A Hyperactive Geomagnetic Field in the Late Visean (Early Carboniferous) From the Late Asbian Stratotype Section in Northwest England, UK

The pattern of geomagnetic polarity changes during the Early Carboniferous (Mississippian) is not known in detail. This information sparsity is addressed by determining a magnetostratigraphy from the late Asbian (late Visean at ∼333 Ma) in Trowbarrow Quarry, UK. This is the stratotype section of the late Asbian and has a detailed foraminiferal zonation based on the same set of paleomagnetic samples, establishing a detailed biostratigraphy. The 195 m‐thick section was sampled at an average spacing of 1.1 m, yielding a detailed magnetostratigraphy comprising nine major magnetozone couplets, and seven submagnetozones. The section dataset has a 78% bias to normal polarity determined from 177 sampling levels. The magnetization is carried by a mixture of hematite and detrital magnetite, with 68% of specimens dominated by hematite magnetizations. The primary magnetization passes a fold test showing its age was prior to the latest Carboniferous. The hematite is inferred to be largely of detrital, eolian origin, although some reddened levels are associated with emergent surfaces, suggesting that a small fraction of hematite is associated with platform emergence. The Mississippian age magnetization is partly overprinted with Kiaman Superchron‐age and Brunhes‐age magnetizations. Using the duration of the section based on astrochronology indicates a reversal frequency of 15.7 ± 0.75 Myr−1, indicating that the geodynamo was in a hyperactive reversing state between 335 and 333 Ma.


Introduction
Stratigraphic changes in geomagnetic polarity (magnetostratigraphy) have become one of the standard tools for stratigraphic study in the Mesozoic and Cenozoic, allowing environmental changes to be calibrated to precise time intervals (Hounslow et al., 2022;Miller & Wright, 2017;Nie et al., 2020).For much of the Pennsylvanian (Late Carboniferous), progress has been made in defining a polarity timescale (Opdyke & DiVenere, 2004;Opdyke et al., 2014), but details require validation, with those currently available largely based on few ageoverlapping studies (Hounslow, 2022).In the Carboniferous, regional differences in stratigraphic scales and biozonations contribute to additional uncertainty in how to correlate between the various studies when compiling a composite global polarity scale (Hounslow, 2022).For much of the Early Carboniferous (Mississippian) and Devonian, changes in geomagnetic polarity are not known in detail or those intervals that have been studied are open to additional uncertainty (Green et al., 2021;Van der Boon et al., 2022).
In addition to their applications in stratigraphy, quantification of geomagnetic reversal rates provides clues to the workings of the geodynamo, evolution of the core and its interaction with the mantle (Biggin et al., 2012;Davis & Buffett, 2022;Hounslow et al., 2018).In the Paleozoic and Ediacaran, intervals of hyperactivity in reversal rates (rates >8 Myr 1 ) seem to be common features (Gallet et al., 2019), along with intervals which do not conform to expectations of the bipolar behavior of the geomagnetic field (Pavlov et al., 2018;Van der Boon et al., 2022).The age distribution of such unusual behavior is also poorly known.For these reasons, it is important to enhance our understanding of geomagnetic polarity changes in the mid Paleozoic (Silurian-Devonian and Mississippian), which is the least well understood interval in this context, since the early Ediacaran.This work addresses the paucity of magnetostratigraphic records in the Visean (∼347-330 Ma), which is addressed here by data from the late Asbian (∼333 Ma), which is a regional substage of the late Visean defined in Britain.The Visean has no internationally agreed subdivisions, yet formal regional divisions exist (Figure 1).In parallel, we have also improved our understanding of the international correlation of the late Asbian and the older Holkerian substages (Cózar et al., 2022a(Cózar et al., , 2022b(Cózar et al., , 2023a)).We detail here the magnetostratigraphy of the Trowbarrow Quarry section in NW England, UK (Figure 2), which is the stratotype section for the late Asbian (Cózar et al., 2022b), and is also important as a reference section for the apparently synchronous emergent surfaces in carbonate platforms which punctuate the late Asbian in western Europe (Cózar et al., 2022c).These emergent  Geochemistry, Geophysics, Geosystems 10.1029/2023GC011282 surfaces are thought to be related to glacioeustatic fluctuations in sea-level, initiated near the base of the late Asbian.We also examine Visean age units at other localities in south and northeast Cumbria to test the relative age of the magnetization with respect to folding (Figure 2).

Geological Background
This work is largely focused on the Urswick Limestone Formation (Figure 1), although our magnetostratigraphic study overlaps with the top of the underlying Park Limestone Formation and the basal few meters of the overlying Alston Fm in Trowbarrow Quarry (Cózar et al., 2022b).The Urswick Limestone Fm is restricted to south Cumbria and northern-most Lancashire (the South Cumbria Shelf; Figure 2) but has coeval equivalents to the east in the Craven Basin, the Askrigg and Alston blocks (Waters et al., 2017(Waters et al., , 2021)), and north Cumbria (Waters et al., 2011a) (Figures 1 and 2).
The Urswick Limestone Fm is a carbonate ramp succession, which passes southwards into the eastern Irish Sea Basin where more shale-rich successions occur, continuous with coeval successions in the Craven Basin (Wakefield et al., 2016).Within the South Cumbria Shelf, the regional facies changes within the Urswick Limestone Fm are generally well understood through the work of Horbury (1987Horbury ( , 1989)), Horbury and Adams (1996), and Adams et al. (1990).A detailed foraminiferal zonation has allowed clear identification of the base of the late Asbian and the base of the Brigantian (Cózar et al., 2022b; Figure 1).The latter faunal records can be clearly related to the Brigantian boundary stratotype at Janny Wood (Figure 2), 55 km to the NE (Cózar et al., 2023b;Cózar & Somerville, 2004, 2005).The Trowbarrow foraminiferal biozonation is based on the same  Burnett, 1987;Metcalfe & Riley, 2010).Also labeled are the basins and blocks in northwest England.CAI data in south Cumbria-NW Lancashire are largely absent and are assumed to be similar to that in north Cumbria.Numbered localities relate to the additional sample set, are: 1 = Dunnerholme Point, 2 = Dalton in Furness road cuttings, 3 = Barker Scar, 4 = Meathop Quarry, 5 = Whitescar Quarry, 6 = Jenny Browns Point, 7 = Grubbins Wood, 8 = Coastguard Quarry, 9 = Sandside railway cutting, 10 = Sandside Quarry, 11 = Ashfell Edge road cutting, 12 = Little Asby Scar, 13 = Ravenstonedale River.14 = Mouse Gill (details in Table S1 in Supporting Information S1).ST = Stainmore Trough.
Geochemistry, Geophysics, Geosystems 10.1029/2023GC011282 set of paleomagnetic samples described here.Although a hiatus has commonly been inferred at the base of the Urswick Limestone Fm (Adams et al., 1990;Waters et al., 2021), this cannot be demonstrated with biostratigraphy, which suggests (at least at Trowbarrow Quarry) that the basal succession is complete, and any truncation is no greater than at other similar paleokarsts in the Urswick Limestone Fm.The inference of this substantial unconformity is in part related to the prior inadequacy of the biostratigraphic definition of the early Asbian (Cózar et al., 2022b).
We have also evaluated samples from fourteen other localities (1-14 in Figure 2) in Cumbria and north Lancashire to provide data for a regional tilt test.These localities range in age from the latest Tournaisian to late Visean (∼347-332 Ma) with locality 14 in the Namurian at ∼325 Ma (Figure 2, details in Table S1 in Supporting Information S1).Here, the late Tournaisian-Visean successions rest unconformably on folded Ordovician-Silurian clastic successions, both of which are affected by extensional faulting, giving east-and south-directed tilted fault blocks in south Cumbria, and north and northeast tilting in NE Cumbria.This faulting is possibly partly of Late Pennsylvanian age (Waters et al., 1994), although more likely, largely related to extension in the Irish Sea Basin during the Permian-Triassic (Johnson et al., 2001).At present, the Late Permian and Triassic rock rests unconformably on the Carboniferous.In the South Cumbria Shelf area, earlier E-W compression produced some compressional structures and gentle folding, probably related to basin inversion in the Late Pennsylvanian to Early Permian.
The Trowbarrow Quarry is situated within one of these compressional structures, an east-facing monocline, N-S oriented, like the Hutton monocline to the east (Brandon et al., 1998).Late Pennsylvanian basin inversion is a more widespread feature of the Craven Basin (Fraser & Gawthorpe, 2003;Waters et al., 1994), possibly coeval with maximum burial of the Visean successions (Metcalfe & Riley, 2010).In south Cumbria, Late Pennsylvanian-Early Permian uplift may have been of the order of 1.5 km or more (Johnson et al., 2001).

