529 Arheološki vestnik 71, 2020, 529–554; DOI: https://doi.org/10.3986/AV.71.19
Interdisciplinary research of the Early Iron Age
iron production centre
Cvinger near Dolenjske Toplice (Slovenia)
Interdisciplinarne raziskave železarskega središča
Cvinger pri Dolenjskih Toplicah
iz starejše železne dobe
Matija ČREŠNAR, Branko MUŠIČ, Barbara HORN, Manca VINAZZA,
Tamara LESKOV AR, Samuel E. HARRIS, Catherine M. BATT, Nejc DOLINAR 
Izvleček 
Prazgodovinski kompleks Cvinger pri Dolenjskih Toplicah leži na strateški točki na prehodu med Dolenjsko in Belo 
krajino. Tukaj odkrite najdbe so imele pomembno vlogo pri opredelitvi mlajšega halštatskega obdobja v jugovzhodnih 
Alpah. Prav tako pomembno je odkritje železarsko-talilniškega območja.
V zadnjih letih pa je bil Cvinger predmet interdisciplinarnih raziskav, ki so povezale sodobne tehnike daljinskega 
zaznavanja, kot so zračno lasersko skeniranje in geofizikalne meritve, s tradicionalnimi arheološkimi metodami. Rezultati 
so pripeljali do novih dognanj o celotnem kompleksu, vključno z natančnejšo datacijo talilniškega območja, pridobljeno 
z metodo arheomagnetnega datiranja.
Ključne besede: Dolenjska; starejša železna doba; gradišče; železarsko-talilniško območje; talilne peči; interdiscipli-
narne raziskave; arheološka geofizika; arheomagnetno datiranje
Abstract 
The prehistoric complex of Cvinger near Dolenjske Toplice occupies a strategic position between the regions of Do-
lenjska and Bela krajina. It has yielded important finds for the understanding of the Late Hallstatt period and holds the 
largest known iron-smelting area in the region.
In recent years, several interdisciplinary research campaigns took place at Cvinger, combining modern remote sens-
ing techniques, such as airborne laser scanning and geophysical surveys, with more established archaeological methods. 
Importantly, the results have brought new understanding about the whole complex, including the chronological refine-
ment of the iron smelting area thanks to archaeomagnetic dating.
Keywords: SE Slovenia; Dolenjska region; Early Iron Age; hillfort; iron smelting area; iron smelting furnaces; inter-
disciplinary research; archaeological geophysics; archaeomagnetic dating

530 ČREŠNAR, MUŠIČ, HORN, VINAZZA, LESKOVAR, HARRIS, BATT, DOLINAR 
INTRODUCTION
The prehistoric complex of Cvinger near Do-
lenjske Toplice occupies a limestone hill between 
the modern towns of Meniška vas and Dolenjske 
Toplice in the Dolenjska region (SE Slovenia). 
It consists of a hillfort (Cvinger), three barrow 
cemeteries (Branževec, Dolgi deli and Gomivnica) 
and an iron-smelting area (Branževec). The set-
tlement holds a strategic position, which enables 
a visual control of the surrounding lowlands, 
with key routes running across this landscape. 
To the north of the complex lies the Krka river 
valley, one of the most important waterways of 
the region. At this specific area, the river changes 
its direction, from NW-SE to SW-NE, and its 
shape, from a narrow valley in the north to a 
broader plain to the east towards Novo mesto. In 
addition, the valleys of the Radešca and Sušica 
streams branch off and run towards the south, 
connecting the Dolenjska and Bela Krajina regions, 
both integral parts of the Dolenjska Early Iron 
Age (EIA) group (Fig. 1).
The history of research of the archaeological 
complex at Cvinger goes back to the end of the 19
th
 
century. It is thoroughly presented by Dular and 
Fig. 1: Cvinger near Dolenjske Toplice and its surroundings on a shaded DTM.
Sl. 1: Cvinger pri Dolenjskih Toplicah z okolico na senčenem DMR-ju.

