Archaeometric investigation of the hoard from Bodrogolaszi, Hungary

Abstract The hoard from Bodrogolaszi is a very significant find, dating back to the 15th-16th centuries. Unfortunately only 73 coins have survived from the original 135. In our study we would like to present the site, the circumstances of finding the hoard, and also a short numismatic introduction to these coins. We used stereomicroscopy to investigate the surface of coins, searching for additional artificial interventions (breaking, cutting, and fillip). We discovered darker reddish brown spots on several coins. We applied Raman spectroscopy to determine the origin of the spots. The results revealed they are part of a lepidocrocitegoethite cover layer caused by exposure to the soil. XRF measurements were used to prove that the purity of the coins differs from historical data.


Introduction
The hoard hoard was found during agricultural fieldwork in 1990 (Map 1; location of coins are signed with a black dot). It became a huge sensation in Hungary; it was among the most significant hoards dating from the Hungarian Early Modern Age (16 th -17 th centuries). Of the 73 gold ducats which remain, 72 are found in the Numismatic Cabinet of the Herman Otto Museum in Miskolc and one is in the Hungarian National Museum's Rákóczi Museum in Sárospatak. Below, we will present this hoard briefly and its archeometric investigation.

About the Hoard
This hoard was found in Bodrogolaszi-Kálnok, Borsod-Abaúj-Zemplén County, Hungary. It is in the northeastern region of the country, an area which suffered a chaotic history during the middle of the 16 th century, because of political wars between King János I of Szapolyai and King Ferdinand I of Habsburg. Continuous Ottoman attacks made life there even worse.
The hoard site is located south of the village, near the river Bodrog (map 1) which has a narrow waterway before its regulation. This area was originally covered with quaternary sediment, and it was a floodplain from the early 16 th to mid-19 th century. The coins were found by local agricultural workers close to the river bank, opposite a Romanesque church and the Bodrogolaszi ferry dock. There was no container in which the ducats were found during a validation excavation in 1990; instead, they were spread out over a few hundred meters. Nothing else was found relating to the hoard during later archaeological field works .
The hoard is one of the biggest coin hoards (among others e.g. Karcag-Jakabszállás, Hódmezovásárhely, Velky Folkmár (Zozuláková 1994); later, but contains coins from this period: Kassa, Újfehértó) (V. Székely 2005;Budaj 2012, Tóth andUlrich 2007) ever found in Hungary which can be dated to the 16 th century. The hoard contains 73 gold coins generated over a 118-year period, between 1438 and 1556 AD. The oldest one was issued by King Albert (reigned 1438-1439 AD); the last coin was issued by King Ferdinand I in 1556. The whole hoard was buried after 1556 but possibly before 1567, when the village was mentioned in a manuscript as being abandoned. Unfortunately, we do not know who buried the coins. We should also emphasize the quite high percentage presence of non-Hungarian coins. They were issued in Salzburg, Carinthia, Austria, and Czech Kingdom, the Duchy of Munsterberg, Silesia and Gdansk, Poland. According to numismatic literature, they were made from almost pure gold, but there are slight differences. We publish their basic numismatic data below in Table 3. We used "n/a", if the coin has got no visible emission year sign on it.

Goals
Our main goals were to answer the following queries. We wanted to: 1) determine the origin of macroscopic anomalies which can be seen on several coins; 2) verify the historical data about the purity of gold florins and ducats (e.g. the Hungarian average is 98.90%); 3) recognize trends in the possible fluctuation of purity and separate them into different groups by their chemical composition. Instrumentation, methods and sampling: We applied three different methods during our study. They were carried out in the Department of Mineralogy, Geochemistry and Petrology, Faculty of Natural Sciences and Informatics, University of Szeged, and Earth Sciences, Hungarian Academy of Sciences, and Institute of Material and Environmental Chemistry, Research Center for Natural Sciences, Hungarian Academy of Sciences. We began our investigation with optical microscopy using an Olympus SZX7 stereo microscope (University of Szeged) with a 1-5,6x magnifying scale. During macro-and microscopic observations, many reddish brown spots were recognized. These were investigated using a Thermo Fisher Scientific DXR Raman microscope from the same department. Measurements were carried out using a 10 mW laser. The spectral resolution was ∼4 cm-1 for each measurement. The chemical compositions were investigated using a Thermo Scientific NITON XL3t GOLDD+ Energy Dispersive portable (handheld) X-Ray Fluorescence (EDXRF) spectrometer. Light elements (Mg, Al, Si, P, S, and Cl) were measured with He purging of the analyzer using refillable and portable He-cylinder. In all, 35 chemical elements were measured simultaneously. The diameter of the measured area was generally 8 mm. The X-ray tube in the EDXRF is equipped with Ag-anode (target) with 50 kV acceleration voltage and Peltier-cooling. The detector is a high-performance GOLDD (Geometrically Optimized Large Drift) detector with 0,185 keV resolution. Quantitative analysis was carried out with 'General Metals' and 'Precious Metals' companypreset calibration packages. The average margin of error is around 5% in case of measuring metallic objects. All coins were examined using XRF, several with Raman spectroscopy-after several tests generating the same result we stopped -and every ducat was examined with stereomicroscopy, which indicated a macroscopically observable anomaly. The gold ducats were not prepared in any way prior to analysis. Statistics were carried on in Statistica 12. Silver/copper scatter plots were made with Sygma-Plot 10. We did not use any sample preparations during measurements.

