Intensity of the Earth's magnetic field: Evidence for a Mid-Paleozoic dipole low

Significance

Variations in past geomagnetic field strength are important indicators of variation in deep Earth processes over hundreds of millions of years. Most other geophysical methods only provide a snapshot of the Earth’s recent interior, and deep Earth materials are poorly represented in the geological record. Recent measurements from Scotland (Northern United Kingdom), in addition to the existing datasets, show the field was relatively weak (less than half the strength of the long-term average field) for tens of millions of years between 332 and 416 Ma. The similarities between this and a later period of low field strength provide further evidence for a ∼200-My cycle linked to deep Earth processes.

Abstract

The Mesozoic Dipole Low (MDL) is a period, covering at least ∼80 My, of low dipole moment that ended at the start of the Cretaceous Normal Superchron. Recent studies of Devonian age Siberian localities identified similarly low field values a few tens of million years prior to the Permo-Carboniferous Reverse Superchron (PCRS). To constrain the length and timing of this potential dipole low, this study presents paleointensity estimates from Strathmore (∼411 to 416 Ma) and Kinghorn (∼332 Ma) lava flows, United Kingdom. Both localities have been studied for paleomagnetic poles (Q values of 6 to 7), and the sites were assessed for their suitability for paleointensity from paleodirections, rock magnetic analysis, and microscopy. Thermal and microwave experiments were used to determine site mean paleointensity estimates of ∼3 to 51 μT (6 to 98 ZAm²) and 4 to 11 μT (9 to 27 ZAm²) from the Strathmore and Kinghorn localities, respectively. These, and all the sites from 200 to 500 Ma from the (updated) Paleointensity database (PINT15), were assessed using the Qualitative Paleointensity criteria (Q_PI). The procurement of reliable (Q_PI ≥ 5) weak paleointensity estimates from this and other studies indicates a period of low dipole moment (median field strength of 17 ZAm²) from 332 to 416 Ma. This “Mid-Paleozoic Dipole Low (MPDL)” bears a number of similarities to the MDL, including the substantial increase in field strength near the onset of the PCRS. The MPDL also adds support to the inverse relationship between reversal frequency and field strength and a possible ∼200-My cycle in paleomagnetic behavior relating to mantle convection.

The evolution of the Earth’s deep interior, although critical to our understanding of the planet’s history, is poorly constrained. Reconstructions of convection patterns, such as the configuration and volume of subduction through time (1, 2) and the occurrence of mantle plumes (often inferred from the occurrence of Large Igneous Provinces) (3), as well as the stability of the Large Low-Shear Wave Velocity Provinces (4), are generally ill defined prior to ∼300 Ma. This lack of constraint is primarily due to poor preservation of materials that formed deep within the Earth and because most geophysical techniques (i.e., seismic tomography, gravity inversion, etc.) can only constrain geologically recent deep Earth processes. Comparatively, the paleomagnetic record, and paleointensity in particular, has the potential to serve as a key indicator of early deep Earth processes, such as the initiation of the geodynamo (5) and inner core nucleation (6, 7). During the Phanerozoic, superchrons (periods of tens of millions of years without magnetic polarity reversals) are suggested to be linked to changes in Earth’s deep interior (8, 9). Three superchrons have been identified (10) and correspond with peaks in field strength (11). The superchrons alternate with suspected periods of frequently reversing, weak dipole moments. If confirmed, this pattern would suggest the existence of a ∼200-My cycle in paleomagnetic behavior, likely resulting from deep Earth processes (8), which alternates between superchrons and these periods of low magnetic dipole moments, such as the Mesozoic Dipole Low (MDL). First proposed by Prévot et al. (12), the MDL is a period of low dipole moment suggested to have lasted for at least the ∼80 My preceding the Cretaceous Normal Superchron (CNS: 84 to 126 Ma). The MDL has since been confirmed by subsequent studies (13, 14), which have potentially placed its origin near the end of the Permo-Carboniferous Reversed Superchron (PCRS; 267 to 315 Ma) (15), also known as the Kiaman Superchron. It has also been suggested that the MDL is actually confined to ∼150 to 170 Ma, while the rest of the “MDL” is biased toward low field values due to rock magnetic effects (16). Recent research from Siberian sites (17, 18) found a similar, persistent low dipole strength magnetic field during the Devonian (359 to 419 Ma), lasting ∼50 My, ∼35 My prior to the start of the PCRS. However, it is still unclear if the behavior of the magnetic field during the Devonian and Early Carboniferous (∼100 My pre-PCRS) is comparable to that of the MDL, as there is very little available data, with only five studies from this age in the Paleointensity Database (PINT15) (19).

