Tarduno等人6,7研究的JH发现地点样品中的锆石相对于BGS的锆石更多地分散在BGS中 ，因此
，使用非磁性重力方法在罗切斯特大学进行了分离。亚甲基碘化物是使用的重液体，并从单独的手工挑选了锆石 。锆石选择遵循Tarduno等人6,7的方法 ，并进行了一些修改
。分离的BGS锆石通常比从JH发现位点研究的锆石略小
，因此，我们专注于大于约70μm的锆石（对150μm）。对于BGS锆石（大约9×10-13 A M2 ，而大约1×10-12 A M2），用于分析的NRM强度选择临界值（应用以确保可以准确测量的电磁值可以准确测量） 。 　　在三个地点（BGS1
，BGS2和BGS5）收集生存的锆石
。分离产生了> 1,000个锆石，但只有大约500个满足尺寸要求。其中 ，可以隔离大约300个晶体
，这些晶体是单个锆石而没有附着的非循环晶粒，并且表面变化明显 。一百三十锆石的NRM值≥9×10-13 A m2。这些晶体中有65个没有可见的裂缝或大夹杂物（可能是多域氧化物）用于565°C的古显度实验
。由于初始NRM强度测量值不可再现性 ，在实验前拒绝了7个锆石。初始值很可能被粘性的隔离磁化污染 。第一次热处理后强度下降
，另外五个锆石失败了，再次表明粘性磁性磁化强度
。六个锆石失败了古显度检查（它们在加热时不会失去NRM ，或者没有获得热量磁化（TRM）或失败的多域尾部检查）。Tarduno等人建立的符合古显度标准的47个锆石受到SEM和阴极发光分析的约束
，而12个因内部晶粒障碍的证据而被拒绝 。与从亚甲基碘化物重型液体分离的35个锆石所占的成功率至少约为3.5％
。尽管技术有所不同，但该成功率高于JH锆石的研究（<2%; Tarduno et al.6), consistent with a better preservation of the BGS zircons. 　　Optical microscopy studies used a Nikon Eclipse LV100POL microscope with both transmitted and reflected-light capabilities, a maximum of 1,000× magnification and a SPOT Insight 4-MP CCD colour digital camera assembly. SEM analyses were also conducted at the University of Rochester using a Zeiss Auriga SEM. For subsurface inclusions, we estimated depth using Electron Flight Simulator Version 3.1E software. 　　Procedures follow those of Tarduno et al.6,7 and are outlined here. All experiments were conducted in the shielded room (ambient field < 200 nT) of the Paleomagnetism Laboratory at the University of Rochester. The 565 °C palaeointensity method6 was used to derive palaeofield strength estimates while limiting the effects of laboratory-induced alteration by reducing the number of heatings. A temperature of 565 °C is chosen for comparison with the JH palaeointensity dataset, which also used this temperature, and to represent blocking temperatures best reflecting single-domain-like magnetite inclusion carriers, while also retaining a measurable demagnetized value. The demonstration that 565 °C falls within the interval of sharp magnetic unblocking (Fig. 2b) supports this choice. In the 565 °C palaeointensity approach, the NRM is first measured, and then the zircon was gradually heated in field-free space with a Synrad v20 CO2 laser12. Specifically, the temperature was held constant for 1 min at 100-°C temperature steps to 500 °C. Next, the temperature was increased to 565 °C for an extra 1 min and then allowed to cool for 3–5 min. Thereafter, a second heating to 565 °C was conducted in the presence of a field. An applied field of 15 μT was used in all field-on treatments. A multidomain check was performed last by reheating the zircon sample to 565 °C in the absence of a field. We tabulate the calculated angle between the TRM vector (565on-565off; Extended Data Table 1) and the applied field. Differences from zero may record small error contributions related to the alignment of the laser beam and a zircon in these challenging experiments and/or an anisotropy of the collection of magnetic particles in a given zircon. 　　