sampled one of the best sections of the Katav Formation, located in the vicinity of the railway station of the town Minyar (Chelyabinsk region), in the western, relatively weakly deformed part of the Bashkirian anticlinorium. The Minyar section consists of ∼200 m of unmetamorphosed reddish, greenish and grayish limestones. Reddish limestones predominate in the lower part of the section, while those with greenish and grayish colors prevail above. The strata are gently tilted by ∼10° to ∼20° toward the north/northeast. About 230 oriented blocks were collected within the section with a sampling interval of ∼1–1.5 m. Only a few stratigraphic intervals were not sampled because of poor accessibility; the thickest gap reaches ∼16 m, but in aggregate the sampling gaps do not exceed ∼10% of the total thickness.
 The age of the Upper Riphean Katav Formation is constrained by several recent radiometric determinations, which were summarized by Bartley et al.  and Kuznetsov et al. . Pb-Pb analyses of limestones from the lower part of the Inzer Formation yielded an age of 836 ± 25 Ma, most probably reflecting the time of the early diagenesis in these sediments [Ovchinnikova et al., 1998]. This result is further supported by Rb-Sr data providing an age of ∼835–805 Ma for the time of early burial diagenesis in the Inzer Formation [Gorokhov et al., 1995]. A Pb-Pb isochron also yielded an age of 780 ± 85 Ma for the middle part of the Minyar Formation [Ovchinnikova et al., 2000]. It is worth noting that K/Ar ages obtained on globular glauconites in the sixties to eighties provided rough determinations of ∼900–800 Ma for the Inzer Formation and ∼940 Ma for the Katav Formation [Keller and Chumakov, 1983]. These ages must be considered with some caution because most of them do not rely on modern analyses. Other age constraints have been derived from the correlation based on paleontological data (stromatolites and organic walled microfossils) between the Zilmerdak Formation, underlying the Katav Formation, and the Lakhanda group of the Uchur-Maya region. The age of the Lakhanda group has been established as roughly 1030 Ma [Ovchinnikova et al., 2001]. The available data therefore indicate that the Katav Formation was deposited between ∼1030 and ∼800 Ma.
3. Paleomagnetic Analyses
 Most paleomagnetic experiments were carried out using a three-axis 2G cryogenic magnetometer housed in the magnetically shielded laboratory at the Institut de Physique du Globe de Paris. Samples were all thermally demagnetized in 12–18 steps. Several rock magnetic measurements were also performed in the Paleomagnetic laboratory of the Institute of Physics of the Earth in Moscow. Components of magnetization were identified using the PaleoMac [Cogné, 2003] or Enkin's  paleomagnetic software packages.
3.1. Paleomagnetic Data From the Malgina and Tsipanda Formations
 The magnetic properties of rocks from the Malgina Formation have already been studied by Gallet et al. . They sampled these deposits in four sections, the Selia, Haahar, Ingili and Emelekeen sections, which are a few tens of kilometers from the Talakh-Khaya section. We observe the same paleomagnetic behavior as described by Gallet et al. . Thermal demagnetization of the natural remanent magnetization (NRM) reveals the presence of two magnetization components (Figure 3). A first component, isolated up to ∼300°C–450°C, has a steep inclination and likely has a recent origin. The second component, which exhibits the two magnetic polarities, is isolated up to ∼570°C–590°C or ∼680°C (Figure 3). These unblocking temperatures indicate that magnetite and/or hematite dominate the magnetic mineralogy in the studied samples. Such variety in magnetic mineralogy is further illustrated by several examples of thermal demagnetization of a differential (0.15 T and 1.5T) isothermal remanent magnetization (IRM) acquired in two perpendicular directions, showing complete or partial demagnetization of those IRM components below or above ∼580°C (Figure 4). Nevertheless, the strong inflexion observed around 580°C in many demagnetization diagrams clearly shows the frequent coexistence of both magnetite and hematite in the same samples (see Figures 3c–3f). Magnetization components were easily isolated in most samples from the Talakh-Khaya section, in particular in the lowest ∼75–80 m, where almost 100% of the samples yielded suitable paleomagnetic results. However, the success rate of analyses was much lower in the upper part of the section mostly because the samples are generally more weakly magnetized (<1–3.10−5 A/m above ∼300°C). Among the 62 samples collected from the Tsipanda Formation, only 18 provided a high-temperature magnetization component (but with a maximum angle of deviation ≥15° for 4 samples).
