Keywords

1 Introduction—Basic Principles of Environmental Magnetic Technique Applied to Archaeological Context

The principles of environmental magnetism are based on the link between the set of (soil) environmental parameters, such as for instance climate (through temperature and precipitation), lithology, time (Schaetzl & Anderson, 2005) and related dynamics of iron (oxy)hydroxides synthesis conditions and transformation pathways (Cornell & Schwertmann, 2003). The impact of human’s influence on the above association is in the core of the application of environmental magnetism techniques in archaeological prospection (Fassbinder, 2015) and artefact research.

The incorporation of environmental magnetism approach into archaeological research has been rather sparse so far, and focused mainly on investigations of anthropogenic deposits in caves (Vergès et al., 2016; Carrancho et al., 2016), while applications to other archaeological contexts are rare (Mooney et al., 2003). Therefore, a need to demonstrate the high potential impact of applying environmental magnetism approach in various topic-oriented archaeological frameworks is welcome. In this contribution, we intend to summarise our experience on the application of magnetic investigations of soils, sediments and archaeological materials of fired clays and show major possible fields of applicability, providing potentially high impact for archaeology. In this respect, this summary fits with the general objectives of the SAGA COST Action to create a better and mutually beneficial link between archaeo-geophysics and archaeology.

The major mineral magnetic parameter in environmental magnetism is volume low-field magnetic susceptibility (k, measured in SI units) or mass-specific magnetic susceptibility (χ in units m3/kg). The concentration of strongly magnetic iron phases in the measured material is the major factor, determining the magnetic susceptibility value, although grain size of the magnetic fraction also plays a role (Thompson & Oldfield, 1986). Magnetic properties of soils are widely utilised as receptive recorders of paleo-environmental conditions during their development (Liu et al., 2012; Jordanova, 2016; Maxbauer et al., 2016). Magnetic grain-size sensitive parameters additionally enlarge the toolbox for extracting relevant paleoenvironmental information from soils. Frequency-dependent magnetic susceptibility (χfd or χfd%) provides important indications of the presence of very fine nanometre-sized magnetic particles which are usually of pedogenic origin in natural soils (e.g. Maher, 1998). Laboratory-induced remanent magnetisations—Anhysteretic (ARM) and Isothermal (IRM)—give further grain-size sensitive traces, linked to (pedogenic) formation of stable single-domain magnetite grains (Maher & Taylor, 1988) and/or presence of weakly magnetic hematite and goethite mineral fractions, respectively. Identification of the mineral magnetic phase(s) in natural materials additionally provides knowledge about the processes and stability of iron containing mineral phases which are related to environmental factors (Cornell & Schwertmann, 2003).

Archaeological remains of burnt clay are composed of fired natural materials (clays, soil, sediments, etc.) which have been strongly influenced by thermal processes. Firing to high temperatures causes various thermo-chemical transformations in the clay/soil mineralogy including dehydration, oxidation, dehydroxylation, decomposition and formation of new phases (Murad & Wagner, 1998; Murad, 1998). Along with the change of all physical/chemical properties, magnetic properties of fired clay are also strongly influenced by heating (Le Borgne, 1955; Jordanova et al., 2001; Beatrice et al., 2008). Thermo-chemical growth of new iron oxides drives the magnetic enhancement of fired materials through formation of strongly magnetic phases like magnetite and maghemite (Murad & Wagner, 1998; Wagner et al., 1998), although prolonged firing in air results in dominance of weakly magnetic hematite (α-Fe2O3) and/or other Fe2O3 mineral phases, such as ε-Fe2O3 (McIntosh et al., 2007).