The geomagnetic field can be as described by the scalar potential (r, θ, ϕ, t): 　　in which are partially normalized Schmidt functions, l and m are spherical harmonic degree and order, respectively, re is the radius of Earth and the Gauss coefficients and describe the spatially and time-varying fields. In the case of a time-averaged field needed to make conclusions on the geodynamo10, the geocentric axial dipole () is represented by a PDM and field strength B is related to palaeolatitude (λ) by: 　　Standard SHRIMP U-Pb instrumental setup40and procedure for U-Pb calibration by reference materials41 were followed. Two 1-inch epoxy mounts (IP985 and IP987) were built containing the JH and BGS zircons as well as zircon reference materials z6266 with an accepted 206Pb/238U age of 559 Myr41 and z1242 with an accepted 207Pb/206Pb age of 2,679.6 ± 0.2 Myr42. Polishing with diamond suspension exposed the internal structure of the zircon grains. After coating with 10 nA of gold, the grains were imaged in cathodoluminescence (CL) and backscattered electron detector (BSD) mode using a MIRA3 TESCAN field emission scanning electron microscope. In order to target pristine areas of zircon grains and avoid cracks and alteration, a 16O- ion beam 13 μm in diameter and with an average beam current of 1 nA was used. The analytical runtable consisted of 11 masses including background with species of Hf, Yb, Zr, Pb, Th, U analysed over six scans. Data reduction used SQUID3 (ref. 43) (note that the name of this software has no relation with SQUID magnetometers; SQUID3 refers to the name of the software and it is not an acronym). Steiger and Jäger44 decay constants were used. Two analytical sessions were carried out on each of the two epoxy mounts. The 1σ external error for the 206Pb/238U calibration relative to reference material z6266 for sessions IP985_1 and IP985_2 was ±0.88% or ±1.01, respectively. The 1σ external error for the 206Pb/238U calibration for sessions IP987_2 and IP987_3 was ±0.75% or 1.07%, respectively. These errors are also specified in the footnote to the data table (Supplementary Table 1). The requirement for a correction due to instrumental mass fractionation of the Pb isotopes was assessed through replicate analyses of reference material z1242. The measured weighted mean 207Pb/206Pb age of those analyses are reported for each session in the footnotes of Supplementary Table 1. No fractionation correction was applied to the Pb- isotope data except for session IP987_2, in which a correction of −0.64% was applied. Concordia plots were generated, and weighted means calculated using Isoplot v. 4.15 (ref. 45). The uncertainties for the weighted mean ages reported in the text and Supplementary Materials are at the 95% confidence level. 　　Palaeomagnetic, reflected-light microscope, electron microscope, microtectonic, geochemical and palaeointensity data indicate the presence of primary magnetite inclusions in select JH zircons and that these have primary magnetic signals6,7. Specifically, zircon microconglomerate tests and the identification of distinct secondary components of magnetization separate from the primary characteristic magnetization exclude magnetic resetting after incorporation of the zircons into the JH host metasediment (figure 2 of ref. 7). 7Li profiling data (for example, figure 6 of ref. 7) argues against thermal resetting of Hadaean and Eoarchaean data, whereas reflected-light microscopy, microtectonic analysis, SEM/EDS, focused ion beam and NanoMOKE investigations document the presence of primary magnetite inclusions (figures 3–5 of ref. 7). Pb-Pb screening of data, documented in refs. 6,7 (figure 7 of ref. 7) argues against a magnetic resetting age older than the age of incorporation into the host sediment in the selected zircons. This is further enforced by the distinct change in palaeointensity data, with values from late Hadaean zircons being higher than those of Palaeoarchaean to Eoarchaean age, inconsistent with