Vegetation History and Archaeobotany

, Volume 14, Issue 4, pp 453–464

Nutrition and environment in medieval Serbia: charred cereal, weed and fruit remains from the fortress of Ras


    • Department of AnthropologyUniversity of Alabama at Birmingham
Original Article

DOI: 10.1007/s00334-005-0092-9

Cite this article as:
Borojević, K. Veget Hist Archaeobot (2005) 14: 453. doi:10.1007/s00334-005-0092-9


This study presents the results of archaeobotanical examination of remains from the medieval complex of Ras in Serbia. The samples were collected from the fortress situated on the hilltop (Gradina) and from a settlement below (Podgradje) during the archaeological excavations of 1972–1984. They were taken primarily from the buildings containing charred cereals dating to the 12th and 13th centuries. The main staple was bread wheat, followed by rye. Grains of barley, oats, and millet were also present. The weeds, including ruderals, were represented by many species. Agrostemma githago (corn cockle) was an important contaminant of the cereal fields. Exceptional finds include a piece of charred round bread, the cereal content of a pot, and peach stone fragments. Previous information about the agriculture and food in medieval Serbia was based solely on documents that were either written after this period or that were not pertinent to the region. This study is the first direct evidence providing information about agriculture and food of the inhabitants of medieval Serbia.


Archaeobotany. Medieval Serbia. Fortress. Stores. Prunus persica. Bread


The site of Ras is located in southwest Serbia, 11 km southwest of the city of Novi Pazar, in the very centre of the medieval Serbian state (Fig. 1). The fortress of Ras is situated on the top of a hill (750 m asl) called Gradina (Fig. 2). Below Gradina, the settlement of Podgradje is situated on a flat surface (620 m asl), close to the confluence of the Sebečevska and Raška rivers. Gradina and Podgradje form a part of the single settlement and fortification complex of Ras (Popović 1999). The geographer Cvijić (1911), when describing the relief of the region, states that two important anthropo-geographic characteristics of this central valley are that it is hidden and well-protected so that it became the economic and political centre of the pastoral population, and several narrow passages that allow exit from this concealed valley in all directions.
Fig. 1

Map of southeast Europe showing location of Ras
Fig. 2

Aerial view of the Ras fortress (photo I. Dimitrijevic, after Popović 1999)

The geographical position of Ras in a protected but still accessible region has been important in the past. According to the principal investigator M. Popović, who is also the author of the monograph The Fortress of Ras (Popović 1999), the complex of Ras is characterized by several stages of occupation and development since prehistory. Remains of a hill fort settlement date to the early Bronze Age and late Iron Age. Excavated remains at Podgradje indicate Roman occupation in the 3rd century a.d. The first fortress, which occupied only a small part of the hill and eastern part of the plateau below the fortress, was constructed in the 4th century a.d. and was abandoned in the 6th century. Construction started again from the mid 9th to the 11th century, and ramparts and palisades were erected on the hill so that the buildings occupied the entire hill and the plateau below. Toward the end of the third decade of the 12th century, the fortress was burnt and destroyed. Shortly after its destruction, the fortress was rebuilt again in 1149 when the Byzantine emperor Manuel I Comnenus started a campaign against the Serbs. Soon after the campaign, the fortress of Ras became the main defensive stronghold in the central part of the newly formed Serbian state under the Nemanjic dynasty that had to defend the ruler and his court. During the reign of the Serbian king Stefan Radoslav (1228–1234) of the Nemanjic dynasty, the first coins were minted at the fortress. Besides the buildings on the hill plateau surrounded by the wall and ramparts (Gradina), a suburb (Podgradje) developed on the eastern slopes of the hill. The fortress of Ras was abandoned after being destroyed in a fire in the fourth decade of the 13th century. The destruction is not recorded in historical sources and can possibly be explained as a result of conflicts among the heirs to the throne. Although the region of Ras remained the centre of the Serbian lands, the fortress was not rebuilt (Popović 1999).

During the systematic archaeological research of the medieval complex of Ras (1972–1984), buildings dating to the 12th and 13th centuries containing charred cereals were excavated. The cereal supplies had been burnt in a sudden conflagration probably connected with the destruction of the fortress in the fourth decade of the 13th century (Popović 1999). M. Popović (Archaeological Institute, Belgrade), the principal investigator of Ras, submitted the samples to the author of this paper for examination.

Until this study, not a single botanical investigation had been undertaken from a medieval site in Serbia. Therefore, the main aims of the analysis were 1) to identify plant remains and to find out what crops were stored at the early Serbian stronghold; 2) to reconstruct possible cereal cultivation and processing; 3) to determine the food of the inhabitants of the fortress; and 4) to compare the results obtained by this analysis those from other Slavonic strongholds, and with the scarce information about agriculture and diet from literary and historical sources of the period.

