Page 45 - Studia Universitatis Hereditati, vol 13(2) (2025)

came in the form of an ASCII grid with a ras- in the northern Adriatic. We treated enclosed
ter size of 1 m and was visualised using the Re- sites as a morphological continuum, incorporat-
lief Visualisation Toolbox (RVT) (Kokalj et al. ing enclosures of every size and type where an ex-
2019; Kokalj and Somrak 2019). The ‘archaeo- clusive modern agricultural or pastoral use could
logical combined cVAT’ visualisation technique be ruled out based on their form and/or relative
was used, blending hillshading from three direc- stratigraphy. The degree of certainty was catego-
tions with slope, positive openness and sky-view rised as one of three levels, based on the litera-
factor. ture and the visible structures in the digital ter-
In order to interpret and enhance this ALS- rain model (DTM). The topographic position
based terrain model, we drew on two comple- was categorised into different classes: knoll/hill-
mentary data streams. Firstly, the published ock, hilltop, hillslope, plateau/cliff-edge, ridge
records of known hillforts and fortified settle- and inland promontory. A short qualitative de-
ments across the archipelago were georeferenced scription categorises the sites into three size 45
to create a point layer containing basic chrono- classes and describes any fortification features.
logical and research history metadata. Second- Some hillforts, mainly on the island of
ly, we conducted a systematic screening of the Lošinj, have been surveyed in recent years
RVT-enhanced DTM itself, mapping the top- (Branković and Benvin 2024). Several sites, par-
ographic signatures of preserved remains and ticularly those near Osor, were surveyed also
identifying enclosed sites that had not previous- during the autumn and winter of 2024/25.
ly been documented. Together, these steps form
the basis of the archaeological interpretation. Results
The data was collected and managed in a A total of 85 locations were systematically re-
spatial database using the desktop GIS-software viewed. Detailed terrain models revealed 35 lo-
QGIS (ver. 3.40.5) in the form of a GeoPackage. cations where simple dry stone wall enclosures
The geometries were organised within a relation- or settlement fortifications had already been
al database scheme, with an entry in the point identified as archaeological sites. As expected,
layer for each site entity (n=85), as well as relat- the terrain models provided a clearer picture of
ed line features for the archaeological interpreta- the shape, structure and complexity of these re-
tion of visible structures. Additionally, a polygon mains. However, no visible archaeological trac-
representing the maximum extent was generat- es were evident in the ALS data at the other 22 Up and Down the Hill: Hillforts and Dry Stone Wall Enclosures on the Kvarner Islands...
ed to provide general information on the size of presumed sites. Furthermore, 28 new locations
the structures. To ensure clear identification, the of enclosures and settlements were added to the
site names are based on the Croatian base map catalogue.
(HOK), although many of the hills have differ- The results are summarised in table 1. The
ent local names to those used on official maps. table compiles ALS results and publications
Additional parameters were recorded as at- by Marchesetti (1924), Mirosavljević (1955;
tributes for each entity, based on the established 1956; 1959; 1960; 1974), Stražičić (1981), Miletić
scheme of the ‘Atlas of Hillforts of Britain and (2002), Šiljeg (2006), Starac (2011), Čučković
Ireland’ (Lock and Ralston 2022). The Atlas of (2017), Ilijanić et al. (2024) and Branković and
Hillforts applied three main inclusion criteria: Benvin (2024). For ease of reading, table 1 pri-
(1) a locally dominant topographic position, (2) marily cites works by Stražičić and Čučković.
enclosing works that were sufficiently substan- The second publication contains detailed ref-
tial (e.g. multivallate ramparts or ditches with a erences, which do not need to be repeated here
width of at least ~4 m), and (3) a minimum in- (Čučković 2017, 21).
ternal area (commonly set at around 0.2 ha). This In table 1 slightly stricter criteria were used for
criteria were simplified and adapted for the use the interpretation of the dry stone wall remains