Introduction

Archaeomagnetic studies of backed artefacts have allowed the reconstruction of the parameters of the geomagnetic field in Europe for the last thousands years (e.g. Gallet et al. 2002; Tema et al. 2006; Pavón-Carrasco et al. 2009; Schnepp et al. 2009; Kovacheva et al. 2014; Genevey et al. 2016; Molina-Cardín et al. 2018; Schnepp et al. 2020; Tema and Lanos 2021). The archaeointensity and archaeoinclination averaged regional curves elaborated for the last millennium display in some parts distinct changes of these parameters i.e. consistent decrease in archaeointensity from 17th to 18th century and increase in archaeoinclination from 14th to 17th century (Le Goff 2002; Schnepp and Lanos 2006; Schnepp 2008, Schnepp et al. 2009). The regional reference secular variation curves can serve as a valuable dating tool for archaeological heat-treated artefacts like bricks, ovens and ceramic tiles. This method can be the most effective when archaeomagnetic data includes all geomagnetic field parameters i.e. the declination, inclination and intensity. The declination of ancient geomagnetic field, needs an in situ position of the backed artefacts with respect to the place where they had been fired. Because of this any brick building cannot provide it. The inclination can be obtained from the bricks because they were positioned ca. horizontally on their narrow long sides during firing (e.g. Płuska 2009). However, not quite horizontal position of the bricks in the oven requires significant amount of them to a better averaging of inclination and improving statistical error. One of the problems related to the use of bricks in archaeomagnetism, apart from their age inaccuracy, is possible long distance between the sites of brick production and brick building. Archaeomagnetic investigations in northern Poland are still scarce and have not been conducted for more than 30 years (Czyszek and Czyszek 1987). In spite of numerous ancient brick buildings, archaeomagnetic data from Poland are single (Brown et al. 2021; Schnepp and Lanos 2006) and methods of their determination do not fulfil current standards of quality.

In our paper, the utility of bricks from N Poland for determination of archaeointensity and its correction factors will be checked. In this part of Poland bricks have been produced from the local Quaternary clays since 13th century (e.g. Wasik 2017). A limited number of localities and samples collected for this work do not allow for construction of any regional archaeomagnetic curve at this stage of investigations but the similarities and differences between our results and archaeomagnetic data from neighbouring regions of Europe will be shown and discussed.

Sampling sites and methods

A set of hand and drill samples were taken from 10 historical buildings: castles, town hall, church and mansion constructed in northern Poland (Fig. 1a, Table 1). In total, 24 bricks from buildings of known age were studied. Single bricks or their fragments were collected during archaeological or conservation works. All drill cores were taken directly from the brick buildings (Fig. 1b) and separate bricks were cut into cylindrical specimens one inch in diameter and 22 mm long (up to 8 specimens per brick). The drilling was perpendicular to the brick’s surface in both types of devices used i.e. the portable drill “Hilti DD 30-W” with movable ring surrounding the diamond bit, that is always perpendicular to the drilling direction, and the bench drill with horizontal table. The ages of particular historical buildings were adopted from the literature (Table 1) with accuracy defined arbitrarily to be no less than 5 years, even if historical data provide more precise ages. This is due to a possible time gap between brick production and their use in the construction of buildings. In cases when the information about the age of a particular building was not direct, e.g. referred to its funding, we assumed an age accuracy of ca. 10 years. The orientation of bricks in the ovens during firing was in each epoch of the last century most probably the same, i.e. all were positioned ca. horizontally on their narrow long sides (Płuska 2009). However, it cannot be excluded that part of bricks deviated in different degree from this position.

Fig. 1
figure 1

a Location of sites selected for archaeomagnetic studies in SE Poland and listed in Table 1. b Taking samples with a drill in one of studied localities

Full size image
Table 1 Brick buildings in SE Poland sampled for archaeomagnetic studies
Full size table