Materials and methods

Archaeological context of the archaeobotanical samples and dating

Based on the archaeological record, it was possible to identify a sequence of four building horizons or phases (BH) for the Medieval fortress (Table 1). Dating of the horizons is supported by the finds of coins (in situ) and ceramic typology (Popović 1999). Twelve archaeobotanical samples were taken from two building horizons (BH III-a and BH IV). Archaeobotanical samples were collected by the excavators from concentrations of charred seeds visible in the soil (M. Popović, pers. comm.). The locations of the buildings and houses from which samples were analyzed are shown in Fig. 3.
Table 1

Building horizons at the medieval fortress of Ras, dating, provenance and the number of plant samples taken

Building horizons


Features (number of samples)


End of 11th century a.d. until fourth decade of 12th century

No samples taken


After a.d. 1130 until mid 12th century

House 36 (1); cultural layer (3)


Second half of 12th century

No samples taken


First decades of 13th century until a.d. 1233

House 49 (3); House 50 (3); House 52 (3)
Fig. 3

Plan of the Ras fortress (12th and 13th centuries) showing location of the buildings from which samples were analyzed: Houses 36, 49, 50, 52 (modified from Popović 1999)

Three samples are from the third building horizon (BH III-a), dated to the second half of the 12th century. One of the samples from BH III-a is a piece of bread found in House 36. The remaining two samples represent charred peach stones which were also recovered from cultural layers of BH III, one from a southeast quadrant of Gradina and the other from a cultural layer excavated in the Podgradje settlement below Gradina.

Eight samples are from stores found in houses 49, 50, and 52, which are situated along the west wall of the fortress and are attributed to BH IV, dated to the first decades of the 13th century. One sample from BH IV represents the contents of a pot found in House 52.

From House 49, three samples were collected from cereal deposits found dispersed on the floor. In the southern part of the house two rotary querns were excavated (Popović 1999). All these finds indicate that the wheat may have been stored in some sort of bags in the northern part of the house, while the milling of the cereals probably took place in the southern part of the house, which is dated to the fourth decade of the 13th century.

From House 50, three samples were collected from the southern part of the house. Two samples were taken from the floor of the house, whereas the third one was gathered from a pit in which a large concentration of cereals was found. The investigators of the site believe that the pit represents remains of a storage pit laid with wood where cereals were stored. In this rather large pit, in addition to cereals, several items were found: wood charcoal, the metal rings of a wooden barrel, and two coins of Stefan Radoslav (1228–1234). The presence of coins dates the building to the first decades of the 13th century (Popović 1999). The wood charcoal from this pit was identified as Acer sp. (maple) by Vilotić (1999, pp 446–447).

From House 52, three samples were collected. Two samples were from the floor and one from a pot. In this house, a well-preserved cooking oven was also excavated (Popović 1999). The house was dated by the partially minted coins of king Radoslav to the first decades of the 13th century.

Preservation, analysis and identification

All plant remains had been charred. Plant samples were labelled with the provenance information of the excavation units, for example G/17, indicating 10×10 m quadrants; the grid is shown in Fig. 3. The letters a, b, c, d (for example G/17-c) designate four smaller 5×5 m units within the large 10×10 m quadrants. The plant samples were stored in the Archaeological Institute in Belgrade until 1997 when M. Popović gave them to the author for analysis. The samples were then analyzed in the Palaeoethnobotany laboratory of the Department of Anthropology at Washington University in St. Louis and in the Palaeoethnobotany laboratory of the Department of Anthropology at the University of Alabama at Birmingham. Samples were analyzed separately and tabulated for each house. After the initial weight of samples was recorded, the plant material was sieved through five U.S. graduated geological sieves to facilitate sorting. The sieve meshes used were 2.8, 2, 1, 0.5 and 0.355 mm. For each sample weighing more than 50 g, half or a quarter of the fraction from each sieve was analyzed in full, whereas the remaining half or quarter was scanned to determine whether the subsamples contained the same spectrum of plants as the analyzed parts. The total numbers of specimens were multiplied (×2 or ×4, respectively) and represent the extrapolated numbers for whole samples. Percentages of counts (presented in pie diagrams) were computed out of the total number of seeds per sample, although quantifying and comparing diverse types of plant remains (for example weed seeds and cereal grains) remain problematic considering different seed production per plant and different preservational biases. However, they can be used to compare the analyzed storage finds.

Identification of plant macro remains was done using low-power (6.3×–63×) microscopes. Scanning electron microscope (SEM) photographs were also used for more precise identification of smaller seeds. The identification of the macro plant remains was based on the morphological characteristics of plant material, using reference collections of the author and of the Palaeoethnobotany Laboratory in St. Louis. A few seeds were unidentifiable because they were poorly preserved or because they exhibited unfamiliar morphological characteristics. Scientific plant names for cultivated plants follow Zohary and Hopf (2000), and other plant names follow Čanak et al. (1978). The term “Triticum aestivum/durum” is used to describe tetraploid or hexaploid naked wheat which is not distinguishable on the basis of grains.