To quantify the length of this potential dipole low, two localities from the east coast of Scotland, United Kingdom, were selected from this time period to augment the previously published studies (Fig. 1). The first of these, lava flows from the Strathmore region (411 to 416 Ma) (20), were initially studied comprehensively by Sallomy and Piper (21), who found paleodirections consistent with this early Devonian age. A follow-up paleointensity study by Kono (22), based on a subset of these sites, gave a mean virtual dipole moment (VDM) of ∼35 ZAm², which is substantially lower than the present-day field strength (∼80 ZAm²). However, developments in the field of paleointensity in the last 40 y mean that the reliability of this study is now uncertain, as it was done prior to the development of modern day paleointensity techniques and selection criteria. No checks for alteration or multidomain behavior were included, and recent studies have shown that the perpendicular protocol that was used therein can give artificially low paleointensities (23). The original paleodirectional study has also since been superseded by Torsvik (24). This updated study argued that the high degree of scatter and the presence of sites with “transitional” directions in the original study, several of which were used for paleointensity, was likely due to bias introduced by the demagnetization techniques used and local tectonic effects.

Fig. 1.

Geological map showing the Strathmore (black circles) and Kinghorn (black stars) localities sampled for paleodirection and paleointensity sites in this study. The geological units come from the 1:50,000 solid geology map from the British Geological Survey (Copyright UKRI 2019), accessed via Edina Digimap, and generalized descriptions are listed in the legend on the right. Key cities are highlighted as white circles and key faults as dashed lines. The Strathmore Group Volcanic (SGV) units and the Kinghorn Volcanic Formation (KVF) are highlighted with dotted outlines. The location of the geological map is outlined in the inset map in the top left corner of the Northern United Kingdom. The metamorphic and igneous geological units are (A) Neoproterozoic metamorphics, (B) Silurian-Early Devonian felsic intrusions, and (C) Silurian–Devonian mafic extrusives. The remaining units are clastic sedimentary rocks from (D) Visean to Westphalian, (E) Arbuthnott-Garvock/Strathmore Groups, (F) other Devonian, (G) Llandovery-Wenlock, and (H) Caradoc to Ashgill.

The second locality, lava flows from the beaches along Kinghorn and Burntisland, Scotland (332 ± 5.6 Ma) (25), has not been studied for paleointensity previously. A paleodirectional study carried out on these lavas by Torsvik (26) found primary directions that were used to determine an Early Carboniferous pole. Paleointensity results from Strathmore and Kinghorn are presented herein alongside a detailed meta-analysis of published datasets dated to 200 to 500 Ma using paleointensity quality criteria (Q_PI) (27). The outcome supports a key feature of long-term geomagnetic behavior: the Mid-Paleozoic Dipole Low (MPDL).

Materials and Methods

Detailed geological backgrounds of the Strathmore and Kinghorn localities, along with a description of the sampling techniques used, are provided in SI Appendix, Geological Background and Sampling, and the locations of the sites sampled are shown in Fig. 1. The suitability of these sites for paleointensity analysis was first determined using paleodirectional and rock magnetic analysis. The majority of sites were initially stepwise thermally demagnetized (see SI Appendix, Methods: Paleodirections for details) to determine if the samples carried a stable magnetic remanence. Selection criteria applied to the individual paleodirections obtained were an anchored, maximum angular deviation (MAD_ANC ≤ 10°) and the angle between the anchored and unanchored directions (α ≤ 10°). Additionally, site directions required some degree of clustering (k ≥ 15) before being compared with those from the previous studies (24, 26) to determine if the magnetization was of the correct age. Rock magnetic analysis was also performed on all sites, including hysteresis, isothermal remanent magnetization, back-field, and thermomagnetic (Curie) measurements (see SI Appendix, Methods: Rock Magnetism for details). These measurements, in conjunction with scanning electron microscopy (SEM) analysis, were used to determine if the remanence carriers were consistent with the sites carrying a primary thermal remanent magnetization (TRM). Sites that passed all the criteria were then used for paleointensity analysis. Both microwave and thermal Thellier-type paleointensity experiments were performed using the IZZI protocol (28) with partial TRM checks (29). Full details of the experimental procedure used, including selection criteria applied, are provided in SI Appendix, Methods: Paleointensity.

Results

Strathmore Paleointensity.

