archaeological investigation at konjuh

This is an extract from:

Dumbarton Oaks Papers, No. 56

© 2003 Dumbarton Oaks

Trustees for Harvard University

Washington, D.C.

Printed in the United States of America

Published by

Dumbarton Oaks Research Library and Collection

Washington, D.C.

www.doaks.org/etexts.html

Issue year 2002

Editor: Alice-Mary Talbot

Archaeological Investigation at Konjuh, Republic of Macedonia, in 2000

CAROLYN S. SNIVELY

The project entitled “Archaeological Investi-gation at Konjuh,” a joint project of GettysburgCollege in Gettysburg, Pennsylvania, and theMuseum of Macedonia in Skopje, Republic ofMacedonia, carried out excavation and surveyat and around the site of Golemo Gradiste, vil-lage of Konjuh, administrative district of Kra-tovo, Republic of Macedonia, from 26 June un-til 21 July 2000. This was the first season of afive-year project. Funding was provided for the2000 season by a Research and Professional De-velopment Grant from Gettysburg College, by aProject Grant from Dumbarton Oaks in Wash-ington, D.C., and by the Museum of Macedo-nia.

This report presents not only the results ofour first season of excavation but also an intro-duction to the ancient urban site of GolemoGradiste at Konjuh including a brief discussionof its features. Although known in local archae-ological literature, the site has been visited byfew archaeologists even from the Republic ofMacedonia and is almost unknown outside the

borders of the former Yugoslavia. Only onemonument from Konjuh, the unique Rotundachurch, has found its way into handbooks onByzantine architecture.

LOCATION OF THE SITE

Golemo Gradiste is located 41 km east ofSkopje. It lies ca. 6 km south of the Kumanovo-Kriva Palanka highway, on the Kriva River,1 ina mining region. The Roman road from Skupi(Skopje) to Serdika (Sofia) ran several kilome-ters to the north of the city, which stood on asecondary east-west road leading to Kratovo.2The ancient name of the site is not known; oneproposed identification is Tranupara.3 In thelate antique or early Byzantine period, the citywas probably located in the province of Darda-nia, whose capital lay at Skupi, although ourknowledge of the exact boundaries of late Ro-man provinces is so uncertain that it mightequally well have been included within theneighboring province of Dacia Mediterranea.

Dr. Dragi Mitrevski, director of the Museum of Macedo-nia, and Dr. Carolyn S. Snively, associate professor of clas-sics at Gettysburg College, served as the codirectors of theproject. Other members of the staff were: Dance Gol-ubovska, archaeologist-conservator, Republic Institute forthe Protection of the Monuments of Culture (hereafterProtection Institute), Skopje; Dr. Virginia Anderson-Stojanovic, archaeologist and ceramics analyst, professor ofclassics and fine arts, Wilson College, Chambersburg, Pa.;Biljana Packova-Kufojanakis, architect, Skopje; MiroslavDimovski, photographer, Museum of Macedonia, Skopje.

In addition to the members of the staff and the fundingagencies, we also wish to thank Dragisa Zdravkovski,deputy director of the Museum of Macedonia, for his helpand kindness, and Jovan Kondijanov and Mitko Prendzov,director and associate director of the Protection Institute,for the excavation permit and other assistance.

1 The modern village of Konjuh is located ca. 2 km southof Golemo Gradiste.

2 B. Georgievski, “Rimskite patista vo Kumanovskiot re-gion,” Macedoniae Acta Archaeologica 10 (1985–86) [1989]:153–59. According to Georgievski, this secondary road di-verged from the main one at the village of Klevcovce andran southeast, passing fortresses near the villages ofDovezence, Konjuh, Sopsko Rudare, and Filipovci on theway to Kratovo. The same author suggests that a local roadconnecting the major east-west route with an easternbranch of the Naissus (Nis) to Thessalonike highway mayalso have passed by Konjuh. To what extent these proposedroutes should be connected with the rock-cut road in theravine east of Golemo Gradiste or with the bridge over theKriva River reported by S. Radojcic (as in note 5 below) re-mains unclear.

3 V. Lilcic, “Razmisluvanja okolu ubikacijata na Tranu-para,” Kulturno nasledstvo 17/18 (1990–91) [1994]: 33–47.

HISTORY OF THE INVESTIGATION

The site at Konjuh first entered the profes-sional literature in the 1940s with the publica-tion of fragmentary inscriptions from the area.4

Svetozar Radojcic visited Konjuh in 1938; in1952 he published a survey of the site and adetailed study of the very unusual church, theRotunda, which had been excavated by localvillagers in 1919.5 In the early 1970s IvanMikulcic surveyed the site and described anumber of its archaeological features,6 as didViktor Lilcic two decades later.7 Borka Drago-jevic-Josifovska collected the inscriptions fromthe vicinity and published them together in1982.8

In 1988 Zivojin Vincic of the Protection In-stitute directed a project focused on the even-tual conservation of the Rotunda.9 The con-struction of the Skopje-Sofia railroad line ledto emergency salvage excavations in 1995 di-rected by Milan Ivanovski of the Protection In-stitute; in the cemetery area of K’sla, across theriver from Golemo Gradiste, a number of pre-historic and early Roman burials were discov-ered as well as a small early Byzantine churchassociated with a large vaulted tomb.10

In June 1998 a Macedonian-American proj-ect, directed by Kiril Trajkovski of the Museumof Macedonia and Carolyn S. Snively of Gettys-burg College, surveyed the site and its imme-diate environs. Their objectives were to find,identify, and describe the features of the site vis-ible on the surface and to record them on a top-ographical plan; to examine the Rotunda for

answers to specific architectural and liturgicalquestions; and to collect and analyze ceramicmaterial from surface survey and test trenchesin order to understand the diachronic occupa-tion of the site and its commercial and culturalconnections. The results of that pilot projecthave been incorporated into this report.