Previous Paleomagnetic Studies in the Visean in Britain and Ireland
Paleomagnetic and magnetostratigraphic studies of Visean volcanic units in the Scottish Midland Valley and Derbyshire have yielded magnetization interpreted to be primary (Wilson & Everitt, 1963;Torsvik et al., 1989;Rother & Storetvedt, 1991;Piper et al., 1991;magnetostratigraphy detailed in Hounslow, 2022), data which has also been widely used in apparent polar wander path construction (Torsvik et al., 2012).Studies on Visean limestones have been less successful due to a strong reverse polarity remagnetization acquired during the interval of the Kiaman Superchron (Late Pennsylvanian-Middle Permian in age), which dominates in many British and Irish Visean age limestones (McCabe & Channell, 1994;Morris, 1972;Wilkinson et al., 2017).
In Britain, early paleomagnetic studies of the Lower Carboniferous focussed on the Craven Basin and Askrigg Block successions (Turner, 1975;Turner et al., 1979;Turner & Tarling, 1975) and largely used spinner magnetometers.These were likely insufficiently sensitive to precisely define the sometimes-weak primary magnetization of these limestones.Nevertheless, the hypothesis of complete re-magnetization of the British Visean limestones advocated by McCabe and Channell (1994) is an over-simplification, since prior measurements clearly showed dual polarity magnetization (Turner et al., 1979;Turner & Tarling, 1975), including downwards-directed SSW magnetization (after tilt corrections), similar to those observed from Visean-age volcanics (Addison et al., 1985;Turner et al., 1979).Even the earliest studies on Irish Visean limestones hinted at the presence of inadequately resolved high-stability components, in addition to the lower-stability Kiaman-age directions isolated during thermal demagnetization (Morris, 1971).Furthermore, in the measurements of McCabe and Channell (1994) from the early Visean and Asbian of the Craven Basin, the characteristic directions have southerly positive and negative inclination groupings with respect to the bedding plane, implying that these "re-magnetizations" are potentially not from a single age interval in the Late-Carboniferous to Permian.
Studies of conodont alteration indices (CAI) have shown that Visean successions in the Craven Basin have CAI values of 2.5-3.5, largely related to maximum burial during the Late Pennsylvanian in the Craven Basin (Metcalfe & Riley, 2010).A CAI value of three corresponds to a possible burial temperature of 110-120°C with assumed geothermal gradient of 30°C/km and 30 Myr burial (Metcalfe & Riley, 2010).On the Askrigg and Alston blocks (Figure 2), larger CAI values of 3-4 occur due to the higher geothermal gradients than in other areas (Burnett, 1987).In the study area, changes in CAI are not known to be related to mineralization or igneous activity.Using both the evidence of some potentially primary magnetization in these limestones, and lower CAI values on the flanks of Cumbria we focussed our study on the areas with lower CAI (Figure 2).CAI values are scarcely known from south Cumbria, with a single CAI value of 2 from the Urswick Limestone Fm near Kendal (Burnett, 1987), and CAI values are assumed to be like the better-studied Visean of northern Cumbria, with a similar burial and uplift history to the South Cumbria Shelf (Kirby et al., 2000;Mitchell et al., 1978).A CAI value of two corresponds to a likely burial temperature of ∼80°C for 30 Myr (Burnett, 1987).

Sampling and Section Details
The main activity was focussed on the Trowbarrow quarry section, where oriented samples were collected with an average stratigraphic spacing of 1.1 m (Figures S1-S3 in Supporting Information S1).These were exclusively hand samples, on which a flat surface was made with a diamond-cup grinder, and oriented using a foot-print style orientation device (Hounslow et al., 2022).Drilling was not allowed at this protected SSSI site.The remaining parts of the samples were used for foraminiferal and petrographic study (detailed in Cózar et al., 2022b) and rock magnetic study.The bedding at Trowbarrow is near vertical and slightly overturned throughout the section (Figure S1 and further details in Supporting Information S1).
Lithologies sampled in the section range from carbonate mudstones to wackestones, packstones and grainstones.Shale intervals and the silty tops of paleosols could not be sampled.Some reddened levels are associated with the emergent surfaces, and 12 samples are in close proximity to these reddened surfaces.Dolomitization is insignificant in this section.However, some emergent surfaces and their paleosols are associated with gray intervals, likely due to the type of paleosol pyritization described by Wright et al. (1997).
Samples from the additional localities 1 to 14 were collected in a similar way (referred to as the "additional set" here), with one to four samples from each locality (details in Table S1 in Supporting Information S1).These localities include a variety of bedding dips, from shallow to steep dips (i.e., localities 7, 8, 14) to evaluate a regional tilt test, in combination with the data from Trowbarrow (bedding dips at Trowbarrow are not sufficiently varied to evaluate an individual fold test).The additional samples were predominantly limestones, and dolomitized limestones, but included some red and brown sandstones (localities 2, 14).

Paleomagnetic Methods
Measurements of Natural Remanent Magnetization (NRM) were made using a 3-axis 2G Enterprises RAPID cryogenic magnetometer at Lancaster University, which has a noise floor ∼2 × 10 12 A.m 2 (∼0.3 μA/m; with holder correction).The magnetometer was housed in a large Helmholtz cage for field cancellation, which together with the Mu-metal shields reduced the Earth's magnetic field to around 0.03-0.08μT at the demagnetizer coil position.Specimen magnetization measurements were performed using a "measurement blank" (i.e., a blank subjected to all the same steps).This bypassed the RAPID "standard holder-correction procedure."The GM4Edit software was used for the blank-type corrections to the data (Hounslow, 2019).Holder magnetizations for NRM measurements have a median of 3.5 × 10 12 A.m 2 (1σ = 7 × 10 12 A.m 2 ).The additional set was measured at Liverpool University, using a similar magnetometer and set-up (but standard holder removal) with a noise floor ∼1 × 10 11 A.m 2 (∼1 μA/m; with holder correction) and median holder magnetization of 13 × 10 12 A.m 2 (1σ = 36 × 10 12 A.m 2 ; data based on these measurements).Some specimens were subjected to stepwise thermal demagnetization, using a Magnetic Measurements Ltd, MMTD1 thermal demagnetizer, in 100-50°C steps up to 700°C.Magnetic susceptibility measurements (using a Bartington MS2 meter, at 0.47 kHz) were performed after each heating step to monitor mineralogical changes caused by the heating, which were often a major problem.Many specimens were measured by a combination of thermal demagnetization up to around 250-400°C followed by static AF demagnetization on the RAPID (up to 95 mT).This combined procedure limited thermal alteration, which often obscured the characteristic remanent magnetization (ChRM).This alteration limited the useful upper limit to thermal demagnetization for many samples.No specimens were measured using AF only.Specimens were held in Mu-metal shielded boxes and larger Mu-metal shields between measurements.At Lancaster University, from the 30 mT AF step onwards, the GRM corrections procedure (ZD-method) of Stephenson (1993) was applied to all data to limit the impact of GRM acquisition.Post-processing of these data was applied using GM4Edit software (Hounslow, 2019).Liverpool University used the standard RAPID set-up and demagnetization (static AF to 95 mT), but many of the Geochemistry, Geophysics, Geosystems 10.1029/2023GC011282 HOUNSLOW ET AL. weaker samples were inverted and re-measured to improve accuracy.Many were also tumble-demagnetized up to 200 mT on an AGICO demagnetizer.
The specimen demagnetization results were analyzed using principal component analysis using LINEFIND (Kent, 1985) as a DOSbox version (Hounslow, 2023a).LINEFIND is more sophisticated than conventional analysis (Heslop & Roberts, 2016;McFadden & Schmidt, 1986), since it has statistical tests for linearity and planarity, uses a fitted model of the measurement variance, and determines α 95 of lines and planes rather than maximum angular deviation values.A tuning parameter (ρ) is used in combination with visual interpretation and Akaike's information criteria in selecting ρ (details in Kent, 1985;McFadden & Schmidt, 1986;Hounslow, 2023a).Great circle type behavior was seen in many specimens during demagnetization, and behavioral classes S (for linear line fits) and T (great circle behavior) were assigned to specimens (like used by Montgomery et al. (1998) and Hounslow et al. (2004)).These were further sub-divided into three qualitative classes (S1, S2, S3 and T1, T2, T3; with class 1 the top quality).For S-class behavior, the directional scatter and length of the fitted line (in the Zijderveld projections) were used for qualitative classification.For T-class behavior, the scatter and approach to the expected ChRM direction was used, with the T1 class having a closer approach but without reaching a stable endpoint (illustrative examples in Figure S8 in Supporting Information S1).The relationship of these qualitative classes to the α 95 of the S-class lines and T-class poles to the great circles is shown in Figure 3a.Directional data for 1.8% of the weakest intensity specimens (often <1 μA/m) was too erratic, and these were assigned as unused data (X-class).This X-class was also assigned to specimen data in which alteration during thermal demagnetization totally obscured the ChRM (3.8% of specimens).
For the specimens from Trowbarrow, virtual geomagnetic pole (VGP R ) latitudes (Hounslow et al., 2022) were determined with respect to the mean VGP direction (S-class data sets) for sub-divisions of the section in Table 1.Values of VGP R latitude near +90°indicate a specimen with normal polarity and values near 90°indicate a reversed polarity specimen.In addition to reversal tests (McFadden & McElhinny, 1990), the site mean VGP A95 min and A95 max thresholds of Deenen et al. (2014) were used as a measure of the capture of secular variation in the directional datasets.Based on the approach to expected Carboniferous directions and the qualitative confidence attached to that polarity inference, specimens were also assigned a polarity quality class R, R?, R??, N??, R??,N??,N? or N, to indicate either reversed or normal polarity status.R? or N? indicate a polarity assignment of middle quality, compared to R or N of top quality, and R??, N?? of lowest quality.A class of "?" indicates a specimen could not confidently be assigned a polarity.The relationships of the VGP R latitude and polarity classifications to the demagnetization behavior classes are shown in Figures 3b-3d.