A compilation of basic magnetic data of archaeological materials from fired clay remains provides a general overview of the distribution and variability of magnetic susceptibility (χ) and frequency dependent magnetic susceptibility (χfd%). Materials from burnt clay (e.g. house destructions; burnt soil; remains of unknown origin), bricks (from AD period) and hearth/kilns (materials from preserved structures (hearths, kilns, ovens)) are compiled from the collection inventory of the Bulgarian archaeomagnetic database (Kovacheva et al., 2014). The pottery data originates from ceramic fragments from six major ceramic manufacturing centres in Bulgaria. Histograms of χ and χfd% distribution are shown in Fig. 1. The data demonstrate that the magnetic susceptibility of different archaeological materials varies over a wide range, but in all materials χ is positively skewed (Fig. 1). The highest median is obtained for pottery and the lowest for baked clay. Positive skewness is frequently observed for environmental variables, which cannot take negative values and are thus constrained by zero (Parkin & Robinson, 1992). Although the highest number of samples investigated from baked clay and hearths/kilns have their χ between (100–300) * 10−8 m3/kg (Fig. 1), a second group in the range (400–600) * 10−8 m3/kg could be noticed. The most scattered is the histogram of χ for brick materials. Percent frequency dependent magnetic susceptibility (χfd%) exhibits differing distribution for the four materials categories. For baked clay the χfd% distribution is close to normal, for hearths/kilns it is negatively skewed (Fig. 1) and for bricks it is positively skewed. Pottery materials show also negatively skewed histogram. Nevertheless, calculations of the median χfd% gives a value of 8% for all kind of materials. The distribution of χfd% values across the various baked materials suggests that firing produces the most significant portion of fine superparamagnetic grains during brick’s production because among all materials more brick’s samples exhibit χfd% > 8 (Fig. 1). This is consistent with the production technology of the materials, requiring high firing temperatures, usually well above 850–900 °C for a prolonged time (Boccalon et al., 2019; Lopez-Arce & Garcia-Guinea, 2005), while ovens/kilns for domestic usage and spontaneous firing (for the class of baked clay materials) commonly do not achieve such high temperatures. In contrast to those fired structures, natural soils exhibit much weaker magnetic enhancement, reaching usually (80–100) * 10−8 m3/kg and depending on the soil type and respective inherent pedogenic processes (Jordanova et al., 2016). It should be noted, however, that anomalously high soil susceptibilities may be obtained for natural soils developed on strongly magnetic parent materials (Grison et al., 2015).

Fig. 1
8 bar charts of frequency versus chi and the percentages of X f d for baked clay, hearths or kilns, bricks, and pottery with different values of N. 4 graphs at the top have decreasing trends. 4 graphs at the bottom exhibit an initial increase followed by a decline.

Histograms of mass-specific magnetic susceptibility (χ) in “10–8 m3/kg” and percent frequency dependent magnetic susceptibility (χfd%) for various types of archaeological materials of fired clay—“baked clay” (fired destructions; burnt soil; remains of unknown origin); hearths/kilns (materials from preserved structures); bricks and pottery. “N” denotes the number of samples included in each category

Full size image

2 Magnetic Susceptibility and Equivalent Firing Temperature of Archaeological Remains of Burnt Clay May Yield Functional Information About Ancient Environmental Settings

In a recent study (Jordanova et al., 2020a) we investigated burnt clay materials resulting from destructive open fires, not related to human house-hold activities (cooking, ware production, etc.). The collection includes materials from 18 Neolithic sites from Bulgaria. The most important characteristic of all sites is that they were affected by an extensive fire which ended their existence. These sites are dated between ~6000 BC and ~4000 BC, spanning the local Early- Late Neolithic and Eneolithic periods (Görsdorf & Bojadziev, 1996). The aim of the investigation was to probe the sensitivity of the basic magnetic property parameters against the most important environmental variables playing role (e.g. climate, lithology, time). In addition to low-field susceptibility and frequency dependent magnetic susceptibility, we involved in the analysis results from the evaluations of the maximum firing temperatures achieved during fire (Tfire). Determinations of Tfire were accomplished utilising the method proposed by Rasmussen et al. (2012). This method is based on the assumption that during ancient pottery firing the fraction of iron (oxy)hydroxides in the initial clay material was chemically transformed due to heating and thus the magnetic mineral fraction would be thermally stable during laboratory re-heating to temperatures lower than the maximum ancient firing temperature. When passing this threshold temperature, iron oxides would continue further transforming along with the appearance of new or the disappearance of existing magnetic minerals. Thus, the abrupt change in magnetic susceptibility during laboratory re-heating tentatively indicates the maximum firing temperature achieved during the ancient pottery production (Rasmussen et al., 2012). More details on the methodological aspects can be found in Jordanova et al. (2020a) while here we would like to focus on the main outcomes related to geoarchaeological aspects. Results for the firing temperature estimates are considered at site level taking the average of all the individual determinations. Further on, data obtained at site level are considered according to the division of the Neolithic period in the Balkan Peninsula, as used in Marinova and Ntinou (2018). The summary of the Tfire estimates for each period is shown in Fig. 2 using bean plot (BoxPlotR software (http://shiny.chemgrid.org/boxplotr/) (Spitzer et al., 2014)). Together with the calculated median Tfire value, individual determinations are also visualised along with the data density distribution.

Fig. 2
a. A bean plot plots T fire versus period. The highest value is plotted by the beans of 5800 to 5500 B C and the lowest value is plotted by the beans of 4900 to 4000 B C. b. An error graph plots chi versus T fire. The North Bulgaria symbols generally plot higher values than the south.