A detailed outline of the results of the paleodirectional and rock magnetism analysis carried out on this locality is described under SI Appendix, Strathmore Results. Based on these results, of the eleven Strathmore sites that gave acceptable paleodirections, six were determined to be suitable for paleointensity experimentation. The five sites that produced thermomagnetic curves consistent with (titano-)maghemite (CB2-CB4, TH1, and SN3) were excluded from further analysis because (titano-)maghemite is unlikely to be the original (primary) mineralogy of these lavas (SI Appendix, Strathmore Results; Rock Magnetism and Fig. S3A). Comparatively, the sites measured for paleointensity were deemed to likely retain a primary TRM. Sites CB1 and SN1 produced thermomagnetic curves that indicated the presence of two magnetic minerals, magnetite and hematite (SI Appendix, Fig. S3B). The unblocking temperature range of the characteristic remanent magnetization (ChRM) supports that the primary remanence is carried by both minerals (see SI Appendix, Fig. S1 B, i and Strathmore Results; Rock Magnetism for details). The remaining sites (WB2-WB5) produced thermomagnetic curves consistent with magnetite (SI Appendix, Fig. S3D) and gave representative SEM that showed no evidence of low-temperature oxidation (SI Appendix, Fig. S3G). Site WB2 was combined with the hematite-bearing (SI Appendix, Fig. S3C) baked sediment (WB1) into a single paleointensity site (WB1/2) because the similar ChRM directions for the two sites (see SI Appendix, Strathmore Results; Paleodirections for details) suggest that WB1 was completely remagnetized by the overlying lava (WB2), which means the two sites would have acquired their TRM in the same field.

From the Strathmore locality, 35 out of 82 paleointensity measurements passed selection criteria (given in SI Appendix, Table S1; pass rate of 43%) across the six sites. All but one site gave low site-mean paleointensity estimates (3.1 to 19.7 μT), corresponding to VDMs between 5.6 and 46.2 ZAm² (Table 1), while site CB1 gave a singularly high site mean estimate of 50.9 μT (98.0 ZAm²). The majority of accepted Strathmore estimates are from thermal experiments because the microwave demagnetization mechanism was largely unsuitable for hematite-bearing sites (CB1, SN1, WB1). The other estimates were split approximately evenly between the two techniques. The pass rate for microwave experiments was 48% versus 36% for thermal results from the sites that used both methods. Characteristic Arai plots (Fig. 2) show the range of behavior exhibited from the six sites and the different techniques used. The hematite-bearing sites (Fig. 2 A and B) showed minimal demagnetization at temperatures below 300 °C and linear Arai plots across the temperature ranges for magnetite or both magnetite and hematite. From Wormit Bay, sites WB2-WB3 (Fig. 2 C and D) behave similarly, as do WB4-WB5 (Fig. 2 E and F), likely because of the similarities in grain size, based on the hysteresis properties of the sites (SI Appendix, Fig. S3I). Sites WB2-WB3 lie close to the MD (multi-domain) range but produce near-linear orthogonal vector and Arai plots, whereas sites WB4-WB5 exhibit some zig-zagging of the corresponding orthogonal plots and the main occurrence of prominently two-sloped Arai plots; however, as seen in the orthogonal plots from the corresponding thermal demagnetization experiments (SI Appendix, Fig. S1 D, i), the samples display more than one directional component, and the temperature range of the ChRM corresponds to the selected component from the thermal paleointensity experiments. Accepted measurements are considered unaffected by anisotropy, based on measurement gamma values (γ), and nonlinear TRM effects, as both the ancient and applied fields used are relatively low (≤60 μT). The Wormit Bay sites are also unlikely to have been affected by cooling rate based on the samples’ grain size distribution [non–single domain (non-SD) grains (30); SI Appendix, Fig. S3I] and any variation between results from techniques with different cooling rates (thermal vs. microwave), or mineralogy with different grain size distributions (magnetite vs. hematite), being inconsistent with the expected effects from cooling rate differences (see SI Appendix, Paleointensity Reliability Assessment; Strathmore Q_PIScoring, the ACN criterion, for further details).

Table 1.

Summary of paleointensity results and Q_PI scores for all the Strathmore and Kinghorn sites