DESCRIPTION OF THE SITE

The hill known as Golemo Gradiste11 risesnearly 100 m above the Kriva River to a heightof 440 meters above sea level (Figs. 1, 2). Itserved as the acropolis and the middle sectionof a fortified ancient city divided into threeparts. This steep hill, nearly 500 m in lengtheast-west, is one of a series of cliffs along thesouth bank of the Kriva River. In contrast to thegenerally steep and craggy terrain of the acrop-olis, the eastern end of the hill consists of a nar-row plateau that slopes gently down from northto south and west to east. Beyond the erodedremains of rock-cut rooms visible along thenorthern side of the plateau, the bedrock dropsto a steep, narrow terrace. The terrain thenfalls precipitously to the northern part of thecity beside the Kriva River.

This second, northern part of the site con-sists of a gentle slope running from the foot ofthe acropolis north to the river. The line of thefortification surrounding this section of the citymay be traced, and the fabric of the wall is vis-ible in places, resting on a bedrock foundationbeside the river (Fig. 3). The location of a northgate and the probable locations of an east andpossibly a west gate can also be seen. Withinthis fortified area, the outlines of several largebuildings are visible.12 When Radojcic visitedthe site, he saw near the foot of the acropolis thewalls of a large early Byzantine basilica,13 which

298 ARCHAEOLOGICAL INVESTIGATION AT KONJUH IN 2000

4 N. Vulic, “Anticki spomenici nase zemlje,” Spomenik 98(1941–48): 96–97, nos. 211–13.

5 S. Radojcic, “Crkva u Konjuhu,” ZRVI 1 (1952): 148–67.6 I. Mikulcic, “Anticki gradovi kod Drenova i Konjuha u

Makedoniji,” Arheoloski pregled 15 (1973): 179–82; idem,“Über die Grösse der spätantiken Städte in Makedonien,”Ziva antika 24 (1974): 207–8; idem, “Dva bezimeni doc-noanticki grada vo Istocna Makedonija,” Zbornik na Arheo-loski Muzej Skopje 6/7 (1975): 115–21; idem, Srednovekovni gra-dovi i tvrdini vo Makedonija (Skopje, 1996), 223–26; idem,Anticki gradovi vo Makedonija (Skopje, 1999), 320–22, 358–61.

7 Lilcic, “Razmisluvanja,” 36–37.8 B. Dragojevic-Josifovska, Inscriptions de la Mésie Su-

périeure VI. Scupi et la région de Kumanovo (Belgrade, 1982),177–80.

9 See the report by Z. Vincic et al., “Elaborat za konzer-vacija i prezentacija na Rotondata vo seloto Konjuh, 1988,”in the archives of the Protection Institute.

10 See the brief note about the burials by M. Ivanovski,“K’sla,” in Arheoloska karta, vol. 2 (Skopje, 1996), 185. Thechurch has not yet been published.

11 In a strict topographical sense, Golemo Gradisterefers to the hill, i.e., the large hill as opposed to MaloGradiste, the smaller hill or long ridge to its south. Theterm gradiste, however, means a town site. Golemo Gradisteis used here to refer to both the large hill and the site gen-erally.

12 A heavy covering of grass and bushes prevented thecollection of pottery from this part of the site in 1998, sothat the date of occupation—except for the 5th- or 6th-century basilica—remains uncertain. In June 2001, how-ever, we examined material from a hole illegally excavatedin this part of the site; although the pottery included anumber of identifiably “Roman” sherds, the two levels ofoccupation dated to the 3rd–4th and 5th–6th centuriesrespectively.

13 Radojcic, “Crkva,” 149.

1 Plan of Golemo Gradiste (plan: G. C. McArdle, after I. Mikulcic, M. Milojevic, and the National Survey Institute)

2 Golemo Gradiste, acropolis, from the northwest

3 City wall foundation, north side along the Kriva River, from the northeast. The city wall is visible in places both aboveand behind the foundation.

4 Room in center of the acropolis, from the east. Note the placement of niches and a window so as toresemble a face.

5 Entrance to rock-cut room in western part of acropolis, from the east

6 Cistern, Sector 1, acropolis, from the east, during 2000 excavation

7 Eastern plateau, acropolis, from the west, before 2000 excavation

8 Western room, Sector 1a, acropolis, from the south

9 Sector 1b, Trench 3a, acropolis, from thenorth. Note the test trench dug (foreground)through the floor north of Wall 1 (center)and nearby steps (left). Wall 2 appears as arounded feature (left middle ground); thefortification wall marks the trench’s southend.

10 Plan of rotunda (plan: M. Milojevic)

➝N

11 North half of rotunda, from the southeast, 1988. Note apse (lower right), large northern pier (center),and modern chapel (upper left).

12 The newly cleaned St. George, from the northeast, 2000

13 The newly cleaned St. George, from the west, interior, 2000

since that time has been almost completely de-stroyed by illegal excavation.

The narrow third section of the city lay be-tween the lower south side of the acropolis anda nearly parallel ridge or outcropping of bed-rock known as Malo Gradiste, which served asthe foundation for a city wall. The eastern endof Malo Gradiste has been destroyed, and theeastern third of this section of the city has beenleveled for the purpose of cultivation.14 A largecity gate near the foot of Malo Gradiste pro-vided access to this part of the city. Two parallellines of city wall ran from the gate to the westend of the acropolis hill.