(a) Bean plot of the distribution of firing temperature estimates (Tfire) for the four time periods. Black bold lines show the medians; red lines represent individual data points; shaded polygons represent the estimated density of the data. Plot created using BoxPlotR software (http://shiny.chemgrid.org/boxplotr/) (Spitzer et al., 2014); (b) bi-parametric plot of magnetic susceptibility (χ) versus firing temperature (Tfire) averaged at site level. Each data is represented with its standard deviation in both parameters. Red symbols denote sites located in south Bulgaria (relative to the Balkans mountain chain), while blue symbols denote sites from North Bulgaria

Full size image

As it is seen in Fig. 2a, a highest Tfire ~ 815 °C median is attained in the period 5800–5500 BC, followed by systematically lower Tfire medians in the succeeding periods of the Neolithic. Plotting the relation of magnetic susceptibility vs. Tfire (Fig. 2b) estimates at site level, two distinct trends are clearly observed—the steeper linear trend is obtained from sites located in North Bulgaria (blue symbols), while all other sites located south from the Balkan chain obey to a different linear tendency. This implies that for the same Tfire estimates burnt clay from the sites located at the Danube plain (north from the Balkan Mountain chain) acquire a stronger magnetic susceptibility when compared to the other sites. The major influencing factor should be sought in the lithological differences in surface geology across the territory. In the Danube area the main lithological unit is represented by loess deposits (Evlogiev, 2006), with significant content of calcium. In contrast, sites from south Bulgaria were built mainly in Quaternary sediments deposited along the river valleys (Spassov et al., 2006). Those sediments are dominantly composed of alluvial- to lacustrine deposits containing mostly sands, sandy clays and clay-shales. As demonstrated in the work of Maniatis et al. (1981), calcium presence strongly affects the clay properties, including formation of high amount of iron oxides upon heating. These authors conclude from their results that destruction of calcareous clays during firing produces aluminosilicate matrix which effectively traps the iron ions and iron oxides do not grow easily upon further heating. In contrast, in Ca-free clays heating to increasing temperatures favours increasing amount and size growth of new iron oxide particles (Maniatis et al., 1981). Other factors, like the wood used for the construction of Neolithic houses, as well as the use of dung into daub preparation (Kruger, 2015), also can influence the firing process and the amount and kind of iron oxide phases formed. Such organic additives in clay aid creation of reducing condition during firing and enhance the process of sintering at temperatures above ~800 °C. Therefore, multiple factors determine the amount and size of iron oxides formed during firing. Therefore, the degree of magnetic susceptibility enhancement upon firing is variable and results in wider changes of χ and Tfire at site level (Fig. 2b). Despite the data scatter, the estimated Tfire medians for the major time intervals during the Neolithic suggest maximum values in the interval 5800–5500 BC with decreasing values towards the Eneolithic (Fig. 2a). Such evolution is consistent with the major climate conditions and palaeofire regime during the Holocene. An increase in global fire activity during the climatic optimum in the Middle Holocene is reported in a number of works (e.g. Brücher et al., 2014 and references therein).

3 Environmental Magnetism as a Tool in Reconstructing Extinction Fire in Ancient Neolithic Settlement

In several Neolithic sites in South-East Europe and Anatolia the phenomenon of burnt houses was discovered (Brami, 2014; Stevanović, 1997). Sintered daub defined as a mixture of clay and organic materials like straw, grass, animal dung, etc. (Kruger, 2015) provides important information about the firing conditions and interpretation of structural burning in the archaeological record (Harrison, 2013). We carried out a detailed and extensive study on a large collection of burned clay materials from the Neolithic site Mursalevo-Deveboaz (Jordanova et al., 2018) with the aim to reconstruct the maximum temperatures reached during the fire which ended the settlement’s life. The prehistoric village is dated to a later phase of the local Early Neolithic (ca. 5700–5500 BC). Remains of 62 houses from the second half of the Early Neolithic and 13 houses from the beginning of the second half of the Late Neolithic, all destroyed by extensive fire were unearthed during archaeological excavations. Mineral magnetic analyses involving a set of techniques were applied to investigate the iron oxide phases and magnetic grain sizes responsible for the enhanced magnetisation of burned clay materials. A large quantity of magnetite/maghemite and hematite, all of very fine (nanometre) grain size were evidenced through detailed multi-parameter rock magnetic analyses (Jordanova et al., 2018). Firing temperature estimates using the magnetic susceptibility method for 148 samples of different colour vary between 680 and 1140 °C (Jordanova et al., 2018). It was established that the obtained Tfire is generally linked to the predominant colour of the daub piece (Fig. 3). The lowest Tfire is characteristic for brown coloured daub, followed by orange-red daub. The highest Tfire were registered in materials from vitrified violet to yellow colour of the daub (Fig. 3). It is well known that the colour of thermally altered clay is primarily determined by the pigmentary iron oxides (Murad & Wagner, 1998; Cornell & Schwertmann, 2003). Further, laboratory combustion synthesis of iron oxides reveals a definitive link between the temperature of solution combustion synthesis and the colour, as well as the grain size of the iron oxides synthesised (Toniolo et al., 2007). The magnetic data obtained from our large-number daub collection suggest an increase in the magnetic grain size from superparamagnetic towards single-domain with increasing firing temperature. Moreover, for samples dominated by strongly magnetic magnetite/maghemite phases a linear relationship between low-field magnetic susceptibility and the Tfire estimates is evidenced (Jordanova et al., 2018). This rule is not obeyed by vitrified violet-coloured and yellow daub which contain high amounts of hematite. Therefore, considering our experimental data and the evidence from laboratory synthesis of iron oxides, we hypothesised that the major mechanism responsible for the extreme firing of Neolithic houses is related to a combustion event.