Palaeoarchaean magnetic resetting. Reference 6 notes that iron oxyhydroxides can be found in JH zircons that are likely related to weathering, but these cannot be the carrier of the magnetic remanence used for palaeointensity determination. Furthermore, an assemblage of magnetite inclusions is needed to account for the magnetizations observed from the zircon, satisfying Maxwell–Boltzmann statistical limits on magnetic recording7. Reference 7 also describes why remagnetization scenarios calling on the neoformation of magnetic grains represent particles too small or too few in number to record stable magnetizations. Reference 7 also notes that the association of magnetic particles with dislocations does not mean that the particles are secondary22; instead, this is expected given the deformation history of the JH conglomerate. 　　Although the zircon studied would not pass our selection criteria, atom probe tomography46 data on a JH zircon illustrate differences in the high unblocking temperature magnetic primary remanences isolated in refs. 6,7 and other potential Fe-bearing minerals. Reference 46 interprets quantum diamond microscope data from ref. 22 on a single zircon as indicating Fe-bearing zones and links those to the carriers of natural magnetic remanence. Atom probe tomography data from one zone within that zircon is further interpreted to contain Fe nanoclusters (which are far too small to carry remanent magnetizations) and a maximum age of approximately 1.4 Gyr is assigned; this is interpreted as a natural remanent magnetization age. A quantum diamond microscope does not have the sensitivity to measure the natural magnetic remanence of JH zircons. Reference 7 explains that the authors instead measured the magnetization of a laboratory-induced isothermal remanence magnetization (0.25 T, >4,000× present-day field; ref. 22); this strong field can enhance the magnetization of Fe oxides/oxyhydroxides that do not contribute to the primary remanence. The zircon microconglomerate tests6,7 discussed above supersede older tests and document that the high unblocking temperature magnetization in JH zircons must be older than the approximately 3-Gyr depositional age of the JH conglomerate. Therefore, the hypothetical remanent magnetization inferred in ref. 46 with an assigned age of approximately 1.4 Gyr cannot be the key high unblocking temperature component of JH magnetization6,7. 　　Samples for this study were collected from the lower approximately 1 m of the approximately 3–5-m-thick BGS. Our samples appear distinct from those of the lowest 20 cm of the BGS, which show extensive shear zones in thin section47. In thin section, our BGS samples used for zircon separation have a greenschist-grade anastomosing foliation in the generally fine-grained matrix. The foliation is cut by several conjugate through-going fractures. The competency difference between the competent zircon grains and the weaker surrounding matrix could result in stress buildups at the grain boundaries that might generate cracks within the zircon grains. Below, reference is made to specific zircons studied by reflected-light and scanning electron microscopy. 　　