 The west/east pointing paleomagnetic directions calculated for the high-temperature magnetization component are very similar to that previously obtained by Gallet et al.  from the Malgina Formation (Figure 5 and Table 1). We remind the reader here that the primary origin of the west/east direction was supported by several lines of evidence, including a positive reversal test, a positive fold test and a positive conglomerate test. The more detailed sampling carried out in the present study provides additional information on the magnetization processes in the sediments from the Malgina and Tsipanda formations and on the evolution of the paleomagnetic directions through the Talakh-Khaya section. First, for taking into account the fact that the magnetization appears principally carried by a mixture of magnetite and hematite, we isolated the high-temperature magnetization component below and above 580°C in order to (roughly) distinguish the paleomagnetic directions mostly recorded by magnetite from those carried by hematite. The results clearly show that the two sets of directions do not statistically share the same mean direction (γ = 5.4°, γc = 2.0° at 95% of confidence) [McFadden and McElhinny, 1990]. The hematite-related directions exhibit shallower inclinations (Figures 5a–5c and Table 1). The difference between the magnetite and hematite-related mean directions, which amounts to 5–6°, may be due to a different response of the minerals to the compaction of sediments during their burial. Flat and elongated grains of hematite may have been more affected by compaction. On the other hand, the inclination shallowing could also be linked to the argillaceous content of the sediments. Hematite grains growing within relatively large flat clay particles would have been preferentially rotated under compaction. In any case, these two possibilities would strengthen the primary (precompaction) origin of the analyzed high-temperature magnetization component. A reversal test was performed on both the magnetite- and hematite-related magnetization components [McFadden and McElhinny, 1990]. Negative results are obtained (Table 1), which will be further discussed in section 5. Furthermore, instead to estimate only general “magnetite” and “hematite” mean directions for the entire section, we also computed submean directions considering sliding subsets, with no overlap, of 40 samples (Table 2). This simple procedure reveals a clear trend, whatever the magnetic carrier, magnetite or hematite (Figures 5e and 5f), toward shallower inclinations from bottom to top of the section. This trend progressively brings the paleomagnetic directions closer to the mean direction previously determined for the overlying Neoproterozoic Lakhanda group (Figure 2) [Pavlov et al., 2002]. This feature does not appear related to any systematic evolution in anisotropy of magnetic susceptibility (AMS) measured on samples well distributed through the section (with AMS always less than 1% and Kmin roughly perpendicular to the bedding plane). The changes in paleomagnetic direction observed in Talakh-Khaya, therefore, most probably reflect the motion of Siberia during the sediment deposition.
Table 1. Paleomagnetic Directions Obtained From the Talakh-Khaya Sectiona
Formation and Outcrops
Polarity, Mineral, and Stratigraphic Level
Reversal Test, γ/γc (deg)
dp (deg)/dm (deg)
Uchur-Maya region, southeastern Siberia. Reversal tests are performed on different data selections and a distinction is made between the magnetization components carried by magnetite and hematite. Mean paleomagnetic directions obtained from two small sections (Nelkan and Ytyrynka) encompassing the Neryuen Formation are also reported.