Fig. 3
Top. A photo of the ground has stones and mud. Bottom. A cluster bar graph plots frequency versus temperature. The bar of brown plots the highest value at (710, 24), followed by bar of orange red at (820, 14). Data are estimated.

Distribution of firing temperature estimates (Tfire) from Neolithic house destructions in Mursalevo-Deveboaz, as related to the dominant colour of the materials, according to the legend shown. The photograph displays variations in colour of the fired materials

Full size image

4 Mineral Magnetic Properties of Archaeological Materials from Mining Archaeology Settings Are Powerful Index for Their Recognition and Allocation

The Balkan Peninsula and the Bulgarian territory in particular, are part of the global-scale collision zone of the Alpine–Himalayan orogenic belt (Marchev et al., 2005). The tectonic units of the Alpine–Balkan–Carpathian–Dynaride orogenic system (Heinrich & Neubauer, 2002) determine the most important structural geology features. Therefore, various ore deposits have been explored since ancient times (Popov & Jockenhövel, 2011; Radivojević & Roberts, 2021). We have studied a collection of samples from the Late Bronze Age open-pit gold mine at Ada Tepe in the Eastern Rhodopes (South-Eastern Bulgaria) (Jordanova et al., 2020b). The most recent radiocarbon dates suggest that gold mining had started in the early fifteenth century BC (Popov & Jockenhövel, 2018). Thus, at this stage of the progress of mining archaeology, Ada Tepe is considered as the oldest known open pit gold mine in Europe. The collection consists of 177 samples, representing materials from waste heaps, pristine rocks, natural soils and soils from cultural layers. Detailed mineral magnetic investigations of the collection aimed to identify the main magnetic minerals and their magnetic grain-size characteristics and classify the materials independently of their archaeological assignment. Thus, results probe the suitability of mineral magnetism itself as a tool for identifying archaeological materials associated with ancient mining activities. A set of mineral magnetic techniques (thermomagnetic analyses of high-temperature behaviour of magnetic susceptibility, decomposition of isothermal remanent magnetisation acquisition curves into coercivity components) was used to identify the major iron (oxy)hydroxides in the materials. Magnetite/maghemite, hematite, goethite and pyrrhotite were identified as an outcome. Magnetic parameters, including magnetic susceptibility and its frequency dependence, remanent magnetisations (anysteretic and isothermal) and a set of bi-parametric ratios were subjected to statistical treatment to retrieve the most reliable number of clusters, describing the variability in the magnetic properties of the materials (Jordanova et al., 2020b). It was found that four clusters explain in the best way the variability of magnetic data. Respective average values with their standard deviations obtained for the four clusters are shown in Fig. 4. As seen from the figure, clusters 1 and 4 display relatively similar magnetic parameters, but in cluster 1 the anhysteretic remanence (ARM) has significantly higher values compared to cluster 4. This suggests that materials in cluster 1 contain higher amount of stable single-domain magnetite/maghemite fraction (Maher, 1988). At the same time, the superparamagnetic fraction, characterising the finest magnetic grains (usually having pedogenic origin (Dearing et al., 1996)) is less represented. Such relationship is characteristic for fire-affected soils (Jordanova et al., 2019) and cluster 1 is thus attributed to burned soils. Cluster 2 is a particular group of samples showing very low susceptibility and high concentration of magnetically hard minerals (hematite and goethite) (e.g. high HIRM values and low S-ratio, as seen from Fig. 4) and is thus ascribed to host rocks. Samples grouped in cluster 3 have the strongest magnetic enhancement, as seen from the maximum values of χ, χfd, χARM (Fig. 4). The hard isothermal remanence (HIRM) is also elevated but it is most probably due to the presence of moderately high-coercivity minerals since the S-ratio is relatively high. Thus, materials from cluster 3 were attributed to burnt sediments/soils and heap material from rock processing with the use of fire. Finally, samples from cluster 4 show magnetic parameters which are similar to those ones for natural soils (e.g. Jordanova, 2016) and therefore these materials were defined as natural soils. After assigning each cluster to a particular type of material, the cluster members were plotted on a spatial map depicting the archaeological observations and attributes (Jordanova et al., 2020b). Comparison between the mineral magnetic results and archaeological information showed an exceptionally good match. This allowed us to conclude that the mineral magnetic approach, combining various rock-magnetic parameters of distinctive natural and human-affected relics of ancient ore mining is a highly promising and efficient tool for identification and classification of materials found in archaeological excavations.