Results
Raman spectroscopy detected characteristic peaks of lepidocrocite (γFeOOH; samples no. 2, 9, 27, 59) ( Figure 7) and goethite (αFeOOH) (Figure 8) on the reddish brown spots which were found on the surface of several coins. XRF results are shown in Table 1. Sample numbers can be found in the column labeled "Sample no". It contains values of duplicates in one averaged row. They are detailed separately (Table 2), but we still need to conduct further measurements of this group. The lowest Au-content is 92.91% in a duplicate copy (sample no d9); the highest is 99.18% (no 23). The average purity of the hoard is 97.80%, but it increases to 98.00% if we only consider Hungarian coins. Detection limits are the followings to explain Lod values: Sn: 0.02, Pb: 0.01, Sb: 0.02. All values are in w/w%. Their mass is around 3.5 g; confirming existing historical data. Márton Gyöngyössy measured the exact mass of gold from this period, which was 3.5215 g. He also published the average purity from written sources: 98.90% for Hungarian ducats (Gyöngyössy 2008). The maximum value of silver content is about 4% (4.036 and 4.018% (samples no 11 and 57)), while copper content is usually below 1% (maximum is 1.318% in sample 54). We detected Fe content as well, but the iron was from the spots of the covering iron-hydroxide layers.

Discussion Stereomicroscopy
Using stereomicroscopy, we realized that the reddish brown spots could not be removed easily, so we decided to use another method. Later we focused on possible signs of additional activities on the surfaces of the coins. We found two cut or broken ducats and   two chipped ducats (Figure 1). The two cut ducats were issued by Archbishop Ernestus in Salzburg. One of them is completely cut (A). As can be observed in the photo, it was put on a solid surface while being cut, as the gold had crinkled at the bottom line of the cut surface. The other one is only partially cut (B). The first chipped ducat was issued by King János of Szapolyai (reigned 1526-1540 AD); it was struck using a sharp tool, creating a hole (C). The second was chipped by an unknown tool; the foot of the portrayed man is missing (D). Summarizing these observations, we can suggest that all of these additional modifications were made by humans. A stress-corrosion-breaking process is a possible solution in some cases, but the average purity of these gold ducats is extremely high, so it would have had to have happened over a long period of time and this cannot be proven.

Raman-spectroscopy
Raman spectroscopy determined the origin and phase composition of reddish brown spots on several coins. It confirmed the soil origin of these spots and also confirmed their iron-hydroxide composition. We are considering using lepidocrocite as a forensic determination factor on other ducats for which the origins are not known. It is a relatively new question, however, and we don't have enough samples, measurements and results to draw any conclusions yet.

