The largest prehistoric mound in Europe is the Bronze-Age Hill of Udine (Italy) and legend linked its origin to Attila the Hun

Prehistoric monuments often constitute evident landmarks and sometimes, after falling into disuse, fascinated local people enough to stimulate speculations about their origin over time. According to legend, the Hill of Udine (NE Italy) was built by Attila the Hun’s soldiers, but its origin (natural or anthropogenic) has been debated until now. Our research analyzed five new 40-m long stratigraphic cores, investigating for the first time the total thickness of the hill and compared the data with the available archaeological information. Moreover, we considered other hills and mounds in northern Italy and other European regions where folklore traditions relate their origin to Attila. The geoarchaeological and ethnographic data prove that the Hill of Udine is a Bronze Age anthropogenic mound erected between 1400 and 1150 BCE and that, later, folklore has transformed the ancestral memory of its origin into legend. By measuring 30 m in height and over 400,000 m3 in volume, the flat-topped hill is the largest prehistoric mound in Europe. This discovery reveals unprecedented skills in earth construction and confirms significant anthropogenic modifications of the environment during Bronze Age.

. Digital Terrain Model (DTM) of the study area at local and regional scale, with indication of the sites mentioned in the text. In (a) and (b) the isolated Udine Castle Hill and the depression of I Maggio Square are evident and are not related to any surface deformation, whereas in (b) the arrows highlight the ridge formed by the blind tectonic thrust of Udine-Buttrio. www.nature.com/scientificreports/ The creation of a large artificial pond in the center of villages was a widespread practice in the gravelly part of the Friulian Plain until the end of nineteenth century 24 . These structures were used for watering animals and people, because the groundwater table is largely unreachable as lying several tens of meters below the surface. In the twentieth century, the idea that a natural relief formed the core of the hill became dominant and was generally related to the likely presence of mid-Pleistocene morainic deposits or older fluvial conglomerates 25 . Moreover, in recent decades, the hypothesis that the hill was formed as a result of tectonic processes become widely accepted, on the grounds of comparisons with some tectonic isolated reliefs located south of Udine [ Fig. 1b and c; 13,26 ]. Thus, the idea that the inner sector of the hill was largely natural was held to be one of the main reasons for the origin of the pre-Roman fortified settlement of Udine 27,28 .

Results
Stratigraphic and archaeological data. The first written version of the legend of the construction of the Udine Castle Hill by Attila the Hun's soldiers back to the twelfth century, when it was recorded by the chronicler Godfrey of Viterbo and Otto, bishop of Freising 11 . Interestingly, both these sources cast doubt on Attila the Hun story, presenting the alternative hypothesis that, instead, the hill was constructed by Julius Cesar. Up to modern times, several historians and chroniclers proposed their own explanation about the formation of the hill, sustaining its natural, fully-artificial or mixed origin, without providing any field data (see Supplementary information). Today the top of the hill features an esplanade covering about 12,000 m 2 (Figs. 2, 3) and its immediate subsoil is occupied by several water tanks, up to 6 m deep (Fig. 2). These facilities were originally built in the first half of the twentieth century and caused the exposure of large sections that were described by Feruglio 12,25 as documenting artificial landfills of unknown age. Despite widespread groundworks, remains of ancient structures were extensively documented in 1986-87 in the eastern sector of the hilltop, where a large excavation investigated over 500 m 2 (Fig. 4). A number of Medieval walls and tombs and some foundation walls of Roman buildings were found below 1-2 m of Modern Age fill. Similar archaeological features were also found in the basement of the palace known as "the Castle" between 1986 and 1989 28 .