Fig. 4
A set of 12 bar graphs with error bars, each with 3, plot 4 clusters versus X, X arm, H I R M, X arm over X, H I R M over X, X f d, X f d percentage, and S. X arm and H I R M over X have the highest values for clusters 1, 3, and 2, respectively. S has the highest values for cluster 4, respectively.

Distribution of mineral magnetic parameters for samples from Ada Tepe according to the four clusters separated by statistical analysis. The following variables are shown: magnetic susceptibility (χ), anhysteretic susceptibility (χARM), hard isothermal remanent magnetisation (HIRM = 0.5 * (IRM2T − IRM0.3T)); ratios: χARM/χ, HIRM/χ; frequency-dependent magnetic susceptibility (χfd), percent frequency dependent magnetic susceptibility (χfd%), S-ratio (S = −IRM0.3T/IRM2T)

Full size image

5 Recovery of Ancient Firing Temperatures of Archaeological Pottery Fragments by Magnetic Susceptibility Method

Pottery fragments are the most common and abundant finds in archaeological sites. They are used to constrain the dating period of the respective site and/or inhabited layer, as well as to project the societal links and technologies through time and space (e.g. Loney, 2000; Shennan & Wilkinson, 2001; Borck et al., 2020; Pawlowicz & Downum, 2021). Determinations of firing temperatures of pottery sherds could provide valuable analytical data for considering archaeologically relevant aspects—provenance studies, characterisation of ceramics technologies, their social links and spatial spread in time (Damjanovic et al., 2014). Pottery firing procedures differ in relation to duration of the firing process, heating/cooling rate, maximum firing temperature, soaking time, firing atmosphere, etc. Complex interplay of all these factors determines the final characteristics of the pottery. Several papers already report data on reconstruction of pottery firing temperatures using the method of magnetic susceptibility (Rasmussen et al., 2012; Goodwin & Hollenback, 2016; Karacic et al., 2016; Kostadinova-Avramova et al., 2018; Lesigyarski et al., 2020; Jordanova et al., 2019) as well as other mineral magnetic signatures (Spassov & Hus, 2006; Tema & Ferrara, 2019; Tema et al., 2022). In this contribution, we summarise data obtained for firing temperature estimates of pottery from several Bulgarian archaeological sites, representing major pottery production centres. Detailed information could be found in the devoted articles (Kostadinova-Avramova et al., 2018; Jordanova et al., 2017, 2019; Lesigyarski et al., 2020). In addition, we summarise here the obtained data on firing temperature determinations in the different sites according to the archaeological periods covered (Table 1). Since for the pottery wares one important classification mark is their purpose of use (serving ware vs cooking ware, decorative, etc.), in Table 1 data for the maximum firing temperature (Tmax) are sub-divided also related to this index. As it is seen from Table 1, the major inference that could be stated is that the firing temperatures for pottery production increase towards more recent times, as also concluded earlier (Kostadinova-Avramova et al., 2018). In order to be able to look for differences related to Tmax and the purpose of use of the pots, it is important to have more numerous sets of single determinations. Nevertheless, as pointed out by Kostadinova-Avramova et al. (2018), the main tendency of indistinguishable Tmax during more ancient epochs (Middle Bronze Age (MBA), Early Bronze Age (EBA)) suggests no selective use of different kilns for pottery baking by ancient potters, while during most recent times (e.g. for example ceramic centres in Pliska and Veliki Preslav, see Table 1) Tmax obtained for the serving wares, and especially the glazed ones are significantly higher compared to Tmax determined for the cooking pots. This finding corresponds well to the technological refining of pottery production with time. The lowest Tmax in our dataset summarised in Table 1, are obtained for black coloured pottery fragments from the Early Iron Age (see sites Dragovishtitsa and Gluhite kamani). As commented in many archaeological studies, this period is characterised by a strong decline in societal and cultural development of human occupation in Eastern Mediterranean and West Asia (Kostadinova-Avramova et al., 2021 and references therein). Various studies show that black-coloured pottery is most often produced in “pit firing”, while red-coloured pots were produced in kilns (a more sophisticated and requiring refined skills technology) (e.g. Maritan et al., 2005).