Outside the fortified city, a deep ravine sepa-rates the east slope of Golemo Gradiste fromthe next cliff to the east, Gagin Kamen. A road,partly cut into the rocky bottom of the ravine,partly paved with large stone slabs, runsthrough the ravine.15 Presumably of Roman orlate antique construction, the road was still be-ing used for local traffic in 1998.

Outside the city and across the Kriva River tothe northwest, on a plateau known as K’sla, afew grave inscriptions and the discovery in1995 of forty-six early Roman burials indicatethe presence of a Roman necropolis. Of the sixpublished Latin inscriptions from Konjuh,16

five appear to be funerary; only two preservecomplete texts. The epitaph of Sabinus Antius,a thirty-five-year-old soldier, has been dated tothe second half of the third or the beginning ofthe fourth century.17 Some or all of the exca-vated early Roman graves had been dug into atumulus; both inhumations and cremationswere found in arched tile graves, that is, gravesconsisting of one or more pairs of pantiles withtheir long sides leaning against each other atthe top to form a kind of arch over the body orthe cremated remains.

At Crkvica, near the southern edge of K’sla,

are the remains of a small early Byzantinechurch. Definitive statements must await itspublication, but the combination of largevaulted tomb and small church is paralleled ata number of sites nearby and further south, forexample, the enormous tomb beside the ear-liest church at Morodvis in the Bregalnica Val-ley, the cemetery church on Karatasou Streetwithin the fortification walls of Beroia, andBasilica B at Argos Orestikon (ancient Diocle-tianoupolis?) near Kastoria.

THE ACROPOLIS OF GOLEMO GRADISTE AND

THE EXCAVATIONS IN 2000

Almost all visitors to the site have commentedon the rock-cut features of the acropolis, whichappear at first to be a confused jumble of cells,walls, streets, water channels, niches, cuttingsfor columns, rooms, staircases, and other fea-tures (Fig. 4). Closer observation has shown atleast two quite different approaches to the uti-lization of the bedrock cliffs and, on the easternportion of the hill, successive phases of use.

The rooms along the north side of the east-ern plateau were created by quarrying and cut-ting down into the bedrock from the top andleaving sections of it standing to serve as foun-dations for walls. The rooms are roughlysquare or rectangular with an occasional apse.Corridors and staircases run parallel with or atright angles to adjacent rooms, and the stairsprovide access to rooms located in the northface of the cliff or to the narrow terrace alongthe north edge of the acropolis. Odd featuresabound, but the assumption of the right andthe ability to adapt the landscape to their needsby removing and shaping large sections ofbedrock, as well as the organization and man-power required, point to a Roman or morelikely a late antique date for the original cre-ation of the architectural complex on the east-ern plateau.

On the western part of the acropolis the best-known feature is the room (3.80 × 3.35 m) thatwas created by digging horizontally into a cliffface (Fig. 5). This room included a rock-cut bedon the south side under a window, a bencharound the west and north sides, and a tomb inthe middle of the floor.18 The doorway opens

CAROLYN S. SNIVELY 299

14 According to local informants, in the early 1980s theKratovo-based company Sileks used heavy machinery todestroy the east end of the ridge and thus made the south-east part of the site accessible to tractors and combines. Theplans of the site drawn by I. Mikulcic in the 1970s showMalo Gradiste with a tower marking its east end and thesoutheast corner of the city.

15 Lilcic, “Razmisluvanja,” 36–37.16 Dragojevic-Josifovska, Inscriptions, 177–80, nos. 234–

39. A seventh inscription found on an impost capital in theRotunda is described below.

17 B. Josifovska, “Jedan novi vojnicki natpis iz Konjuha,”Ziva antika 13/14 (1964): 166–70.

18 Despite references to rooms or cells in the plural, thisis the only one of its kind. It was described and drawn by

onto a terrace, and cuttings in the rock abovethe entrance suggest a porch or a second roombuilt in front of the rock-cut one. Around andabove this terrace are niches of various shapesand sizes, a number of curving staircases thatconsist of footholds and even handholds carvedinto vertical rock faces, and cuttings for beamsthat point to rooms built against the cliffs. Thepeople who occupied this part of the acropolisseem to have had both less ability to cut andquarry bedrock and possibly a more benign at-titude toward the utilization of the cliffs ofGolemo Gradiste. The evidence suggests that asmall community once lived around the ter-race. The hypothesis we inherited, that it was amonastic community in the medieval period,awaits further investigation.

The collection and analysis of surface potteryon the acropolis in 1998 had indicated occupa-tion in the prehistoric, Hellenistic, late Roman/late antique, and medieval periods.19 Pottery ofthe fourth–sixth centuries was by far the mostcommon. With the exception of several types of late Roman amphorae from Carthage, how-ever, imports were rare, not only on the ac-ropolis but everywhere ceramic material wascollected. The usual Roman finewares were en-tirely absent, and no pottery dating betweenthe third century B.C. and the third century A.D.could be identified.

Two test trenches dug on the acropolis in1998 revealed debris from a late sixth- or earlyseventh-century destruction. In the first one,near the south edge of the central part of theacropolis, numerous fragments of large storagevessels and cooking pots were found in a roomformed by two rock-cut north-south walls. Inthe second test, in the southwest part of theeastern plateau, part of a poorly preservedoven or kiln came to light.20

On the basis of information from the surveymaterial and the test trenches, we expected tofind substantial remains from the late antiqueperiod on the hill. In summer 2000 we there-fore excavated in three sectors on the easternhalf of the acropolis. In each area, rock-cut ar-chitecture was visible on the surface, and thetrenches were placed to investigate that archi-tecture and, if possible, to elucidate its use.