	Strathmore						Kinghorn
Site	CB1	SN1	WB1/2	WB3	WB4	WB5	KH2	KH1	KHA	KHB	KH 10	KH4	KH7	KH8/9
Paleointensity results
nINT	9	11	22	12	15	13	20	28	12	21	17	10	12	23
NINT	5	5	11	7	5	2	6	12	3	9	4	6	4	9
NT	5	5	8	3	2	1	1	1	—	1	—	2	1	—
NMW	—	—	3	4	3	1	5	11	3	8	4	4	3	9
Mean (μT)	50.9	12.6	16.8	19.7	3.1	6.3	6.6	6.1	3.7	6.7	5.2	5.3	10.9	8.6
SD (μT)	15.9	3.6	5.8	6.4	0.3	4.3	3.0	1.6	0.5	2.5	1.1	0.7	4.3	4.0
SD/mean (%)	31	29	34	32	10	68	45	26	13	38	22	13	40	47
VDM (ZAm2)	98.0	20.2	36.6	46.2	5.6	12.0	16.4	15.7	9.3	16.8	13.4	10.4	26.6	20.2
QPI scores
AGE	1	1	1	1	1	1	1	1	1	1	1	1	1	1
STAT	0	0	0	0	1	0	0	0	0	0	0	1	0	0
TRM	0	0	1	1	1	1	1	1	1	1	1	1	1	1
ALT	1	1	1	1	1	1	1	1	1	1	1	1	1	1
MD	1	1	1	1	0	0	1	1	1	1	1	1	1	1
ACN	1	1	1	1	1	1	1	1	1	1	1	1	1	1
TECH	0	0	1	1	1	1	1	1	0	1	0	1	1	0
LITH	0	0	1	0	0	0	0	0	0	0	0	0	0	0
MAG	1	1	1	1	1	1	1	1	1	1	1	1	1	1
QPI	5	5	8	7	7	6	7	7	6	7	6	8	7	6

n_INT, number of samples measured; N_INT, number of measurements that passed selection criteria; N_T, number of accepted measurements from thermal IZZI; N_MW, number of measurements from microwave IZZI. The nine Q_PI are described in full in Biggin and Paterson (27); 1 is a pass and 0 is a fail to meet the qualitative criteria, and Q_PI is the total score of all these criteria. Site longitude and latitude and the corresponding site directions are available in SI Appendix, Table S1.

Fig. 2.

Representative Arai plots from the six Strathmore sites, (A) CB1, (B) SN1, (C) WB(1/)2, (D) WB3, (E) WB4, and (F) WB5, illustrating the different Arai plot behaviors observed. All the measurements were done using microwave and thermal Thellier-type experiments using the IZZI protocol, with the thermal plots showing the highlighted temperature steps (°C) and the microwave plots showing the highlighted power steps (Ws, watt-seconds). The thick black lines connecting measurement steps are the partial TRM checks. The k’ value is the curvature of the selected component on the Arai plot (49). The corresponding orthogonal plots are inset in the top right corners of the Arai plots. Plots A and B are examples from the sites where there are only thermal measurements as the components come from both the magnetite and hematite temperature ranges (SI Appendix, Fig. S3A). The remaining plots (C–F) come from the magnetite-only samples (SI Appendix, Fig. S3D) and show microwave and thermal examples from sites that have similar hysteresis properties (SI Appendix, Fig. S3I), WB2-WB3 (C and D), and WB4-WB5 (E and F).

Kinghorn Paleointensity.

A detailed outline of the results of the paleodirectional and rock magnetism analysis carried out on this locality is described in the SI Appendix, Kinghorn Results. All the Kinghorn sites that passed the paleodirectional selection criteria (six out of nine sites) were deemed suitable for paleointensity, along with two additional sites (KHA-KHB; SI Appendix, Fig. S2C) for a total of eight paleointensity sites. Some sites had curves that were consistent with relatively Ti-rich titanomagnetite (KH1 and KHA-KHB; SI Appendix, Fig. S3F), while the rest (KH2, KH4 and KH7-KH10; KH8/9 is a single site) were consistent with magnetite or low-Ti titanomagnetite (SI Appendix, Fig. S3E). Representative SEM showed primary igneous textures (i.e., coarse exsolution structures) and no evidence of low-temperature oxidation (SI Appendix, Fig. S3H), which is consistent with the sites carrying a primary TRM.

Of the Kinghorn locality samples, 53 out of 143 measurements passed selection criteria (given in SI Appendix, Table S2; pass rate of 37%). All sites produced very low site mean estimates (3.7 to 10.9 μT; Table 1), corresponding to exceptionally low VDM estimates (9.6 to 27.0 ZAm²). The majority of the accepted measurements were made using the microwave system, as it had a much higher success rate (54% success rate vs. the 9% success rate for thermal experiments). This may be because the relatively Ti-rich titanomagnetite is prone to altering more in the thermal, rather than in the microwave experiments, due to reduced bulk heating of the samples in the latter (31). The appearance of the Arai plots varies, with some sites producing near-linear plots with minimal overprints (Fig. 3 A and F), whereas others exhibit varying degrees of two-slope behavior (Fig. 3 B–E). Like sites WB4-WB5, the two-slope Arai plots all show a corresponding change in direction, which is also indicated in the thermal demagnetization orthogonal plots (SI Appendix, Fig. S1 E, i and F, i). These plots suggest that the steep, low-temperature slope is likely an overprint rather than due to nonideal behavior. This interpretation is also supported by a lack of zig-zagging when using IZZI protocol and the low degrees of curvature (which was evaluated using curvature, |k’|). There is no clear correlation between the appearance of the Arai plots and either the titanium content of the titanomagnetite or the apparent grain size (SI Appendix, Fig. S3J). Accepted measurements are considered unaffected by cooling rate due to the samples’ non-SD grain size (SI Appendix, Fig. S3J) and the lavas being “fast-cooled” (27) as well as nonlinear TRM effects (for the same reason as the Strathmore sites). Anisotropy checks on the Kinghorn samples, performed due to relatively high γ values, show that some of the Kinghorn sites are slightly anisotropic, but the anisotropic corrections required (which vary depend on the applied field direction) are negligible and average out to give nearly identical results to the uncorrected site means (see SI Appendix, Paleointensity Reliability Assessment; Kinghorn Q_PIScoring, the ACN criterion, for further details and quantification).