Table 1 Summary of the data for the firing temperatures of pottery from several archaeological sites
Full size table

Furthermore, complementary to determinations of firing temperatures using the low-field susceptibility, more advanced mineral magnetic analyses aiming to establish also magnetic grain-size and colour characteristics of pottery and burned clay materials were carried out on a pilot collection of materials (Jordanova et al., 2019). The main results demonstrated that the grain size of the secondary iron oxides vary from superparamagnetic to stable single domain or pseudo-single domain. Pottery sherds were shown to contain more often hematite in comparison with burnt clay from house destructions. Importantly, combination of mineral magnetic analyses and spectroscopic colour measurements demonstrate that the ratio “value/chroma” shows an inverse relation with the ancient firing temperatures, as determined using magnetic susceptibility method (Jordanova et al., 2019). Specific regressions were obtained for different sites (e.g. different source clays), suggesting potential use of this relation for provenance studies.

6 The Power of Classical Archaeomagnetism—Bulgarian Master Curves of Geomagnetic Field Variations

The beginning of archaeomagnetic studies in Bulgaria was set in 1967 by Prof. Mary Kovacheva. Due to continuous work and successful cooperation between Bulgarian archaeomagnetists and archaeologists, a unique archaeomagnetic database of geomagnetic field determinations has been accumulated for more than half a century. It represents an extended local series of the three main geomagnetic elements—declination (D), inclination (I) and intensity (Fa), recovered from the same baked clay materials and covering almost completely the last 8 millennia. Bulgarian archaeomagnetic data are part of the global database GEOMAGIA50.v3 (http://geomagia.ucsd.edu) and are crucial for the construction of the geomagnetic reference curves of the Balkans and South-Eastern Europe (Pavón-Carrasco et al., 2010; Tema & Kondopoulou, 2011). Over the years, the secular variation curves (also called reference curves) for Bulgaria have been calculated several times (e.g. Kovacheva et al., 2009, 2014) since the database undergo regular updating and revision. In the most recent compilation (Kovacheva et al., 2014) the D, I and Fa reference curves were smoothed by Bayesian statistics (Lanos, 2004) applied to 310 reference points obtained by archaeomagnetic investigation of numerous, relatively evenly distributed in space, archaeological sites of different age (Figs. 5, 6 and 7).

Fig. 5
A map of Bulgaria highlights archaeological sites of the Neolithic, Eneolithic, Transition L E to E B, Bronze age, Iron age, Roman period, late Antiquity, First Bulgarian Kingdom, Byzantine period, Second Bulgarian Kingdom, Ottoman period, and New times.

Distribution of the archaeomagnetically studied sites at Bulgarian territory. Different archaeological epochs and historical periods are shown in different colors. Abbreviations: LE - Late Eneolithic; EB - Early Bronze age

Full size image
Fig. 6
3 error bar graphs. Top. The early middle age, Neolithic, and late middle age have high declination values. Center. The Iron age and early middle age have high inclination values. Bottom. The Iron age, early middle age, late Antiquity, and Roman period have high intensity values.

Archaeomagnetic data with the corresponding errors over the time scale

Full size image
Fig. 7
3 line graphs. Top. The declination line plots a fluctuating curve peaking at (negative 5,000, 20). Center. The inclination line plots a fluctuating curve peaking at (negative 1,000, 70) and (1,000, 70). Bottom. The intensity line plots a fluctuating curve peaking at (negative 500, 90). Data are estimated.

Bulgarian secular variation curves of the three geomagnetic field elements—inclination (I), declination (D) and intensity (Fa) (Kovacheva et al., 2014)

Full size image

The most ancient traces of settled life in Bulgaria date back to the 7th millennium BC. The Bulgarian Neolithic is divided into several stages differing in the time of onset and duration. The beginning of the early Neolithic is considered to be at 6300/6200–6100 BC, and the end of the late Neolithic is constrained to 4950–4850/4800 BC (Görsdorf & Bojadziev, 1996). There are plenty of multilayered Neolithic settlements that have been inhabited for centuries. Many of them are well stratified and possess a sufficiently precise chronologies based on detailed archaeological studies and series of radiocarbon dates. Additionally, there are almost always burnt building levels, rich in well baked clay remains (dwelling walls, floors, roofs, hearths, ovens, etc.). Therefore, the Neolithic is particularly favorable for archaeomagnetic purposes (Kostadinova-Avramova et al., 2014, 2020). A total of 51 archaeomagnetic determinations belong to this period.