All Talakh-Khaya Section: Comparison Between Magnetization Carried by Magnetite or Hematite
Malgina, Talakh-Khaya, latitude = 58.56°N, longitude = 134.39°E
hematite normal and reversed
magnetite normal and reversed
lower 24 m, normal
lower 24 m, reversed
lower 24 m, MAD ≤ 8°, normal
lower 24 m, MAD ≤ 8°, reversed
lower 24 m, MAD ≤ 3°, normal
lower 24 m, MAD ≤ 3°, reversed
samples from interval MLG347v–MLG381v, ∼2.6 m (387–651 cm), all, normal
samples from interval MLG347v–MLG381v, ∼2.6 m (387–651 cm), all, reversed
samples from interval MLG347v–MLG381v, ∼2.6 m (387–651 cm), MAD ≤ 3°, normal
samples from interval MLG347v–MLG381v, ∼2.6 m (387 cm – 651 cm), MAD ≤ 3°, reversed
lower 24 m, without interval MLG347v –MLG381v, 21.4 m, normal
lower 24 m, without interval MLG347v –MLG381v, 21.4 m, reversed
Tsipanda, Talakh-Khaya, latitude = 58.56°N, longitude = 134.39°E
Other Nearby Sections
Neryuen, Milkon subformation, Nel'kan, latitude = 57.60°N, longitude = 136.07°E
Neryuen, Milkon subformation, Ytyrynda, latitude = 58.92°N, longitude = 134.57°E
Table 2. Evolution in the Paleomagnetic Directions Observed Across the Talakh-Khaya and Minyar Sections Using Sliding Windows of 40 and 20 Samples, Respectively
Ordinal Sample Numbers
Number of Samples
Talakh-Khaya Section (Malgina and Tsipanda Formations)
Minyar Section (Katav Formation)
3.2. Paleomagnetic Results From the Katav Formation
 The Katav Formation was the subject in the early sixties of one of the first paleomagnetic studies of Precambrian rocks in the former Soviet Union [Komissarova, 1970]. Further work was conducted in the seventies to early eighties [Shipunov, 1991] but these paleomagnetic data, obtained only using partial or single-point thermal demagnetization procedures, do not satisfy enough modern experimental requirements. The stepwise thermal demagnetization routine used in the present study allows detection of two magnetization components. A lower unblocking temperature component is first isolated up to ∼300°C, but sometimes up to ∼400°C (Figure 6). The directions obtained for this component are close to that of the present-day field at the site, and likely have a modern origin. At higher temperatures, most samples show a second magnetization component, which decays toward the origin of the demagnetization diagrams (Figure 6). This component possesses the two magnetic polarity states (Figure 7) and is generally destroyed around 680°C. From the evolution of the magnetic moments (NRM and IRM) during heating, there is clear evidence for a mixture of both magnetite and hematite (Figures 4d, 4e, and 6b–6d). The frequent coexistence of both moderate and high-coercivity minerals is confirmed by hysteresis loop measurements exhibiting clear wasp-waisted shapes (Figures 8a and 8b). In contrast with the previous results, however, the directions obtained for the magnetite- and hematite-bearing magnetization components are identical, thus showing no differential inclination shallowing effect which may be related to the higher carbonate (lesser argillaceous) content in the Katav Formation.
 Several lines of evidence support a primary origin for the high-temperature magnetization component isolated in the Minyar section. The first was obtained from scanning electron microscope (SEM) observations made on several samples. These samples display titanomagnetite grains with exsolution textures, supporting their detrital origin. (Figure 8c). The second point is the presence of numerous magnetic polarity reversals within the section. Several samples collected at different transitions between intervals of opposite magnetic polarities yield unusual paleomagnetic directions (Figures 6e and 6f). The latter probably reflect either the “true” record of a transitional geomagnetic field or an artifact induced by the simultaneous lock-in of both normal and reversed polarity magnetizations in those strata (or perhaps a combination of these two possibilities). After their removal, the two groups of remaining directions with opposite polarities successfully pass the reversal test designed by McFadden and McElhinny  (γ = 2.6°; γc = 3.3°; Table 3). Positive reversal tests are also obtained when the section is divided in two subsections (lower ∼80 m and upper ∼80 m; Table 2). Note that in this case, the two mean paleomagnetic directions (combining normal and reversed polarity data) are different at the 95% confidence level. Moreover, although only slight changes in dipping attitude are observed within the Minyar section, a positive fold test, according to the modified version of Enkin's  fold test, is obtained at 95% when using the high-temperature magnetic directions [see also Shipunov, 1991].
Table 3. Paleomagnetic Directions Obtained From the Katav Formationa
Reversal Test, γ/γc (deg)
dp (deg)/dm (deg)
Minyar, Pervomaysky, and “Ufa-Beloretsk road” sections, southern Urals. Reversal tests are performed on different data selections. Mean paleomagnetic directions obtained from three sections encompassing two the Inzer Formation and the other the Zilmerdak Formation are also reported.
Mean site 1 + site 2; N = 25.