Fig. 3.

Representative Arai plots from six of the Kinghorn sites (A) KH1, (B) KHB, (C) KH10, (D) KH4, (E) KH7, and (F) KH8/9, illustrating the different Arai plot behaviors observed. All the measurements were done using microwave and thermal Thellier-type experiments using the IZZI protocol, with the thermal plots showing the highlighted temperature steps (°C) and the microwave plots showing the highlighted power steps (Ws, watts-seconds). The thick black lines connecting measurement steps are the partial TRM checks. The k’ value is the curvature of the selected component on the Arai plot (49). The corresponding orthogonal plots are inset in the top right corners of the Arai plots. All the Arai plots represent (low-Ti titano-)magnetite apart from (A) KH1 and (B) KHB.

Discussion

Reliability of 200 to 500 Ma Sites.

To further assess the reliability of these site-mean paleointensity estimates and to provide a framework for comparing them to others from the Paleozoic (∼252 to 541 Ma), all sites were evaluated using Q_PI criteria. Biggin and Paterson (27) proposed these nine criteria to acknowledge and mitigate the potential biases that affect the interpretation of paleointensity data and are applied in a similar way to Q criteria for paleomagnetic poles (32). Sites that have published information addressing a criterion pass (score a 1) and if not, they fail (score a 0). The Q_PI score for the site is the sum of the individual criterion scores. Detailed descriptions of these nine Q_PI criteria and how they have been assessed for this study are included in SI Appendix, Paleointensity Reliability Assessment; Q_PIcriteria, and explanations for how the Kinghorn and Strathmore sites were scored are provided in SI Appendix, Paleointensity Reliability Assessment; Strathmore Q_PIScoring and Kinghorn Q_PI Scoring. A summary of the individual criteria scores for each site is provided in Table 1 and in Dataset S3. The Strathmore sites in this study received Q_PI scores ranging from 5 to 8 (median: 6.5, mean: 6.3) out of a possible 9. The Kinghorn sites similarly received Q_PI scores ranging from 6 to 8 (with a slightly higher median of 7 and mean of 6.8). STAT and LITH are the Q_PI criteria that the majority of sites failed. The failure to meet STAT is largely because the sites gave low paleointensity estimates, which are less likely to pass STAT due to its definition (see SI Appendix, Paleointensity Reliability Assessment; Strathmore Q_PIScoring for details), and LITH because only one site had a suitable contact lithology that could be sampled. MD and TRM failing for some of the Strathmore sites is why its Q_PI scores are slightly lower than those of the Kinghorn sites. However, the sites are not considered unreliable, as their failure to pass these criteria need not require that they have been remagnetized and/or affected by MD behavior. We note that our results are also very similar to those obtained from Siberian rocks of a similar age (7 to 39 ZAm² with a single outlier site of 98 ZAm²) (17).

Integration of the Strathmore and Kinghorn estimates with the existing Paleozoic dataset first requires determination of what published data are sufficiently robust for meta-analysis. All the data in the PINT15 database (19) from 200 to 500 Ma were checked against their corresponding study to fix any errors. Ages were recalculated where possible (e.g., stratigraphic ages were revised to be consistent with the most recent timescale [ICS2020/v1], isotopic ages were replaced where superseding ages were known, etc.). The biggest reassessment of site ages comes from the apparently Middle-Late Carboniferous sites from Uzbekistan (33, 34). The relative ages between sites and the single inclination sign across multiple sections, with 13 to 40 sites per section, indicate that these sites could only come from the part of the Carboniferous during the PCRS (i.e., Moscovian-Gzhelian; 298.9 to 315.2 Ma). Sites from five studies published since the last PINT15 update were also added (references listed in Dataset S4). Q_PI criteria were applied to all the sites based on the published information from the corresponding studies. This time period covers both the PCRS and the surrounding time periods, which allows paleomagnetic field strength during the superchron to be compared with that from a reversing field. This time period also complements two other Q_PI studies that assessed the PINT15 database for 500 to 3,500 Ma (6) and 65 to 200 Ma (16).