Similar to the Neolithic, there are also a variety of artefacts from the Eneolithic and Bronze Age that are suitable for archaeomagnetism. The discovered settlements remain largely multilayered, frequently with extensive fired building levels. The earliest findings from the Eneolithic date to ~4900 BC, and the latest to ~3800/3750 BC (Boyadziev, 1995). The archaeomagnetic results are summarised in 42 reference features. It follows that a transitional period separating the Late Eneolithic from the Early Bronze Age (~4100–3200/3150 BC) providing only 8 reference points due to the scarce archaeological sites of that time (Fig. 6). The most plausible explanation is palaeoclimatic environment with relatively high average annual temperatures and severe drought assumed for the end of the Eneolithic (Todorova, 1995). The problem of missing occupation traces within this period was extensively examined through a multidisciplinary study, both for Greece and Bulgaria (Tsirtsoni, 2016 and references therein). It is believed that Bronze Age culture began to spread in Bulgarian lands around 3200/3150 BC and developed until ~1100/1000 BC. A total of 44 archaeomagnetic determinations, concentrated mainly in the Early Bronze Age, cover these two millennia (Fig. 6). The Middle and Late Bronze Ages are much less studied because of the small number of archaeological sites discovered.

The Iron Age spans the last 1200 BC and is perhaps the most unfavourable for archaeomagnetism (Kostadinova-Avramova et al., 2021). Many researchers (e.g. Weninger et al., 2009; Drake, 2012) discuss the decline of the ancient societies in the Eastern Mediterranean and Western Asia between the 13th and 9th c. BC as a consequence of drastic climatic and environmental changes. This affected the life in the Balkans as well. The archaeological finds from the first Iron Age phases testify to lower population density, human migration, tribe incursions, weak trade relations, etc. The settlements are generally short-lived, with thin cultural layers, poorly stratified and often located at high naturally protected and almost inaccessible places. Due to all the above, well-studied archaeological sites with clear stratigraphy and stable chronology that can serve as benchmarks, are rather an exception. Additionally, the materials collected are quite often unsuitable for palaeointensity determinations and fail to restore this element because the Iron Age thermal constructions (hearths, household ovens and non-specialised production furnaces) are usually poorly preserved, short-term used, and the attained firing temperatures rarely exceed 600 °C. Therefore, only for a half of the 40 reference features available, the full geomagnetic field vector was determined.

The Antiquity age in Bulgaria is divided into early Roman Period (I–III AD) and Late Antiquity (IV–VI AD). It is attested by numerous historical and archaeological monuments, well studied in terms of material culture, evolution, and chronology. Workshops with furnaces specialising in various types of production were common during this era. The usual firing temperatures are tending to exceed 800 °C, which favours the success of archaeomagnetic analyses. A total of 81 reference features belong to the Antiquity—39 from the Roman period and 42 from the Late Antiquity. Unfortunately, about 65% of the materials studied are bricks for which only I and Fa were restored i.e., the secular variations of magnetic declination are based on a smaller data number.

The Middle Ages can generally be divided into Early (VII–XI AD) and Late (XII–XIV AD). The remains of Early medieval settlements indicate that the technological progress in the Antiquity was followed by a decline in many aspects of life affecting not only the quality of the discovered ceramic products, but also the nature of the thermal structures. Nevertheless, baked clay objects with sufficiently precise chronology are not lacking due to the presence of numerous archaeological and historical evidence. The reference features involved in the archaeomagnetic database are a total of 50 (29 early medieval and 21 late medieval) and 30% of them are bricks.

For the Ottoman period (XIV–XIX AD) the bricks studied are the majority—80% or 22 out of 27 reference features. As a result, the declination variation curve is based only on 5 data (Fig. 6). Mainly bricks are collected also from the period covering the time after the Liberation to the present day, and the five archaeomagnetically studied archaeological sites produce single result for D.