Parameters of hysteresis for selected brick samples (one per each site, one or two small sub-samples for each sample with mas up to 30 mg) were obtained with AGM 2900–02 Micromag alternating gradient force magnetometer (Princeton Measurements Corporation, USA) at Laboratory for Paleomagnetism and Environmental Studies at Institute of Geophysics Polish Academy of Sciences (IGF PAS). For the same samples changes of magnetic susceptibility were monitored with the KLY5A kappabridge with CS4 high temperature unit (Agico S.r.o., Czechia) at IG PAS. Powdered samples were subsequently heated and cooled in the range of 20–700 °C in air atmosphere. The additional cyclic heating–cooling procedure was applied for one sample with increasing maximum temperature of heating in steps of 50 °C (starting from 20–100 °C range, 20–150 °C range etc. until 20–700 °C range) to monitor any possible mineralogical changes upon heating. For that sample the susceptibility after each cooling step at 50 °C was also plotted to monitor the susceptibility changes at room temperature after each cycle. The SAFYR and CUREVAL software (Agico, Czechia) were used for data processing. To monitor the changes of magnetic remanence due to heating for 10 samples (one per site) SIRM was imposed for samples below 5 mm in size with MMPM10 pulse magnetizer (Magnetic Measurements, Great Britain) at the effective field of 9 T. In the next step each sample was heated in the magnetic field free space in the STEPS device (TUS, Poland) in 20–700 °C temperature range in air atmosphere and SIRM (T) changes upon heating were recorded. Two curves were determined–for fresh unheated and for heated to 700 °C and both SIRM(T) curves were compared.

The anisotropy of TRM was evaluated following the procedure used by Veitch et al. (1984) and Chauvin et al. (2000) at temperature of 450 °C when 20–50% initial intensity of NRM was left (op. cit., see also Schnepp et al. 2009). Higher temperatures were not chosen to avoid any influence of eventual oxidation of magnetic minerals. After heating in + Z and − Z direction further heating in + X, − X, + Y, − Y and finally again in + Z direction. Due to the limited duration of the project this was done for 25 selected specimens with well-defined Arai plots up to 450 °C only. The anisotropy of TRM parameters (Jelinek 1981) was computed using the ANISOFT software (Agico S.r.o.). The 95% confidence uncertainty ellipses were determined using the linear perturbation analysis described by Jelinek (1978). For the same set of specimens a slow linear cooling time of around 16 h after heating up to 650 °C in MMTD nonmagnetic oven with fan switched off was chosen to approximate the natural cooling and to introduce a cooling rate correction following the method as described e.g. by Hill et al. (2007).

All drill cores cut into cylindrical specimens one inch in diameter and 22 mm long (up to 5 specimens per brick) were demagnetized thermally in a nonmagnetic oven MMTD (Magnetic Measurements Ltd.). The intensity of particular magnetization components was measured after each demagnetization step using a JR-6A spinner magnetometer (Agico S.r.o.) in the new palaeomagnetic laboratory of Maria Curie-Skłodowska University in Lublin. The MT4 protocol (Leonhardt et al. 2004) using a conventional thermal technique was applied to determine the palaeointensity values on the basis of Thellier and Thellier (1944, 1959) method. 38 temperature steps between 20 and 650 °C were measured, starting the experiment with a step in zero-field followed by heating and cooling. The protocol included 5 pTRM checks, 5 tail checks (Riisager and Riisager 2001), and 5 additivity checks (Krása et al. 2003). The intensity value of the laboratory field was 57 µT. The measurement data were elaborated using the Thellier Tool 4.0 software and the quality of each palaeointensity value was defined as it is presented in the ‘‘Criteria Dialog’’ (Leonhardt et al. 2004). These criteria are subdivided into several groups. Quality criteria on intensity determination, like the number of successive measurement steps (N), standard deviation of the linear fit (Std), fraction of NRM used (f), and the quality factor (q) (Coe et al. 1978) define the first group. The selected temperature segment for intensity analyses must match the temperature range which carries the characteristic remanent magnetization of the sample. Using a principle component analysis (PCA) (Kirschvink 1980), such directional aspects are monitored using the parameters of the second group (MAD, alpha). Depending on the applied experimental checks further criteria for classifying δCK (difference between the pTRM check and original TRM value at a given temperature normalised to the TRM), TR (pTRM tail checks) were applied. Results were subdivided into three different quality classes using specific criteria for classes A and B, as well as an user’s choice C, if at least one parameter violates both class A and B criteria (Leonhardt et al. 2004). In this work, all samples with a palaeointensity estimate being qualified as class C, were rejected. The following values of critical parameters for determining palaeointensity as class A and B (parameter criteria given in the brackets) were applied: number of points (N) ≥ 5 (≥ 5), standard deviation (Std) ≤ 0.1 (≤ 0.15), fraction of NRM used (f) ≥ 5(≥ 3), quality factor (q) > 5 (≥ 0), mean angular deviation anchored (MAD) ≤ 6 (≤ 15), 95% semi-angle of confidence alpha ≤ 15 (≤ 15), relative check error d(CK) ≤ 5 (≤ 7), cumulative check difference d(pal) ≤ 5 (≤ 10), difference ratio (Drat) ≤ 999 (≤ 999), normalised tail of pTRM ≤ 3 (≤ 99), relative intensity difference ≤ 10 (≤ 15), relative additivity check error d(AC) ≤ 5 (≤ 10).