SEM sample BGS2-z51. The fractures appear to be in three main systems (Extended Data Fig. 2a). The ‘vertical’ or N–S fractures are parallel to the c axis (crystallographic axis based on geometry). The ‘east-dipping’ and ‘west-dipping’ sets appear to be part of the same (112) crystallographic system as close-packed planes (for example, noted by Reddy et al.48). The melt inclusions (Extended Data Fig. 2b,f,d) are interesting because that in panel d does not have a fracture going through it, further indicating that the inclusions are not being formed by material diffusing in through the fractures. Instead, these patterns suggest that the inclusions are primary and many of the fractures are localized on the inclusions. The competency difference between the host crystal and the inclusions likely result in stress concentrations at the edges of the inclusions (depending on their composition). The fractures would tend to propagate along close-packed crystallographic planes (that is, the (112) planes). 　　Reflected-light sample BGS5-z30, Extended Data Fig. 1h,i. From the form of the crystal, the c axis is approximately vertical. There are two sets of conjugate fractures with the acute angle between them facing approximately E–W (they form the V pattern in the upper part of the grain), which would suggest an E–W compression direction when those fractures formed. The E–W fracture on either side of the ‘football-shaped’ inclusion probably formed during the same compression. The fractures coming off the tapered ends of the ‘football-shaped’ inclusion are likely guided by the inclusion itself. The tapered ends of an inclusion have the smallest radius of curvature and, therefore, generate the highest stress concentrations, causing the fracture to propagate out from the tapered tip. Those fractures do not appear to be guided by the crystal structure of the zircon—if the fractures had propagated farther, they might have reoriented to follow a crystallographic close-packed plane. There is a possible fracture within the inclusion (Extended Data Fig. 1j) that could reflect propagation from a stress concentration outside the inclusion. 　　Reflected-light sample BGS2-z39, Extended Data Fig. 1. Although the fractures look random at first glance (Extended Data Fig. 1d), a few fractures seem to be parallel to the c axis or 001 (along the length of the crystal), a set is perpendicular to the c axis (on the bottom of the crystal) and there is a pair of conjugate sets (perhaps the (112) set) indicating compression approximately in the E–W direction. The globular form of the quartz inclusion highlighted in Extended Data Fig. 1e,f suggests that it is a melt inclusion. The adjacent fracture does not pass through the inclusion. The fractures likely originate at the inclusion boundary as a result of stress concentration created by the inclusion. The fracture below the inclusion is clearly following a crystallographic plane parallel to the c axis. 　　SEM sample BGS2-z36, Fig. 1l and Extended Data Fig. 2k–l. The inclusion is not connected to fractures in the 2D view available. Lattice diffusion (Nabarro–Herring creep) requires temperatures greater than half the melting temperature. In the case of zircon, that would be approximately 900 °C, unless the crystal structure was sufficiently disturbed that the lattice was more open. There is no evidence of metamictization that might signal such an open structure. The distance from the edge of the crystal to the inclusion (in 2D) is approximately 2 μm and it is unlikely that the inclusion and interior Fe particles could form by diffusion through the crystal lattice. Instead, it is likely that this is a melt inclusion formed during the original formation of the crystal (at temperatures >1,000°C）。 　　在数十至数百个纳米尺度上的磁性夹杂物在硅酸盐内得到很好的确定（例如 ，Tarduno等人10）；Trace Fe信号可以反映锆石内的这种夹杂物