531 Interdisciplinary research of the Early Iron Age iron production centre Cvinger near Dolenjske Toplice ...
Križ (2004); a short summary is presented here. 
The first investigator of the site was Jernej Pečnik, 
who excavated seventeen barrows in 1898 and 1899, 
one at Dolgi deli and the others at Branževec to 
the south of the hillfort (Teržan 1976, 393–413, 
Pl. 1–93; Dular, Križ 2004, 209–210). His work 
was supervised by Josef Szombathy from Vienna, 
who initiated the research of the settlement. They 
excavated 18 trial trenches in 1899, some on the 
rampart and the majority in the interior of the 
hillfort. These excavations yielded important data 
regarding the construction of the fortification and 
they present one of the first attempts of systematic 
settlement research in the region. Later, in the 
1930s, W. Schmid also excavated in the settle-
ment, but little is known about his work (Teržan 
1976, 413, Pl. 92–93; Dular, Križ 2004, 209–215). 
The finds from the barrows, as well as from the 
Schmid’ s research of the settlement, were integrally 
published. The grave finds in particular had an 
important role in refining of the EIA chronology 
for the region (Teržan 1976, 385–413, Pl. 1–93).
Significant for the current knowledge about 
the Cvinger complex is the fieldwork of Križ be-
tween the years of 1986 and 1991. He lead several 
campaigns, excavating six trial trenches on the 
hillfort and one on the adjacent iron-smelting area 
(Dular, Križ 2004). The latter was subsequently 
prospected with geophysical methods, presenting 
the first study of this kind in Slovenia (Mušič, 
Orengo 1998).
The current research shows that the first occu-
pational phase of Cvinger can be dated to the Late 
Bronze Age (Ha B). At that time, the settlement 
was already surrounded by a rampart constructed 
out of wood and soil, which was destroyed in a 
fire. After a hiatus, the hillfort was fortified by a 
dry stone wall in the Late Hallstatt period, most 
probably in the late 6
th
 century BC. It remained 
occupied until the end of the Late Hallstatt period, 
that is until the end of the 4
th
 century BC (i.e. 
the Certosa fibula and Negova helmet horizons) 
(T eržan 1976, 385–393; Dular, Križ 2004, 215–228, 
231–232).
Fig. 2: Cvinger near Dolenjske Toplice. Overview of the main part of research activities 
carried out between 2014 and 2019.
Sl. 2: Cvinger pri Dolenjskih T oplicah. Pregled večjega dela raziskav med letoma 2014 in 2019.