Rock Magnetic Methods
For the Trowbarrow set, magnetic mineralogy was investigated using one sister specimen from each sampled horizon.This used static anhysteretic remanent magnetization (ARM) imparted with a 100 μT DC field, and 80 mT AF field and subsequently its static AF demagnetization at 40 mT on the RAPID (all along Z-axis only).The hardness of the ARM (Peters & Thompson, 1998) was calculated as %d.ARM 40mT = 100*(ARM -40mT / ARM80 mT ).The RAPID blank-type corrections remove the ARM magnetization and demagnetization of the quartz glass rod during these experiments.Subsequently determined on the same specimen was isothermal remanent magnetization (IRM) at 1000 mT, with 40, 100, 300 and 1000 mT backfield measurements to determine the S-ratio (-IRM 0.1T /IRM 1T ), the HIRM (IRM 1T -IRM 0.3T ) and the coercivity of remanence (B cr ) was estimated by linear interpolation.IRM was measured on a JR6 spinner magnetometer (with cross calibrated intensity to the RAPID) using procedures in Walden (1999).Also, rotational remanent magnetization (RRM) and rotational ARM (ARM ROT ) were measured on nine specimens at a rotation rate of 2.5 Hz, using the procedures in Hounslow et al. (2023) to characterize the single domain (SD) magnetic mineralogy.In these cases, the diagnostic indicator determined is the effective gyromagnetic field Bg.The RRM of two of these specimens could not be reliably measured.Two specimens from the additional set were investigated with a Quantum Design Magnetic Properties Measurement System (MPMS-5S) at the Institute for Rock Magnetism at the University of Minnesota (detailed in Figure S5 in Supporting Information S1).Additional details of the stratigraphic variation of the rock magnetic data at Trowbarrow are in Hounslow, Cózar, et al. (2024).A subset of the Trowbarrow samples were measured for carbon isotopes to aid in the longer-term correlation to other sections without such a detailed foraminiferal biostratigraphy (Figure S4; details in Supporting Information S1).

Magnetic Mineralogy
Some 82% of the limestones have magnetic susceptibility <0 (Figure 4c), so they are diamagnetic.Positive susceptibilities are not particularly confined to any particular carbonate lithology other than the two rudstone samples.Instead, positive magnetic susceptibility is mostly related to more shaley and argillaceous stratigraphic intervals, such as the Woodbine Shale and those in the Alston Fm (Figure 4c).NRM intensity is largely independent of magnetic susceptibility, suggesting that these largely express different magnetic phases (Figure 4b).Specimens show a large range in S-ratio and B cr , indicating that there is a strong mixture between hard IRM behavior with B cr >100 mT and soft behavior with S-ratio 0.4-0.8(Figure 5a).The largest proportion of samples have S-ratio < 0.6 (histogram in Figure 5a) There is no strong correspondence with lithology, but rudstone (only 2 samples), mudstones, and argillaceous packstones have the larger positive S-ratios, and grainstones the smaller (i.e., negative) S-ratios on average (Figure 5a).Since pure hematite has B cr typically 100-800 mT (Peters & Dekkers, 2003), and there is limited evidence in thermal demagnetization of goethite (but see Figure S5 in Supporting Information S1), the relationship in Figure 5a is inferred to reflect a variable magnetic mixture of hematite and a softer phase.Some 68% of specimens have B cr >100 mT, which shows that samples are largely hematite dominated.Since only some 7% of the sample set can be directly associated with observed reddening at the emergent surfaces (Figure 4b), the hematite has been contributed to these limestones in other ways, in addition to directly at emergent surfaces.The HIRM and IRM 40mT are proxies for the total contributions from the hematite and soft component (Walden & White, 1997) and are weakly co-related (Pearson R 2 = 0.41, power-relationship in Figure 5b), indicating some connection between their supply.Samples associated with the emergent surfaces have variable but generally larger HIRM (Figure 5b), confirming the field data and petrographic observations of Horbury (1987) that the reddening at these surfaces is associated with hematite.Wright et al. (1997)    one hypothesis, namely that this may come about by "terra-rossa-type" insoluble residuals during karst formation, although evidence for this is sparse (Hounslow, Cózar, et al., 2024).