Minyar, latitude = 55.06°N, longitude = 57.52°E
lower 79.9 m (samples KT1–KT64), reversed
lower 79.9 m (samples KT1–KT64), normal
upper 72.9 m (samples KT104–KT105), reversed
upper 72.9 m (samples KT104–KT105), normal
Pervomaysky, latitude = 55.05°N, longitude = 57.54°E
 We studied two additional outcrops of the Katav Formation, each yielding the possibility to perform a fold test. In those cases, the age of folding may be as old as Vendian to Ordovician [e.g., Maslov et al., 1997]. The first outcrop is located in the vicinity of the village of Pervomaysky, some 1.5 km southwest from the Minyar section. It shows evidence of a meter-scale postdepositional rock deformation event. The Fisher parameter K computed from 23 samples increased by a factor of 5 after bedding correction, yielding a positive fold test (Figures 9a–9d). The second outcrop was sampled along the road Ufa-Beloretsk, ∼90 km to the south of the Minyar section. It shows a two-limb fold exposed over a distance of ∼40 m. One paleomagnetic site was collected from each limb. The two groups of samples reveal dual magnetic polarities (Figures 9e–9g). Their mean directions are clearly different in in situ coordinates, whereas they become identical at the 95% level after bedding correction. It is worth pointing out that the latter are close to that observed from the Minyar and Pervomaysky sections (Figure 9h and Table 3).
 Submean paleomagnetic directions were also computed for the Minyar section from a successive nonoverlapping series of 20 samples. As in the Talakh-Khaya section, a systematic trend toward lower inclinations (by ∼10°) is observed, bringing the paleomagnetic directions obtained in the upper part of the Katav Formation closer to the mean directions we found from two near-located outcrops of the overlying Inzer Formation (Figure 7d and Table 3). This trend, which is not related to any evolution in AMS measurements carried out through the section (again with AMS of less than 1% and Kmin roughly perpendicular to the bedding plane), likely describes the plate motion during sediment deposition.
 At this step, it is worth recalling that widespread remagnetization was reported in rocks of different ages from the Ural-Mongolian fold belt [e.g., Komissarova, 1970; Pechersky and Didenko, 1995; Grishin et al., 1997]. This remagnetizaton, consistently showing reversed polarity, was thought to be of Upper Paleozoic age based on the proximity of the obtained paleomagnetic poles with the Upper Paleozoic segment of the eastern European apparent polar wander path (this plate being located in the Northern Hemisphere at this time) [e.g., Smethurst et al., 1998; Cocks and Torsvik, 2007]. Although the paleomagnetic pole derived from the Minyar section is also close to that segment, a remagnetization of the section appears very unlikely [see Pavlov and Gallet, 2009, Figure 6]. The preceding arguments coupled with the numerous magnetic polarity reversals observed from that section are clearly incompatible with the occurrence of the late Carboniferous-Permian Kiaman reversed polarity superchron. Moreover, the paleomagnetic poles obtained for the Katav and Inzer formations (Table 3) are rather similar to the ∼700–800 Ma mean pole proposed by Meert and Torsvik  and Walderhaug et al.  for the eastern European platform, which was assumed to be located in the Southern Hemisphere. These arguments indicate a primary origin for the high-temperature magnetization component isolated in the Minyar section.
4. Magnetostratigraphy of the Talakh-Khaya and Minyar Sections and Supplementary Results
 As a preliminary remark, we acknowledge the problem of the polarity option for our two magnetostratigraphic data sets. In both cases, the choice between normal and reversed paleomagnetic directions is not trivial and is currently under discussion [e.g., Gallet et al., 2000; Pavlov et al., 2002; Pisarevsky and Natapov, 2003; Cocks and Torsvik, 2007; Meert et al., 2007; Walderhaug et al., 2007]. It is particularly critical for the results from the Minyar section as there is still ambiguity on the hemispheric position of Baltica during the Late Neoproterozoic and Vendian [Meert et al., 2007; Walderhaug et al., 2007]. The hemispheric solution for the Talakh-Khaya results strongly depends on the location of Vendian and Early Cambrian Siberian poles. Recent paleomagnetic data obtained by Shatsillo et al.  from Vendian red beds of the southern Siberian platform favor an “Indian Ocean trend” of the Siberian apparent polar wander path (APWP) during the Neoproterozoic, rather than a “Pacific” trend. If correct, the north paleomagnetic pole of the Malgina and Tsipanda formations should be located in the eastern hemisphere and thus the west pointing paleomagnetic directions found in these formations should be considered of normal magnetic polarity. This option is different from that chosen by Gallet et al. .