A significant find was a pit of ca. 40 m 2 , named "fossa bronzo", containing thousands of potsherds of different cultural phases, which have been recently studied in detail (Figs. 2,3,4 and 5). The pit can be interpreted as an erosive gully filled with an organic-rich silty matrix and potsherds. The oldest finds date to the transition between MBA to RBA 29 . Anyhow, according to the typo-chronology assemblages, the findings span from Bronze Age to Early Iron Age (ca. 1350-600 BCE), with most fragments dating to RBA and FBA [ Fig. 5 and Supplementary material, Fig. S5; 29 ]. In 2021 a trial trench of 12 × 8 m was opened near the 1987 excavation, up to a depth of 3 m (Fig. 4, Supplementary Fig. S3). Floors, walls and two tombs dating between the fifth and eighth centuries CE were found just below the present ground surface. As visible in Fig. 4a and Supplementary Fig. S3, these features partly reworked and lay on a sequence of silty gravels, gravels and reddish pedogenized clays of 5-15 cm in thickness. Moreover, traces of a vertical buttress were documented, most likely originally made of wood and Map of archaeological structures according to Buora 28 . The new stratigraphic cores drilled from the hilltop are indicated in yellow, while the blue dots are the geotechnical cores carried out in 1976 to restore the castle palace after the severe earthquakes that occurred that year. The red star indicates the point of the tunnel A, where a wooden tool had been found in 1943, and we radiocarbon dated it. The map was drawn with software Adobe Illustrator (www. adobe. com). www.nature.com/scientificreports/ now decomposed ( Fig. 6a and Supplementary Fig. S3b). The reddish fine sediments are comparable to the neoforming clays accumulating in the Bt horizons of the surrounding alluvial plain 13 . Core CAST-2 was drilled at the eastern boundary of the 2021 excavation (see Supplementary Data S1 for details). Below about 6 m of reworked historical debris, the core detected repeated alternations of clays and silty gravels, comparable to those documented in the 2021 excavation. Fragments of millimetric charcoals, found in rare slightly organic horizons along the sequence of alternations, have been radiocarbon dated (Fig. 6, Table 1). The obtained ages range between 4330 and 1770 years BCE (calibrated to 2σ; Table 1) , but they are in reverse chronological order. The alternation of lenses stops at a depth of 30.8 m (109.9 m asl), where an organic-rich soil was found and two fragments of charcoal collected at depths of 30.90 and 31.30 m have been radiocarbon dated to 1510-1310 BCE and 1201-920 BCE, respectively. Below the buried soil, the sediments consist of silty gravels and weakly cemented conglomerates which match with the natural fluvioglacial sediments forming the subsoil of the city center 13 . Paleobotanical investigations carried out on pollen grains extracted between 30.90 and 31.90 m of depth in core CAST-1 documented a fairly stable vegetation, with the dominance of hazel and alder, with secondary percentage of deciduous oak ( Supplementary Fig. S7). On the western side, cores CAST-1 and CAST-3 reached 20 and 25 m in depth, respectively (Supplementary Figs. S1, S6) and both cores encountered a sequence that is comparable with the lithological alternation described in CAST-2 up to 30.8 m. In CAST-3 the base of this unit was found at 21.3 m-depth It sealed slightly organic silty clays with wood remains dated to 1427-1292 BCE and a pottery fragment at 21.87 m-depth, detected just above a large wood fragment dated to 1495-1310 BCE. Also in CAST-4, in the northern sector, at 31.51 m a centimetric potsherd has been found within the soil buried by alternations of silty gravels and pedogenized clays. These sediments are also documented in CAST-4, between the depths of 3.5-31.1 m and in CAST-5, near the castle palace, between 1.5 and 30 m.
During the Second World War, in 1943, four anti-aerial tunnels were dug in the eastern flank of the hill of Udine (Figs. 2, 3). The work started on the east side at the ground level of I Maggio Square and the tunnels were dug horizontally up to 30 m into the hill. Their excavation documented the clear eastern dip of large stratigraphic units, that consisted of alternations of gravelly and clayey layers 28 . No conglomerate or bedrock lithology was encountered but a wooden artifact was found in tunnel A at 25 m from the entrance, included in an organic-rich clayey matrix; we radiocarbon dated it to 1492-1283 BCE, comparable to the age of 1454-1111 BCE obtained from another wooden remain found in another tunnel (Fig. 6). As depicted in Fig. 6c, the sequence is very similar to the anthropogenic landfill documented at the top of the hill by the 2021 excavation. The work in 1943 stopped into few weeks, when the buildings on the eastern hilltop started to suffer significant damage due to the subsidence induced by the tunneling. This strongly suggests there are no lithified units within the hill. In 1943 another bunker (Tunnel Malignani in Figs. 2, 3) was dug in the western flank of the hill, starting at about 10.3 m above the ground level of the neighboring city center (section d in Fig. 6). It was dug 8.5 m horizontally into the hill and, also in this case, the stratigraphic sequence was characterized by inclined successions of gravels and clayey lenses 27 .