The results obtained by the analysis of plant macro remains from different archaeological features—Houses 49, 50, and 52—are shown in Table 2 and in pie charts in Figs. 46.
Table 2

Counts (extrapolated numbers for the whole samples) of identified charred seeds from burnt deposits from Houses 49, 50 and 52 from the fortress of Ras

House number/finds

House 49 Rotary querns

House 50 Two coins (1228–1234); pit (wooden barrel); few pottery sherds

House 52 Cooking oven; cache of coins (1228–1234); many pottery sherds (pot with porridge not included)

Building horizon/dating

BH IV: Fourth decade 13th c

BH IV: First decades 13th c

BH IV: First decades 13th c


Floor H/ 8-d

Floor H/ 9-b

Floor H/ 9-by t3

Floor H/ 11-a

Floor H/ 11a-b

Pit H/ 11-b

Floor G/ 17-c

Underneath plaster floor G17-b/ H/17-a

Weight (g) of sample









Species (count)

1. Cereals (grains)

Triticum aestivum/durum L








Hordeum vulgare L






Secale cereale L







Avena sativa L







Avena sp


Panicum miliaceum L



Setaria cf. glauca


Setaria sp











total cereal grains









2. Legumes (wild)





Vicia sp


total small seeded legumes






3. Weeds and ruderal plants

Agrostemma githago L








Anthemis cf. arvensis






Bifora radians M.Bieb






Bromus sp



Bromus secalinus L



Centaurea cyanus L



Chenopodium sp


Cirsium sp



Daucus carota L



Galeopsis sp





Galim spurium L


Lapsana communis L


Neslia paniculata (L.) Desv


Polygonum sp. (3–sided)


total weeds and ruderal plant seeds







4. Fruits

Rubus fruticosus L



total fruit seeds



5. Varia indeterminata

unidentified Type “A”








total of all seeds








Fig. 4

Proportions of the taxa (%) found in three samples (a–c) in House 49 based on the total number of seeds per sample
Fig. 5

Proportions of the taxa (%) found in three samples (a–c) in House 50 based on the total number seeds per sample (legend see Fig. 4)
Fig. 6

Proportions of the taxa (%) found in two samples (a–b) in House 52 based on the total number of seeds per each sample (legend see Fig. 4)

The sample from Quadrant H/8-d from House 49 (Fig. 4a) consisted primarily of grains of Triticum aestivum/durum (wheat) (88%). This sample contained the largest proportion of weeds (9%), of which Agrostemma githago (corn cockle) constituted 6%. The number of Avena sativa, Secale cereale, Hordeum vulgare grains is small enough to be considered an unintended mixture in the main cereal. After corn cockle, the most numerous were Daucus carota seeds. In this sample, a small number of short brome seeds of Bromus secalinus were also identified, as well as seeds of Anthemis cf. arvensis, Centaurea cyanus, Galeopsis sp., Lapsana communis, Neslia paniculata, and Polygonum sp. The seeds of Polygonum sp. are three-sided and probably belong to Polygonum convolvulus L. (syn. Bilderdykia convolvulus (L.) Dum.), a common weed in crops. From this sample, small unknown seeds (Type “A”) approximately 3 mm wide × 2 mm long and triangular in cross-section were identified.

The sample from Quadrant H/9-b from House 49 (Fig. 4b) was similar to the sample from Quadrant H/8-d, described above, consisting primarily of wheat grains (94%). This sample, however, contained a smaller percentage of weeds (5%). Corn cockle was again the dominant weed (4.5%). Twelve rather small seeds of legumes were identified (Fabacaeae) from this sample, but more precise identification was not possible. Again, the number of them is small enough that they may be considered weedy intrusions in the cereal crop.

The third sample from Quadrant H/9 by t3 (Fig. 4c) was from the same quadrant as the sample described above. This sample was the purest of all three. It contained 97% wheat grains and less than 3% weeds, with 2.6% corn cockle being the major contaminant. In this sample, there were two seeds of Galeopsis and a few Bromus sp. seeds attributed to the short-seeded type, but because of the incomplete preservation, positive identification was not possible. A new weed species, Cirsium sp., was not identified in two previous samples from the same house. The seeds of various species in the genus Cirsium are very similar, so precise determination to species level was not possible.

All three samples from House 49, where the rotary querns were found, (Fig. 4a–c) consisted primarily of Triticum aestivum/durum grains which constituted 88–97% per sample. In all samples, Agrostemma githago at 2.6–6% was the dominant weed. However, the samples differed in the percentage of total weed seeds present (3–9%) and in the weed spectrum present per sample. These differences may indicate that the stored wheat from three samples was not harvested from the same field or during the same year.