Additionally, Arai Plot curvature (k) was calculated using the method suggested by Paterson (2011) and Paterson et al. (2014). The CircleFit-ByPratt (XY) function (Chernov 2020) that fits a circle to a set of data points in the plane following the method proposed by Pratt (1987) was used. The limit value of k ≤ 0.164 as indicative of SD particles as the principal carriers of remanence, and k ≤ 0.270 as indicatory of PSD particles as the principal carriers of remanence (Paterson 2011).

Demagnetization results were used for determination of archaeoinclination. They were analysed using orthogonal vector plots (Zijderveld et al. 1967) and the directions of the linear segments were calculated using principal component analysis (Kirschvink 1980). The linear segments were tied to the origin of the coordinate system.

Results

Magnetic carriers and anisotropy of TRM

Changes of magnetic susceptibility during subsequent heating and cooling differ between sites indicating varying contribution of strongly magnetic phases. The phase with high Curie temperatures in the range of about 560–580 °C is present in all samples (Fig. 2a–e). The high temperature phase of about 700 °C was detected in one site (Człuchów, Fig. 2e). The substantial decrease in susceptibility at low Tc (starting below 200 °C) is also observed for 6 of 10 sites (the most prominent for Ostróda, Fig. 2d). However K(T) cooling curves do not show a large difference from heating curves. The magnetic susceptibility at room temperature in the end of the heating–cooling procedure increased by up to 15% when compared to the initial value before heating (Fig. 2a–e). The curves of cyclic heating–cooling procedure for magnetic susceptibility with increasing maximum temperature of heating in steps of 50 °C and the curve of susceptibility after each cooling step at 50 °C after each cycle indicate no significant changes of susceptibility during the multiple heating–cooling runs (Fig. 2f, g).

Fig. 2
figure 2

a–e Changes of magnetic susceptibility monitored with the KLY5A kappabridge and CS4 high temperature unit. Powdered samples were subsequently heated and cooled in the range of 20–700 °C in air atmosphere, Red—heating curve, blue—cooling curve. f The cyclic heating–cooling procedure for magnetic susceptibility with increasing maximum temperature of heating in steps of 50 °C (starting from 20–100 °C range, 20–150 °C range etc. until 20–700 °C range). g The magnetic susceptibility after each cooling step at 50 °C plotted to monitor the susceptibility changes at room temperature after each cycle. The SAFYR and CUREVAL software (Agico, Czechia) were used for data processing. h–l Changes of SIRM upon continuous heating in air in the 20–700 °C range in the absence of the magnetic field. Red line—fresh samples, blue line—previously heated sample. Intensities are normalised to its maximum (maximum values are listed in uV, mV). No significant changes of susceptibility and SIRM are observed are observed as the effect of multiple heating–cooling runs. m Hysteresis lop for a two phases rich sample—with low and high coercivity components. Maximum field is 798 kA/m (1 T), magnetic moment is normalised per mas (25 mg), high field linear component is subtracted, initial curve is also plotted, and Micromag AGM 2900 was used. n Hcr backfield and IRM acquisition curve measurement for the same sample, maximum field is the same. o Day plot (Day et al. 1977) for 13 brick samples from historical buildings of northern Poland from 10 sites. Hysteresis parameters: Mrs/Ms versus Hcr/Hc (SD/MD and SD/superparamagnetic (SP) grains) foe magnetite are shown after Dunlop (2002) for reference only. The contribution of highly coercive phase is observed