The archaeomagnetic data accumulated over more than half a century of research allows to trace the patterns in the three geomagnetic characteristics during the last 8 millennia (Fig. 7). A general trend of increasing intensity from the Neolithic to the middle of the first millennium BC followed by a subsequent decrease is undoubtedly evident. The lowest Fa values (~30 μT) were observed circa 5400 BC (first half of the late Neolithic), and the highest (~96 μT) circa 525 BC (Late Iron Age). Three consecutive maxima for the intensity can be assumed at the beginning (~5800 BC), the middle (~5550 BC) and the end (~5000 BC) of the 6th millennium, while during the Eneolithic, Fa changes do not seem so significant. At the same time, the magnetic declination turns from strongly western, at the beginning of Neolithic, to eastern, at the end of the epoch. The subsequent smoother transformation in D during the Eneolithic shifts again to more western values. Magnetic inclination decreases significantly from Neolithic to Eneolithic with a maximum around the middle of the 5th millennium. Despite the few results from Late Eneolithic—Early Bronze Age transition, as well as from the Middle Bronze Age, a relatively smooth increasing intensity accompanied by D and I values close to the present ones can be admitted. Only additional high-quality archaeomagnetic data could help to define the secular variations of the 4th millennium BC and the Middle Bronze Age in detail. The Iron Age is characterised by high to maximum intensity. A significant decrease in Fa, accompanied by an inclination minimum and a strongly western declination was observed around 400 BC. During the Antiquity, archaeomagnetic data suggest rapid changes in the geomagnetic field intensity and a significand variation for I, which at the beginning of the Roman period is of ~65°, then decreases to ~45° and increases again up to ~70° in the end of Late Antiquity. Due to the large percentage of the bricks studied, magnetic declination for these periods is insufficiently defined. The early Middle Ages are characterised by an intensity decrease at the beginning and a relatively rapid increase and a subsequent decrease at the end of the period, while for the late Middle Ages Fa does not change much. The inclination tends to decrease from the early to the late Middle Ages, and D values turn from western to eastern. Despite the large percentage of bricks from the Ottoman time, a new shift in D from east to west is evident. Along with this, I values at the beginning of the period are very similar to those in the early Middle Ages and then another decline occur. A well-expressed Fa peak exceeding 70 μT, is defined in the middle of the Ottoman period, while at the beginning and the end of it, the geomagnetic field magnitude is around 40 μT.

The main Bulgarian geomagnetic secular patterns are generally consistent with those observed in the Central Mediterranean and in Western and Central Europe (e. g. Hervé et al., 2017; Osete et al., 2020; Schnepp et al., 2020a, b; Rivero-Montero et al., 2021, etc.). In the last years, more and more archaeomagnetists (Gallet et al., 2003; Shaar et al., 2011; Osete et al., 2020; Rivero-Montero et al., 2021; Tema et al., 2021 etc.) focused their scientific interest on proving and explaining the short-term geomagnetic events called “spikes” (Shaar et al., 2011) and “jerks” (Gallet et al., 2003). So far, the Bulgarian secular variation curves do not show indication for any of these events (Kostadinova-Avramova et al., 2020, 2021).

The potential of archaeomagnetism in archaeological research is well-known. The Bulgarian geomagnetic secular variation curves are successfully used for absolute dating since more than 40 years (e. g. Kovacheva, 1989; Jordanova et al., 2004; Kostadinova & Kovacheva, 2008; Kostadinova-Avramova & Kovacheva, 2015, etc.). The dating compares the obtained archaeomagnetic results for a site (or a feature) with the local reference curves representing D, I and Fa variations in the corresponding period, and several possible intervals are usually distinguished (Fig. 8a). The final solution is a combination of all defined intervals (Fig. 8b) and depends not only on the accuracy of the experimental results and the reference curves used, but also on the rate of the geomagnetic field change during the period in question. Furthermore, the magnetic analyses are applied not only to compare and synchronise archaeological structures (Kostadinova-Avramova et al., 2014), but also to draw conclusions about the firing conditions and temperatures reached, which are informative for the technological development of the respective society (Jordanova et al., 2018; Kostadinova-Avramova et al., 2018).

Fig. 8
2 set of charts. b. The declination, inclination, and intensity have fluctuating trends, peaking at negative 5250, negative 5460, and negative 5250, B C, respectively. The archaeomagnetic dating is highlighted from 5324 to 5227 B C.

Archaeomagnetic dating for the Neolithic site Sharkov Chiflik (Kostadinova-Avramova et al., 2020): (a) comparison of the I, D, Fa results obtained with the corresponding reference curves and (b) combination of the individual dating results and final archaeomagnetic solution with 95% probability density

Full size image

7 Conclusions

In this overview of the combined archaeomagnetic and environmental magnetic research carried out in Bulgaria, we present and discuss several major milestone areas of application in archaeology. Case studies on advantageous utilisation of magnetic signature of iron oxides retained in fired archaeological materials—house destructions, pottery, ovens, etc.—demonstrates the vast array of potential applicability of rock magnetism. It is shown that mineral magnetic parameters of fired clay materials, combined with estimations of the maximum firing temperature using magnetic susceptibility method, can provide valuable information for paleoenvironmental conditions in case of open-air settlement fires or ancient mining activities, as well as technology-related aspects in case of pottery. Furthermore, advanced state of archaeomagnetic database compiled for the territory of Bulgaria allows performing rigorous archaeomagnetic dating and synchronisation of archaeological sites.