The sample from Quadrant H/11-a (Fig. 5a) from the floor of the House 50 consisted of cereals and no positively identified weed seeds. The main cereal was wheat (82%), followed by rye (12%), oats (4%), barley (2%), and millet (0.4%).

In the second sample from Quadrant H/11 a–b (Fig. 5b) from the floor of House 50, wheat was the main cereal (82%), followed by barley (8%), rye (7%), and oats (3%). The wheat grains from this sample were the largest in size compared with those from other samples. The majority of the wheat grains were retained within the sieve with the mesh size 2.8 mm that was used in the laboratory for sorting. This sample had a very few weed seeds: four Agrostemma githago seeds and two Bifora radians seeds. Six small round legume seeds were identified as Vicia sp. and were probably weeds in the main crop.

The third sample from Quadrant H/11-b (Fig. 5c) was from a storage pit in House 50. It contained the largest proportion of rye grains (67%), followed by wheat (28%), oats (3%), and barley (2%). Among all the samples analyzed from Ras, this is the only sample in which rye grains were dominant.

The three samples from House 50 (Fig. 5a–c) were different. Two samples were collected from a floor (Fig. 5a, b), and the third from a storage pit where the remains of the wooden barrel were found (Fig. 5c). In the two samples from the floor, the dominant grains were Triticum aestivum/durum, representing 82% of both samples, but seeds of other cereals—rye, barley and oats—were present in different proportions. In the sample from the pit, the main cereal was rye (Secale cereale), constituting 67% of the sample. Bread wheat constituted 28% of this sample. In all three samples, less than 1% of weed seeds was identified, and only a few seeds of Agrostemma githago were present per sample, in contrast to the large number of these seeds present in the samples from House 49.

The sample from Quadrant G17-c (Fig. 6a) from House 52 contained almost entirely wheat grains (95%). There were very few grains of other cereals present. Among the weeds seeds (5%), the seeds of A. githago dominated (4%). From this sample, four unknown Type “A” seeds were identified, the same as those identified from Quadrant H/11-b, House 50. Otherwise, the weed spectrum of the wheat sample from this House was similar to that identified from samples from House 49.

The sample from Quadrant G/17-b H/17-a of House 52 (Fig. 6b) consisted of a lump of carbonized Panicum miliaceum (hulled millet grains). Only few seeds of Setaria sp. were identified, which are common weeds in a millet crop.

The third sample from Quadrant G/17 in House 52 consisted of a type of charred substance recovered from a pot (Figs. 7, 8), found next to the cooking oven within House 50. The substance resembled some sort of porridge, in which the cell structure of the pericarp of cereal grains could be seen, which showed that it had been made of cereals.
Fig. 7

Drawing of a pot (×4) that contained charred porridge from House 52 (after Popović 1999)
Fig. 8

Charred porridge from the pot in House 52: a fragment; b magnified detail of transverse cereal cells (×10)

From House 52 (Figs. 6a, b, 7, 8), where a cooking oven was excavated, the three samples differed from each other again. One sample consisted almost completely (95%) of Triticum aestivum (bread-type wheat) grains. The second sample consisted entirely of Panicum miliaceum (millet) grains. The third sample represented the contents of a vessel that contained the remains of a cereal porridge.

The larger fragment of bread was excavated between the destruction layers of House 36 and underneath House 42 from the central sector. These two houses occupy the same general space, with 42 built after 36 was destroyed in a fire. The diameter of the piece of bread was approximately 30 cm (Popović 1999). The analyzed fragment was only a few centimetres large and 2 cm thick. In a cross section of the bread, it was noticeable that it was made of dough that had been folded several times (Fig. 9a). On the basis of microscopic analysis, it seems that the bread was made from coarsely ground whole flour including the pericarps of the cereal grains (Fig. 9b). It is possible that the cereals were ground in the rotary querns discovered near the sample from House 36.
Fig. 9

Charred round bread from House 36 (×10): a fragment; b magnified detail of wheat pericarp (×63)

Another rare find from the cultural layer of BH III, a horizon at Gradina, are fragments of charred Prunus persica (peach) stone dated to the last decades of the 12th century (Fig. 10). Charred fragments of peach stone were also retrieved from the cultural layer in Podgradje below the fortress, but their precise dating was not possible.
Fig. 10

Fragment of charred Prunus persica (peach) stone (scale bar = 1 mm)


Cereal cultivation

Triticum aestivum/durum (wheat) was identified in seven out of eight samples, in all three houses (Table 2). The wheat grains resembled bread wheat grains of the T. aestivum-type (so hexaploid naked wheat), but in the absence of chaff and rachis fragments, the grains were classified as T. aestivum/durum. Proportionally, T. aestivum/durum was the main cereal in six of the samples. The highest percentage (97%) of wheat grains was from the sample recovered from House 49.