Sector I “Cisterna”

Near the highest point of the acropolis,above and west of the eastern plateau, we in-vestigated a large room, ca. 4 × 10 m, dug downmore than 2 m into the rock21 (Fig. 6). Two testtrenches were dug to the floor of the room, onein the northwest corner along the north wall,and a second one running north-south acrossthe room to the east of its midpoint. In Trench1, two bedrock features projecting from thesurface of the north wall and the floor nearbysuggest their deliberate use in the plan or func-tion of the room when it was hewn from therock. In Trench 2, two roughly circular depres-sions near the north wall of the chamber couldhave been created at any time during the use ofthe building, perhaps for storage jars. The ex-cavator concluded that the room had func-tioned as a storage chamber, a cellar below abuilding constructed of large, carefully workedstone blocks. After the structure went out ofuse, the walls collapsed into the undergroundroom; a section of wall composed of sizableblocks was excavated where it had fallen intoTrench 2.

No evidence for the original date of con-struction was found, but the underground


Radojcic, “Crkva.” The other rock-cut spaces opening fromthe terrace look more like storage spaces, complete withcuttings along the front edges for the insertion of boards,than monastic cells. One other, fairly sizable room, dug hor-izontally into the south face of the acropolis, has been dis-covered further to the east and at a lower level; althoughpartly filled with earth, it appears to be a simple room with-out amenities.

19 V. Anderson-Stojanovic provided this summary of theceramic material in her 1998 report. Fields south and westof the acropolis yielded pottery of prehistoric to Turkishdate. Pottery from a field on K’sla was dated to the 3d cen-tury. A number of sherds found in 2001 in late antique ormedieval contexts have now been identified by VojislavSanev of the Museum of Macedonia as local wares of theNeolithic and the Iron Age.

20 For both test trenches we took advantage of holes dugby illegal excavators, cleaned their profiles, and then dugstratigraphically beside them.

21 Virginia Anderson-Stojanovic was the excavator inthis sector; this account is based on her season report. As in-dicated by the name given to this area, we had originally as-sumed that the room was a cistern. Anderson-Stojanovicpoints out that no traces of plaster or mortar exist on any ofthe already exposed or newly excavated surfaces, and thatseveral deep cracks through the bedrock argue against awater-tight container. Our original identification as a cis-tern raises the problem of water supply on the acropolis.There is and probably was no source of water on the hill. Allwater must be carried up either from the river or fromthree springs outside the city or collected during rains andstored for later use. An archaeologist from the KumanovoMuseum has mentioned evidence for an aqueduct bring-ing water to the site, but we have not yet verified details andlocation.

room and its superstructure seem likely to haveformed a part of the late antique complex onthe acropolis. After the collapse of the super-structure, the room was used as a garbagedump, as suggested by many animal bones andmuch fragmentary pottery. All of the materialfound in the room was later than the late an-tique period and remains to be studied and achronological determination made.

Sector IA

To the east of the subterranean room, the ter-rain drops abruptly to the eastern plateau (Fig.7). Sector IA is located near the foot of thatdrop, at the western end of the plateau, andstretching back from its northern edge. Withinan excavated surface area of ca. 100 m2, tworooms of a large building, possibly administra-tive but more likely residential, were found.22

Room 1 at the west had interior dimensionsof 3.60 × 9.40 m; it displayed an apse at thenorth and a doorway in the south wall (Fig. 8).The dimensions of room 2, adjoining the firstroom on the east side, were 4.0 × 9.0 m. Near itssouth wall a round hearth (diameter 1.20 m)was cut into the bedrock floor.

The western wall of room 1 consisted of abedrock socle above which a wall of stones andmortar once rose. The north walls of bothrooms had been almost completely destroyed,but their impressions could be traced in themortar substructure. The south wall was care-fully built of similarly sized stones bonded withlime mortar. Evidence for two phases of con-struction are visible in the building. The eastwall of room 2, constructed of stones and mudmortar, appeared to the excavator to have sub-divided an originally larger room. Two door-ways once connecting rooms 1 and 2 had beenclosed in the second phase.

The floor of room 1, almost completely pre-served, showed two circles formed from smallpieces of gray or white limestone against a back-ground of red tile fragments. In room 2, a lev-eled bedrock surface formed part of the floor,with pieces of tile filling the intervening spaces.

The depth of fill in Sector IA did not exceed0.50 m. The stratigraphy was relatively simple:(1) a surface layer of grass and earth with agreat deal of post-late antique pottery, and (2) a

layer of debris heavy with large stones, whichrested on the floor.

The small finds included sherds of coarsepottery, metal belt buckles, knife blades, and anearring. Unfortunately nothing found in thetwo rooms provides any evidence for their func-tion. A first half of the sixth-century date hasbeen tentatively proposed for the building.23

The two excavated rooms lie at the north-ern edge of the plateau. The building almostcertainly extended to the east and the west.The discovery of this building demonstratesthe presence of monumental architecture onthe acropolis of Golemo Gradiste. Several frag-ments of stone moldings, including one ofmarble, point to a degree of provincial elegance.

