Updated archeomagnetic data set of the past 8 millennia from the Sofia laboratory, Bulgaria



[1] This data brief reports the latest updates of archeomagnetic data obtained at the Sofia palaeomagnetic laboratory of the Geophysical Institute, Bulgarian Academy of Sciences. The current data set consists of measurements from 284 Bulgarian archeological sites covering the past 8000 years. There are also 54 sites from other European regions, namely, Serbia, Kossovo, Greece, Spain, Switzerland, Finland, and Russian Karelia, as well as five sites from Morocco in North Africa. The update of the archeomagnetic results consisted of a thorough revision of all geomagnetic field measurements as well as dating these measurements that were published in the original papers or in previous compilations. The updated results can be found in GEOMAGIA (http://geomagia.ucsd.edu) or as an Excel spreadsheet at the EarthRef.org Digital Archive (http://earthref.org/cgi-bin/erda.cgi?n = 946).

1. Introduction

[2] The necessity of well-established archeomagnetic data sets from different parts of the world is vital for the development of our knowledge of geomagnetic field behavior during the Holocene, its peculiarities, analyses, and modeling. During the past decades several global compilations appeared [e.g., McElhinny and Senanayake, 1982; Burlatskaya et al., 1986; Daly and Le Goff, 1996; Yang et al., 2000; Korte et al., 2005; Donadini et al., 2007b; Korhonen et al., 2008; Genevey et al., 2008], each of which served for different analyses of the field. However, a continuous update of the compilations is necessary to refine these analyses. It is well known that archeomagnetic studies, together with analyses from lava flows, are unique in their capacity to determine all three components of the Holocene geomagnetic field when oriented materials are used. These measurements represent spot readings of the field, and so the elaboration of a good succession of geomagnetic determinations strongly depends on the abundance of archeological discoveries in a particular region. The collaboration between geophysicists and archeologists also plays an important role. In fact, archeologists often discover structures and sites (e.g., in situ kilns, or ovens) that are potentially good for paleomagnetic investigations. Unfortunately, almost as often, structures are destroyed before being sampled for archeomagnetic studies if the geophysicists are not present at the right moment during the excavations, and so precious material is lost. This normally happens because the archeological excavation needs to proceed. Another aspect concerns the dating of a particular site. When the excavation is ongoing and geophysicists collect archeological material for magnetic investigation, archeologists normally give a preliminary age. Often, after detailed archeological analyses, this date change, and so the better the cooperation, the more likely the date associated to the archeomagnetic study is going to be updated. In this sense the recently concluded European training network contributed significantly to the enlargement of archeomagnetic determinations in Europe and strengthened these collaborations (Archaeomagnetic Applications for the Rescue of Cultural Heritage (AARCH), project HPRN-CT-2002-00219, http://dourbes.meteo.be/aarch.net/index.html).

[3] Archeomagnetic measurements at the Sofia laboratory have a long tradition and were started in 1967 [Kovacheva, 1969]. During the past four decades the methods used to determine the geomagnetic field have gradually been optimized, and so measurements that were performed at the beginning of the 1970s might not fulfill the current requirements for a reliable determination. This data brief describes the updates of the measurements performed at the Sofia laboratory.

2. An Updated Data Set From the Palaeomagnetic Laboratory at Sofia, Bulgaria

[4] The current data set (Sofia_archmag_2009.xls) mainly consists of measurements from Bulgarian archeological sites (284 sites) covering the past 8000 years. However, there are also 54 sites from other European regions, namely, Serbia, Kossovo, Greece, Spain, Switzerland, Finland, and Russian Karelia, and five additional sites from Morocco. Results from Serbia and one site from Kosovo coming from the same geographical region are included in the main body of the updated data set and will serve for future construction of reference curves. Results from other European countries and Morocco are listed at the end of data set. All these data, previously published in other compilations [e.g., Kovacheva, 1980, 1992, 1997; Genevey et al., 2008], have undergone a thorough revision by considering site by site the available measurements. Additional paleointensity experiments were carried out to check the older determinations under more reliable criteria and in general the site average value was confirmed. The new experiments showed good agreement with the older ones in particular when the material investigated had good rock magnetic properties. Directions obtained from studies performed a few decades ago were reconsidered in the light of their reliability criteria such as number of samples used, internal consistency (mainly for inclinational results from bricks), and magnetic cleaning applied. Some unreliable sites were rejected from the previous compilation. Additionally, existing gaps in the data set have been partly filled with the results of newly studied sites [e.g., Herries et al., 2007, 2008; Donadini et al., 2007a, 2008; Kostadinova and Kovacheva, 2008]. In the Excel file, results from new sites are in bold, whereas corrected ones are in italic. Table 1 also shows the individual sites that have been updated (corrected, merged, or removed).

Table 1. Modifications Within the Bulgarian Data Seta
LABNOArcheoIntNBModifications/ReasonsNew Assigned Site
  • a

    Laboratory numbers (LABNO) refer to entries in the data set of Kovacheva [1997] that are now attributed to new sites. Additionally, the corresponding ArcheoIntNB number of Genevey et al. [2008] is given when available. Modifications/reasons indicate what happened to that particular study and why. The new assigned site explicitly tells which entry is now related to the LABNO.