Besides wheat, grains of barley, rye, millet (Fig. 11), and oats were present in various quantities in different samples (Figs. 46). The differences in the quantities of various cereals in each sample might indicate that some cereals were sown together, such as wheat and rye from samples in House 50 (Fig. 5c). Sowing of two crops together as a mixed crop or maslin is traditionally explained as a risk-reducing strategy (Jones and Halstead 1995). However, secondary mixing of cereal grains during the burning of the house cannot be excluded, but it is unlikely that the rye grains stored in a wooden barrel within a pit in the southern part of the house became mixed with the wheat grains on the floor above in the northern part of the house, although it is possible that some of the wheat grains from the floor were spilled into the pit.
Fig. 11

Charred cereal grains: a Triticum aestivum/durum from House 49 H/8-d; b Secale cereale from House 50 H/11a-b; c Hordeum vulgare from House 50 H/11a-b; d Panicum miliaceum from House 52 G/17-b (scale bar = 1 mm)

Later historical sources dating to the 14–16th centuries indicate that in medieval Serbia, winter and spring wheat, oats, barley, and millet were grown (Blagojević 1973). Historical documents from Dubrovnik archives from the 15th century also indicate that different cereals were grown together; in fact, there is a special local term that indicates growing an intentional mixture of cereals called “sumješica” (Blagojević 1973). There is also a folk term for the rye-wheat maslin called “suražica”, also known as “napolica” that was historically documented in the 19th century in Voyvodina (Hegediš and Čobanović 1991).

A lump of charred millet was found in one sample from House 52. Millet is a spring-sown crop that needs the light and warmer temperatures of summer. The sowing of a spring crop might indicate the practice of crop rotation, if millet was sown in spring in the same fields where the winter crops, for example wheat, had been sown in the previous year. The charred lump contained millet grains with glumes that would have needed further dehusking if intended for human consumption. Medieval documents from monasteries indicate that millet was grown intensively, and Blagojević (1973) believes that millet was equally important as oats and that only wheat was a more important cereal grown on the monastic lands.

Oats were found in small percentages (3–4%) in samples where wheat or rye were the principal cereals. Oats and rye are traditionally more commonly grown in the northeast parts of Europe than in southern Europe. According to Lefort (2002), in southern Europe, oats were primarily grown in medieval times as animal feed and in the Byzantine Empire as horse feed. At Ras, bones of horses and parts of horse equipment were rare (Blažić 1999; Popović 1999); therefore, it is difficult to estimate the importance of oats as horse feed.

Cereal processing and storage

The analyzed samples of cereals did not contain chaff or internodes, so they had already been threshed and winnowed before being stored in the fortress. As expected at a consumer site like Ras, very few agricultural tools were discovered, only two metal covers for hoes (Popović 1999); agricultural tools were probably stored somewhere outside the fortress, perhaps closer to the fields where the cereals were grown. It is not exactly known who grew the cereals, but most probably there were farms nearby (see also below, weeds).

Some of the samples from House 50 contained a very low percentage of weed seeds (less than 1%); it is possible that these cereals were already cleaned for human consumption by handpicking out the undesirable weed seeds. In the Medieval documents from Serbian monasteries, there are special terms used for dehusking “plevljenje” and for cleaning by handpicking “žitotrebljenje” (Blagojević 1973).

Several rotary querns were discovered at the fortress of Ras, some of them dating to building horizons BH III-b and BH IV, which makes them contemporary with the finds of cereals. According to the information of the investigators (Popović 1999), cereals were found dispersed on the floor of House 49, indicating that the wheat may have been stored in some sort of bags in the northern part of the house, while the grinding of the cereals probably took place in the southern part of the house where two rotary querns were found in situ. The discovery of rotary querns indicates that at least some of the cereals were hand ground at the fortress, probably on a daily basis. Rotary querns were also found in the houses in phase BH II, at which time the fortress was a Byzantine stronghold. Because the stone rotary querns would be very heavy to transport, they could have been left in the fortress and could have been used consequently by its various occupants. Considering the defensive purpose of the fortress, it is not surprising that its inhabitants had cereal supplies and the grinding equipment available there in case of a siege. Watermills were used in Byzantium since the 10th century, and they became common by the 12th century (Lefort 2002); however, rotary querns remained in use there along with traditional agriculture (Bryer 2002).

Concentrations of cereals discovered both on the floor, for example the wheat from House 49, and in pits, for example the rye from House 50, indicate that cereals were stored above ground, probably in some sort of bags, but also underground in special pits. In the northern part of the fortress, along the northern wall, a large underground space (room) was discovered. The investigators believe that this served as a large cereal storage place rather than a water cistern (Popović 1999). Although there was a layer of ash, the investigators did not find any cereal concentrations in this room. Unfortunately, not a single sample was collected, so it is impossible to confirm the actual function of this feature.