Full size image

The SIRM (T) curves show no substantial change of the intensity of SIRM as the result of heating (Fig. 2h–l) except Boguszewo sample (Fig. 2h). Small difference in intensity is due to calibration of the instrument. The SIRM (T) curves for fresh and heated samples are very similar. It indicates that ferromagnetic (s.l.) phases responsible for archaeomagnetic record are in general very stable upon heating. In addition to the phases with blocking temperatures of 520–550 °C (9 of 10 samples) and up to 700 °C (1 sample from Człuchów, Fig. 2e, l) the prominent phase with low blocking temperature (below 200 °C) is observed for all samples. The hysteresis curves for all samples indicate the presence of two phases–one with low coercivity and the other with very high coercivity but with varying proportion of them in each locality. As the results for samples with dominating contribution of high coercivity phase wasp-waisted hysteresis lops are observed with high Hcr values (above 100 mT) and SIRM steeply increasing up to the maxim applied field of 1 T (Figs. 2a, n). The distribution of Mr/Ms vs Hcr/Hc hysteresis parameters reflects this varying proportion of both phases (Table 2, Fig. 2o). No conclusion of the domain states can be formed because magnetite is not the only or dominating magnetic phase present in studied samples. The results of K(T) and SIRM (T) as well as the hysteresis are consistent with the presence of two stable ferromagnetic phases–one low temperature and high coercivity phase (epsilon iron oxide) and the other of high temperature and low coercivity (magnetite or/and maghemite) (see e.g. López-Sánchez et al. 2017).

Table 2 Hysteresis parameters for samples from ancient bricks sampled in N Poland
Full size table

The degree (K1/K3) of the TRM anisotropy of most of samples ranges from 1.02 to 1.3. This parameter is significantly higher in four specimens only (Fig. 3). These values and their distribution are the same as noted in the backed bricks from France (Chauvin et al. 2000). The magnetic fabric in ca. 50% of the specimens is dominated by flattening (Fig. 3b). The limit of acceptance of alteration as not higher than 15% was assumed (see Hill et al. 2007). The alteration parameter was obtained from the difference between the two pTRM acquired along + Z.

Fig. 3
figure 3

Distribution of degree of anisotropy P of the TRM measured for 25 samples of bricks from SE Poland (a) and comparison of degree of anisotropy of the TRM and shape parameter T for the same sample set (b)

Full size image

Results of Thellier–Thellier experiment

The evaluation of Thellier experiments for 48 specimens by ThellierTool4.2 software (Leonhardt et al. 2004) using the default values for classification, provided 28 specimens that passed the automatic routine and were classified as category B. Category C was defined for 20 specimens that were not taken into consideration for further statistical evaluation. It means that all specimens passed the automatic routine of used software. Figure 4 summarises the criteria used for the determination of A and B categories of quality. Limits between them are marked by vertical lines. It is clearly visible that the majority of specimens passed the linear fit and direction criteria with values characteristic for category A. Lack of specimens of category A is mainly due to poor values of parameter d(t*) characterising repeated demagnetization steps. Data points in the Arai-diagrams, as well as in the Zijderveld-diagrams form straight lines (Fig. 5a, b). Most of samples (63%) display a significant drop of intensity of NRM at 200 °C (Fig. 5c). This fact may indicate the presence of the epsilon iron oxide in the studied specimens (e.g. López-Sánchez et al. 2017). The created pTRM is aligned parallel to the laboratory field and the pTRM check difference is generally less than 5% (Fig. 5d). A repeated demagnetization shows no substantial change (Fig. 5c). This observation corresponds well with the magnetic susceptibility behaviour during heating/cooling cycles (Fig. 2).

Fig. 4
figure 4

Distribution of quality criteria calculated for individual specimens that allowed to distinguish category A and B (boundary marked by vertical lines) of the Thellier results (see Leonhardt et al. 2004 and text). Number of points–number of successive measurement steps, Std—standard deviation of the linear fit, q—quality factor (Coe et al. 1978), f—fraction of NRM used, ALPHA—95% confidence limit, MAD—mean angular standard deviation, d(CK)—relative check error i.e. the difference between the pTRM check and original TRM value at a given temperature normalised to the TRM, d(t*)—normalized tail of pTRM, d(TR)—relative intensity difference, d(AC)—relative additivity checks error, dpal—cumulative check difference

Full size image
Fig. 5
figure 5

Results of Thellier experiments obtained from representative specimens of category B showing a the Arai plot with pTRM checks as triangles and additivity checks as squares; b orthogonal vector components; c NRM intensity versus temperature, black squares indicate the results of the repeated demagnetization (TR); d parameters related to quality of Thellier experiments (see Leonhardt et al. 2004)

Full size image

The Thellier method results for individual specimens ranged from 54.2 to 69.3 µT (Online Appendix 1). The correction factor related to the anisotropy of TRM falls between 0.82 and 1.06 (Online Appendix 1). In total, the Thellier data for 25 specimens were corrected for the anisotropy of TRM.