The revised PINT15 data for 200 to 500 Ma, including Q_PI scoring, is included in Dataset S3. A workflow for the scores provided is outlined in Dataset S4, and the age distribution, coverage, and reliability of the revised 200 to 500 Ma PINT15 data are illustrated in Fig. 4. Given that most studies from this time period were published before Q_PI criteria existed, their Q_PI scores tend to be lower because there is insufficient information published to confirm that a potential issue has been addressed rather than it being clear that an issue has affected the estimates. Only sites with Q_PI scores of 0 will be excluded entirely; these either have no published information to support the reliability of the site means or they have been confirmed to be unreliable. All the highest scoring sites (Q_PI ≥5) are found in the time periods immediately before (16 sites) and after (26 sites) the PCRS, which comprises the data from this study along with recently published studies (17, 18). While there are numerous sites with PCRS ages, 144 of the 195 sites (74%) covering this period come from just four studies (33, 35–37). The Q_PI scores for these are low because these publications use outdated paleointensity methods (i.e., calculating the ratio of total NRM [natural remanent magnetization]/total TRM) (38) and include very little supporting information.

Fig. 4.

The age distribution of all the virtual (axial) dipole moment [V(A)DM] values with Q_PI >0 between 200 and 500 Ma. Summaries of the Q_PI scores applied to each of the studies from this period are outlined in Datasets S3 and S4. The size and the color of the circles representing the V(A)DM values corresponds to the Q_PI scoring as outlined in the key. The PRE, PCRS, and POST sections refer to the same age bins used for the Kolmogorov-Smirnov tests in Fig. 5 and SI Appendix, Table S2. The dashed lines represent (A) the present day field strength (50), (B) CNS at Q_PI ≥ 3 (16), (C) JHAP Q_PI ≥ 3 (16), (D) JHAP Q_PI ≥ 0 (16), (E) the maximum possible, and (F) the minimum possible reversal frequency from the Canning basin magnetostratigraphy (45).

Paleozoic Field Variation.

Based on the pattern of field strength variation from the sites from this study and the existing PINT15 dataset, weighted by Q_PI score (Fig. 4), a relatively long period of low dipole moment presents itself in the period preceding the PCRS, followed by a substantial increase in field strength during the superchron relative to periods of reversing field. To evaluate whether this variation is similar to that observed during the Mesozoic, an analysis was performed, following the methodology of previous studies (6, 16), by comparing the field strength distribution of different periods, filtered by Q_PI scoring. The combined dataset was grouped into three bins using the superchron as an anchor: PRE (315 to 416 Ma), PCRS (267 to 315 Ma), and POST (200 to 267 Ma). Kulakov et al. (16) was able to identify periods of distinct dipole moment during the MDL based on reversal frequency. However, the reversal record prior to the PCRS is too sparse to apply the same technique (Fig. 4), so the PRE bin was not divided further. Its maximum age bound was set to 416 Ma to avoid the Ordovician Reversed Superchron (ORS; 461 to 480 Ma) (39) and because there are no estimates with Q_PI ≥ 1 between 416 to 461 Ma. There is no analysis of the ORS, or the period before it, as between 461 and 500 Ma there are only three available estimates. The age distribution of the POST bin is also substantially skewed (skewness = −6.14 at Q_PI ≥ 3; see SI Appendix, Table S2) as there are only 13 site-mean results between 200 to 250 Ma and an abundance of data around ∼250 Ma. This peak in the data are almost entirely the result of a large number of studies from the Siberian Traps [see Anwar et al. (15) for details]; however, the paucity of data between 200 to 250 Ma means it cannot be connected to the Early bin from Kulakov et al. (16) and should be considered independent of it.