，但是在Feldspar的情况下，需要进一步的工作才能将它们与Fe内在到晶体区分开 。当在反射光下观察时 ，我们注意到孤立的夹杂物在0°至90°极化角度旋转后通常会显示出灭绝
。对于足够在晶体中通过SEM EDS分析取样的浅的人，Fe与灭绝的存在为磁铁矿7提供了证据。 　　我们强调的是
，我们研究中的锆石是使用鲁棒方案来分离原代磁性记录器的，这是在锆石的首次古磁研究中建立的 。我们的NRM和Demagnetization数据以及我们的传输/反射光和SEM观察结果表明 ，BGS锆石“几乎没有铁磁矿物质 ”的主张49是不正确的
。 　　为了证明各个锆石晶粒的同位素系统的鲁棒性
，在可能的情况下进行了复制分析。具有可重复性PB-PB年龄的锆石晶粒被推断为由于扩散，重结晶或其他热过程而损失PB的封闭系统 。当加权平均年龄大于0.05时 ，该年龄被认为是磁化年龄。在存在锆石过度生长的情况下
，我们将最年轻的健壮年龄分配给磁化时间
。补充信息中可以提供每种谷物结果的描述。总体而言，我们注意到我们的BGS锆石缺乏3.4个轮辋 。这与我们的解释是一致的 ，即这些锆石没有经历过3.4-GA高温事件
，否则可能会影响它们记录的磁性历史记录
。 　　为了比较JH和BGS的锆石古大度数据，我们首先检查以下假设：H0 ，JH和BGS数据样本样本相同的基础场分布
，指定为零假设；和H1，JH和BGS数据样本样本不同的基础场分布 ，被指定为替代假设。我们专注于三个统计测试：学生t检验的Welch版本（Welch50），两样本KS Test51和Mann – Whitney U Test52 。学生t检验与KS和Mann-Whitney U检验不同
，假设每个样品正态分布都具有均等的差异
。考虑到每个人群中的一些比较和不平等差异的样本量（见下文），两尾韦尔奇的t检验版本是适当的50 ，将每个样本均值和差异与t分布进行比较。非参数两样本KS测试比较了两个样本
，并测量其经验累积分布（DKS）之间的最大距离，并拒绝如果PKS超过一定的临界阈值 ，则拒绝零假设 。Mann – Whitney U检验是另一个非参数测试
，该测试检查了两个样本（例如X和Y），而无效假设是 ，来自样本X的随机值具有相等的或小于人口y的随机值的可能性相等
。如果两个样本共享相同的基础分布
，这是预期的结果。所有三个测试都比较了假设H0和H1并返回P值 。如果P值大于定义的显着性阈值（α），通常为α= 0.05 ，则H0假设将无法拒绝
。如果p值小于定义的显着性
，则可以在（1 -α）置信度以H1拒绝H0。对于我们的假设，如果p> 0.05 ，那么可以合理地推断出JH和BGS锆石古显度分布无法在95％的置信阈值下区分 。 　　我们首先比较3.4至3.9 GA之间的JH和BG，并证明年龄范围的选择范围更低；我们发现Welch的t检验
，非参数两样本KS测试和Mann-Whitney U测试均产量P值> 0.05，这表明我们不能拒绝零假设H0（扩展数据表2）。 　　在Tar​​duno等人7中 ，使用100兆元的非重叠箱在移动窗口模型中平均JH锆石古显度数据
，以估算每个100兆间隔间隔的PDM
。垃圾箱边缘是通过在其JH锆石古敏度中的可用数据分布来定义的，该数据的分布为3,258 MA ，3,358 MA高达4,258 MA。我们遵循这种方法
，使用相同的垃圾箱定义定义使用BGS数据的100兆平均箱，以允许在JH和BGS古大压数据之间进行直接比较 。最后 ，使用相同的bin定义组合了来自JH和BGS数据集的古敏性
，以产生扩展数据中使用的100兆平均值
。图4。在扩展数据表3中提供了每个100-MYR BIN的古强度统计数据 。 　　接下来，我们将分别考虑在移动窗口模型中定义的100兆年龄箱（图3和扩展数据表2）
。JH和BGS数据集只有两个间隔包含五个以上的观测值（Welch的t检验的最小阈值） ，以3,408 MA和3,508 MA为中心。每个间隔的Welch的t检验分别得出0.17和0.09的p值
，因此零假设H0是JH和BGS数据记录相同的场强度不能拒绝 。非参数测试产生相似的结果，两个间隔的P值超过0.05（扩展数据表2） ，支持对Welch t检验的解释
。 　　为了定义JH和BGS数据样本相同的基础领域的最长年龄间隔，对数据进行了一系列Welch的t检验
，以用于3.9 GA及以下的数据。对于每次测试，通过对年轻绑定的时间进行调整后 ，年龄间的间隔逐渐缩短
，并记录了所得的P值（扩展数据图4A） 。我们发现，对于年龄超过3.4 GA的年龄 ，P值很高并支持零假设
。我们还注意到
，对于小于3.4 ga的年龄界，韦尔奇的t检验p值下降到0.05以下 ，导致拒绝零假设H0对H1有利于H1（扩展数据图4A）。这表明当比较间隔扩展到年轻时代时
，JH和BGS数据开始采样不同的基础场分布 。 　　接下来，我们考虑JH古显着历史是否预测了从3.9至3.4 GA之间观察到的 ，我们检查了JH古生力的100兆平滑窗口模型与用于定义它的数据之间的残差（也就是说
，是JH Zircons的古质数据；我们对BGS古代密度数据集（即BGS古代数据与JH模型；扩展数据图4C，E）这样做。考虑到这两个分布 ，我们应用了两个样本KS测试（扩展数据图4F），并评估新数据残差是否看起来像输入到JH平滑窗口平均值中的数据
。 　　零假设是
，两个样本（即残留种群）是从相同分布中得出的。在模型很好地预测新数据的情况下，KS测试应返回p值> 0.05 ，如果新数据的预测很差
，则P值为 <0.05 (that is, in this case, the hypothesis that the residuals are similar can be rejected). When applied to the new BGS, the P value is 0.32, which supports the hypothesis that the JH model can predict the Barberton zircon results. Although the P value is above the 95% threshold, the distributions nevertheless merit some discussion. The JH palaeointensities are clearly more scattered than the BGS results. This could signal one, or both, of the following:(1) the BGS zircons are, in general, better preserved, having less amorphous iron oxides in cracks than the JH and, therefore, they might be better recorders; (2) the JH might record the field on shorter timescales than the BGS, averaging higher-frequency secular variations of the past geomagnetic field. Palaeointensities from the SCP analyses of Nondweni dacite samples are within the BGS values of the same age, but SCP of Barberton dacites at 3.45 Ga are higher. Both the Nondweni and Barberton SCP values are plotted as individual results from two relatively shallow intrusions, and these might be expected to sample higher-frequency variations of the geomagnetic field than the zircons. 　　