Sector IB

In three trenches in Sector IB we investi-gated the north-south width of the easternplateau, from the eroded rock-cut remains ofan apsidal room and a corridor at the north, tothe line of the fortification/terrace wall markingthe south edge of the plateau, a distance ofca. 17.5 m. The surface of the plateau dropsca. 1.5 m from the north to the south side.24

Excavation in the first 5 × 5 m trench at thenorth quickly revealed a very rough surface ofbeaten stones and earth covering the leveledbedrock in the apsidal room; east of the roomran a rock-cut corridor, whose north end con-nected with two staircases carved into the cliffforming the northern edge of the plateau. Thedepth of earthen fill in Trench 1 nowhere ex-ceeded 0.20 m.

Trench 2, also 5 × 5 m, was located one meterto the south of the first one. As a test latershowed, the level of the bedrock dropped ca.0.50 m in the space between the two trenches;bedrock was not reached in Trenches 2 or 3.

The third trench was located one meter tothe south of Trench 2 and extended to thesouth edge of the plateau. The north face of thefortification/terrace wall formed the effectivesouthern edge of Trench 3, whose length be-came 5.60 m along the east side and 6.20 malong the west. Excavation in a 2 m wide stripalong the west side, Trench 3a, reached a final


22 Dance Golubovska was the excavator in this sector;her season report is the basis for this account.

23 The pottery on which Golubovska bases this datecomes from within the building and on or above the floor;it does not therefore provide reliable evidence for the dateof construction.

24 Carolyn Snively was the excavator in this sector.

depth of ca. 2.6 m beside the fortification wall(Fig. 9).

Wall 1, running east-west across Trench 3aparallel to and ca. 3.5 m north of the terrace/fortification wall, appears to have separated apaved room to the north from a cellar at thesouth. A deposit of debris with roof tiles andlarge fragments of smashed pottery covered afloor of stones and packed earth north of Wall1. Testing below this floor revealed a very solidpacking and substructure. Three stone steps ofa stairs, built beside the north face of Wall 1, led from this floor up to the east, perhaps to anupper story above the cellar on the south side of Wall 1.

South of this wall, the western face of a sec-ond wall appeared in the scarp of Trench 3a.Wall 2, which lack of time did not permit us toinvestigate, was associated with deposits of clay.Below those deposits and the bottom of Wall2, a layer of debris ca. 0.90 m deep appeared in the space between Wall 1 and the terrace/fortification wall. It included, among otherthings, many roof tile fragments and largesherds of broken pots, window glass, and vesselfragments, five loom weights, 5 small millstones(?), iron fragments, and a small marble slab. Noidentifiable floor or use level was found be-neath this debris, a circumstance which sug-gests that the artifacts fell from an upper floorinto this subterranean space. Although the bot-tom of the debris and of the fortification wall layat nearly the same depth, we continued to ex-cavate another ca. 0.60 m of earth, of which thefinal 0.30 m was sterile soil.

The fortification wall is preserved in Trench3a to a height of ca. 2 m. Although its north facedisplays a solid construction of fieldstones andlime mortar, the southern exterior face of thewall could not be found and appears to havebeen destroyed to its foundation.

Given the very limited area of excavation inSector IB, particularly Trench 3a, we can reachonly preliminary interpretations and conclu-sions. The ceramic material from the destruc-tion debris above the floor north of Wall 1 andfrom the 0.90 m deep deposit of debris at alower level south of Wall 1 appears on initial ex-amination25 to be similar and probably to rep-resent the late sixth- or early seventh-century

destruction encountered elsewhere on the site.This conclusion, which may require revision,would indicate that Wall 2 and associated de-posits represent post-late antique activities onthe acropolis, behind the shelter of the still ex-isting terrace/fortification wall.

Our present hypothesis, to be tested in futureseasons, states that the gently sloping plateauon the eastern third of the acropolis is an artifi-cial construction of the late antique period. Theremains of rock-cut walls along the north sideof the plateau show that the bedrock once roseto a greater height. Because it drops off towardthe south, however, a relatively level space wascreated by building a series of terrace walls andone major wall that probably served (1) as ter-race wall, (2) as the supporting south wall ofbuildings, and (3) as a fortification along theedge of the newly created plateau.

THE ROTUNDA CHURCH

Ca. 160 m south of Malo Gradiste and thusoutside the fortified city, the remains of an un-usual early Byzantine church in the form of aRotunda are visible (Figs. 10, 11). Local vil-lagers excavated the building in 1919, and ithas stood exposed to the elements for morethan eighty years with only minimal conserva-tion. The use of large, cut stone blocks in thewalls and piers of the Rotunda is typical of theregion and is probably a factor in the partialsurvival of the excavated structure. Radojcic’sarticle,26 describing the state of the church in1938, remains an essential source for scholarswho wish to study the Rotunda, because hedocumented and discussed elements of thechurch that no longer exist.

The exterior walls of the church form atrapezoid, ca. 21 m long, whose east and westwalls are parallel. A rectangular apse protrudesfrom the east wall and buttresses in the form ofsmall towers from the east corners. A centraldoorway in the west wall gave access to a smallnarthex flanked, within the west corners of thetrapezoid, by apsidal rooms. A modern chapelwas constructed over the northwest apsidalroom in 1955, so that Radojcic’s photo27 pro-


25 Anderson-Stojanovic did not have the opportunity toanalyze the pottery from the last week of excavation, so that

the preliminary conclusions about chronology here and inSector IA are those of Snively and Golubovska respectively.

26 Radojcic, “Crkva,” 152–67.27 Radojcic, “Crkva,” fig. 16.

vides the only view of it. Cleaning in 1998 veri-fied that the east end of the southwest room wasapsidal and that a platform had occupied theapse.