At the south-western foot of the hill, near Palazzo Dorta, an archaeological excavation in 2020 uncovered the ground floor of a hut in use up to 1300-1200 BCE 30 , which overlays the original natural topography ( Fig. 6b and Supplementary Fig. S4b). Thus, the pre-existing surface lies at 110 m asl and continues almost horizontally westward. It was exposed until the second part of the RBA, but it is now covered by 4 m of later deposits that are marked at the base by a gravelly colluvium which included Iron Age potsherds and likely originated from the hill 30 . The natural soil is characterized by well-developed Bt horizons, evolved in drained conditions over the  Conglomeratic layers have been widely documented at Udine including in I Maggio Square below 2-6 m depth 13 and constitute impermeable layers that allow the occurrence of superficial perched water tables, fed by the local rainfall. This stratigraphic setting is likely responsible for the stagnation of water in I Maggio Square and the historical presence of a lacustrine environment (Fig. 1a).

Discussion
Characteristics and chronology of the mound. The available data demonstrate that the Udine Castle Hill is comprised of loose sediments, consisting of repeated alternations of silty gravels and pedogenized reddish clays for almost its entire volume. These sediments are comparable to those forming the natural deposits and soils existing in the surrounding plain, but their layering within the hill does not correspond to a natural pattern. In particular, throughout the cores and excavations, the reddish clays have been often found out of any pedogenetic sequence. Moreover, the radiocarbon samples from the stratigraphic sequence forming the hill turned out to be chronologically reversed, suggesting that the sediments had been reworked from older deposits. These data demonstrate the artificial origin of the hill, created through the accumulation of sediments that had been likely collected in the surrounding area.
The anthropogenic mound is almost 30 m high and its construction buried the previous topography, located at an elevation between 110-111 m asl. The sealed surface is in continuity with the natural soil on which most of the city is built and corresponds to the top of the LGM fluvioglacial sediments, prone to pedogenesis since www.nature.com/scientificreports/ about 18,000 BCE. In the soil sealed by the mound, ancient human activity is clearly documented by prehistoric artefacts found in CAST-3, CAST-4 and in wartime tunnel A, and by the disturbance on the original forest cover, as inferred through paleobotanic analyses. The radiocarbon dates available for the buried soil testify that the original surface was exposed until the end of the MBA and it is likely that earlier significant archaeological structures existed somewhere at the base of the mound. In fact, inside the area of castelliere the first significant archaeological documentation dates back to MBA 23,30-32 .
The finding of prehistoric structures in the 1987 and 2021 excavations on the hilltop, as well as the lack of evidence for brick structures during the construction of the modern water tanks, indicates that the mound had already achieved a shape somewhat similar to its current form in prehistoric times. In particular, the eastern  30 . The RBA deposit overlays the original natural topography; (c) 1943 air-raid shelter tunnels, according to data in Someda De Marco, where a wooden tool was found 27 and radiocarbon dated to RBA; the section in the upper box shows the stratigraphic sequence characterized by inclined successions of gravels and clayey lenses; (d) tunnel "Malignani" dug in 1943 into the western flank of the hill, starting at about 10.3 m above the ground level of the neighboring city center. It was dug 8.5 m horizontally into the hill and, in this case too, the stratigraphic sequence was characterized by inclined successions of gravels and clayey lenses 27 ; a detailed section is not available (cf. 27 ), see the text for details; (e) Archaeological excavation in Via Mercatovecchio; the detailed section is not displayed in this figure, see the text for details 23 www.nature.com/scientificreports/ portion already had its present morphology around 1350-1300 BCE. All the radiocarbon dates available for assessing the age terminus post-quem for the beginning of the mound erection are almost coincident, showing a median result slightly after 1400 BCE ( Table 1). The only exception is represented by sample 31.30 m in core CAST-2, that is 1201-920 BCE. Its reliability is weakened by the occurrence of an age of 1510-1310 from a charcoal sampled 40 cm above. The technique of constructing earthworks by alternating gravelly layers and clay lenses strongly recalls that adopted during the Bronze Age in the alluvial plains of northern Italy to build both the tumuli and the ramparts of terramare and castellieri settlements 15,20,33 . This building method was clearly described in Udine in the perimeter rampart excavated in Palazzo Mantica ( Fig. 2 and Supplementary Fig. S8), dating to the RBA 22 .