Q_PI filtering was applied based on the total Q_PI score for the site, up to Q_PI ≥ 5 (because the PCRS bin does not include any sites that have a total score greater than 5). Fig. 5 A–E illustrates the distribution of site VDMs in these bins for different Q_PI minima, while details of the data included in these bins are included in SI Appendix, Table S2. However, not all the Q_PI criteria are considered to be of equal weight. As discussed in Kulakov et al. (16), AGE, STAT, TRM, MD, and ALT are considered the priority criteria; however, it is not possible to make all these criteria mandatory and still have sufficiently large datasets for statistical analysis. AGE and ALT are considered essential criteria to retain (16), but the way the TRM criterion is judged (SI Appendix, Paleointensity Reliability Assessment; Q_PIcriteria) means that a site can carry a TRM but not successfully meet the criterion. Too few PCRS sites pass TRM (11 sites) for further analysis to be statistically meaningful (especially when requiring AGE and ALT to pass as well). The inverse problem exists for the STAT criterion: very few sites pass the criterion for the PRE (18 sites) and POST (16 sites) bins, probably because of the weak site mean values, which is discussed in SI Appendix, Paleointensity Reliability Assessment; Kinghorn Q_PIScoring. However, requiring the inclusion of the MD criterion still retained a statistically rigorous number of sites across all three age bins (171 sites passed for AGE, ALT, and MD). The inclusion of the MD criterion is also particularly important because MD effects on paleointensity experiments have been cited as a primary cause for the apparent long-lasting weak strength of the paleomagnetic field, compared to the present-day field value (40). The distribution of VDMs that pass all these criteria (AGE + ALT + MD) is shown in Fig. 5F, and details of the data included in these bins are included in SI Appendix, Table S2. Finally, to check that the bin statistics are not being biased by oversampling of the field at certain time periods (i.e., 250 Ma), the subsets of sites with total Q_PI scores ≥1 and ≥3 were binned in 10 Ma and 5 Ma intervals before averaging (details in SI Appendix, Table S2).

Fig. 5.

Boxplots showing the V(A)DM distribution of the PRE, PCRS and POST bins (same as in SI Appendix, Table S2). The boxplots are filtered based on the total Q_PI scores applied to the sites, between Q_PI ≥ 1 and Q_PI ≥ 5 (A–E) and for sites that pass the AGE, ALT, and MD criteria (F). The numbers over the boxes display the number of sites in each of the age bins. On each box, the central mark is the median, the edges of the box are the 25th and 75th percentiles, the dashed lines extend to the most extreme data points not considered outliers, and outliers are plotted individually (+).

A series of Kolomogorv-Smirnov (K-S) tests were run to compare the bins pair-wise (SI Appendix, Table S2). The tests reject the null hypothesis that the datasets come from the same distribution at the 1% significance level up to Q_PI ≥ 5 (the PCRS and POST bins have too few data to be significant if a higher Q_PI cutoff is applied; see SI Appendix, Table S2). All three time periods, therefore, have distinct VDM distributions. Fig. 5 A–F indicates that the dipole moment during the PCRS is substantially higher than the surrounding time periods, regardless of Q_PI filtering, although its average strength is unclear. The median values for the PCRS range from 39 to 94 ZAm² depending on the Q_PI filters and binning used. Although the median CNS values (48 to 59 ZAm²), based on similar analysis (16), fit within this range, the large range in average field strength values for the PCRS points to issues with the current PCRS dataset. The difference emerges from the greater number of low estimates recorded for the CNS than the PCRS. The higher median values are generally from subsets with lower levels of reliability (low Q_PI scores/number of bins) because the four studies that make up the bulk of the PCRS are older studies that give generally high field values and have Q_PI scores ≤3. There are a greater number of, and more recent, studies available for the CNS, so the median values for it are more likely to represent reliable estimates of field strength during a superchron. Thus, while the difference in average field strength between the PCRS and the rest of the Paleozoic is likely to remain, further studies are needed to evaluate the average strength of the field during the PCRS and if the potential differences between the average field strength during the two superchrons is valid or due to data bias.

Both the PRE and POST bins gave consistently low values across all Q_PI filters (Fig. 5), including AGE + ALT + MD, suggesting that MD behavior is not responsible for the appearance of these periods of low field strength. Comparatively, the AGE + ALT + MD filter provided one of the lowest median values for the PCRS (48 ZAm²), suggesting MD behavior could instead be biasing the PCRS values high (if taken from the lower temperature slope). The aging of TRM has also been considered as a potential cause of long-term field strength underestimation. While this process is less well understood than MD effects, experiments have shown that specimens affected by aging tend to show more pronounced MD-like behavior (41), so they should also be excluded by the MD criterion. Dividing the POST bin into 5 and 10 Ma intervals gives marginally higher field values. This result suggests that the low average field strength of the POST bin and, potentially, the distinct VDM distribution from the PRE bin (despite their similar median values), could be due to under-sampling of the average field behavior during this time bracket. As demonstrated by the highly skewed age of the bins and the low N values for 5 and 10 Ma subbins, almost all the sites from this time period were emplaced over ∼800,000 y (the Siberian Trap sites) (42). The low N value bins, while they may still be underestimating the long-term strength of the field, are starting to approach the strength of the field following another superchron, the CNS (Late period; 36 ZAm² at Q_PI ≥ 3) (16).

