< back


  • Excavation of small 13th century Byzantine statio on the ancient Via Egnatia east of Kavala
  • Transformed into Ottoman kervanseray in the 15th century
  • Director: Prof. Siri Sande (University of Oslo)
  • Ten seasons 1993-2002

1992 the Norwegian Institute was offered a small Byzantine fortress near the village of Petropigi 20 km east of Kavala as an archaeological project by Charalambos Bakirtzis, then ephor of Byzantine antiquities at Kavala. The next year the field work started, going on for the seven summers. Of these the first five consisted of excavations, a further two were dedicated to the measuring and conservation of the monument, and the final three to the study of the finds.

The fortress takes its name after the neighbouring village of Petropigi, as its ancient name is not known. The monument is situated on the southern side of the main road between Kavala and Xanthi, 9.23 m. above sea level, on a plain stretching down to the sea. North of the road the terrain starts to rise towards the Rhodope mountains, which form a splendid backdrop when one sees the fortress from the south. Originally the shoreline must have been closer to the fortress than it is today, and the distance to the  Petropigi village was longer as the old village was located higher up on the mountain slope. With the draining of the marshlands near the coast new land for cultivation became available, and a new village was constructed closer to the main road. Only traces of the old Petropigi can be seen today.

The fortress is oriented north/east-south/west, and it lies exactly parallel to the modern road. Because of this parallelism it is probable that the modern road at this point follows the course of the ancient Via Egnatia. The remains of the latter may either be buried under the tarmac, or possibly further south, where a dirt road now runs. Unfortunately it has not been possible to lay out a trench northwards to try to locate the road, as only the plot immediately adjacent to the fortress belongs to the Greek state. The surrounding fields are privately owned.

Though the Petropigi fortress lies in the midst of cultivated fields, no trace of human occupation such as pottery shards has been found in the neighbourhood. There appears to have been no settlement in the immediate vicinity. Already before the excavation started, Bakirtzis suggested that the fortress was a statio, that is, a fortified staging post. Our investigations so far have corroborated this hypothesis.

The inner measurements of the fortress, 29,6x29,3 m, correspond to 100x100 Roman feet. Its walls are constructed of mortar, stones and bricks, a technique used from the early Byzantine period onwards. One band of bricks runs continuously through all four walls at the same height, while the rest of the bricks are distributed more unevenly for shorter stretches. There are some slight differences in the size of the bricks, which may be attributed to different building stages. 

The building technique is in itself not sufficient to allow a precise dating, since it was used over a long period. As usual, the walls of the fortress were constructed around a wooden scaffolding. When the work was finished, the beams were cut and the scaffolding removed, leaving the timbers in the walls to decay there. During our excavation we found a tiny piece from the scaffolding in the mortar in the south/western tower. It was subjected to a C 14 analysis at Uppsala, Sweden, which suggested a date between A.D. 1275 and 1350. For historical reasons it seems most reasonable to opt for a construction date in the 13th century, when there was a certain building activity following the recapture of Constantinople from its Latin occupants.

The layout of the fortress was changed in the course of time. Originally it had two gates, one in the south/east and one in the north/west, and two towers, one in the south/west and one in the north/east. The projecting walls of the gates were probably vaulted. At a later stage these walls were extended and fitted with a portcullis, the slots of which are still visible. Contemporarily a tower was added to the south/eastern corner. The north/western tower was enlarged, and a room inside added on an upper level. There is a “seam” in the wall of the south/western tower suggesting that this, too, may have been strengthened. 

It is likely that these alterations were made not long after the construction of the fortress, during the turbulent 14th century, a period characterized by Byzantine civil wars and the incursions of various powers such as the Serbs, the Catalan Company and finally the Ottoman Turks, whose conquest of Thrace and eastern Macedonia started in the 1360’s. If the Byzantines themselves strengthened the fortress, they probably did it during the first half of the 14th century, while they still exercised  some control, but it is, of course, also possible that the decision to fortify it was made by one of the temporary masters of that particular stretch of the road during the second half of the century. The fortress could not have withstood an army with siege equipment, but it could probably repel attacks from brigands, of whom there were many in the 14th century. The fortress is, in fact, very conservative in its typology, and differs little from its early Byzantine prototypes.