From the narthex, through a tribelon, oneentered a U-shaped aisle or corridor whose eastends formed pastophoria beside the presby-terium. A circle of four piers and six large mul-lion columns divided the aisle from the roundnave, which could be entered at the westthrough a second tribelon and through two en-trances beside the eastern piers. Screen slabsresting on a low wall closed the other four in-tercolumniations, at north and south.

The two large, irregularly shaped piers at theeast side of the nave marked the west corners ofthe presbyterium; between them stood a chan-cel screen of slabs supported by six posts. Thesouthern half of the base for the screen is stillpreserved. Radojcic found no trace of an altaron the east-west axis in front of the apse. Hisplan showed the destruction of the centralstone slabs of the presbyterium floor, suggest-ing that the reliquary normally to be found un-der the altar had been removed in antiquity orbetween 1919 and 1938. Cleaning in 1998 re-vealed one slab of the floor still in situ at thenorth side of the presbyterium.

Clergy benches at the north and south sidesof the presbyterium abutted the piers markingthe west ends of the apse; some blocks of thesouthern bench are preserved. Between themand the piers at the west corners of the presby-terium, narrow openings allowed access to thepastophoria, from which doors in the northand south wall of the church opened to the ex-terior.

A unique feature of the Rotunda was theblind corridor within the apse. As Radojcicshowed both in plan and in photograph,28 theinner side of the rectangular apse was semicir-cular, but a wall—straight on its west face, con-vex on the east—ran across the chord of theapse, except at the south side where the en-trance to the blind annular corridor formed bythe two concentric apsidal walls was located.Three steps cut into the front of the wall on thechord of the apse presumably led up to a plat-form on top of the wall or above the corridor;there the episcopal throne or the seat of the

presiding clergyman once stood. The only partof this construction now preserved consists ofseveral precariously balanced blocks at thenorth corner of the apse. Cleaning carried outin 1998, in the hope of finding traces of a foun-dation for the vanished inner apsidal wall, indi-cated that no foundations had been employedin the church for any features except load-bearing walls and piers. Small rectangularspaces were noted at the corners of the apse, afeature not shown on Radojcic’s plan. The an-nular corridor may be related to the kykleianoted in sixth-century churches further south,for example, in Basilica A at Amphipolis, or con-ceivably even to the apsidal crypts found inchurches in Macedonia.29 But the precise pur-pose of the blind annular corridor in the Ro-tunda at Konjuh remains and probably will re-main a mystery.

Between the northwestern part of the aisleand the apse of the northwest room, where thewall was quite thick, Radojcic noted a doorwayopening to a staircase leading to an upper story.In 1998 no evidence for the staircase was ex-tant. A shallow niche occupied a similar posi-tion in the southwest wall of the corridor.

Radojcic assumed that the dome, the archesabove the colonnade, and the vault coveringthe aisle had been constructed of brick, ofwhich many examples were still visible in thedebris at the time of his visit; hexagonal bricksalso paved the floors of nave and aisle. Not evensmall fragments of brick or tile now remain inthe building.30 A reconstruction of the upperportions of the Rotunda, including the dome,the roof above the narthex and aisle, and pos-sible galleries, awaits further study of the re-mains.31

Radojcic found a large number of pieces ofarchitectural sculpture in the Rotunda; somewere removed to the Archaeological Museum,


28 Radojcic, “Crkva,” figs. 11, 12.

29 C. Snively, “Apsidal Crypts in Macedonia: PossiblePlaces of Pilgrimage?” JbAC, suppl. 20.2 (1995): 1179–84.Also of relevance may be P. Chevalier’s discussion of the“synthronos libre” in Dalmatian churches in her book, Ec-clesiae Dalmatiae. L’architecture paléochrétienne de la provinceRomaine de Dalmatie (Rome, 1996), 2:117–18.

30 A large number of ancient bricks are visible in a 20th-century building near the Rotunda. Workmen from the vil-lage of Konjuh have told us that bricks and rooftiles fromthe site are very valuable for the construction of ovens.

31 One of the goals of the Konjuh Project is to carry outadditional minor investigations in and around the Rotundaand to publish a definitive study of this unique monument.

now the Museum of Macedonia, in Skopje,while others remain on site.32 In addition to thebases of the mullion columns of the colonnadeand pieces of the columns themselves, Radojcicrecorded three impost capitals; one carried abrief, enigmatic inscription. The function offour smaller capitals remains uncertain. Frag-ments of the posts from the chancel screen basealso appeared, richly carved in a soft, green lo-cal stone. Radojcic divided the pieces of flatand curved screens into three groups: (1) thosewith relief carving, mostly of animals, on onlyone side; (2) fragments from the chancelscreen, carved on both sides with variations ofcrosses in circles;33 and (3) pieces from theambo.34

From the time of its initial publication, thedate assigned to the Rotunda on the basis of itsplan and its architectural sculpture has beenthe sixth century.35 I. Nikolajevic dated thesculpture to the middle of the sixth century, cit-ing its connections with Justinianic monumentsin Serbia, for example, at Caricin Grad.36 Nei-ther the investigations of the Protection Insti-tute in 1988 nor ours a decade later revealedany new archaeological evidence for dating thebuilding; almost no pottery was found. Never-theless, the plan and the architectural sculp-ture are not consistent with the fourth-centurydate recently advanced for the Rotunda byBlaga Aleksova.37 Radojcic posed the questionwhether there had been more than one phaseof construction. The investigations of the Pro-tection Institute and ours confirmed Radojcic’sown conclusion that no major renovations had

taken place in the church; it was basically a one-phase building. The only piece of evidence forchange or renovation was the walling-up of thenorth exterior doorway.38