The cooling rate experiments have provided acceptable data (see Chauvin et al. 2000; Hill et al. 2007) for all the specimens with measured the anisotropy of TRM. Because the correction factor is negative in part of samples, their magnetic remanence cannot be carried out by SD magnetite particles only (Dodson and McClelland-Brown 1980).

Online Appendix 1 summarises the curvature results for 28 specimens. Finally, eleven specimens exceed the limit value of k ≤ 0.270 (Paterson 2011). Most of them (8) displayed concave-up shape but three of them gave concave down shape (Fig. 6). It should be stressed, however, that in the same sample specimens with k higher than 0.270 have provided very similar values of archaeointensity than those with k below this limit (Online Appendix 1).

Fig. 6
figure 6

Example Arai plots from bricks sampled in northern Poland. NRM and TRM have been normalised by the respective maximum values. Circles represent the measured TRM–NRM data and the solid lines are the best-fit circles to the measured data. The values of curvature parameter (k) are also presented

Full size image

Discussion

The magnetic fraction of studied bricks is composed of magnetite and the epsilon iron oxide (López-Sánchez et al. 2017). It cannot be excluded also a certain contribution of maghemite in part of them. The archaeointensity data corrected for anisotropy and cooling rate were defined for 25 specimens cut from 14 bricks that were taken from 9 sites (Table 3). Data from the site Lidzbark Warmiński (Table 1) were not corrected. Relatively large number of specimens with category C of archaeointensity (20 specimens, Online Appendix 1) is due to high value of additivity check error d(AC) (Krása et al. 2003) that exceeds in these specimens 10 (see Leonhardt et al. 2004). These specimens contained most probably a significant amount of multi-domain grains of magnetite, which can lead to ambiguous interpretations of the archaeointensity experiment (Krása et al. 2003). The presence of multi-domain magnetite can be concluded from the Arai plots data with concave-up shape (Fig. 6) (Paterson 2011). Bricks from N Poland were made of Pleistocene varved clays or glacial tills, sometimes with an admixture of sand (Wasik 2017). Therefore, the multi-domain magnetite can occur at least in part of them.

Table 3 Summary of archaeomagnetic data from northern Poland
Full size table

One of the problems of interpretation of archaeomagnetic data from ancient bricks is credibility of their age because some of them could be older than studied building being reused from other ancient constructions. On the other hand, during any later reconstructions younger bricks could be also used. Fortunately, in the state of the Crusades any reconstructions of castles and other brick buildings were precisely noted in the written sources. However, to avoid the risk, a significant part of sample collection studied here was taken from the foundation of the buildings or their basements (Table 1, localities 1, 2, 3, 6 and 10). Samples from Ostróda (Table 1, locality 7) were taken from the bailey that was built during the initial phase of the construction of the castle. Samples from Olsztyn (Table 1, locality 8) were collected directly under the roof truss dated by the dendrochronological method. Brick samples from Boguszewo (Table 1, locality 9) were taken from uniform wall containing on its uppermost part the inscription with the date of construction. In the case of locality nr 4 (Table 1) samples taken from the castle tower and from the floor revealed almost identical archaeointensity values (Online Appendix 1). It supports thesis about age homogeneity of the whole construction. Two bricks from Lidzbark Warmiński (Table 1, locality 5) were collected during archaeological prospection from the outcrop. Only in this case the risk of age inaccuracy is high.

The mean archaeointensities for studied localities are compared with the archaeointesity curves with an error envelope for western and central Europe (Schnepp et al. 2009) and the archaeointesity curves for models BIGMUIh.1 (Arneitz et al. 2021) and SCHA.DIF.4k (Pavón-Carrasco et al. 2021), and for NW Russia (Salnaia et al. 2017). It is clearly visible that all corrected data from northern Poland fit into first curve (Schnepp et al. 2009) best (Fig. 7a). The introduction of correction for anisotropy of TRM and cooling rate improve substantially internal homogeneity of data at brick level in most of localities (Table 3) and their convergence with the reference curves (Fig. 7a).