Inside the fortress several structures came to light. In the southern area the somewhat flimsy substructures of two buildings remain. Between them there is a channel which runs through the outer wall of the fortress. The buildings give the impression of being utilitarian structures such as stables or the like. The substructures are not datable, but they are probably as old as the outer walls of the fortress, since they take the channel into account.

The structures in the middle and the northern area of the fortress are better preserved and more characteristic. The most important one is a long building which practically divides the courtyard in two. It is oriented according to the points of the compass contrary to the fortress walls, which, as remarked above, deviate slightly in accordance with the course of the Via Egnatia.  The differences in orientation are only perceptible when measured. The building is divided into one longer and one shorter unit with an aperture between them, which gives access to the northern part of the courtyard. The shorter unit is in line with the south/eastern  gate. Apart from having a slightly different orientation from the fortress itself, the two units are constructed in a different technique, the so-called cloisonné, where the stones are framed by horizontal and vertical bricks. The floor of the long building was stuccoed. No door was found, neither in the longer nor in the shorter unit. This suggests that the lower part of the building was used as a basement, and that there was a second story with access from outer staircases which probably led to a walkway or gallery, all in wood.

The cloisonné  technique was also used in a structure running along the inside of the northern fortress wall in its full length. Since it was demolished more thoroughly that the building in the middle of the courtyard, very little of it remains. Though it stretches from one end of the fortress to the other, there is no trace of a bond higher up on the walls. This may mean that the structure was no building, but simply a low platform.

The northern fortress wall is marked by eight recesses at about equal distance from each other. The one in the west, which is better preserved than the rest, shows that the recesses were fireplaces with chimneys. They seem to be secondary in relation to the fortress wall, and were probably inserted into it when the platform along it was built.

The platform bordered on the north/eastern tower and the north/western gate, both of which were closed. The gate was closed twice, both towards the courtyard and outwards. The outer wall is in cloisonné  technique, indicating that the builders that constructed the platform and the long building in the middle of the courtyard were also responsible for the closing of the tower and the gate. 

The cloisonné technique finds its closest parallels among early Ottoman buildings in Western Thrace, such as the han in Traianoupolis and the imaret in Komotini. A C 14 dating of charcoal found inside the long building in the courtyard, also gave a date in accordance with the Ottoman occupation of Thrace and Macedonia, since it suggested the period A.D. 1410-1425. The Ottoman invaders obviously turned the Byzantine statio into a Turkish kervansaray. Since these constructions have the same function, they needed only to make few changes. The closing of one of the gates was one of these, as Turkish kervansarays have normally only one gate, not two. It may seem odd that the gate nearest to the Via Egnatia was closed, but the new masters probably preferred to have the gate in the same area as the utilitarian buildings, and the sleeping quarters out of sight.

A platform with fireplaces in the back wall is a typical feature of many Ottoman kervansarays, and can be seen illustrated in miniatures, for instance. The travellers slept wrapped in their cloaks or blankets with their mounts or tethered  to the platform or immediately in front of it. The fireplaces in the wall behind them gave them warmth, and also possibilities to cook their food. There were few, if any facilities in these kervansarays, which are mostly described as being completely empty. A French traveller who visited the Ottoman Empire  in the 17th century and undertook a journey with a caravan, was advised by an Armenian friend in Constantinople to bring with him everything he might need, even down to the shoes of his horse.

There was, however, the possibility of having one’s horse shod. The blacksmith was an important feature in Byzantine stationes  and Turkish kervansarays. We did, in fact, find his forge in the Petropigi fortress. It was located close to the westernmost chimney in the northern wall.

Since the travellers brought with them pots and pans and cutlery and departed with it again, few objects came to light inside the fortress. This is a marked difference from settlements. What we found, was mostly pottery which had been accidentally broken by the travellers. Most of it was cheap “kitchen ware” such as people are not afraid to lose, though a few sherds of late Byzantine glazed pottery turned up. The coins we discovered during the excavation, were all Ottoman, the majority cheap and badly preserved copper coins. One silver coin from the reign of Suleiman the Magnificent could be dated to the middle of the 16th century A.D. The earliest coins seem to be datable to the beginning of the 15th century A.D., while the latest coins bore a characteristic star pattern which is found on some of the coins of Sultan Murat the third (A.D. 1574-1595). Judging from the finds, the fortress of Petropigi would seem to have been used as a kervansaray at least during the 15th and 16th centuries.