In contrast to the tells existing in large number in Asia and Middle East, but also documented in the Danubian plains [e.g. [34][35][36] ], the Udine Castle Hill was not formed by sedimentary units consisting of overlapping anthropogenic layers deposited through time. On the contrary, according to the available chronostratigraphic data, the Udine Castle Hill was constructed slightly after 1400 BCE and before 1300 BCE. This probably corresponds to the significant expansion of the castelliere of Udine, also signaled by the reinforcement of the perimeter rampart documented at Palazzo Mantica 31,32 .
Roughly at the same time (i.e. RBA1-RBA2 not advanced), numerous settlements in the alluvial plain of northern Italy flourished or were founded 15,21,37 . This phase is clearly documented also at several sites located in the distal plain and in the coastal sector of NE Italy; however, these were definitely abandoned before FBA, while those in the inner plain, including Udine, survived this crisis [( Fig. 1C; 16,17,21,22,37 ]. In Udine the continuity of the settlement during FBA and Iron Age is documented at several locations in the city center 23,30 , as well as on the top of the mound, as testified by the findings in the pit "fossa bronzo" 31 . Although later additions could have been made, shaping its current form, our data support the hypothesis that the Udine Castle Hill already existed at the beginning of RBA and had a flat-topped surface. Neither its shape or dimensions are compatible with the Bronze Age tumuli documented in the Friuli Plain (Fig. 7) and, Table 1. Radiocarbon dates available for the mound of Udine. Some more data are reported respect the information included in this Table along    www.nature.com/scientificreports/ although buildings or other structures could have been built on its large top, there is currently no evidence for any features. In fact, the original surface has been almost entirely eroded by the later occupation of the area. At the same time, the occurrence of any natural hilly landform (e.g. hard rock formation) in the core of the mound can also be rejected, thus excluding the hypothesis of its neotectonic origin. Thus, the Castel Hill is completely artificial and was built as a monumental construction, displaying the power of the community with its size and visibility. In spite of the rather flat landscape, the pre-existing topography below the eastern sector of the mound seems to dip about 3 m towards I Maggio Square, as documented by the stratigraphy of wartime tunnels (Fig. 7c). This is likely compatible with the fluvial incision eroded by the local drainage in the post-LGM. Thus, a shallow natural depression was already present in the area, allowing the emergence of the local perched water table at the surface. This is an extraordinary situation in the gravelly sector of the Friulian Plain and may justify the interest of ancient communities for this site since remote periods. This setting is further confirmed by paleobotanic and paleopedologic data, supporting that some humid environment was previously present in the area of I Maggio Square whereas, west of it, soils were well drained.
Considering the pre-existing topography of the area, today the volume of the entire mound is about 600,000-650,000 m 3 , while the volume of the basin corresponding to I Maggio Square is assessed between 400,000 and 450,000 m 3 (see Supplementary information and Fig. S9). The mound had a sub-circular plan since its earliest phase and the occurrence of a rather flat top already in that first period is suggested by the presence of a structure on the hilltop with Bronze and Iron Age findings. This hypothesis is further supported by the occurrence of historical deposits just in the very shallow subsoil and only along the rim of the hill. The prehistoric mound may have had a maximum volume approaching 550,000 m 3 , but, even considering a cautious assessment of its dimensions and limiting the hilltop radius at 35 m, the minimum volume ranges between 400,000 and 450,000 m 3 . The fairly good correspondence between the original volume of the mound and the sediments potentially extracted from I Maggio Square support the idea that both features are anthropogenic landforms of later Bronze Age, with a connected origin, as postulated by the first modern scholars 11,25 . Moreover, it is likely that the two actions were part of a combined project, aiming at creating a water reservoir for the settlement and using the extracted material to build a monument. It is worth noting that the excavation of the basin likely occurred during the RBA, when the site of Udine underwent a significant expansion 23 . Moreover, it matches with the onset of a severe drought occurred in northern Italy, as testified in the area of the Terramare at two main excavated settlements of Santa Rosa di Poviglio and Fondo Paviani 16,37 .