Fig. 7
figure 7

a Mean archaeointensities obtained from part of studied brick buildings from N Poland (numbers of sites as in Tables 1 and 2) presented for the raw data and after correction for the anisotropy of TRM and cooling rate on the background of Bayesian curve with error envelope (grey area) for western-central Europe (Schnepp et al. 2009). The paleosecular variations curves for models BIGMUIh.1 (Arneitz et al. 2021) and SCHA.DIF.4 k (Pavón-Carrasco et al. 2021), and for NW Russia (Salnaia et al. 2017) are also presented b Mean archaeoinclinations defined for historical brick buildings from N Poland (raw and corrected for the anisotropy of IRM data) on the background of archaeointensity curve (with its error limit) defined for Austria and Germany (Schnepp et al. 2020). All data are recalculated for geographic coordinates of the city of Olsztyn. Vertical bars define standard deviation errors. Mean values of archaeointensities and archaeoinclinations were calculated as arithmetic averages

Full size image

The distributions over time of the archaeointesity and archaeoinclination mean values for particular sites (recalculated for geographical coordinates of Olsztyn: 22° 27′ E, 53° 48′ N) on the background of reference curves (Fig. 7a, b) confirm two main features of geomagnetic field defined in other parts of Europe i.e. only small drop of archaeointensity between the end of 13th and beginning of 17th century, followed by a significant increase in archaeoinclination at that time (see Chauvin et al. 2000, Schnepp et al. 2009; Kovacheva et al. 2009; Molina-Cardín et al. 2018).

The mean values of inclinations, although convergent with the reference curve (Fig. 7b) mostly display big values of standard deviations, even up to 10.7° (Table 3). This is most probably due to not quite horizontal position of the bricks in the oven during their firing and not sufficient quantity of them. This may not apply the data from the bricks obtained in Lidzbark Warmiński (Table 1, locality 5) where large age inaccuracy cannot be excluded. The orientation of bricks during firing was in each epoch of last century most probably the same, i.e. all were positioned ca. horizontally on their narrow long sides because they are stronger in this position (Beamish and Donovan 1993; see also Lanos 1987 and Płuska 2009). However, it cannot be excluded that part of bricks deviated in different degree from this position. The averaging of inclinations obtained from more bricks should limit the problem of inaccurate horizontal position of them making the mean inclination more credible. However, this is not so obvious in the case of mean inclinations from 2 to 5 bricks where this improvement of convergence is not visible as is shown on the example of data from N and SE Poland (Fig. 8). Most likely significantly more than five bricks from one site studied for this purpose can lead to a better averaging of inclination should improve this statistical parameter and credibility of mean values.

Fig. 8
figure 8

Standard deviations of inclinations obtained from bricks sampled in northern (this paper) and south-eastern (Nawrocki et al. 2023) Poland versus number of them used for calculations of this parameter. Data are presented on the locality level

Full size image

Further studies of Gothic to Baroque bricks from N Poland for increasing statistical representativeness and fulfilling whole time period from 13th to 18th centuries will be conducted.

Conclusions

  1. 1.

    The results of petromagnetic studies are consistent with the presence of two stable ferromagnetic phases in studied bricks–one low temperature and high coercivity phase i.e. epsilon iron oxide and the other of high temperature and low coercivity i.e. magnetite. A certain content of maghemite is also possible.

  2. 2.

    The evaluation of Thellier experiments provided 28 brick specimens (58% of whole sample set) that passed the automatic routine and were classified as category B. In total, the Thellier data from 9 sites and 25 specimens were corrected for the anisotropy of thermal remanent magnetization and cooling rate. The curvature parameter k exceeds the limit value in 11 specimens. However, in the same sample specimens with k higher than 0.270 have provided very similar values of archaeointensity than those with k below this limit.

  3. 3.

    A large number of specimens classified as category C is due to a high value of relative additivity check error d(AC) that exceeds 10. A significant multi-domain magnetite content in these specimens is therefore very likely. Its presence can be inferred from the Arai plots data with concave-up shape.

  4. 4.

    Corrected data are convergent with the Central European master curve of archaeointensity. The corrections of raw data reduce their dispersion at specimen/sample level in most of sites.