have inferred
The Silverdale-Arnside area around Trowbarrow is not impacted by the hematite-mineralization like that seen in the Furness Peninsula in southwest Cumbria (Rose & Dunham, 1977; Figure 2).However, trace amounts of hematite mineralization do occur in the nearby area as: (a) micron-scale coatings along some joint surfaces (probably mostly from Holocene karstification, since it is a near-soil-surface feature).This is fairly sporadic in its distribution, but occurs in parts of the Park Limestone at Trowbarrow in those parts of the sections below that sampled.Since Trowbarrow is a deep quarry, this is not seen in the sampled parts of the faces; (b) Rarer mm-sized pods and veins of hematite, which can be associated with dolomite veining, are sometimes seen to amalgamate and grow from joints.This type of veining seems to be younger than the more common white ferroan-calcite calcite veining described by Horbury (1987).This dolomite-hematite veining is common in sections in the Furness Peninsula (localities 1, 2; Figure 2) but has not been seen in the Trowbarrow samples; (c) More substantive hematite mineralization in fault networks and calcite-dolomite-filled fracturing/faulting.Sections at Trowbarrow are not affected by this to any extent.Hence, these macroscopic forms of hematite mineralization are unimportant in the Trowbarrow samples.Also, the rare partial dolomitization in thin sections concurs with the largely primary nature of the carbonates.
The soft magnetic component appears to be in part associated with argillaceous material, with the median IRM 40mT being larger in the argillaceous wackestones and packstones (Figure 5b).Since hematite does not significantly acquire ARM (Peters & Thompson, 1998), this measurement provides a window into characterizing the soft component in these mixed-mineralogy samples.%d.ARM 40mT is typically 30%-40%, as seen in harder magnetites found in deep-sea sediments, older sediments and reference rocks (Figure 6a).These are unlike the soft (%d.ARM 40mT <20%) ARM behavior seen in multi-domain (MD) pyrrhotite (Figure 6a) or in MD titanomagnetite, as seen in less-oxidized basalts (Figures 6a and 6b).Unoxidized titanomagnetites are fairly uncommon in sediments due to pre-sedimentation oxidation and diagenetic modifications (Roberts, 2015) and seem unlikely in these limestones deposited under oxidizing conditions.Likewise, greigite and hard pyrrhotite have %d.ARM 40mT larger than 45% (Hounslow et al., 2023;Peters & Thompson, 1998), indicating that these phases do not occur in significant proportions in the Trowbarrow limestones.Therefore, the soft behavior in these samples is either due to magnetite or an intermediate stability pyrrhotite (<20 μm in grain size; Figure 6b).McCabe and Channell (1994) also detected no pyrrhotite low temperature transition in their Craven Basin samples.
The ratio of saturation IRM and χ ARM is indicative of particle size and interaction in low Ti magnetites (Moskowitz et al., 1993;Oldfield, 1999;Peters & Dekkers, 2003).However, because of the additional presence of hematite in the Urswick Limestone Fm, we use the IRM 40mT /χ ARM as a comparable proxy and compare this to available reference materials (Figure 6b).The bulk of the Urswick Limestone Fm samples fall at the lower IRM 40mT /χ ARM end of the natural magnetite and Indian Ocean reference magnetite materials.Reference material of intermediate and high stability pyrrhotite likely has IRM 40mT /χ ARM > 2 k A/m between the synthetic pyrrhotite and natural detrital pyrrhotite <20 μm in size (Figure 6b), perhaps similar to four samples from Trowbarrow.
The few successful samples measured for RRM indicate small positive and negative Bg values, and ARM ROT / ARM less than 0.6 (Figure 6c).These are similar to reference materials of magnetite-bearing siliciclastics and magnetofossil-bearing chalks (Hounslow et al., 2023).The Trowbarrow data indicate distinctively different RRM behavior to the Taiwanese pyrrhotite (ARM ROT /ARM >1.2), and the Taiwanese greigite with Bg < 10 μT.However, the unusual behavior of the Urswick Limestone Fm samples is the dominance of specimens with Bg > 0, which was restricted to only a few magnetite-bearing samples tested by Hounslow et al. (2023).This may be because of the incompletely known range in Bg for natural magnetite under the test conditions.One sample was also tested for variation with rotation rate, which did not change this conclusion (Figure S6 in Supporting Information S1).The two specimens from the additional set tested with low temperature measurements suggest a weak Verwey transition indicative of some unoxidized magnetite (Figure S5 in Supporting Information S1).
In summary, the magnetic mineralogy in the Urswick Limestone Fm is dominated by hematite, with a variable contribution probably from fine-grained non-interacting magnetite, which is probably detrital in origin.The hematite is Carboniferous in age and likely deposited with the carbonates, but in a small number of cases may be generated via emergent surface formation (Hounslow, Cózar, et al., 2024).Based on correspondence with carbonate cycles at Trowbarrow, and a feasible environmental model of the non-carbonate inputs, the bulk of the hematite was likely delivered as eolian dust from far-field sources (Hounslow, Cózar, et al., 2024).There is a Geochemistry, Geophysics, Geosystems 10.1029/2023GC011282 HOUNSLOW ET AL.
speculative possibility that the soft component may in part be of magnetofossil origin (i.e., with low ARM ROT / ARM; Figure 6c).Approximate estimates of the relative mass contribution to the IRM can be obtained using the hematite-magnetite mixing experiments of Frank and Nowaczyk (2008) (see scale on upper Figure 5a).A detrital origin of magnetite for Asbian limestones from the Craven Basin was demonstrated by Turner (1975).Their optically observable magnetite was intimately associated with ilmenite and hematite and had B cr of 33-37 mT and showed the Verwey transition.Pyritization of these was also typical.The same may apply to the soft phase here, since pyrite is common in the Urswick Limestone Fm (Horbury, 1987).The additional set likely has a mineralogy similar to those at Trowbarrow.

Paleomagnetic Results
The response of the specimens to demagnetization depended on their magnetic mineralogy.For those that were dominated by hematite, thermal demagnetization was used up to about 650-700°C.For those not dominated by hematite, a combination of thermal demagnetization from 250 to 450°C (300°C most commonly) followed by AF demagnetization worked best, since this allowed better isolation of the small ChRM (see Figure S9 in Supporting Information S1 for demagnetization diagrams).Thermal demagnetization alone on these types of samples often obscured the ChRM.This was the prime reason Palmer (1987) failed to identify any primary magnetization in Visean limestones from this region.The alteration crisis identified by the susceptibility measurements was commonly around 350°C, but lower (∼200-250°C) in darker wackestones, and up to 400-500°C in some more thermally stable grainstones.

Trowbarrow Quarry Set
The specimens generally displayed three components, first a low stability (LT) component was generally removed by ∼180-200°C with some larger unblocking temperatures for a few samples (Figure 7a).Some 80% of specimens contained this component.Second, an intermediate stability component (K-component) was often the strongest component present, displaying southerly directed magnetization of shallow inclination (∼±10°) in stratigraphic coordinates and mostly negative inclinations in geographic coordinates (Figure 7b).In the 66% of specimens in which hematite dominated the magnetization, the K-component was weaker or absent.Some 62% of all specimens displayed the K-component clearly enough for a PCA line fit.Third, a dual polarity ChRM (Figure 8a) was usually the highest stability component.Some 34% of all specimens displayed a component suitable for a PCA line fit (i.e., S-class data), with the remainder (55% of specimens of T-class) displaying this component as great circle trends toward the ChRM directions (Figure 8b).The strength of the S-class ChRM component is largely related to the initial NRM intensity (Figure 3b), although in many non-hematite-dominated specimens the initial appearance of the ChRM was typically below an intensity of 100 μA/m.
The LT component is typically quite small in intensity, and most often shows a very well-defined break on Zijderveld diagrams from the intermediate stability K-component.The LT component has a fair scatter, with a mean direction (geographic coordinates) of 356, +73, α 95 = 2.2, k = 14, n = 316 (Figure 7a).This is interpreted as primarily a Brunhes-age component (geocentric axial dipole inclination at site = 70°).A very similarly behaved Brunhes-age component was reported by McCabe and Channell (1994) from the Craven Basin.
The intermediate stability K-component is evenly distributed throughout the section (Figure 4d) and this dominates in 4.3% of specimens to the extent that no ChRM could be detected.Using thermal demagnetization, the Kcomponent was most often seen between 200 and 400°C, but lower and higher unblocking ranges were also seen (Figure 7d).The upper temperature range of the K-component is around 300-400°C with the median for hematite dominated magnetization at ∼400°C and for magnetite-dominated magnetization ∼300°C.McCabe and Channell (1994) report a similar typical 200-450°C unblocking range for their inferred Kiaman-age component from the Craven Basin limestones.For those subjected to the composite demagnetization scheme, the K-component is most commonly seen from the 5-20 mT steps until AF steps to 60 mT or higher.However, the upper AF step that removes the K-component has a wide range in the AF field (Figure 7c).The K-component does not have a strong bias with S-ratio, having a distribution similar to the overall sample set (Figures 7e and 5a).This suggests that it is carried by both hematite and magnetite.The K-component has a mean direction (in geographic coordinates) of 182.3°, 14.4°, α 95 = 1.9°(k = 24.3,n = 244; VGP = 43.1°,354.4°; Figure 7b) and is interpreted as a Kiamanage partial remagnetization, similar to that widely reported from other Visean limestones and pre-Carboniferous rocks in Britain (Channell et al., 1992;Palmer, 1987;Piper & Crowley, 1999;Setiabudidaya et al., 1994).The paleopole derived from this mean is similar to those from other Kiaman remagnetizations elsewhere in Europe (Figure 9), which are mostly a little to the east of the stable Europe apparent polar wander path of Torsvik et al. (2012).Geochemistry, Geophysics, Geosystems