Even considering the minimum volume estimate, the size of the mound is without a parallel in any other structure of the period in northern Italy (Fig. 7) and must have been the most prominent landmark in the late Archaeology and folklore. In many folktales, both in Italy and other regions of Europe, Attila the Hun is identified as "the scourge of God" and his name has become a topos for describing the destruction and, in a wider sense, the collapse of late antique society 38 . His name has often been associated with the creation of peculiar features, both natural or anthropogenic, which mark the landscape. In Udine the king of the Huns has been used to explain the formation of the huge mound. It is likely that the legend evolved within the local community, through the reworking of the ancestral memory that the hill had a distant artificial origin. Another mound named after Attila the Hun was located at the boundary of Caorle Lagoon, near Lugugnana (Fig. 1d, Supplementary Fig. S12). It was 5-m high, about 20 m in diameter and was levelled after the First World War. Despite its unknown age and function, people used to call it the "Bastione di Attila", explaining its origin as part of an overnight camp of the Huns during their raid. A "Colle d' Attila" is located also at the end of the Piave River valley, near Farra di Soligo, where two isolated hillocks rise above the surrounding plain ( Fig. 1d and Supplementary Fig. S11). Their position does not fit the geological context and their origin has been queried [e.g. 39 ] but, according to popular tradition, they are the tumuli built by the Huns for their king.
The practice of assigning the origin of ancient mounds to the Huns is also found in the plains of Poland and Lithuania, where local people used to call Bronze and Iron Age tumuli Hunnenbette, literally "Hun beds", considering them to date to that period 40 . South of Reims (France), near Châlons, where Attila fought with the Romans, the feature described by the local tradition as Attila's Camp is actually a large Roman fort, already built in the first century CE 38 . In this light, at European scale, the Huns' invasion was so traumatic to become a topos for explaining the occurrence of older archaeological artefacts whose memory had been lost.

Relevant comparisons and social implications.
Our multidisciplinary approach allowed the investigation of sedimentary bodies and ancient surfaces that are largely inaccessible with conventional archaeological excavation. In particular, as already shown in studies of the origins of other prehistoric mounds and urban contexts [42][43][44] , the detailed study of stratigraphic cores was key to distinguishing natural and anthropogenic sediments. Our data show that the size and shape of the Udine Castle Hill are unparalleled in later Bronze Age communities of Europe. This opens new perspectives on the study of labor organization and social complexity of these prehistoric communities. Beyond northern Italy, fortifications constructed with wooden gabions filled with rubble are attested also in central-eastern Europe at several settlements of the Lausitz culture 45 . It may be that this technique developed as a result of the unprecedented growth of network of trade and interconnections among the communities in the second part of the Bronze Age at European scale 42 .
Our data suggest that the prehistoric mound of Udine achieved its final shape during RBA and, according to the recent literature, in that phase the settlement area of the castelliere enlarged up to reaching over 20 ha 23 . This extent is much larger than the terramara of Santa Rosa di Poviglio, that had a maximum area of 7 ha, but is a reference for the RBA because of the detailed available data about its stratigraphy and plan map 15,16 . Differently, Udine is rather comparable to the largest RBA Terramare settlements of Northern Italy, as Case Cocconi, Case del Lago and Fondo Paviani 15,37 , whereas, is slightly larger than Frattesina, that was not a terramara, flourished in the FBA and reached 15 ha 47 [Figs. 1d and 7j; 37,41,47 ]. The reorganization of settlements that occurred at the end of the RBA in Friuli points to the resilience of this area to the general collapse of the Eastern Mediterranean societies in the twelfth century BCE 46 .
The mound of Udine is far larger and higher than the burial mounds and kurgans attested elsewhere in continental Europe from the Neolithic to the late Iron Age, such as in Brittany, Poland, Ukraine and Hungary, where the largest reach 12-19 m in height 48 . Otherwise, its dimensions are similar to the Neolithic mound of Silbury Hill in Wiltshire (England), built around 2400-2300 BCE, which has an overall volume of nearly 350,000 m 3 . As at Silbury, at Udine too the quarrying left an enormous depression in the landscape [ Fig. 7b; 49 ]. This comparison makes Udine Castle Hill the largest prehistoric mound of Europe.