The strength of the field prior to the PCRS is the most reliable period herein due to the consistency between the median values for the PRE bin, regardless of filtering and age binning. The paleointensity data presented in this study are also in close agreement with previous studies, with median field estimates for the PRE bin ranging from 12 to 19 ZAm², based on the same filtering of Q_PI scores as used in SI Appendix, Table S2. This range of median values for the PRE bin remains the same if sites from the Kinghorn locality or the Strathmore locality, or both, are excluded. The consistency between these estimates and the legacy weak-field data, despite differences in reliability between the sites and those from the older studies, is encouraging because it indicates the field strength during this time period is not being underestimated. At ∼17 ZAm² (Q_PI ≥ 3), the PRE bin is considerably lower than the average for the Phanerozoic, which is estimated to be 42 ZAm² (13) to 50 ZAm² (6) and is probably the typical strength of the field outside of the Bruhnes (43), that is, the period since the last geomagnetic field reversal (the last ∼770 ka). The PRE bin values are closest in strength to the Jurassic hyperactivity period [JHAP; 155 to 171 Ma; 26 ZAm² for all data points, 35 ZAm² at Q_PI ≥ 3 (44)], which had an average reversal frequency of ∼11 reversals/My. In comparison, the other periods of reversing field during the Mesozoic (Early, Mid, Late) had median field strengths of 36 to 48 ZAm² with reversal frequencies of 1 to 3 reversals/My. The low average field strength may be partly due to recent studies (with higher Q_PI scores) tending to sample periods of very high reversal frequency, like the JHAP. This tendency is difficult to constrain because magnetostratigraphic records before the PCRS are generally too sparse to provide a reliable record of reversal frequency. A recent magnetostratigraphic study from the Canning Basin (45) suggests reversal frequencies of a minimum of 2 to 5 reversals/My around the same time that the Viluy sites cooled (∼360 Ma) (18). In addition, an evaluation of reversal frequency ∼8 to 14 My before the PCRS (46) suggests reversal frequencies of ∼12 reversals/My, very similar to the high JHAP values, occurring ∼5 My after the Kinghorn lavas (this study) erupted. These studies produced notably weak site mean values in the range of 4 to 27 ZAm². A greater nondipole component to the field, relative to the present day, is also indicated for some of the Siberian sites, such as the Minusa (17) and the aforementioned Viluy Traps (18), although the sites from this study provide no indication of a greater nondipole component, despite comparable timing and field strength estimates.

A Mid-Paleozoic Dipole Low and Its Implications for Deep Mantle Variation.

Despite some potential biases of the dataset, as discussed in the previous sections, the evaluation of the period of low dipole moment leading up to the PCRS provided by this study suggests it is a significant and distinct feature of the Paleozoic paleomagnetic record (see K-S tests in SI Appendix, Table S2). The proposed term for this feature, the MPDL, is based on its similarities to the MDL. These similarities include the weak field (discussed in the previous section) and the ∼80 My duration from ∼416 to 332 Ma. In both cases, however, gaps in the record make it difficult to confirm if the period of low dipole moment extends back further in time (18). The average field strength has been shown to vary throughout the MDL (16), and there is also evidence for a difference (Fig. 4) between the relatively strong early part of the MPDL in the interval 390 to 416 Ma (median of 36 ZAm²), based on the Minusa (17) and Strathmore (this study) sites, and the rest of the MPDL (317 to 390 Ma; median of 14 ZAm²). Unlike the MDL, the low field strength is difficult to relate directly to reversal frequency due to the paucity of magnetostratigraphic records at this time. Finally, there is a clear increase in field strength around the onset of the PCRS, estimated to be 3 to 4 times the strength of the pre-PCRS field if the average for the PCRS is reliable. We point out, however, that a gap in the dataset exists from ∼20 My prior to the PCRS to an unknown point in the superchron (assumed to be within the Carboniferous part of the PCRS; 299 to 315 Ma). The large age uncertainties associated with the early PCRS sites prevent a clear determination of this transition.

The assessed paleointensity record provided in this study gives an improved indication of patterns in Phanerozoic paleomagnetic field behavior across 10 to 100 My timescales (8) back to ∼415 Ma. The similarities observed herein between the MPDL and the MDL prior to their respective superchrons provide more evidence for the proposed inverse relationship between field strength and reversal frequency. There are insufficient site-mean paleointensity estimates prior to (three site means) and during (zero site means) the ORS to test this theory further back in time. There have been several mechanisms proposed for this variation in field behavior relating to mantle plumes (9, 47), subduction (2), and True Polar Wander (8). The extension of our reliable Phanerozoic paleointensity record will assist future studies in linking these processes to paleomagnetic evolution.

Data Availability

Raw thermal demagnetization and paleointensity data have been deposited in MagIC database on Earthref.org (https://earthref.org/MagIC/17067/8cfb6f3d-fc3b-46c7-8995-ec55be815859) (48). The rest of the data are all included in Datasets S1–S4.