10.1029/2023GC011282
The dual polarity ChRM is nearly always the highest stability component (Figure 8c) seen in thermal demagnetization.In geographic coordinates, this dual polarity direction has an axis at ∼026°with inclination of ∼35°.
The ChRM is commonly observed going to the Zijderveld origin, particularly so for isolation using AF demagnetization (Figure 8c).Thermal alteration at >600°C may obscure origin fits, so end-ranges in this case are more commonly 500-650°C (Figure 8c).The line-fit (S-class) ChRM data is somewhat biased toward the most negative S-ratios (Figure 8d).A subset of 7 specimens shows S-class line fits for the ChRM prior to isolation of the K-component.Some 39% of specimens show no clear K-component but have the ChRM in the mid stability range (i.e., thermal upper range 200-400°C and AF upper range <40 mT; Figure 8c).This somewhat unpredictable overlap in the stability range of the K-component and the ChRM is likely responsible for the large number of specimens with great circle type behavior.Notably, those few showing only the K-component (4.3% of total) or the T3 class of great circle trends (17.1% of total) tend to be biased toward the magnetically softer-end (most positive) of the S-ratio classes (Figure 7e).For a normal polarity ChRM, this stability overlap is characteristically expressed as a great circle trend toward the NE, and sometimes an increase in intensity, as the approximately opposite K-component is removed.A particular feature of some specimens subjected to combined AF demagnetization is a high stability tail (Figure 3d), in which an AF-resistant phase (i.e., hematite) is evidently carrying the ChRM but cannot be fully demagnetized with the composite scheme.This behavior also contributes to the dominance of the great-circle type behavior.
The Trowbarrow data has been divided into three stratigraphic intervals for the purpose of defining directional sets to be used in the fold test of the ChRM (Table 1) and two sets for fold tests on the K-component (Table 2).This was to enable rather closer sample-group sizes for the fold test and parametric bootstrap-based estimates of the likely unfolding percentage.The whole-section mean is probably more applicable to a paleopole estimate (Table 1).
Fisher mean directions using the S-class data generally fail the reversal tests, with inverted dual polarity directions ∼10°different.Nevertheless, the mean VGP, A95, is within the thresholds of Deenen et al. (2014), indicating that dispersion is within the expected secular variation range (Table 1), and the dispersion (precision parameter k) is similar in the dual polarity sets (Figure 8a).The reversal test failure could be due to a small fraction of the Kcomponent remaining in the ChRM directions (see Section 3.3).A combined great circle mean (method of McFadden & McElhinny, 1988) was also determined for the three intervals independently (Table S3 in Supporting Information S1).For the T-class specimen data, the directions from this were used to define site (i.e., sample) means, since generally, 2-3 specimens were measured from each sample.This mean displays similar but slightly shallower mean directions than the S-class means (Table 1, Table S3 in Supporting Information S1) and is closer to antipodal.The paleopoles derived from the S-class directions (and site means) are similar to other Visean volcanic poles from stable Europe (Figure 9), supporting the reliability of the directional analysis.2012) partitioned into age intervals (using their timescale).Other paleopoles are selected Mississippian (m1 to m11) and Pennsylvanian (p1 to p7) poles and inferred Kiaman-age reverse remagnetizations (k1 to k7).Also shown are the paleopoles of the Cumbrian and Isle of Man hematite mineralization's (H1 to H3) of Rowe et al. (1998) and Crowley et al. (2014).Inverted azimuthal equidistant projection.See Table S5 in Supporting Information S1 for sources of other data in this figure.

The Additional Sample Set
Demagnetization behavior of the additional set was similar to that from Trowbarrow Quarry (Figures S7, S10 in Supporting Information S1).In these also, both the LT and K-components were present alongside a dual polarity ChRM, which in most cases, was only a small fraction of the initial remanence (Figure S10 in Supporting Information S1 shows demagnetization diagrams).The remanence intensity was similarly often dominated by the Kcomponent.Of the 103 specimens measured (42 samples), 51% displayed evidence of a Carboniferous magnetization (17% S-class; 34% T-class specimen data).When grouped by localities, the ChRM mean directions are similar to those at Trowbarrow (Table 2; Table S3 in Supporting Information S1).Of these specimens, 47% were reverse polarity and 53% were normal polarity.
Some 25% of specimens displayed only evidence of the LT component or the K-component (9% and 16% respectively) without any evident ChRM.These overprints were scattered throughout the localities but were dominant at localities 1 and 10.The remaining 22% of specimens were strongly impacted by strong thermal alteration during thermal demagnetization or erratic directional behavior.The mean direction (in geographic coordinates) of the LT component is 351°/+75°(n = 71, k = 11.6,α 95 = 5.2°), much like that seen at Trowbarrow.
These additional localities give a (sample) mean K-component direction (without locality 2) of 190°/ 20°(α 95 = 3.9°, k = 16, n localities = 5) some 9°d ifferent from that seen at Trowbarrow (Table 2; Figure S11b in Supporting Information S1), which is significant using the test of McFadden and McElhinny (1990).In addition, the sample-based means of the K-component show a fair degree of dispersion (Table 2; Figure S11c in Supporting Information S1), which together could suggest some post-Permian tilting or rotation.However, fold tests on the K-component indicate that this component was acquired close to zero percent untilting (Table 3), with 95% confidence limits on tilting, which include zero untilting (Figures S14 and S15 in Supporting Information S1).The Fisher block rotation fold test of Enkin and Watson (1996), which accounts for vertical axis rotations, gives better grouping at zero untilting.
Fold tests were performed using ChRM directions recovered from the additional and Trowbarrow sets combined.These used different data groupings to the K-fold test, since; (a) ChRM and K-components were not evenly distributed across the localities, (b) bedding dips varied at most localities, and (c) the requirements of the parametric bootstraps on the fold tests require minimum numbers of data.Consequently, some localities were regrouped.The fold tests support the conclusion that the ChRM was acquired pre-tilting (Table 3; Figures S12 and  S13 in Supporting Information S1).The rocks from locality 14 are notably younger than the remaining, largely Visean localities, but excluding this locality does not significantly impact the fold test.A fold-test was not possible using only the Trowbarrow dataset, since bedding dip dispersion is too small.

Magnetostratigraphy at Trowbarrow
The section is dominated by normal polarity with the 182 sampling levels comprising 76.4% normal, 20.9% reverse and 2.8% of unknown polarity (Figure 10).The inferred polarity quality is generally unevenly distributed with demagnetization behavior with N, R and R? groups containing the bulk of the S-class behaviors, and the other groups the bulk of the T-class data (Figure 3b).Conversely, the VGP R latitudes are only weakly related to the polarity quality groups (Figures 3c and 3d) due to their dependence on angular dispersion from the mean pole, something which is more weakly represented in the qualitative polarity quality groups.The VGP R latitudes have a similar dispersion between normal and reverse polarity (Figures 3c and 3d; 10) suggesting that any possible directional contamination is equally represented in both polarities.Nine reverse magnetozones (TQ0r to TQ5r.1r, TQ5r.2r to TQ8r) are defined with two or more adjacent samples, each with multiple specimens (Figure 10e).Seven sub-magnetozones (TQ2n.2r,TQ4n.1r,TQ5n.1r,TQ5r.1n,TQ7n.1r,TQ7n.2r and TQ9n.1r) are defined with multiple specimens from a single sample.Two rather more tentative submagnetozones are defined with one or two lower quality specimens (TQ2n.1r,TQ9n.2r; Figure 10e).Cózar et al. (2022c) have proposed a set of regionally correlated (between Ireland, northern England and south Wales) emergent surfaces, with Trowbarrow as the reference section (Figure 10a).Independent means of assessing the missing time in the rock record at these emergent surfaces could be based on a variety of generalized modeling and observational data (Barnett et al., 2002;Rygel et al., 2008), and cyclostratigraphy (Hounslow, Cózar, et al., 2024).If the magnetic polarity changes across the emergent surface, this suggests a possible missing interval, since there is no reason to suppose that changes in polarity and drops in sea-level should be coincident (unless by chance).Four of the emergent surfaces (O'a, II, Iva and XII) display a switch in polarity across these surfaces, suggesting that missing intervals may be present at the bases of magnetozone TQ2n, TQ2n.2n,TQ4n.1r, and TQ9n.2n.Also, the complex of paleosols around emergent surface X, show a change in polarity across the overlying paleosol at 129.5 m (base of TQ7n.2r).Other emergent surfaces do not coincide with polarity changes, so polarity offers no evidence of likely missing intervals.The base of TQ8n occurs some 1.5 m above surface XI, possibly coincident with a prominent burrowed surface some 70 cm below sample TQ142.Emergent surface VII is the most intensely mamillated surface and has the thickest silty paleosol in the quarry; therefore, it may Note.Set is using either the specimen-data or grouped data (partitioned by bedding dips) as in Table 2. %U = most-likely unfolding percent and 95% confidence limits on unfolding displayed in [ /+.].p 0 , p 100 are the probability of exceeding the f-statistic for the 0% and 100% unfolding cases respectively, that is, values <5% indicate the 0% or 100% unfolding scenarios are unlikely.%s = the acceptable unfolding percent for a synfolding solution.95% confidence limits determined with 1,000 bootstrap simulations.Ng = number of site groups (with similar dips), N = number of specimens.The ChRM data from Trowbarrow (TQ site) was grouped into three sets, the Lower Urswick Limestone Fm, Upper Urswick Limestone Fm and the Humphrey Head Mb + Alston Fm.For the ChRM tests locality 2 was not used in the McFadden (1998) fold test since it has too few points.Using Pmagtool v.5 (Hounslow, 2023b).The DC foldtest of Enkin (2003) cannot be easily applied here since the bedding dips vary in the groups.Progressive unfolding is the test of Watson and Enkin (1993), and Fisher block rotation that of Enkin and Watson (1996).
represent the longest hiatus using the criteria of Wright (1994), although the polarity does not change at it.Sedimentation rate models for the section based on cyclostratigraphy suggest that surfaces II, VI and VII are the most important hiatuses in the section (Hounslow, Cózar, et al., 2024) with missing intervals of ∼30 ka, ∼50 ka and ∼19 ka, respectively.However, the eccentricity-modulated cyclostratigraphy is unable to reliably detect hiatuses <∼10 ka.Hence, of these three, only surface II could be indicative of a more substantial truncation of TQ2n.2n or TQ2n.2r (Figure 10e).