Considering its flat-topped surface, the hill of Udine is a unicum that has no clear analogues in Europe, but it displays a convergence with the mounds conceived to host ceremonial and political structures on their top, best illustrated by the earthen platforms erected by the Mississippian culture of the SW United States between the tenth and sixteenth century CE. In particular, a relevant comparison can be established with the imposing quadrangular Monks Mound at Cahokia, in Illinois, where the final construction resulted in a terraced structure of over 730,000 m 3 of soil [ Fig. 7c; 6 ].
At the moment, the available archaeological data for the Udine mound are few and do not allow to make solid hypotheses about the structures and activities that characterized its hilltop during Bronze Age. However, it is likely that, besides representing the power of the community through its monumental size, the mound would have been the site of some key buildings in the life of the settlement and of its related territorial system.

Methods
Research methodologies are described in this section, with a detailed characterization for the most peculiar techniques, while for the conventional ones some more specific details are reported in the Supplementary information online. Typo-chronology of pottery finds and reference archeological periods. For the Bronze and Iron Age, the relative and absolute chronological phases and sub-phases presented in this work (see Supplementary  Table S1) follow the chronological system developed by Cardarelli 17

Topographic analysis and volume estimation of Udine mound and other sites. Topographic
information about the study area was essentially derived from the analysis of digital terrain models (DTMs) produced from LiDAR data. For this purpose, we used LiDAR data gathered from the 2006-2007 survey carried out by the Civil Protection of the Regione Autonoma Friuli Venezia Giulia. The DTM of the area was freely downloaded from the official website of the Region (https:// irdat. regio ne. fvg. it/ CTRN/ ricer ca-carto grafia/) and had a cell size of 1 m 2 and a nominal vertical accuracy of ± 0.15 m. The elevation is referred to as above sea level (asl), based on the official Italian cartographic sea level, IGM1942. Calculation of the current mound volume was derived from the altimetry of the natural ground surface detected around the Castle Hill and below it. Its identification was carried out in both archaeological excavations and stratigraphic cores, based on considerations derived from fresh information and data extracted from previous literature [e.g. 23,27 ]. With this aim, we also reviewed unpublished material regarding the archaeological excavations carried out in the city center. This included sketches, photographs, and stratigraphic plans and sections) stored in the Raptor database, that is the archive of the Ministry of Culture-Soprintendenza Archeologia Belle Arti e Paesaggio of Friuli Venezia Giulia (https:// www. raptor. benic ultur ali. it/; web access on November 2022), for the areas of Palazzo Dorta, Palazzo Contarini, Palazzo Mantica and Joppi Public Library (Figs. 2, 3 and 4 and Supplementary Figs. S1, S2).
The perimeter of the present mound was reconstructed through the analysis of topographic changes (from convex to flat) in the DTM, as indicated in Supplementary Fig. S9. At the base, the hill covers an area of about 44,500 m 2 . To obtain several estimations of the volume of sediment forming the present mound from the DTM, we subtracted the hypothetical horizontal surfaces corresponding to the base of the Castle Hill, which could be located at an elevation of 110 m, 111 m and 112 m asl. A similar calculation was made to assess the volume of the basin corresponding to I Maggio Square. In this case, the perimeter was drawn joining the points where, in the DTM, the topography changed from flat to concave.
To assess the volume of the prehistoric mound (i.e., without the historical additions detected along the rim of the hilltop), the geometry of the mound was simplified and different possible simulations were performed. On the basis of its present setting, we assumed that the mound had a round base and, as today, was symmetric also in ancient times. Nowadays, the circle inscribing the present base of the mound has a radius between 110 and 115 m. The stratigraphic data indicated that the prehistoric mound was similar to the present one. Nonetheless, we cautiously considered a radius of 100 m. In this, we took into account that the base of the Castle Hill could www.nature.com/scientificreports/ have expanded through millennia, although little evidence of colluvial material at the base has been documented to date. Considering the data gathered in the 2021 excavation on the eastern sector of the hilltop, data from the CAST-5 core and information from the construction of the water tank in the mid-twentieth century CE (Fig. 6c), we hypothesized that the hilltop was also flat during prehistoric times, as the reworked deposits and historical debris in these areas are found within 2 m of depth from the present flat surface, which was located at about 141 m asl.