Here we consider the probability of observing two sites, located on different plates, experiencing both little to no latitudinal motion as well as little to no relative latitudinal distance separation during an interval of 600 Myr. To construct this test, the plate-reconstruction model of Merdith et al.53, a continuous full-plate reconstruction model spanning the past 1 Gyr, is used. A set of sites are defined for the present day and the site palaeolocations are reconstructed back from the present day to 600 Ma using GPlates54. The distribution of palaeolocations is described and comparisons between sites located on different plates is made. From this empirical dataset, a set of statistical tests is constructed. 　　The sampling grid is defined using a Fibonacci spiral55, which yields a uniform distribution of locations distributed globally with a median separation of approximately 6°. From this grid, 1,000 sites are assigned to plates using a built-in GPlates function. Sites that do not fall within the boundaries of a plate are removed before the analysis. Using the plate-motion model of Merdith et al.53, site palaeolocations are reconstructed in 1-Myr steps from the present day to 600 Ma. From the initial set of sites, only those assigned to plates that existed 600 Ma are preserved, resulting in a set of 228 locations (Extended Data Fig. 5a). 　　We model a plate-motion path for each site from the GPlates reconstruction. Using the plate-motion path, the absolute maximum latitudinal distance travelled, Δλ can be determined as: 　　in which λ0 is the present-day site latitude, is the palaeolatitude at time t and ‘max’ is the maximum distance for the set considered. Δλ distances range from 33° to 127°, with a median Δλ distance of 76° (Extended Data Fig. 6b). 　　From the distribution of Δλ for each of the 228 sites located on 66 unique ‘plates’ as designated by GPlates, the general trend of latitudinal motion can be described (Extended Data Fig. 5c). Broadly speaking, there is a weak positive correlation between the magnitude of Δλ and the age at which the maximum latitudinal distance is observed (Pearson’s correlation coefficient, r = 0.49);, however, sites with Δλ falling in the lower 5% of the distribution (Δλ ≤ 40°, n = 12/228) have ages ranging from 150 to 480 Ma. We note that Δλ estimates can be biased downward by as much as approximately 10° when sites are downsampled at 100-Myr versus 1-Myr intervals. 　　With these general trends in mind, we next consider whether any sites experience little to no latitudinal motion. No sites could be identified with near-zero latitudinal motion irrespective of the sampling level (Extended Data Fig. 6a,b). If we arbitrarily define ‘little motion’ as ≤30° of latitudinal motion, no sites at 1-Myr or 20-Myr sampling meet the criterion, whereas only a single site (located on the Paraná/Pampia plate) met the criterion with 100-Myr downsampling. If the definition of ‘little’ motion is further relaxed to a threshold of ≤35° and the sampling interval to 100 Myr, 11 of 228 sites meet the criterion. The 11 sites showing ≤35° motion are observed on seven plates. We note that we only consider the maximum amount of latitudinal motion in this analysis; if both northern and southern latitudinal motion are considered separately instead, the lower 5th percentile of the revised distribution indicates a minimum total latitudinal motion of approximately 50–60°. 　　