History of the research
The Dubrovnik Cathedral, as it stands today, is a remarkable Baroque structure erected between 1671 and 1713 on the grounds of its predecessor, destroyed during the 1667 earthquake (1). After the 1979 seismic event, structural reinforcement works were undertaken on the Cathedral’s foundations, including archaeological investigations (2, 3). Commencing in 1981 within the Cathedral interior, the research extended to the adjacent square (Bunićeva Poljana), ultimately encompassing an area of approximately 1,200 m² (Fig. 1) (3, 4, 5, 6, 9).[1]
As anticipated, the inquiry unveiled the foundation walls of the former Romanesque cathedral, whose existence is corroborated by the written sources records, historical depictions in paintings and graphics, as well as by the sculptural representations of the city of Dubrovnik (4, 9, 10, 12).[2] The cataclysmic earthquake of 1667, followed by fire, obliterated its remains, but aside from its foundations, among other archaeological findings, numerous architectural elements, fragments of stone furnishing and sculpture (4, 9, 7, 10, 11), and, based on recent studies, various fragments of wall paintings were found (4, 21).
A surprising discovery emerged when the remnants of an even earlier church were uncovered within the foundations of the Romanesque basilica, at some 2,40 to 2,60 meters beneath the baroque Cathedral pavement level. The investigation revealed the well-preserved architectural assembly of its sanctuary, the adjoining nave, and the sections of southern and northern aisle. On the sanctuary apsidal wall, stone-built and plastered two-levelled subselia with central bishops’ cathedra, as well as remains of a painted scene depicting the Church Fathers were still well preserved (Fig. 2) (4, 9, 10, 19, 21).
As the research further indicated, at the time when the construction of the Romanesque cathedral was well advanced, the interior of the ‘first church’ was filled with rubble and construction waste that served as the foundation for the new pavement. The rubble contained numerous fragments of liturgical stone furniture and sculpted architectural elements of late antique and medieval period, including secondarily carved stone and fragments of pottery and ceramic ware, glass material, and diverse metal objects (4, 7, 9, 19, 22). Among these findings, the most abundant were fragments of wall paintings, which can be classified in different historical and stylistic periods and attributed to different workshops and workshop circles (8, 10, 21).
Given the extraordinary discoveries, in 1983, the archaeological investigation expanded to the adjacent square (Bunićeva Poljana). Noteworthy findings included remnants of a defense wall, which bounded from the west the area encompassing the architectural remains of the ‘first church’, and those of the newly excavated quatrefoil building, situated next to the church’s façade. The latter has been identified as a memoria in its early stages, with the hypothesis that it may have later served as a baptistery (4, 9, 19). In different stratigraphic strata numerous burials that encircled the entire church complex were further discovered (4, 20). Finally, at various zones within the site, ruins of medieval dwellings were identified, and the remains of the late medieval bell tower-baptistry, which had been dismantled during the planned extension of the square in the 19th century, were uncovered (4, 9, 13, 19). By the end of the archaeological investigation, it was obvious that the excavated complex represented one of the most compound and layered ensembles of the Eastern Adriatic Coast, whose further research would significantly enrich the knowledge and the understanding of the earliest development of the historic city of Dubrovnik.
Consequently, during the final research campaigns (1987–1988), the site was covered with reinforced concrete slabs, allowing the restauration of the Baroque Cathedral to continue, while keeping the accessibility of the archaeological site for future exploration (3, 5, 19, 23). Unfortunately, at that time all further research had to be stopped due to the war in Croatia (1991–1995), and the majority of the movable artifacts remained stored in the attic and the underground chambers of the Cathedral for almost thirty years, where they were at risk from the war damage and later exposed to unfavorable microclimatic conditions.[3] Following that period, only in 2018, a comprehensive revision of the 1980s research was initiated through the project Discovering the Old Dubrovnik Cathedrals, and the systematic processing of all movable finds finally began, inevitably including re-examination of the in situ masonry structures, while respecting data given by the archaeological documentation and the written records from the 1980s (16, 17).[4]
Dating of findings
Due to the scarcity of historical records, the dating of the masonry structures within the “first church”, which was named by the first researchers as Byzantine basilica (3, 4, 9), relied greatly on the stylistic features of the fragmented architectural sculpture and liturgical stone furnishing, as well as in situ preserved wall paintings (7, 8, 10, 11). Nevertheless, the first published reports elucidated greatly its phases, as they included the analysis of the spatial and structural relationships between excavated walls and/or defined architectural units, also indicating traces of rebuilding and reconstruction activities, even some less discernable architectural modifications (3, 4, 5, 9, 19).