In assessing the volumes and dimensions of Silbury Hill (UK) and Cahokia Monks Mound (Illinois, United States), we reviewed relevant archaeological literature and compared information with the DTMs of both areas. For Silbury Hill, we processed the DTM of the area at 50 cm resolution, freely available from https:// envir onment. data. gov. uk/ Defra DataD ownlo ad/? Mode= survey. In this case, data were acquired in 2017 with a vertical accuracy of ± 15 cm root-mean-square error (RMSE) (https:// data. gov. uk/ datas et/ 977a4 ca4-1759-4f26-baa7-b566b d7ca7 bf/ lidar-point-cloud).
For the Cahokia Monks Mound we produced the DTM of the area by downloading the original data that are freely available from https:// clear ingho use. isgs. illin ois. edu/ data/ eleva tion/ illin ois-height-moder nizat ion-ilhmp. The data were acquired in 2011 with a nominal pulse spacing (NPS) of 0.7 m. Raw LAS data were downloaded and processed to obtain a 50 cm resolution DTM.

Total viewshed analysis.
To evaluate the visibility from and to the Udine Castle Hill we performed the so-called total viewshed analysis. The analysis, performed in the software QGIS 3 through the Visibility Analysis plugin 52 , allows quantification of the extent to which each location within the area is visible from all other pixels of the grid according to a visibility index. This ranges from 0 to 1, where 1 is equal to 100% visibility and implies that a point can be seen from all of its neighbors [53][54][55] . The investigated area encompasses the piedmont strip, the upper, the lower and the coastal plain of Udine approximately between the Tagliamento and the Torre-Isonzo rivers. A 10-m DTM of the area, obtained from the re-sampling of both Italian and Slovenian LiDAR data at 50 cm resolution, was considered suitable for this purpose. Slovenian LiDAR data were downloaded at https:// gis. arso. gov. si/ evode/ profi le. aspx? id= atlas_ voda_ Lidar@ Arso& cultu re= en-US and processed following the methodology developed in other areas of the Trieste Karst and described in previous publications (e.g. 56 . The radius was set at 10, 20 and 30 km; the sampling line of sight was at 8 and the observer height was at 160 cm. The analysis considered 41 sites including 8 fortified settlements, certainly coeval to the hill of Udine and 33 definite as Bronze Age tumuli documented in a recent survey 57 . To test whether or not the position of sites was affected by a more extensive view across the landscape, a robust statistical approach, known as the Monte Carlo simulation, was adopted 58 . This test has been applied to visibility studies in archaeology since the late 1990s [59][60][61] . In our case, the Monte Carlo test was performed through the software RStudio and consisted of taking several random samples of points within the study area and comparing these with the archaeological samples of the considered sites (tumuli and castellieri). With this aim, in the first stage, we first set 'the convex hull' of all burial mounds as the window of analysis; second, we created a set of 99 samples of 41 random locations corresponding to the total number of considered sites, from which we extracted the mean of the visibility index of a 200 m buffer area. The results were then observed by visually comparing the empirical cumulative frequency distribution (ECDF) of the mean of the visibility index in each of the random sample with the sample of sites. Lastly, we obtained the p-value by ranking the mean of the slope of the 100 total samples in descending order and dividing the tumuli mean slope by the total number of means.
At a later stage, we considered the entire window of analysis to compare the visibility index for total viewshed maps at 30 km of 100 random points located on the top of the Castle Hill of Udine (observed sample) with the background population consisting of 99 samples of 100 points each, randomly located within the entire study area (the Friulian Plain).

Toponyms and folklore traditions relating to Attila the Hun and other hills in NE Italy. Besides
the Udine Castle Hill, the name of Attila the Hun is connected with several other places in NE Italy and reported in many tales, toponyms, or other folklore traditions. In particular, we selected places where a hill was present and analyzed the available sources of information. We also considered also the stories that are very well known by the local people, but are generally not described in any scientific paper, whereas they are reported only in the so-called "gray literature".