X-Ray Fluorescence (XRF)
Determining the chemical composition of a Medieval or Early Modern Age ducat is really difficult, mostly because of the extremely high purity. Of course, it is not the only, but the most relevant reason, why it is an obstacle. We can also mention the lack of relating data about gold coins from this region and age, so our chances to real comparisons were limited. XRF is one of the best methods to detect the gold-silvercopper content of ducats, but it has limitations when talking about trace elements. That is the main reason why we had problems with tracing the travel of the raw material or determining its provenance. It is true that gold artifacts may have been reminted many times. Finally, we must mention that the fluctuation  of purity is mainly related to the al marco type of issuing. It means that regulations governed only the amount of coins which needed to adhere to a standard unit of gold, not the exact mass of each pieces. In the followings, every data will be related to the known remaining part of the hoard. One of our main goals was attempting to separate issues into different groups based on their silver and copper content. Lack of enough data about trace elements in these coins, we have chosen minor elements, because of the high purity of gold. Unfortunately, we needed to exclude duplicates, because their differences from first copies had been remained unknown. We should study them more in the future, including a comparison with otherstill unmeasured hoards from this age.
Hungarian issues can be separated into two subgroups by their silver content ( Figure 2). The tendency is that oldermostly from the 15 th centurycoins contain less silver. According to the scatter plot, our statement is not true if we take Ferdinand's issues into consideration. Here, we ought to focus on their copper content either. Only 3 of them have got 0.2 or less percentages of copper. Most issues by Ferdinand have got a measured value of 0.3% or higher. As it can be seen in the plot, the purity of gold is still around the average of other coins from the 16 th century, which is a bit lower, than it was during the 15 th century. The ratio of the silver and copper content is different, but we cannot state, that Ferdinand's coins had the same or better purity than older ones. We have got no information from historical written sources about a reform or development of material preparation or minting and striking technology under Ferdinand's reign, so we will have to apply different analytical methods and measure more coins to find out the background of this difference.
We have been carried out the same visualization of silver/copper content of non-Hungarian issues ( Figure 3). Here, we observed that coins from Salzburg can be separated into a subgroup. They contain more than 0.7% of copper, while the silver percentage is between 1 and 2.5. Two issues from the Czech Kingdom were struck under Ferdinand I's reign, but their chemical composition is the most different in the hoard. The only exemption is one of them. Generally, we can suggest that Czech gold ducats of this hoard had relatively poor quality that possibly meant a worse exchange rate in the 16 th century comparing with other issues. The unmarked subgroup of coins from other regions is too various and mixed to observe any exact conclusions.
We used two more statistical methods during our investigation. The first is Principal Component Analysis   (Figures 9 and 10) we chose the certain range of values between 5 and 10. We found the same subgroups (coins from Salzburg and Jagellonian age), but the results are not as clear or obvious as they were using PCA. XRF has limitations when investigating the provenance of these gold coins. Its inability to detect a wide range of trace elements and determination protocols meant we needed to treat each case study individually. Here, we can present only assumptions about Hungarian issues. They were sorted into three different chambers: Körmöcbánya (Kremnica), Nagybánya (Baia Mare) and Nagyszeben (Sibiu). The last two chambers used Transylvanian mine products. Vasilescu et al investigated mines from this area and they detected silver and copper in gold. They found tin (100-300 ppm in the case of alluvial presence in cassiterite phase) and antimony (50-500 ppm in jamesonite and/ or stephanite phase) as trace elements. Lead and tellurium were also detected (Vasilescu et al 2011). XRF analysis of these coins is not detailed enough to confirm presence of those trace elements, and tellurium is overly sensitive to heat: it can be volatilized easily during various types of measurements. Dana Pop et al found significant silver, copper and tellurium contents either in raw gold during the Roşia Montana project (Pop et al 2011). Coins from Transylvanian chambers in the hoard usually had a higher Agcontent than issues from Körmöcbánya, but the difference is not significant. We cannot state this is a proper methodology to separate Hungarian issues by chambers, but it can be a basic hypothesis for further research. Despite a significant Cu presence in Transylvanian-mined gold, the highest values of copper were detected in ducats from Salzburg and one from the Czech Kingdom. There is a slight decrease in Cu purity if we consider from the oldest issue to the youngest.

Summary
Generally we can state that archaeometric methods and its new aspects offered new perspectives to better understand historical coinage. We set up a protocol of investigating gold coins in a hoard. We started with macroscopic observation, looking at reddish brown spots on the surfaces of several coins. We used stereomicroscopy to search for additional modifications to determine the origin of those spots. Sample of cutting and chipping were found. Raman spectroscopy showed that spots came from soil-originated crystallized iron-hydroxide layers. XRF provided good data about the chemical composition of coins. We separated the ducats into issuing area-based groups and subgroups according to gold-silver-copper ternaries. PCA showed us coins from Salzburg and Jagellonian periods were different from others. The first subgroup was confirmed by the ternary about non-Hungarian issues. Both of them can be observed in the dendrogram about agglomerative cluster analysis results, but not as clearly as on a PCA diagram. Studying Medieval or Early Modern Age Central European-mostly Hungarian-gold coins is a unique and relatively new field of Archaeometry. However, even though there are results from studying Hungarian florins from the Anjou age (14 th century) we know almost nothing about those issued in the 15 th -16 th centuries. Our main goal was to conduct an investigation which could be a good starting place for further research.