541746Combined with LABNO24Büso Island, Finland
3101723Combined with LABNO309Helsinki, Finland
318Combined with LABNO274Reinach, Switzerland
3171700Combined with LABNO274Djadovo, XVI hor.
1391651Combined with LABNO108Jabalkovo, older hor.
3081629Combined with LABNO233Valaam, Russian Karelia
2481518Paleointensity results merged with LABNO203, Directions discardedVeliko Tarnovo, Medieval kilns
1381646Combined with LABNO126La Maja-Calahora, Spain
1591641Results distributed between LABNO154 and 195 after reassessment of archeological agesDrustur, Medieval furnace
100, 155, 2101578, 1625, 1563All coming from a single site and merged in LABNO155, Medieval constructions with reused Roman material (excavator's opinion)Halka Bunar, ceramic center, and Zlatna Livada, Medieval furnace
3201691Merged with LABNO99Veliko Tarnovo, ovens
258Results not reliable and discardedSerdica, oven No 1
431788Merged with LABNO34Sarovka, trench H18
681742Directional results revised and reevaluated 
251799Directional measurements revised and reevaluated 
1751605Directional results discarded: small number, not reliable 
2341549Inclinational results obtained by bricks discarded: bad grouping 
631753Directional results discarded: bad grouping, probable displacements 
2221562Inclinational results obtained by bricks discarded: bad grouping 
1341652Dating refined, direction corrected 
1351680Dating refined, inclinational result (from bricks) discarded, unreliable 
1581639Paleointensity result added 
1641618Paleointensity result reevaluated 
1851582Doubtful measurements discarded 
1961589Dating and results refined 
200Paleointensity result added 
201Reevaluation of directional results, paleointensity results added 
2683245Combined with LABNO269; paleointensity results from Oxford laboratory also addedPlochnik, Serbia
3291525Dating interval (and culture) corrected OxCal4 calibration 