 A sequence of 33 magnetic polarity intervals is obtained from the Talakh-Khaya section. Each interval is represented by several samples (Figure 10; the paleomagnetic directions reported in Figure 10 are those determined in the magnetite-unblocking temperature range). All magnetic polarity reversals are found within the lower 24 m of the section (see enlargement in Figure 10), while the upper 174 m are a single normal polarity zone. Taking into account the change in polarity option, the new data appear to be in good agreement with the poorly resolved (because of insufficient sampling density) magnetostratigraphic sequences obtained by Gallet et al. . They also indicate the presence of a stratigraphically thick interval of normal magnetic polarity, encompassing the middle-upper parts of the Malgina Formation and the lower part of the Tsipanda Formation. This thick interval was not seen from the previous results and is a striking feature of the Talakh-Khaya magnetic record. Other paleomagnetic data yield some constraints on the location of its upper (younger) limit. Whereas the upper part of the Tsipanda Formation is not suitable for paleomagnetic investigation (this study), the upper lying formations from the Lakhanda and Ui groups are more favorable for this purpose [e.g., Pavlov et al., 2000, 2002]. Thermal demagnetization analysis of samples collected from several sections of the Neryuen Formation yielded a high-temperature magnetization component of normal polarity (Figure 11) [Pavlov et al., 2000]. The same polarity was observed in samples from the lower part of the Ignikan Formation [Pavlov et al., 2000]. The end of this magnetic interval may in fact occur somewhere between the upper part of the Ignikan Formation and the lower part of the Kandyk Formation, as samples with a reversed polarity magnetization were found by Pavlov et al.  from the Kandyk and Ust-Kirba formations (Figure 12). All these data therefore suggest a long duration for the normal polarity interval (see below), which begins within the Malgina Formation and contains the Mesoproteroic-Neoproterozoic boundary traced in the Uchur-Maya region (Figure 2).
 Following Meert et al.  and Walderhaug et al. , we assumed a Southern Hemisphere position for the eastern European platform around 800 Ma (i.e., a normal polarity for the southwest pointing paleomagnetic directions obtained from the Minyar section). The Minyar section contains 43 magnetic polarities intervals, 8 of which are defined by one sample (Figure 13). 34 magnetic polarity intervals occur within the upper ∼70 m, while a ∼40 m thick interval immediately below this does not show any magnetic reversal. This new record encompassing the Katav Formation is much more detailed than the sequences previously determined by Shipunov , Komissarova , and Komissarova et al. . We also obtained preliminary magnetostratigraphic data from the underlying terrigeneous Zilmerdak Formation sampled near the village of Inzer, about 100 km to the south of the Minyar section. The directions of the high-temperature magnetization component are rather scattered, but a positive fold test attests to the primary origin of the magnetic signal, and at least 23 additional magnetic polarity intervals are observed (Figure 14 and Table 3). Considered together, the data reported in the present study thus show evidence for approximately one hundred magnetic polarity reversals around the Mesoproterozoic-Neoproterozoic boundary, although a continuous magnetic polarity sequence cannot be established yet.
 The numerous geomagnetic polarity reversals observed in the Talakh-Khaya and Minyar sections provide a rare opportunity to better constrain the behavior of the Earth's magnetic field at two distinct (but relatively close) time intervals of the Proterozoic. In particular, our new results shed further light on the possible occurrence of asymmetric polarity reversals during the Proterozoic that might indicate a greater contribution of nondipole sources during this period than during the Phanerozoic [e.g., Pesonen and Nevanlinna, 1981; Nevanlinna and Pesonen, 1983; Kent and Smethurst, 1998]. In addition, rough approximations can be made on the sedimentation rates across the studied sections to obtain important new information on the changes in magnetic reversal frequency during Precambrian time.
 The occurrence of asymmetrical polarity reversals during the Proterozoic has been suggested on the basis of a very limited data set, mainly obtained from the Keweenawan lavas in Canada dated at ∼1.1 Ma (with differences in inclination as high as ∼25° between normal and reversed polarity directions) [Pesonen and Nevanlinna, 1981; Nevanlinna and Pesonen, 1983]. This possibility was discussed and challenged by Gallet et al.  from their magnetostratigraphic investigation of the two coeval late Mesoproterozoic Malgina and Linok formations from the southeastern and northwestern Siberian platform, respectively. These authors did not confirm the persistence of such a property of the Precambrian geomagnetic field [see also Gose et al., 2006; Dunlop and Yu, 2004]. Nevertheless, the new data reported in the present study, together with results from other sections not available in 2000, provide additional constraints on this issue.