Weeds and ruderal plants

Agrostemma githago (corn cockle) was the dominant weed (Fig. 12) and was identified in six samples at Ras (Table 2). The seeds cannot be removed by sieving because they are large and heavy and mimic in size the grains of the main crop. On the other hand, the smaller seeds of weeds were not numerous (Table 2), suggesting, together with the lack of cereal chaff, that the crops had been winnowed and sieved before being stored. Thus, the weed spectrum of stored cereals is an artifact of crop processing, reflecting only a limited number of the weeds that were originally present (Jones 1987).
Fig. 12

Charred seeds of Agrostemma githago (scale bar = 1 mm)

Corn cockle is a typical weed of cereals, and there is an old folk saying, “nema žita bez kukolja,” meaning that there is no wheat without cockle. The seeds were most abundant (2.6–6.1%) in samples from House 49. Having a rather high percentage of corn cockle seeds present in wheat can have negative effects if it is not eliminated by handpicking before human consumption. Flour that contains more than 0.5% of cockle seeds can be dangerous for human health because it contains toxic glycosides (Mowszowicz 1955, cited in Szydlowski and Wasylikowa 1973), and if present in smaller quantities, it causes headaches (Slavko Borojević personal communication, 1997). A. githago is the most common weed in other Slavonic medieval sites where rye and wheat were the principal crops, such as Starigard/Oldenburg and Gross Lübbenau (Kroll and Willerding 2004; Medović 2004).

Together with other typical cereal weeds, Bromus (brome grass) grains were identified in the samples from Houses 49 and 52. The seeds belong to the short type of Bromus sp., B. secalinus. These seeds, like corn cockle seeds, are rather long and difficult to eliminate from the main crop, but, unlike corn cockle, they are not poisonous. Among other typical cereal weeds, the grains of Centaurea cyanus and Neslia paniculata were also identified (Fig. 13a). Together with A. githago, these belong to the cereal weed class Secalinetea, which is typical of winter crops and was common during the Middle Ages (Ellenberg 1996).
Fig. 13

SEM photographs: a Neslia paniculata; b Daucus carota; c an unidentified seed labelled Type “A” whole seed; d cross-section of Type “A” (scale bar = 500 μm)

The majority of the identified weed species grows well in warm, calcareous and sandy soils (Šarić 1989; Hanf 1990), such as are present in the vicinity of Ras; therefore, we may assume that the cereals were locally grown. The presence of typical cereal weeds in large quantities indicates that the fields were cultivated for at least several years. The weed spectrum at least partly resembles those typical of the three field rotation system in the Middle Ages in other parts of Europe (Karg 1995).

Some of the seeds in the cereal samples that were classified among the weed taxa were not typical weeds infesting cereal fields, for example Daucus carota (Fig. 13b), which may have grown as a ruderal plant on the edge of fields, and thus got harvested together with the wheat. Similarly, Rubus fruticosus (blackberry) is not a weed of cereals, but grows on the edges of fields and woods, and perhaps a few fruits accidentally ended up among the cereal grains. Eight unidentified seeds temporarily labelled Type “A” (Figs. 13c–d) could perhaps be classified as weeds because they are small and were found among cereals.

A few small Fabaceae seeds and several rather small round seeds (diameter 3–5 mm) were found and identified as Vicia sp. (vetches). It is interesting to note that at Ras, not a single concentration of legume crops such as lentils, peas, or broad beans was recovered, although they were found at a late Roman site, Svetinja, north of Ras (Borojević 1988), and at many Medieval sites in central Europe, including several Slavonic strongholds (Kroll and Willerding 2004; Medović 2004; Wasylikowa et al. 1991). Historical resources also indicate that legumes were grown in Byzantium (Lefort 2002). It is possible that at Ras, by chance legumes were not stored in large quantities before the final destruction of the fortress, or that they were all consumed.


In the wheat sample from House 52, four Rubus fruticosus (blackberry) seeds were retrieved. Blackberries can be found growing in the region of Ras, and the wild fruits are gathered today. However, an unintentional admixing cannot be excluded.