Radojcic also asked whether the Rotundahad served as a martyrium, adducing its centralplan and the inscription on an impost capital,DOMATRIRS, interpreted as domus martyris. Hesuggested that relics might have been kept inone of the western apsidal rooms or that thegrave of a martyr might have been located un-der the altar.39 Aleksova identified the buildingas a martyrium and assumed that members ofthe local elite had been buried in the westernrooms.40

The most significant discovery made by theProtection Institute in 1988 was that the build-ing had been constructed on a nearly sterilesite. Everywhere, except in the southwest room,a layer of yellow clay appeared below the sur-face humus and broken pieces of the substruc-ture of the now vanished floor. Additionalcleaning along the east-west axis of the presby-terium in 1998 exposed undisturbed yellowclay immediately below the surface deposit andindicated that only a small and shallowly placedreliquary could have occupied the space underthe altar. The modern chapel makes investiga-tion in the northwest room impossible for theforeseeable future. Below the southwest room,however, the Protection Institute uncoveredtwo walls forming the right-angled corner of anearlier structure. The walls were smaller thanthose of the Rotunda and of different construc-tion and orientation. No evidence of burialswas reported. The Protection Institute’s trenchalong the outer face of the church walls was ex-cavated through yellow clay without finds.

The central plan and the inscription remainthe only evidence for a martyrium. No burialswere found in the church or immediately


32 Radojcic, “Crkva,” 154–55; I. Nikolajevic-Stojkovic,Ranovizantijska dekorativna plastika u Makedoniji, Srbiji, iCrnoj Gori (Belgrade, 1957), 47–50, 91; R. F. Hoddinott,Early Byzantine Churches in Macedonia and Southern Serbia(London, 1963), 220–26. Snezana Filipova, from the Uni-versity of Cyril and Methodius in Skopje, is restudying thearchitectural sculpture from Konjuh.

33 K. Petrov, “Staurodekoracija od Konjuh,” Zbornik naArheoloski Muzej Skopje 2 (1957–58): 31–45.

34 K. Petrov, “Rekonstrukcija na ambonot od rotondatavo Konjuh,” Godisen Zbornik na Filozofski Fakultet, Skopje 22(1970): 271–302; J.-P. Sodini, “La sculpture architecturale àl’époque paléochrétienne en Illyricum,” in Actes du Xe Con-grès International d’Archéologie Chrétienne, Thessalonique, 1980(Thessalonike–Vatican City, 1984), 1:293–94.

35 Radojcic, “Crkva,” 163.36 Nikolajevic-Stojkovic, Dekorativna plastika, 50.37 B. Aleksova, “Konjuh: Golemo Gradiste,” in Arheoloska

karta, vol. 2 (Skopje, 1996), 184–85; eadem, Loca SanctorumMacedoniae. The Cult of Martyrs in Macedonia from the 4th to the9th Centuries (Skopje, 1997), 259–60.

38 Radojcic, “Crkva,” 155. As indicated above, on theacropolis of the city there is clear evidence of a late 6th- orearly 7th-century destruction; an extramural church wouldnot have escaped that destruction. We found no evidence tosupport Aleksova’s contention that the church had func-tioned again in the 9th and 10th centuries, although thegeneral area including the Rotunda to the south of MaloGradiste is known as Seliste and seems to have been occu-pied for some period between late antiquity and moderntimes. We shall never know if evidence for later reuse ex-isted in 1919 and was swept away.

39 Radojcic, “Crkva,” 154–56.40 Aleksova in Arheoloska Karta, 185; eadem, Loca Sancto-

rum, 256, 259–60.

around it.41 Although a church of the sixth cen-tury would certainly have included a relic un-der the altar, the data are insufficient for us tosay whether the Rotunda served as a mar-tyrium in any additional sense.

Ca. 150 m southeast of the Rotunda a ha-giasma, of unknown antiquity but still in use in1998,42 is located within a grove of trees. Theshrine consists of an enclosure focused on asmall structure that stands above the mouth ofa well. The springhouse, if it may be so called, isbuilt of stone blocks.43 Icons decorate the fa-cade of the structure, and niches in the enclo-sure wall hold cups and other containers forwater. The mineral water from the well is said tobe a panacea.44 Although the distance betweenthe Rotunda and the hagiasma leaves any con-nection between them uncertain, the existenceof the latter reminds us that numerous paganshrines may have stood in the countrysidearound Golemo Gradiste and offered sites andreasons for the construction of churches.

CLEANING AND DOCUMENTATION OF THE

CHURCH OF ST. GEORGE

Ca. 300 m southwest of the site and ca. 400 mwest of the Rotunda, within the functioningcemetery of the village of Konjuh, stands theroofless church known as St. George (Fig. 12).The building and the graveyard are isolatedfrom the surrounding fields by ravines and canbe easily approached only from the southeast.St. George stands a few meters east of the deepravine through which a small stream runsnorth to the Kriva River. A spring rises in theravine near the church.

St. George is a one-aisle building with a pro-truding eastern apse. It is approximately 7 mlong by 4.5 m wide. It was built in part of largestone blocks, presumably taken from the ruinsof the ancient city; small stones, brick and tilefragments, and mortar fill in the spaces be-tween the blocks. A fragment of a Roman tomb-stone was built into the outer face of the south

wall. Wooden beams, visible on the interior, re-inforced the structure. The use of smallerstones and different construction, especiallynear the top of the walls, suggests interventionsand perhaps a new roof at some point. Theremnants of the roof indicate that it too was ofstone.