 The paleomagnetic directions determined for the high-temperature magnetization component from the Minyar section yield a positive reversal test (Table 3). This result holds both for the entire data set and for successive subsets of ∼60 samples isolated within the section. We thus find no evidence for an asymmetrical geomagnetic field during the time of deposition of the Katav Formation (i.e., during ∼20 Myr between ∼1030 Ma and ∼800 Ma; see below). Gallet et al.  previously obtained a positive reversal test in three sections from the Uchur-Maya region encompassing the lower part of the Malgina Formation. The data from the Talakh-Khaya section, which encompasses the entire Malgina Formation and the lower part of the Tsipanda Formation, are much more numerous but, considered as a whole (438 directions), they fail the reversal test (Table 1). The same negative result is observed when only the data obtained from the lower ∼24 m of the section are taken into account, thus avoiding the data defining the long normal polarity interval within the upper part of the Malgina Formation and the Tsipanda Formation and limiting any potential effect due to the motion of Siberia. However, the interpretation of this characteristic in terms of geomagnetic field behavior appears unsatisfactory for two reasons. The first relies on the quality of the data. If only the best defined paleomagnetic directions with a maximum angle of deviation ≤3° are retained, the reversal test then becomes positive (with classification A according to McFadden and McElhinny  (Table 1)). The second reason springs from the observation that the main source of the nonantipodality between the normal and reversed polarity directions in the Talakh-Khaya section arises from a restricted stratigraphic interval, between ∼3.9 and ∼6.5 m from the base, which only contains a pair of normal and reversed polarity intervals. If this zone is excluded from the computations, the reversal test is positive whatever the data selection (Table 1). Moreover, if only the directions with a maximum angle of deviation ≤3° are considered for this interval, they pass the reversal test. This means that evidence for an asymmetrical geomagnetic field from this section is tenuous, and most probably originates from very local lithological disturbances. This conclusion is further supported by the positive reversal test also obtained from the Linok Formation, in the Turukhansk region [Gallet et al., 2000], considered as coeval with the Malgina Formation [e.g., Semikhatov, 1995].
 Recent paleomagnetic and magnetostratigraphic results dated to different time intervals of the Late Archean and of the Proterozoic also provided positive reversal tests, which again contradict the persistent or even the temporary occurrence of an asymmetrical geomagnetic field during this period. Among those studies, there are the data obtained from the ∼2.7 Ga flood basalts in the Pilbara craton (Western Australia) [Strik et al., 2003], from the ∼2.0–1.6 Ga mafic dykes of the eastern Bushveld Complex (South Africa) [Letts et al., 2005], from red beds in the ∼1.8 Ga Shoksha Formation (Russia) [Pisarevsky and Sokolov, 2001], from the ∼1.4 Ga Belt-Purcell Supergroup (North America) [Elston et al., 2002], from the ∼1.1 Ga Umkondo dolerites in the Kalahari craton (southern Africa) [Gose et al., 2006] and from the ∼0.8 Ga Aksu dyke swarm (Tarim basin) [Chen et al., 2004]. We further note that positive reversal tests were also obtained from eastern Canadian formations and carbonatite complexes coeval with the ∼1.1 Ga Keweenawan lava flows [e.g., Costanzo-Alvarez et al., 1993; Symons, 1994], which argues against a regional consistency of the asymmetrical geomagnetic reversals found in the Keweenawan lavas. The latter observation was then interpreted as resulting either of nonaveraged secular variation [Costanzo-Alvarez et al., 1993] or of unremoved secondary magnetization [Dunlop and Yu, 2004]. More recently, two additional paleomagnetic studies conducted on the Keweenawan volcanic rocks favor the view that the differences in inclination observed between the normal and reversed polarity lavas would be principally due to age progression and rapid continental motion of the North American plate between the two groups of lava flows [Tauxe and Kodama, 2009; Swanson-Hysell et al., 2009]. We therefore consider that there is currently no convincing evidence for a long-standing asymmetric geomagnetic field during the Proterozoic that would make the Precambrian geomagnetic field markedly less dipolar than during the Phanerozoic [Gallet et al., 2000; Dunlop and Yu, 2004].