However, a broader dating range for the initial phases of the ‘first church’ was suggested, spanning from the early-Christian period of the 6th century, to different periods in the early Middle Ages (from the middle of the 7th century to the middle of the 9th century) (3, 4, 5, 7, 9). Additionally, at least four different phases of adaptations were proposed, notably visible in the well-preserved sections of the sanctuary and the eastern portions of the side aisles. Noteworthy alterations would include two different stages of the masonry clergy bench (subselia) in the sanctuary apse that was at one time extended to a double bench, and four stages of the bishop’s throne, which would later be expanded and elevated. Rich stratigraphy on the apse wall was identified, as layers of various plasters, limewashes, and wall paintings, spanning an extensive timeframe, were preserved. Significantly, the researchers also pointed out the later additions of columns and T-pylons in the nave and side aisles for the construction of the dome and vaults, as well as changes in stone furnishing (10th century). The persistence of the ‘first church’ culminated at the time when the construction of the Romanesque cathedral was initiated, when, in researchers’ opinion, the “Byzantine basilica” was first spatially reduced, and then partly demolished and partly covered with rubble (4, 9, 19).
Based on its construction techniques, an unearthed segment of a defense wall at Bunićeva Poljana was dated to Late antiquity (the 5th -6th century). The quatrefoil building posed a dating challenge. While its elevated position suggested an early Medieval origin, with discernable later modification (4, 9, 19), an alternate theory proposed its status as the oldest building within the complex. This hypothesis drew parallels with more reliably dated monuments from the broader Mediterranean area, as well as its proximity to the oldest excavated burials, which would typologically correspond to Late antiquity (24, 9).[5] However, more recently performed 14C analyzes of osteological material from one of those graves, as well as the burning layer under the quatrefoil building, point to the 7th to 9th centuries (14, 20), which together open up several research questions.
These considerations were later augmented by the authors delving into a more detailed analysis of the architectural evolution and the alterations of the church predominately based on the ornamented fragments of stone liturgical furnishing, including those from the broader area of Dubrovnik’s historic center (7, 10, 25, 26). But movable archaeological finds of the site in total, stored in the Cathedral’s attic, that could have played a pivotal role in enhancing the contextual understanding of the discoveries – more than 3000 pieces of diverse stone fragments and tens of thousands of wall painting fragments and ceramics, with numerous metal and glass objects – were until 2018 in greater extent excluded from all scientific considerations.[6] As they remained largely unprocessed for more than thirty years, the collaborative effort was made through the project DiODuCat to initiate the first-ever expert analysis, systematization and cataloguing of all the finds, while including proper storage procedures (Fig. 3) (16, 17, 21, 22).
As the research of the stone fragments and the fragments of wall paintings advanced, a re-examination of the architectural structures of the ‘first church’ was prompted, considering that different layers of plaster and murals, as well as built foundations of the liturgical furniture, including the altar, were preserved in situ. Structural transformations, in parts, evidence of micro adaptations of the church furnishing have been newly identified. Additional late antique and early Christian marble sculptures, complemented by many heterogeneous marble fragments, which could be dated to the 4th to 6th centuries, were subjected to a more detailed analysis (16, 17, 22).[7] On the other hand, by correlating the stratigraphy of wall painting fragments with that of in situ murals, efforts were made to categorize them based on different types of plaster, revealing a relatively diverse array. Six samples of plaster layers were investigated more thoroughly to determine the type of aggregates and binders. The mineralogical-petrographic composition of the samples was determined by optical microscopy and scanning electron microscopy with energy dispersive spectroscopy (SEM/EDS). In addition, one sample was analyzed by X-ray powder diffraction and thermogravimetry/differential thermal analysis (TG/DTA) to verify the presence of hydrated phases (21).[8] These analyses represented the base on which further methods of plaster dating were later planned.
At the Ruđer Bošković Institute in Zagreb, we conducted analyses using the 14C AMS method.[9] These results not only confirmed earlier hypotheses but also provided a new set of data that allowed us to identify numerous remains of wall paintings from the Romanesque cathedral that were previously unidentified. The most significant aspect for our discussion, however, was the results of the examinations of the samples of plaster from the sanctuary apse wall of the ‘first church’, connected with the oldest phase of its wall paintings, as well as of the coal from the remains of the lowest preserved floor level in its sanctuary (18).
Although the results largely corroborated some of the assumptions and hypotheses of the 1980s researchers, they also offered more detailed insights into the chronological relationships of certain masonry structures. Notably, as the applied radiocarbon analysis shows, the grey plaster associated with one of the earliest phases of the sanctuary, featuring a single clergy bench, could be dated to the second half of the 7th to the early 8th century (18). However, a noteworthy observation of the sanctuary apse structure indicates that the presumed oldest layer of grey plaster, which would correspond to the phase prior to the construction of the clergy bench, was applied later than previously considered. Namely, as we have established, the initial layer of the wall levelling in the apse interior comprised surface plaster resembling mortar used in the masonry of the apse wall (21).
Despite these intriguing discoveries, hypotheses proposing earlier structures in the ‘first church’ area still lacked support from natural science research.
To fill this gap, we turned to the IPERION HS consortium's additional dating methods. The project activities involved the application of the OSL for dating plaster and soil, along with the analysis of soil samples (red earth) using palaeomagnetism (archaeomagnetism).[10] OSL yielded effective results, with the outcomes of these analyses presented in this text.