A band of decorative brickwork is still pre-served across the east wall above the apse.Three niches break up the monotony of the ex-terior face of the north wall. A narrow windowopened at a relatively low level in the apse, andan apparent second window was located in thesouth wall. The only entrance is centered in thewest wall.

The floor had been paved with large stoneslabs of varying shape; they rested on a stonefoundation. At some time, probably after WorldWar II, treasure hunters pried up the stoneslabs of the floor and left them projecting atangles here and there with gaping holes be-tween. Two trees took root within the ruinedfloor and grew higher than the preserved walls.The date when the roof collapsed is not knownbut predated Radojcic’s visit in 1938.

Two vertical slabs of the chancel barrier re-main in situ, as does the base for the altar withinthe chancel area. Severely damaged remains offresco are still visible on the east wall (Fig. 13).

The church was declared a “monument ofculture” and placed under legal protection in1954 under the name of St. George.45 Never-theless, neither the original dedication nor thedate of construction is known; the proposeddates run from the fourteenth to the sixteenthcentury. Radojcic, in his survey of the site, de-scribed and briefly discussed this church, pro-vided a plan and photograph, and identified itas a church of the Holy Archangel. He cites asevidence for this identification the fresco in thelunette above the entrance, now vanished with-out a trace; he was, in fact, arguing against anearlier identification as St. Nicholas.46

In July 2000 we took the following measures.


41 A pile of stone slabs in a field some distance to thesouth of the Rotunda is the only evidence for burials in thevicinity.

42 In summer of 2000 the shrine appeared to be neg-lected and no longer in use.

43 We observed a similar little stone building built over aspring in the ravine southwest of the church of St. George.

44 Radojcic, “Crkva,” 152, mentions the spring briefly asdoes Aleksova, Loca Sanctorum, 252.

45 Information provided by D. Golubovska.46 Radojcic, “Crkva,” 148–49, figs. 6–7. There has been

considerable controversy about the dedication of thischurch, including the question whether Radojcic mighthave confused its saint with that of another church stillstanding in the vicinity in 1938. We know from local in-formants that a church once stood near the modern bridgeover the Kriva River, only a few hundred meters from theone now referred to as St. George.

The two trees inside the shell of the churchwere removed without damage to the standingwalls. The floor slabs, torn from their places,were replaced. Loose earth was removed fromthe interior of the building. The architecturalpieces and spolia found inside the church weredocumented and arranged along the walls. Atrench immediately outside the west doorway,presumably dug by treasure hunters, wascleaned, examined, and backfilled. Architec-tural pieces and spolia found outside andaround the church were examined, separatedfrom heaps of unworked stone, and drawn orphotographed. The terrace wall near the southside of the church and the two stone “tables” forfunerary meals, located to the southeast andnorth of the building, were cleaned of vegeta-tion.

The present condition of the church was doc-umented both photographically and by meansof architectural drawings. It will now be pos-sible to study the building, which presents sev-eral interesting features. We also hope that doc-umentation of the present precarious conditionof St. George will move it up in the long queueof churches awaiting conservation.

CONCLUSIONS AND FUTURE DIRECTIONS

Our excavations on the acropolis of GolemoGradiste, together with observations aboutother parts of the site and its environs, have ledto several conclusions, some of which may re-quire revision in the future, and to a numberof hypotheses to be tested in coming seasons.It now seems clear that the general area wasoccupied, at least sporadically, from the lateBronze Age to the period of Turkish domina-tion, although the natural fortification ofGolemo Gradiste and the site of the late an-tique city were not always the focus of activities.The locations of any prehistoric, Hellenistic,and early Roman settlements remain to be dis-covered. Although no stratified, post-late an-tique deposits were found on the acropolis in2000,47 the amount of later pottery points to

long-term or intensive occupation in the me-dieval period. Analysis of the excavated mate-rial and its correlation with pottery from otherparts of the site will eventually provide a cleareridea of the shifting patterns of activity on andaround Golemo Gradiste through the cen-turies.

In late antiquity a heavily fortified city occu-pied the site. To what extent this was a rebuild-ing, a fortification, and an enlargement of anearlier town remains unclear. The number ofsixth-century coins found on and around thesite point to major activity here in the sixth cen-tury.48 Very likely this city participated in theJustinianic building program described by Pro-copius (De aedificiis 4.4). One might then ask ifreconstruction was needed because the Scupiearthquake of 518, described by MarcellinusComes 100, had devastated this region.

The discovery of relatively large and sub-stantial buildings on the acropolis in all threesectors investigated points to intensive adapta-tion and use of at least the eastern part of thehill in the late antique period. The time and ef-fort required to carve rooms into the bedrockand to construct a fortification wall and terracewalls in order to create the eastern plateau donot reflect emergency response to a threat butrather a planned and deliberate program ofbuilding. But intentional and long-term occu-pation of an acropolis so inaccessible raises a se-ries of questions. What segment of the popula-tion chose or was compelled to live there? Localofficials? A military garrison? And why? For de-fensive or strategic reasons? The answers tosuch questions are not available yet, but ourwork in 2000 has begun to fill in with facts anddetails the general picture of the site painted bysurvey. We expect that investigations in futureseasons will allow us to answer some of the ques-tions about this fascinating site.

Gettysburg College


47 Medieval levels are being excavated in 2001. 48 E.g., Lilcic, “Razmisluvanja,” 37 and figs. 7–9.

archaeological investigation at konjuh

Documents