[5] The data in the present compilation are described by 26 parameters, explained in the notes of the Excel spreadsheet and in the following. The geomagnetic field measurements for the 343 sites are reported in terms of declination (DEC), inclination (INC), and intensity (PI). Results are given as feature's mean values, where feature denotes a volume of material that can be considered to have been magnetized at the same time [Tarling, 1983]. Additionally the paleointensity result divided by the dipole value for the corresponding geographical latitude [Creer et al., 1983] is also given (PI_PIDIP). One of the peculiarities of the measurements at the Sofia laboratory which are reproduced here is that in general all the geomagnetic field components are determined, and so the full geomagnetic field vector is available. The uncertainties associated with the geomagnetic field determinations are reported as α95 (ALFA95 [Fisher, 1953]) for directions, and as weighted standard deviation (SDPI [Kovacheva and Kanarchev, 1986]) for intensities. Also the number of accepted samples (or specimens) (NDIR for directions, and NPI for intensities) are reported. In case only one PI result is accepted, the uncertainty is given by the standard error of the best fitted line from the Arai diagram (e.g., laboratory LABNOs 102, 153, 180, 183, 190, 255, 282, 300 and most of the Greek results). As an indicator for the reliability of the paleointensity result the total number of measured samples (N0PI) is also given. The paleointensity method applied is shown in column TYPE. The Thellier and Thellier [1959] method with monitoring of the Natural Remanent Magnetization (NRM) direction and the linearity of the Arai plot (THDL), was subsequently replaced by the one with additional monitoring of the susceptibility (THkD). More recently, pTRM checks (THpT) were introduced (maximum percent of alteration accepted as 8%). During the years this maximum value has been very rarely encountered, usually pTRM check steps have shown less than 5% difference in the accepted temperature interval. Note that starting from 2004 the Coe [1967] method has been used in Sofia to determine paleointensities. Sometimes the anisotropy correction (AN_cor) was applied using the Anhysteretic Remanent Magnetization (ARM), the ThermoRemanent Magnetization (TRM), or the Isothermal Remanent Magnetization (IRM). The demagnetizing method (DEMAG) used to determine the directions relies on thermal (T), or alternating field (A) demagnetization. The maximum step used in the demagnetizing process is reported in the STEP column. Another very important information is the dating method (DATING) used. The associated minimum (LOWAGE), the maximum (HIAGE), and the midpoint (DTPOINT) are reported. At the beginning of our archeomagnetic studies (1960s and 1970s) the so called “short chronology” [Quitta and Kohl, 1978] of prehistoric cultures was believed to be valid. Later, the development of radiocarbon dating, dendrochronological calibration, etc., refined many of the previous dates, and led to crucial updates. In some of our publications the archeomagnetic dating is performed, but in the database the independent archeological dates are given. The archeological dates related to the archeomagnetically studied sites are one of the most critical point especially for the prehistoric past. In our work the determination of the chronological interval of the existence of a specific site is based on analysis of several factors: the stratigraphy, the thickness of layers, 14C dating, archeological material, cultural features. The highest precision can be obtained for multilayer settlements, discovered in the territory of Bulgaria, a great part of which have been studied archeomagnetically. For them the stratigraphical sequence determines the chronological position of different layers (horizons). When 14C dates are available, they determine the absolute chronology of the settlement, and its precision is directly proportional to the number of dated horizons. In this case not only the absolute values of 14C dates are taken into consideration but also their development in time. The graphs of these trends elaborated using the stratigraphical sequence are juxtaposed to the calibration curves (so-called “Archeological Wiggle Matching” method [Weninger, 1986; Boyadziev, 1995]). In this analysis the type of the material from which the 14C date is obtained is taken into account (being perennial or annual plants, as grains for example), as well as the place of the sample in the settlement's existence (beginning, middle or end). The limits of particular periods and phases of Bulgarian prehistoric cultures are clearly defined [Boyadziev, 1995]. In these frames the place of a specific horizon is determined on the basis of the number and thickness of layers, being informative about their duration. The chronological juxtaposition of the sites from Eneolithic period (5000–4000 B.C.) appeared particularly difficult because the dates from the second half of that period repeat, to a great extent, those from the first half of the Eneolithic, or even dates from the Late Neolithic sites [Boyadziev, 1995]. Both geomagnetic field variation and sunspot activity influence the cosmic ray flux, and so long- and short-term anomalies in the concentration of 14C nuclides can be linked to these processes. Still Süss [1980] has noted a minimum in their concentration during the middle of 5th millennium B.C., which can be partly related to these contradictions of the radiocarbon dates. Thus, the unreliable dates for the second half of the Eneolithic and the transitional period toward the Bronze epoch are given in parentheses in the DATING column. These are the cases when the calibrated absolute values do not match with the chronological order of the site. Single-layer settlements are ordered chronologically on the basis of comparative analyses of archeological artifacts (relative chronology), and on the basis of series of 14C dates which are related to multilayer settlements of the same culture. When 14C dates do not exist for a given site, on the basis of archeological material's analysis, the chronological frames of the period are taken. Additional data that characterize a particular site are defined as LABNO (Sofia laboratory number), SLAT and SLONG (site latitude and longitude), COUNTRY, PLACE, and SITE HORIZON, FIELD NUMBER (identification number of each individual sample collected on site), and type of MATERIAL collected for the archeomagnetic investigation. We also included the reference to the original publication in the REFERENCE column. We consider that the revised data set will be much more valuable for further worldwide compilations and geomagnetic field models. A very important fact is the inclusion of new sites from time periods with few data. Figure 1 shows the time variation of raw data without any reduction, together with the model prediction of the CALS7k.2 model of Korte and Constable [2005] at Sofia's latitude. Considering that we included also measurements from Serbia and Kosovo, Sofia appears to be in the center of the region investigated. Figure 1 also shows evidence for a substantial gap in the data set around 3500 B.C., which can be interpreted as a lack of artifacts, or as a lack of research involving that particular period. If we assume no discontinuity of the prehistoric development, how can we explain the lack of discoveries within this period? Todorova [1995] assumed a link with the paleoclimate. The climatic optimum around mid-5th millennium B.C. was followed by a Black sea transgression, and so a number of coastal Late Eneolithic settlements lays today submerged. Nevertheless, the problem of this chronological hiatus during the transitional period between the Late Eneolithic and Early Bronze is not yet completely resolved. Another period with sparse data lie within 2200 B.C. and 1000 B.C. (Middle and Late Bronze Age). Obviously, the progress of archeomagnetic studies is tightly related to the progress of archeological discoveries within the region of interest. There is a general good agreement between the Bulgarian data and the model long-term trends. The CALS7k.2 [Korte and Constable, 2005] is derived using a global data set, and so differences between the model and the Bulgarian data set can arise from discrepancies between the data within the European region [e.g., Lodge and Holme, 2009]. On the other hand, the CALS7k.2 is rather smooth compared to the observations. The choice of Korte and Constable [2005] to utilize damping parameters that would better reproduce the long-term variation of the geomagnetic field, rather then its fine structure, was motivated by the fact that the distribution of the archeomagnetic data is inhomogeneous in space and time. The updated results can be found in GEOMAGIA (http://geomagia.ucsd.edu), or as an Excel document at the EarthRef.org Digital Archive (http://earthref.org/cgi-bin/erda.cgi?n = 946).

Figure 1.

The evolution of the geomagnetic field in Bulgaria, Serbia, and Kossovo, according to the results from the Sofia laboratory. The data are not reduced to a common location. The red line represents model predictions of the CALS7k.2 model of Korte and Constable [2005] at Sofia.


[6] The Academy of Finland partly supported Fabio Donadini's work at the Sofia laboratory. Ian Hedley from the University of Geneva (Switzerland) is thanked for providing the directional results for the Swiss sites (laboratory LABNOs 251, 327, 318, and 323). The directional results for the Spanish site (laboratory LABNO 138) are from Parés et al. [1993]. We are most grateful to two anonymous referees whose valuable comments and remarks significantly helped improving the text and the data presentation.