Given the rarity, or absence, of plates showing no motion over 600 Myr, the likelihood of observing two sites on separate plates with little to no latitudinal motion is very low. We next consider the bounds of Δλ and relative latitudinal distance for plate pairs provided by the palaeointensity data (Fig. 4). The case of a plate backtracking to the north is automatically recognized as being inconsistent with plate motion over the past 600 Myr by the single-site Δλ distributions (Extended Data Fig. 6a,b). For a plate backtracking to the south, the Δλ bound is 48° (see main text). The bound on relative distance between the plate pair is the combination of this backtracked southward motion for one plate and the maximum backtracked northward motion of another plate consistent with the palaeointensity data (5°), for a total of 53°. 　　To determine the probability of sampling sites from two separate plates with these Δλ and relative distance characteristics, a hypergeometric distribution is appropriate. A hypergeometric distribution models a sampling without replacement, in contrast to the more common binomial distribution, which samples trials with replacement56. Sampling without replacement is appropriate for this scenario because the same site (plate) cannot be selected twice. The probability mass function (p(k)) for a hypergeometric function is defined as: 　　in which N = 66 is the number of plates, K is the number of plates in the 600-Myr dataset known to record latitudinal motion and relative distance less than or equal to a specified threshold, n = 2 represents the number of unique plates selected in a single random sampling and k is the number of plates in that sampling showing limited latitudinal motion. The resulting probability of identifying a single pair of plates (k = 2) that both show limited latitudinal motion, here defined as Δλ ≤ 48° and relative latitudinal distance ≤53° at 100-Myr downsampling, is <1% (Fig. 4f). 　　Finally, we note that palaeolongitudinal trends characteristic of plate tectonics can also be inferred from plate-motion models spanning the past 600 Myr. To avoid biasing of longitudinal motion owing to geometric effects (that is, less area at higher latitudes), we normalize longitudinal motion by analysing the fraction of maximum angular travel from the initial position that is not explained by latitudinal change. We note that plates having dominant longitudinal motion (for example, >50％）很少见（约5％） 。在这些情况下，至少有40°的纬度运动
。 　　一些作者提出了涉及约3.1至2.7 GA之间的Pilbara和Kaapvaal craton的“ Vaalbara
”超级克拉顿 ，其中一些作者提出了57,58名
，但受到其他作者的挑战59。可以说，最强的地质相关性是在这些craton上的大约2.7–2.8-ga火山之间（Kaapvaal craton的Ventersdorp超级组和Pilbara craton的Fortescue Group） 。Kaapvaal craton的古磁数据 ，应用了倾斜的解释59
，这意味着在很高的纬度（75.4°）中进行了扩展
。这种火山关联与BGS和JH锆石记录开始差异的时间之间存在约700米长的年龄差距，这可能会暗示板块构造的水平运动特征的开始。该持续时间比我们用来评估板块构造的纬度运动特征的持续时间长100个MYR 。然而 ，我们的分析可以用作保守的措施
，以评估板块构造的典型板动作率是否与Neoarchaean中的高纬度kaapvaal craton兼容，还是在古代安学中的低纬度位置
。鉴于中位数最大纬度位移的0-600 Ma为76° ，很明显，板态速率是兼容的。假设仅在一个（n或s）半球中仅在一个（n或s）中的kaapvaal craton的位置产生的平均运动纬度成分为8 mm年-1 。假设一个赤道交叉会导致运动的平均纬度成分16毫米-1
。 　　长期以来
，长期以来，一直使用古磁变化 ，主要是逆转和看似不转变的时期（超金子）的发生（超金子）
，但同样是对世俗变化的推断，以争辩核心 - 摩托车边界结构的变化 ，从而影响了GeodyNamo60,6161的核心边界结构。超级与逆向间隔的古大度变化也被解释为反映这些变化
，这是一些数字Geodynamo Models的支持 。最近，有人提出 ，这种变化可能会扩展到Devonian67和/或可能在Ediacaran时期开始。如果我们的3.9-3.4-GA数据中存在类似于在植物学中观察到的古生地的变化
，那么我们可能无法将发电机效率的变化与古质量变化分开
。但是，由于古显度记录是恒定的 ，我们推断出古质量是恒定的
，地幔过程并未创建纬度核心 - 摩托车边界热量升华模式，从而极大地影响Geodynamo效率。