Of special interest is the discovery of charred Prunus persica (peach) stone fragments recovered from cultural layers of the fourth building horizon (BH IV) at the fortress and from below the fortress. Peaches originate from Asia and are not native in the natural vegetation of the Balkans or the Mediterranean (Zohary and Hopf 2000). Peaches are soft and difficult to transport and cannot be dried like apricots or raisins, so we can suppose that the peach trees were brought to Serbia and were grown locally. The Romans could have introduced peaches into the area during their occupation of the fortress of Ras and the settlement below. However, the area was abandoned between the 6th and 9th centuries, and the peach trees could not have survived unattended for several centuries. Peach stones were recovered from the samples dated to the 8th–10th centuries from the early Slavonic stronghold of Mikulčice in Moravia (Opravil 1998) and from several medieval sites in eastern central Europe (Wasylikowa et al. 1991). Byzantine sources mention that peaches were grown in Macedonia in the 14th century (Lefort 2002). Blagojević (1973), using historical documents, does not mention peaches in his book about agriculture in medieval Serbia. The actual finds of peach stones from Moravia, dating to the 8th–10th centuries, and from the fortress of Ras, dating to the end of the 12th century, point to a discrepancy between archaeological evidence and historical records.

There were probably more fruits consumed at Ras in the past than were actually found, but fruit remains are usually underrepresented when not dealing with waterlogged material (Willerding 1991).

What was the food like at Ras in the 13th century?

Based on the analysis of only a few plant samples, our knowledge about the diet of the inhabitants of Ras is limited, but it may indicate what the main staples were. Wheat was the most ubiquitous and the dominant cereal at Ras, whereas rye was dominant only in one sample. Wheat was also the major cereal identified in archaeobotanical samples from the medieval fortress at Silistra in Bulgaria (Popova 1990). In contrast, rye was the dominant cereal in many other Slavonic sites from eastern central Europe. In Slovakia, for example, rye and barley were dominant after the 8th century (Wasylikowa et al. 1991). Rye was also the main cereal in the archaeobotanical samples from the early medieval site of Lubomia, Poland (Szydlowski and Wasylikowa 1973) and in the Slav strongholds Starigard/Oldenburg and Gross Lübbenau in Germany, dated to the 9th and 10th centuries (Kroll and Willerding 2004; Medović 2004). These sites all lie north of Serbia, indicating that the food of southern Slavs, including Serbs, differed from that of northern Slavs primarily in the main cereal consumed, wheat being more common in the south and rye in the north (Dembinska 1999). In later medieval documents, wheat is also mentioned as the most commonly grown cereal in the southern regions (Blagojević 1973). However, in northern Greece, at the multi-layer site of Agios Mamas, a total of 424 rye grains were recovered from a 10 l sediment sample within Byzantine debris (12th–14th centuries) left as the intrusion into the Bronze Age layers from a Byzantine church and small graveyard (Kroll 1999).

Rye was also found at the fortress at Ras and other Slav strongholds. It is possible that wheat predominates at sites where inhabitants were of higher rank, and that rye was used for provisioning the ordinary people. The fortress of Ras was temporarily used as the ruler’s refuge, and rye may have been intended for his subordinates. At Ras, another explanation could be that rye was sown together with wheat as a risk-reducing strategy, in case the wheat yields were low.

The contents of a pot from House 52 at Ras represents the remains of some sort of cereal porridge. Additionally, in the same house, a lump of millet was found that could also have been prepared as porridge. As mentioned previously, an oven was discovered in House 52, and it could have been used for various food preparations, including the cooking of porridge in pots. Porridge was commonly eaten in Byzantium (Hill and Bryer 1995) and was eaten by the Slavs (Lunt 1997) and other peoples. Lefort (2002) believes that millet was not a highly valued food in Byzantium, citing Anna Comnena (12th century) who mentions millet only twice and regards it as the food of barbarians (Comnena 1928). Millet is typically a crop of regions with warm summers, but it was readily consumed and grown by the Slavs in northern regions as well. During the Middle Ages, several millet varieties were identified from the excavations of medieval Wroclow dating to the 10–12th centuries (Kosina 1995), as well as in the medieval fortress at Silistra in Bulgaria (Popova 1990). Millet is the second most important crop after rye at the Slav stronghold at Gross Lübbenau in Germany (Medović 2004).

A piece of round bread, found in House 36 and probably made of wheat, indicates that bread was probably made and baked at the fortress. A type of bread made of coarsely ground wheat (psomos pithyrodes) was common during the Byzantine period (Kislinger 1999), and it is possible that a similar tradition of making bread continued at Ras. Bread, the so called “kleb”, is also the most commonly mentioned food (17 times) in the Russian Primary Chronicle, indicating the importance of bread among the Rus (Lunt 1997). The round, rather flat, leavened bread has a local term, “pogača”, and is commonly consumed in the wider region today. Leavened bread, in contrast to unleavened bread, was considered the “real” bread and was consumed readily among the Eastern Orthodox in the 11th century (Montanari 1999).