When it was rebuilt, great care was lavished on it. Apart from the structures in cloisonné technique, we excavated a stucco floor in the north/western area of the courtyard. Every stone in this building had been removed, so it is impossible to date it by technical means, but as it would not have been built before the north/western gate was closed, it must been Ottoman in date. We have not yet determined its function. In the north/eastern part of the courtyard the lower parts of two thick stucco pillars (diam. 1 m.) and the imprint of a third one came to light. They were probably four in  number originally, and must have carried an elevated structure, possibly a small mescid, a prayer room. 

At some point after the 16th century, all the Ottoman structures were dismantled. This clearly happened by public decree and under public control, since the work was very thorough. The walls were all at once taken down to the same level, presumably to get hold of the bricks, which could be reused in other public buildings. We found much larger quantities of stones than of bricks, which suggests that the latter were carted off. With regard to the outer walls of the fortress, the picture is quite different. They were left to decay and were gradually taken down by the local farmers, when the latter needed building materials. They, incidentally, wanted stones, not bricks.

Because they were left to themselves, the fortress walls are preserved to different heights. The southern and western walls have always been visible to a considerable height, whereas the northern wall is especially badly preserved. This is due to the presence of fireplaces.  Long after the fortress had ceased to function as a staging post, it was used by the local farmers and herdsmen, which almost up to the  present day used to lighten fires in the old fireplaces. The rapid changes in temperature weakened the wall and caused it to crumble gradually, a process which probably lasted for centuries.

As suggested by the coins, the Petropigi fortress was used for its original function at least till the end of the 16th century. We do not know when the dismantling of its inner structures took place, but the reason was probably that a better staging post had been found. The distance from Kavala is only 20 km, rather short by Ottoman standards (30 to 40 km seem to have been normal, depending on the road). By moving 5 km further east, one arrives at Khryssoupolis, an old town which is known to have been a statio already in the Byzantine period. It is therefore likely that, at a certain point, the Petropigi fortress was abandoned as an official staging post in favour of Khryssoupolis.

In the centuries following the Ottoman conquest, life was difficult in Thrace and Eastern Macedonia. Large segments of the original population were transferred to other parts of the Ottoman Empire, and towns and villages were depopulated and in a sorry state. The decline probably started already in the late Byzantine period, otherwise the Petropigi fortress would not have been built. Its construction suggests that Xryssoupolis was not safe enough as a statio, a state of affairs which continued well into the Ottoman period. In the 17th century the conditions of the Christian population ameliorated somewhat, and repopulation of sparsely populated areas was encouraged. With towns and villages more populous and with increased peace and prosperity, it would have been easier for travellers to find accommodation in inhabited areas, and the isolated kervansarays began to go out of use.

Having ceased to function as a staging post, the Petropigi fortress continued to be used by the local population. In the south/western tower we found a fragment of an 19th century Turkish pipe together with several bones from sheep and goats, evidence that it was used as a lair, probably by herdsmen. There were also a number of 18th-19th century pottery sherds in the upper layers in the courtyard  of the fortress. There we also found small, rectangular flints. In all likelihood they belonged to a sort of sledge used for threshing, called tykane in Greek. It was employed in Greece and Anatolia till well into the 20th century, and specimens can still be studied in various local museums. It is easy to see that when the ruins inside it had become covered with earth, the courtyard of the Petropigi fortress would have been ideal for threshing. We also found several horse-shoes. Some of these are recent and should be connected with the various agricultural activities that have been going on in the fortress until the 20th century, while the oldest ones were possibly lost by travellers in the period when the fortress still retained its original function.

The Petropigi statio or kervansaray is a good example of acculturation, showing, among other things, how the Ottoman invaders simply took over the Byzantine infrastructures and adapted them to their own use. Neither the Ottoman nor the Byzantine name of the staging post is known, but Charalambos Bakirtzis suggested that it is mentioned in historical sources as a place where members of the Catalan Company came together in 1307 in connection with the disbanding of the Company. Perhaps studies of old maps and documents may yield more information with regard to identity of the fortress.

Text by Siri Sande