The Origin of the K-Component
Fold tests on the K-component suggest that it was acquired close to zero untilting.This test is principally one using the steeper bedding dips (associated with the Silverdale monocline at Trowbarrow and locality 9) compared to the less steeply dipping strata in other sections.However, there are several potential phases of tilting in this region.These are principally related to Late Pennsylvanian-Early Permian basin inversion, and later extensionrelated tilting in the Permo-Triassic, associated with the formation of the Irish Sea Basin and the Craven Fault systems (Kirby et al., 2000).If the K-component was acquired before the Permian-Triassic, it is possible Permian-Triassic tilting introduced some additional dispersion in the K-component locality means.Any Permo-Triassic tilting has the most impact at localities 1 and 2, since these occur in the Furness Peninsula, which is more severely affected by the boundary faults of the Irish Sea Basin (Kirby et al., 2000).In the Craven Basin, fold tests of the K-component are principally associated with datasets from the Clitheroe anticline (McCabe & Channell, 1994) and the Skipton anticline (Palmer, 1987; dataset reworked here), with both indicating that the Kcomponent was acquired synfolding (at 63% and 50 ± 17% unfolding, respectively).Since both these anticlinal structures are Variscan in origin (Kirby et al., 2000), it seems unlikely that Permo-Triassic extensional faulting in the Craven Basin added further additional tilting to these structures.
To examine the mineralogical origin of the K-component, NRM intensity and rock magnetic datasets were partitioned into specimen sets which contained no K-component, contained a ChRM and the K-component and contained only the K-component (Figure 11).Many proposed remagnetization mechanisms involve introduction of new magnetic phases (Elmore et al., 2012), which should give enhanced NRM intensity and magnetic mineral abundance proxies for specimens with remagnetizations.For the Trowbarrow specimens the NRM intensity for those containing the K-component is not enhanced over those without (Figure 11a), although the intensity from magnetite-dominated samples is rather more focused around 10-60 μA/m and 100-400 μA/m, with the highest intensity specimens more represented in the group without the K-component.The χ ARM shows similar relationships, with more of a bias toward middling to large χ ARM for those specimens containing the K-component (Figure 11b).Significantly, the K component is present in both magnetite-and hematite-dominated magnetization (Figure 11a).These relationships which would tend to favor a thermoviscous (TVRM) origin (Kent, 1985;Dunlop et al., 1997) of the Figure S16 in Supporting Information S1).However, it is conceivable that remagnetization could result from replacement or transformation of existing particles without modifying the Fe-oxide abundance.Processes such as annealing of vacancies (in hematite or magnetite) or magnetite to maghemite oxidation during prolonger burial at low temperatures could conceivably produce burial-related magnetization.For those containing the K-component, IRM 40mT /χ ARM values are more focused within the mid part of the range and conversely for those specimens without the K-component in the larger IRM 40mT /χ ARM values (Figure 11c).This indicates that the grains that may be carrying the K-component are not focused on either end of the magnetite particle size range.For the magnetite-dominated magnetization, the NRM intensity is related to magnetite abundance (as measured by either IRM 40mT or χ ARM ; Figure 11d), which is in turn strongly related to the siliciclastic content in the limestones (Hounslow, Cózar, et al., 2024).Perhaps importantly, NRM intensity to IRM 40mT correlation improves (larger R 2 ) and steepens through the three categories in Figure 11d.These conditions suggest that the K-component is preferentially held by magnetic grains associated with the siliciclastic fraction.
One of the pitfalls of using laboratory remanences with sediments is that the laboratory magnetization may activate that proportion of the magnetic fraction which are not carrying the natural remanence.Here, the natural remanence is <0.4% of the IRM 40mT (Figure 11d) and <∼0.2% for χ ARM (χ ARM = ∼2x IRM 40mT ).In addition, magnetic particle populations in siliciclastics tend to occur as both discrete fractions (not included in other detritals), and an included fraction, which is found in the much more abundant silicates (Franke et al., 2007;Hounslow & Maher, 1996;Hounslow & Morton, 2004), although the included fraction may not carry much natural remanence (Chang et al., 2016).These will respond differently to sulfidic diagenesis, with the discrete fraction being preferentially removed and the included fraction preferentially retained with greater sulfidicrelated losses.The included fraction could therefore be a reservoir of Fe-oxides grains, which carry little or no natural remanence, but a potential larger reservoir of grains to acquire TVRM.Hence, a working hypothesis for the magnetization in these limestones, is that the ChRM component is largely carried by a discrete magnetic fraction (perhaps with some addition from Fe-oxide inclusions), and the K-component is disproportionately carried by the silicate-hosted fraction of hematite and magnetite as a TVRM.Diagenetic and siliciclastic sourcecontrolled variations in the relative abundance of discrete and included Fe-oxides could therefore explain the variable preservation of the ChRM and K-component.Cózar, et al. (2024).Histogram intervals were chosen to give an approximate even spread through the whole respective datasets.

The Mineralogical Origin of the ChRM
The fold tests indicate that the ChRM component was acquired close to 100% untilting.This together with the VGP locations that are comparable to Mississippian volcanic VGP's (Figure 9), and the recovery of dual polarities suggests that the ChRM component is an Early Carboniferous magnetization.
A surprising feature in the Trowbarrow succession is the dominance of hematite as the carrier of the magnetization, which is more widespread than could be inferred from the localized reddening observed at many of the emergent surfaces.An estimate of the average weight% of hematite is 0.07% (min, max = 0.001, 1.1%; with Sratio < 0.5) using a saturation remanence of 0.18 A.m 2 kg 1 (Peters and Dekkers, 2003).There are three possible sources for the hematite.First, as "terra-rossa type' residues associated with emergent surface development, as described by Wright et al. (1997).Such sourced grains would need to have percolated downwards through the carbonates via the pore networks during emergence to account for its presence meters below the emergent surfaces.However, a carbonate-hosted dissolution source for hematite as a residual phase has been discounted for modern carbonate systems, where the amount of limestone dissolution required is beyond realistic limits (Jones, 2021;Muhs and Budahn, 2009).Second, as eolian sourced material either from volcanic dust or as dust supplied from nearby desert margins (Jones, 2021).Sources of terrestrial dust could be nearby Visean uplands (Wakefield et al., 2016) or more far-field sources (McGlannan et al., 2022).Hounslow, Cózar, et al. (2024) have proposed this as the primary source of hematite in the Trowbarrow carbonates.Third, dispersal from coeval laterally present paleosols, since the carbonate ramp probably had some topography (e.g., Eberli, 2013), and also likely shallowed to the north against the Cumbrian high (Mitchell et al., 1978).In this scenario, the silty paleosoltops would be prone to reworking during the immediately overlying flooding event, from which siliciclastics would be subsequently dispersed laterally into developing carbonate deposits.
Either of the latter two possibilities could produce detrital hematite carrying a primary Carboniferous magnetization.In contrast, the first option would generate a post-depositional (but early) magnetization timed to the overlying paleosol and emergence.Where the polarity changes across the emergent surface (i.e., at base of TQ2n, TQ2n.2n,TQ4n.1r, and TQ9n.2n), the underlying magnetozone is often rather thin.In this case, using the emergent-surface sourced scenario would suggest that any down-wards percolation of hematite was of limited vertical extent.