Another food difference between the northern and the southern Slavs was the preference of the northern Slavs for game meat (Lunt 1997; Dembinska 1999). At Ras, only a small percentage of wild animal bones was identified. Bones of domestic animals dominated the samples and represent 96.7% of the total bones identified in BH III and BH IV (Blažić 1999). Although bones of boar, deer, and hare were identified at the fortress of Ras, it seems that the inhabitants preferred the meat of domestic animals. Among domestic animals, the bones of sheep and goat represent over 60% of the samples, followed by cattle and pig bones. It is not surprising that in the mountainous region of Ras, sheep and goat husbandry provided an important source of animal protein. Also, these rather small animals could have been easily kept in the fortress in case of a siege.

Archaeologically, foreign influences at Ras are shown by the evidence of imported items from various parts of Europe, including pottery from workshops in Byzantium, southern Italy, and Thessalonica (Popović 1999). Together with imported pottery, the knowledge of other types of food and serving practices may have been passed along. During the early Middle Ages, the courts often moved from place to place, including the courtiers and the cooks. According to Popović (1999), the fortress of Ras served as a temporary refuge and one of the ruler’s residences. The national identity was not yet precisely defined and thus the so-called ‘ethnic foods’ were not yet common. Nutrition was not based only on the locally available plant and animal sources, but also included various cultural influences of different regions. It is possible that the wives of the Serbian rulers who were of Byzantine and Venetian origin also influenced the nutrition and the culinary practices of the courts. For example, Stefan Radoslav of the Nemanjic dynasty, who ruled the region at the beginning of the 13th century, was a son of the Serbian king Stefan and a Byzantine princess, and he was later married to the daughter of Theodore I Angelus Comnenus, the despot from Nicea (Ostrogorski 1957). Perhaps the find of peaches at Ras offers evidence of southern cultural influences, including food. During the 12th century and on several occasions in the previous centuries, Ras served as a Byzantine stronghold, and it is possible that Byzantine soldiers had already brought with them different culinary preferences from the southern regions where they originated.


The analysis of plant macro remains has shown that cereals were stored above ground and in the pits at the fortress of Ras during the 12th and 13th centuries. Triticum aestivum/durum (wheat) was the major staple crop stored there, supporting the written and meagre archaeobotanical evidence for the preference for wheat among southern Slavs. However, more archaeobotanical research is needed in several sites in the region of the southern Slavs to confirm this. Besides wheat, Secale cereale (rye), Hordeum vulgare (barley), Avena sativa (oats), and Panicum miliaceum (millet) were also found. Three samples from House 50 provide the best example for crop diversification and/or different storage within an individual house. Wheat and rye were probably sown together in a maslin, unless post-depositional processes contributed to the mixing. Millet adds to the evidence for growing a wide range of different grain crops. The find of millet grains underlines the discrepancy between the diversity between the archaeobotanical record and the narrow range of grain crops mentioned in Byzantine records (cf. Lefort 2002).

Agrostemma githago (corn cockle) was the dominant large-seeded weed. It was present in a significant percentage in several cereal supplies, indicating that these large seeds could not be removed by sieving and had to be handpicked out before human consumption in order to avoid negative health effects. The presence of large quantities of annual cereal weeds shows that the cereals were cultivated on long-lasting open fields. The majority of weed seeds belong to the Secalinitea group, typical of winter-sown wheat and rye crops.

Exceptional finds include a piece of charred round bread and cereal porridge from a pot, demonstrating that bread was baked and porridge was cooked in the ovens at the fortress. Hand rotary querns found in the fortress indicate that cereals were probably milled on a daily basis.

The find of Prunus persica (peach) stone fragments reveals that peaches were eaten at the fortress already in the 12th century, two centuries earlier than indicated in the written sources (cf. Lefort 2002). The presence of peach demonstrates that the food at Ras included not only readily available local products but that, together with pottery imports, some foreign culinary and gardening practices were already being introduced into early medieval Serbia.

Previous information about the agriculture and food in medieval Serbia was based solely on written documents that were either of later date than the analyzed samples or were not relevant to the region. This study provides the first direct evidence about food and agricultural activities of the inhabitants of medieval Serbia and points to discrepancies between written documents and archaeological finds, in this case plant remains, as in other parts of Europe (see, for example, Irniger and Kühn 1997, 1999; Irniger 1999).


I would like to thank M. Popović for entrusting me with the plant samples from Ras for this analysis. I am grateful to him and the Archaeological Institute in Belgrade for granting permission to reproduce and adapt illustrations for Figs. 2, 3, 7 in this article from the publication The Fortress of Ras by Popović (1999). I am also thankful to H. Kroll who helped me with the identification of ambiguous seed specimens. SEM photographs were done by M. Vieth from the Department of Biology at Washington University in St. Louis. All other photos of plant remains were made by the author. Special thanks to S. Jacomet, H. Kroll, an anonymous reviewer for their comments on the manuscript, and J. Greig for copy editing.

Copyright information

© Springer-Verlag 2005