 Computing magnetic reversal frequencies during the Proterozoic from our magnetostratigraphic data is clearly not an easy task as this requires the use of constraints on the duration of sediment deposition in the studied sections (where short-time-ranging biostratigraphic markers are virtually absent). And only very rough estimates of the mean sedimentation rates across the Talakh-Khaya and Minyar sections can be proposed considering the available isotopic, lithological and paleomagnetic data.
 The middle-upper Malgina and the middle Kandyk formations were dated at 1043 ± 14 Ma and 1005 ± 4 Ma, respectively [Ovchinnikova et al., 2001; Rainbird et al., 1998]. In the Uchur-Maya undeformed zone, this time interval of ∼40 Myr is represented by ∼700 m of sediment accumulation, which yields a mean sedimentation rate of ∼1–2 cm/1000 years. To first order, such an estimate appears reasonable when compared with other lithologically similar sequences of Phanerozoic shallow water carbonates [e.g., Bosscher and Schlager, 1993; Altermann and Nelson, 1998; Satolli et al., 2007]. In the case of the paleomagnetic directional trend observed within the ∼180 m of the Talakh-Khaya section (with a difference of paleolatitude of ∼7° between the top and the base of the section), if a “reasonable” velocity of ∼3 cm/year (i.e., the rough mean value for the present plate network) is considered for the Siberian plate, the sampled section would have a rough duration of ∼25 Myr. This would yield a sedimentation rate on the order of 1 cm/1000 years, consistent with the previous estimate. The same rough value is also obtained from the directional trend found from the Minyar section through the Katav Formation (duration of ∼20 Myr for a thickness of ∼200 m, again considering a plate motion of ∼3 cm/year). For the computations below, we will therefore consider this conservative, rather low, mean sedimentation rate of ∼1 cm/1000 years.
 In the Talakh-Khaya section, we find a sharp transition between a period of ∼2 Myr, at the base of section, characterized by a magnetic reversal frequency of ∼10 reversals per Myr (22 reversals within ∼20 m) and a period above of ∼16 Myr of constant magnetic polarity (Figure 10). Data obtained from the overlying formations [Pavlov et al., 2000, 2002; this study] indicate that the duration of this interval of fixed polarity could have been on the order of 30 Myr or possibly longer (Figure 12). It is worth pointing out that even decreasing the sedimentation rate by a factor of 2 (which is on average not supported by the available isotopic ages) would maintain a very rapid transition between a period of high magnetic reversal frequency (∼5 reversals/Myr) and a superchron. By contrast, if the sedimentation rate within the Talakh-Khaya section, in particular in its lower part, is higher than ∼1 cm/1000 years, the magnetic reversal frequency would then reach a rate much higher than during the past, well-documented, ∼150 Myr. Although less extreme, the magnetostratigraphic results from the Myniar section also indicate a relatively sharp contrast in magnetic reversal frequency between a rate of ∼4.9 reversals/Myr during the upper third of the section (34 reversals during ∼7 Myr) immediately after a ∼4 Myr long interval with no reversal (Figure 13; for comparison, this duration is equivalent to that of chron C33r succeeding the Cretaceous Long Normal Superchron [Cande and Kent, 1995]).
 We acknowledge the crude nature of our computations. Nevertheless, they reveal the same feature as found in the detailed magnetostratigraphy of the ∼1.4 Ga North American Belt–Purcell Supergroup [Elston et al., 2002]. In the latter study, one long reversed polarity interval with a duration of ∼30 Myr was found in the middle part of the section, followed first by a thin zone with frequent geomagnetic reversals, next by a ∼5–10 Myr long normal polarity interval and finally again by a zone with many reversals. This sequence also suggests sharp transitions between periods of frequent reversals and superchron(s). This striking feature may well be an important property of the Precambrian geomagnetic field. However, from the analysis of the geomagnetic polarity time scale over the past 150 Myr, Hulot and Gallet  also showed that no special behavior in reversal rate, such as a medium-term decreasing trend, could be confidently identified in the period of ∼40 Myr immediately preceding the Cretaceous long normal superchron (with an average reversal frequency of ∼2–3 reversals per Myr). This superchron lacked a clear precursor. As a result, Hulot and Gallet  questioned the validity of linking the origination of superchrons to change in core-mantle boundary conditions. On the contrary, the close juxtaposition or alternation of long periods without reversal and periods with frequent reversals may attest to the nonlinear nature of the fluid geodynamo processes, that could induce sudden transitions between reversing and nonreversing states of the geodynamo. Our results and those of Elston et al.  tend to support such a view, at least for Precambrian time.