Magnetostratigraphic Comparisons
In Europe, the only detailed coeval magnetostratigraphy to that seen at Trowbarrow is from the lava succession at Kinghorn, near Kirkcaldy, Scotland (Wilson & Everitt, 1963;Torsvik et al., 1989;summarized in Hounslow, 2022).This lava succession is 390 m thick, with the First Abden Limestone immediately below the top-most lava, and the second Abden Limestone some 20 m above the uppermost lava.The First Abden Limestone is near the top of the early Brigantian, and the Second Abden Limestone is near the base of the Serpukhovian (Cózar & Somerville, 2020, 2021).The lowest part of the miospore Bellispores nitidus-Reticulatisporites carnosus (NC) Zone can be placed above the lava succession within the Second Abden Limestone (Brindley & Spinner, 1989;Owens et al., 2005).In northern England and Scotland, the base of this zone is typically in the mid Brigantian, with the underlying Tripartites vetustus-Rotaspora fracta (VF) Zone extending into the latest Asbian (McLean et al., 2018).A secure underlying marker at Kinghorn is the Birdiehouse Limestone (some 190 m below the base of the lavas; Figure 12), which is a well-defined marker in this region (Guirdham, 1998) in the lower part of the NM biozone (Brindley & Spinner, 1989;Rex & Scott, 1987), with the base of the NM biozone placed in the early Asbian (Owens et al., 2005).The strata between the Birdiehouse Limestone and the lower part of the Kinghorn Volcanics belong to the Raistrickia nigra-Triquitrites marginatus (NM) Biozone (Brindley & Spinner, 1989).
The NM biozone is dated to most of the Asbian.Therefore, the top part of the lava succession should be early Brigantian with the Asbian-Brigantian boundary probably within the lava succession (Figure 12).The base of the lava succession is also within the Asbian, but at an uncertain position above or within the early Asbian.
The normal polarity dominance in the mid part of the Kinghorn Volcanic Fm concurs with that observed at Trowbarrow, although the likely sporadically erupted lavas are unlikely to express fine-scale polarity changes as at Trowbarrow.The reverse polarity in the upper part of the lava succession (inferred as MI5r in the GPTS; Figured 12) is not seen at Trowbarrow, which suggests that this reverse polarity interval is younger than the Brigantian limestones at Trowbarrow.It is probable that the reverse polarity intervals in the lowest part of the Upper Visean age lavas studied for paleomagnetism in Derbyshire are stratigraphically widely separated, and the normal polarity lavas in the base of the Monsal Dale Limestone Fm are probably coeval with the normal polarity interval seen at Trowbarrow in the basal Alston Fm (Figure 12).The lower and upper units of the Monsal Dale Limestone Fm contain corals indicating early Brigantian zones H and I of Mitchell (1989).Hence, the overlying reverse and normal polarity lavas in the Monsal Dale Limestone are early Brigantian in age (Waters et al., 2011b).The underlying Bee Low Limestone Fm contains goniatites, indicating the Goniatites globostriatus Subzone (B2b Subzone; Waters et al., 2011b) of the late Asbian.Likewise, foraminifera indicate the Cf6γ1b assemblage from the base of the formation, with probably assemblage Cf6γ2c missing and possibly the upperpart of Cf6γ2b also absent (Chisholm et al., 1983;Strank, 1985) The two levels of lava in the upper 20%-30% of the Bee Low Limestone show reverse polarity (Figure 12), and may relate to the reverse polarity intervals TQ6r and TQ7r.However, without a study of the remainder of the Bee Low Limestone the magnetostratigraphic correlation remains uncertain for these two lower lavas.
In eastern Canada, the magnetostratigraphy from the Middleborough Fm (Figure 12) is inferred to be Brigantian in age based on the late Asbian age for the underlying Limekiln Brook Fm (Giles, 2008;Jutras et al., 2016;Utting et al., 2010).However, since the Limekiln Brook Fm is laterally transitional into the lower part of the Middleborough Fm (Jutras et al., 2016), it is possible that the oldest part of the magnetostratigraphy of Opdyke et al. (2014), consisting of magnetozones NN1-NN2, may extend into the late Asbian rather than being entirely Brigantian (Figure 12).This latter option is compatible with the magnetostratigraphy indicating that the NN1r-NN2n magnetozone interval could be coeval with an interval low in the early Brigantian with NN1n probably equivalent to the basal Brigantian as seen at Trowbarrow (Figure 12). .These are the short (e.g., Brezinski & Kollar, 2021) and long (e.g., Ettensohn et al., 2022) duration options for the Mauch Chunk Fm of NE Pennsylvanian discussed by Hounslow (2022).Based on the arguments presented by Opdyke et al. (2014) and Hounslow (2022), the most likely position for the top of the magnetostratigraphy from the Mauch Chunk Fm (not shown in Figure 12) is near the base of the early Bashkirian.Based on similarity in the reversal pattern between the palynologically dated Brigantian from eastern Canada and that from the upper part of the Indian Run Mbr of the Mauch Chunk, Hounslow (2022) suggested that the most likely option was that the mid parts of the Kinghorn Volcanic Fm were coeval with the predominantly normal polarity interval referred to as magnetochron MI5n, derived from the Mauch Chunk dataset (Figure 12).The data from Trowbarrow now more firmly assign this thicker normal polarity interval to the late Asbian.The equivalent interval to magnetochron MI5n was sampled in the Mauch Chunk Fm at two sections in the lower part of the Indian Run Mb (Figure 12).It was defined by results from 19 levels in one section and 18 levels in the other.In comparison, the late Asbian succession in Trowbarrow displays a much more detailed polarity pattern due to our higher sampling density.This additional complexity of MI5n is perhaps partly reflected (but there unresolved) in the multiple single-sample submagnetozones detected in MI5n by Opdyke & DiVenere (2004).
The astrochronology from Trowbarrow (Hounslow, Cózar, et al., 2024) indicates that the late Asbian has a duration of 1.976 ± 0.086 Myr (1σ), including 3 significant hiatus levels (marked on Figure 12).This duration is within the uncertainty from radioisotopic dates spanning this interval.The duration estimate and its uncertainty include three magnetic proxies from two separate equally likely sedimentation rate models (i.e., six estimates).Uncertainty in the period of the eccentricity targets used in the astrochronology adds some 1.5%-1.9%additional uncertainty (∼0.037Myr; Hounslow, Cózar, et al., 2024), which gives a combined 1σ uncertainty of 0.094 Myr.
Quantifying duration uncertainty beyond this is challenging but could come from using multiple astrochronologic techniques (e.g., Da Silva et al., 2020) or multiple coeval overlapping sections (Sinnesael et al., 2019).
This duration translates to a polarity reversal frequency of 15.7 ± 0.75 Myr 1 (TQ1n-TQ9n.2n;excluding the tentative submagnetozones).Evidently, the lower reversal frequency, largely estimated with the datasets from the Mauch Chunk Fm through this interval, is an underestimate (black curve in Figure 13).However, the data from the Mauch Chunk Fm does not indicate a high reversal frequency in either under or overlying strata.Hence, it is possible that the interval with hyperactive-reversals is a stratigraphically restricted interval, like that which occurs in the Triassic and Lower Jurassic at around [246][247][248]respectively (Figure 13).This brevity is rather different from the Cambrian and Ediacaran, which seem to have more sustained intervals of high reversal frequency and corresponding weak paleointensity, which particularly characterize the Middle Jurassic (Kulakov et al., 2019) and the Late Ediacaran (Levashova et al., 2021;Thallner et al., 2022;Figure 13).
However, a middle Visean high reversal frequency seems compatible with current datasets from the Late Devonian (gray curve in Figure 13; Green et al., 2021;curve from Hounslow, 2022).It is further consistent with the very low paleointensity estimates recovered from the coeval Kinghorn Volcanics that define the youngest point of the "mid-Paleozoic Dipole Low" (Hawkins et al., 2021).The fact that the mid-Paleozoic is temporally equidistant to the Middle Jurassic and the Late Ediacaran is suggestive of a ∼200 Myr quasi-cyclicity in paleomagnetic field behavior (Biggin et al., 2012).Such long-term variations remain to be convincingly explained but may be linked to changes in subduction flux (Hounslow et al., 2018), true polar wander (van der Boon et al., 2022) and/or mantle plume events Biggin et al. (2012).As highlighted by Van der Boon et al. (2022), the Earth's magnetic field behavior during the Early Carboniferous and Devonian is key to consolidating hypotheses about the causes and implications of the Phanerozoic behavior of the geodynamo.

Conclusions
A magnetic polarity stratigraphy has been evaluated through the late Asbian type section in Trowbarrow Quarry, covering the top-most early Asbian through to the earliest .This is tied to a detailed foraminiferal biostratigraphy and an astrochronology.The magnetostratigraphy yields nine major magnetozone couplets, seven minor submagnetozones and two tentative submagnetozones through the 195 m of section.
The magnetization is carried by hematite and in some samples a likely detrital magnetite phase.The hematite is probably of aeolian origin but also perhaps with some pedogenic-derived hematite associated with some of the 13 major emergent surfaces.A regional fold test using data from fourteen, other largely Visean localities studied in NW England, indicates that ChRM magnetization was of pre-tilting age so prior to the Late Pennsylvanian to Early Permian.
The Mississippian age magnetization is partly overprinted with Kiaman Superchron-age (ca.320-270 Ma) and Brunhes-age (0.77-0 Ma) magnetization, which are inferred to be thermoviscous in origin.Regional fold tests using the Cumbrian and north Lancashire sections indicate that the Kiaman age partial remagnetization is younger than the principal Variscan age folding, so probably of the latest Pennsylvanian to early Permian age.
The polarity stratigraphy defines many short-duration submagnetochrons in the Mississippian magnetochron MI5n.A reversal rate of 15.7 ± 0.75 Myr 1 is inferred for the late Asbian, indicating that at ca. 335-333 Ma the geodynamo was in a state of hyperactive reversal.

Figure 3 .
Figure 3.For the Trowbarrow dataset: (a) the partition of LINEFIND α 95 values with demagnetization behavioral classes, (b) the partition of demagnetization classes with polarity quality, and (c) and (d) the partition of reverse and normal polarity quality classes with VGP R latitude.Similar data for the additional set are shown in Figure S7 in Supporting Information S1.S1, S2, S3 are descending quality classes of PCA lne fits, and T1, T2, T3 are descending quality classes of great circle PCA fits.Arrows in (a) are the mean α 95 , and the mean ρ value of the respective class.
6)Note.Ns = number of levels (sites), Nl = number of specimens used with fitted lines, and Np = number of specimens with great circle planes used in the determining the mean direction.α95, Fisher 95% cone of confidence.k, Fisher precision parameter.G o is the angular separation between the inverted reverse and normal directions, and G c is the critical value for the reversal test.In the reversal test the G o /G c values flagged with * indicate common K values, others not flagged have statistically different K-values for reverse and normal populations, in which case a simulation reversal test was performed.Plat and Plong are the latitude and longitude of the mean south virtual geomagnetic pole1.A95 (min, max) = Fisher 95% confidence interval for VGP-based site mean (Ns sites), and A95 min and A95 max threshold values ofDeenen et al. (2014).Statistics determined with Pmagtool v.5(Hounslow, 2023b).a = conventional Fisher mean using specimens.b = using method ofMcFadden and McElhinny (1988) to find great circle intersection points (T-class data) and combined with S-class line fits for each sample mean.c = great circle combined mean using the method ofMcFadden and McElhinny (1988) on specimen level data.

Figure 4 .
Figure 4. Specimen magnetic data for the Trowbarrow Quarry section.(a) shows a simplified log (sampling details in Figures S1-S3 in Supporting Information S1), and (e) shows the foraminiferal subzones and British substage boundaries (from Cózar et al., 2022b).In (b) the NRM intensities are partitioned into hematite and non-hematite magnetizations, and the ChRM intensity is also shown at the first demagnetization step at which the S-class ChRM is isolated.(d) The occurrence of the K-component and high stability tail during AF demagnetization.PLF = Park Limestone Formation, WS = Woodbine Shale, e.As = early Asbian.FZ-SZ = foraminiferal zone and subzone.

Figure 5 .
Figure5.Isothermal remanent magnetization (IRM) data for samples from Trowbarrow Quarry, segmented according to the petrographic data inCózar et al. (2022b).Key applies to both (a) and (b).The %magnetite scale on the top of (a) is based on the hematite-magnetite mixing relationships ofFrank and Nowaczyk (2008) applied to the S-ratio (-IRM 0.1mT /1T).The medians of the petrographic groups are indicated with a crossed-square and labeled with group numbers 1 to 8 (as in key).Specimen corresponding to extremes in the S-ratio (in a) and extremes in (b) are indicated.

Figure 7 .
Figure 7. Directional and unblocking data for the low temperature LT and K-components from Trowbarrow Quarry specimens (in geographic coordinates).(a) the LT component and its upper unblocking temperature for magnetite (mean ∼180°C) and hematite-dominated (mean ∼ 200°C) magnetizations, (b) the K-component directions.(c) K-component alternating field (AF) ranges and (d) lower and upper thermal demagnetization ranges, with the upper-range set partitioned into dominant type of magnetization, and for the combined demagnetization scheme, if this temperature matches the uppermost thermal step.These data also include composite thermal-AF ranges (24 specimens); n = amount of data in each graph.(e) The S-ratio of those samples in which the K-component is defined (upper) and in which the magnetization are T3 class or contain no ChRM and are entirely dominated by the K-component (bottom).Directional data for additional sample set in Figure S9 in Supporting Information S1.

Figure 8 .
Figure 8. Directional data for the ChRM component, with line fits (S-class data) in (a) and poles to the fitted great circles in (b).The single great circle plane in (b) is that normal to the mean of the S-class data.(c) S-class ChRM lower and upper thermal demagnetization ranges (in upper bar graphs) and alternating field (AF) ranges (in lower bar graphs).These data also include composite thermal-AF ranges (20 specimens); n = amount of data in each graph.(d) The S-ratio of those specimens which have an S-class ChRM.

Figure 9 .
Figure 9.South paleomagnetic poles for the three S-class means (and their A 95 confidence cones) from Trowbarrow Quarry (T1, T2, T3) and that of the K-component from Trowbarrow (TK) placed alongside the stable Europe apparent polar wander path of Torsvik et al. (2012) partitioned into age intervals (using their timescale).Other paleopoles are selected Mississippian (m1 to m11) and Pennsylvanian (p1 to p7) poles and inferred Kiaman-age reverse remagnetizations (k1 to k7).Also shown are the paleopoles of the Cumbrian and Isle of Man hematite mineralization's (H1 to H3) ofRowe et al. (1998) andCrowley et al. (2014).Inverted azimuthal equidistant projection.See TableS5in Supporting Information S1 for sources of other data in this figure.

Figure 11 .
Figure11.The K-component within the Trowbarrow specimens with respect to magnetic properties.In each column from left to right are displayed the specimens with left: no inferred K-component, mid: K-component was present with the ChRM, and right: specimens containing only the K-component and no ChRM.The first row, (a) shows specimens partitioned into NRM intensity and hematite and magnetite-dominated demagnetization behavior, second row, (b) specimens partition into χ ARM , (c) specimens partitioned into the grain size proxy IRM 40mT /χ ARM and (d) mass specific NRM intensity and IRM 40mT , with respect to hematite and magnetite-dominated demagnetization behavior.In (d) grayed values are the NRM/IRM 40mT as a percentage, using the regression fitted line (and Pearson R 2 ) to the magnetite-dominated data at IRM 40mT of 16, 64 and 256 × 10 7 A.m 2 /kg.Rock magnetic data for (b) and (c) fromHounslow, Cózar, et al. (2024).Histogram intervals were chosen to give an approximate even spread through the whole respective datasets.

Table 1
Trowbarrow Quarry Directional Means (With Tectonic Correction and Converted to Normal Polarity), Reversal Tests and South VGP Poles

Table 3
Fold Tests on the ChRM and K-Components