 From numerical simulations of the magnetohydrodynamical processes in Earth's core with a smaller inner core than today, Coe and Glatzmaier  recently argued that the magnetic reversal frequency should have been lower on average during the Precambrian than during the Phanerozoic (see also Roberts and Piper  and Biggin et al. [2008, 2009] for a discussion). Magnetostratigraphic data allowing a direct test of this hypothesis are obviously not numerous enough. For instance, Dunlop and Yu  proposed that the period between ∼1050 Ma and ∼820 Ma (the “Greenville time”), which comprises the time interval of deposition of the Malgina and Katav formations, was principally characterized by four “dominant” polarity chrons. Although this oversimplified picture cannot be disregarded because a continuous magnetic polarity sequence spanning this entire period is not yet available, we note that the Malgina Formation contains 33 magnetic polarity intervals, the Katav Formation 43 intervals and the underlying (older) Zilmerdak Formation 23 additional intervals. Furthermore, simply considering the total number of the magnetic polarity intervals found in the Minyar and Talakh-Khaya sections (76) and their cumulated estimated duration (∼45 Myr) yields an averaged reversal rate of ∼1.7 reversals per Myr, a value very close to the long-term magnetic reversal frequency which has prevailed over the past 150 Myr (with the occurrence of a superchron in both cases). Taken as a whole, our data therefore do not indicate a particularly low magnetic reversal frequency around 1 Ga that might be considered as a consequence of a smaller inner core during the Proterozoic. Further research on the frequency of superchrons (rather than magnetic reversal frequency) during the Earth's middle age could provide crucial constraints on the results of 3-D geodynamo simulations such as those of Coe and Glatzmaier .
 Our concluding remarks are twofold. The first point directly concerns the behavior of the geomagnetic field during the Precambrian. Our magnetostratigraphy of two Proterozoic sections from Siberia and the southern Urals dated between ∼1100 Ma and ∼800 Ma does not reveal a geomagnetic field markedly less dipolar than during more recent time [see also Gallet et al., 2000; Dunlop and Yu, 2004]. We observe sharp transitions and alternations between long periods without reversal and periods with frequent reversals (with frequencies possibly higher than 5 to 10 reversals per Myr). However, this characteristic may not be restricted to the Precambrian geomagnetic field. Our study shows that conducting detailed magnetostratigraphic work in order to better constrain the geomagnetic field behavior during Precambrian time is an attainable and promising goal. In particular, the Siberian platform possesses many other sedimentary sections that may be suitable for investigating the frequency of superchrons during the Precambrian.
 The second point concerns the need for numerical geodynamo simulations in order to decipher the origin of superchrons and the cause of the sharp transitions observed between the reversing and nonreversing states of the geodynamo. Do they principally originate from the nonlinear nature of the geodynamo (as supported by Hulot and Gallet ) or from medium-term changes at the core-mantle boundary? Our new magnetostratigraphic data clearly show that the first option, although rarely favored, should be considered with greater care. For this reason, there is a pressing need for additional numerical experiments of geodynamos with a variety of boundary conditions, different sizes of the inner core and, most importantly, long intervals of simulated years (>100 Myr). Such experiments could be crucial to improving our understanding of the past behavior of the geomagnetic field.
 We thank A. Shatsillo and V. Vodovozov for their help in collecting samples, N. Kuztetsov and V. Puchkov for constructive discussion on the field, and V. Tselmovich for his assistance with microscope investigation. We are grateful to D. Evans, A. Gogichaishvili, and S. Gilder for their comments on an early version of the manuscript. We also thank J. Tarduno, J. Feinberg, L. Tauxe, C. Mac Niocaill, and A. Maloof for their very constructive reviews. The manuscript benefited from the careful reading of B. Black. This study was supported by RFBR grants 07-05-00880 and 09-05-91056, CNRS PICS 4816, and IPGP. It was also partly financed by the INSU-CNRS program “SEDIT.” This is IPGP contribution 2550.