532 ČREŠNAR, MUŠIČ, HORN, VINAZZA, LESKOVAR, HARRIS, BATT, DOLINAR 
Recent research campaigns on Cvinger near 
Dolenjske Toplice complex started in 2014 as a 
part of the ENTRANS project and continued after 
2017 as a part of the Iron-Age-Danube project. 
In the first step of our research we employed the 
airborne laser scanning (ALS) in order to build 
the base document of the site and the broader area 
around it. The preliminary desk-based analysis of 
the ALS data revealed a range of features, showing 
that the landscape surrounding the hillfort is more 
complex than previously thought. Features identi-
fied in the ALS data study were then combined 
with field observations and refined by a large 
scale multi-method geophysical measurements 
(magnetic method, low-frequency electromag-
netic method, electrical resistivity tomography, 
and magnetic susceptibility of surface layers), 
as well as an intra-site surface collection (Fig. 2) 
(Mušič et al. 2015; Črešnar, Burja, Vinazza 2017; 
Horn, Mušič, Črešnar 2019). To provide valuable 
information on certain landscape features and/
or probable archaeologically relevant geophysical 
anomalies, a series of trial trenches were excavated 
in selected locations. As the research is still in 
progress, some of the results presented here need 
to be considered as preliminary.
THE HILLFORT
AND ITS LANDSCAPE
Although a thorough study of the site was pre-
sented before (Dular, Križ 2004), a detailed analysis 
of the ALS data produced some additional findings. 
It led us to the recognition of some previously 
unidentified or not fully understood features and 
offered a broader picture of the landscape around 
and the hillfort structures.
The hillfort has an irregular trapezoidal form, 
with the interior consisting of several settlement 
terraces, sloping gently from the centre in all direc-
tions. Its form is influenced by the karstic landscape 
with several dolines that are partly incorporated 
in the fortification. It is completely surrounded by 
the remains of an impressive 730 m long rampart, 
the ruins of which are up to approx. 11 m wide 
and have an average height difference of 0.5 m 
towards the inside and of 3 m to the outside.
1
 The 
1
  The average values were calculated from 39 profiles 
over the rampart on the DTM from all around the settle-
ment. The minimum/maximum values are 0.1/1.6 m towards 
the inside and 1.7/4.7 m to the outside.
outside of the rampart is accompanied by a shal-
low depression or levelling of the terrain (Fig. 3). 
Its course cannot be followed clearly around the 
entire perimeter of the settlement; however, it was 
assessed with different ALS data visualizations. It 
is most obvious on the south-western side of the 
settlement, where it was further investigated by 
two ERT profiles and the trial Trench 5 (Fig. 2). 
The ERT profiles showed that the upper soil lay-
ers, as well as part of the limestone ground, were 
removed/lowered in the areas around the rampart. 
The Trench 5, excavated inside this depression, 
did not yield any further conclusive information. 
After the removal of turf and humus only a layer 
of transported material eroded from the rampart 
covered the solid limestone, partly interrupted by 
natural clays. The levelling can be understood as an 
anthropogenic feature, resulting from the removal 
of building material, i.e. earth and limestone, for 
the erection of the rampart and the dry stone wall, 
and not as a defensive structure.
Two entrances, one to the south and one to 
the north, originally led into the settlement.
2
 The 
embanked approach path indicates that the south-
ern entrance was more important, although there 
is a clear holloway leading towards the northern 
entrance as well, extending into the interior of 
the settlement.
The approx. 180 m long embanked approach 
path
3
 leads to the settlement from the smelting 
area on the saddle called Branževec. It makes 
a series of slight turns and combines with two 
transverse features with a maximum width of 8.5 
m and 6.5 m respectively (see below). They seem 
to have functioned as additional reinforcement 
of the fortification, as the transverse wall closest 
to the hillfort path narrows from 4–5 m to 2.5 
m
4
 (Fig. 3).
The rampart partially extends around the northern 
side of the iron-smelting area. The latter can be 
clearly recognized by the different surface texture 
on the digital terrain models (Fig. 3). It covers an 
area of approx. 0.6 hectare. This area was detected 
by surface finds of slag and burned clay. It was first 
excavated already in the late 1980’s and surveyed 
2
  Other entrances lead into the settlement today, but 
they are of modern date.
3
  The distance here is considered a walking distance 
along the path and not air distance between two points.
4 
 The width of the path is an approximation from 12 
measurements along the path.

533 Interdisciplinary research of the Early Iron Age iron production centre Cvinger near Dolenjske Toplice ...
by geophysics a few years later (Dular, Križ 2004, 
228–230; Mušič, Orengo 1998).
From the iron-smelting area to the east, south-
east and south we can identify a broad corridor of 
commingled holloways (Fig. 3). This corridor, as 
already observed elsewhere (e.g. Mason, Mlekuž 
2016, 104–109; Mlekuž, Črešnar 2014), is connected 
to the barrows in the Branževec area, which seem 
also to border this corridor (Fig. 3). Although no 
direct evidence for the dating of the holloways 
exists, it is very probable that they originate in 
the time of all the other identified features, i.e. 
in the Early Iron Age.
Additionally, a basic landscape study of the 
three barrow cemeteries, known after their location 
names as Branževec, Dolgi deli and Gomivnica 
was conducted. The largest barrow cemetery at 
Branževec was researched in 1898–1899 and al-
together 16 barrows were excavated (Teržan 1976, 
393–413, Pl. 1–93; Dular, Križ 2004, 209–210). 
The grave finds were later used for the previously 
mentioned modification of the Late Hallstatt period 
chronology of the Dolenjska region (Teržan 1976, 
391–393). Locations of 26 barrows were published
5
 
(Dular, Križ 2004, 209–210); however, the ALS 
study combined with field observations indicates 
that there might be more possible barrows in the 
area (Fig. 3).
The location of the barrow excavated by Pečnik 
at Dolgi deli, north of the hillfort, is only known 
by the plot number
6
 (Teržan 1976, 395–396, Pl. 
1–3; Dular, Križ 2004, 210). With the analysis of 
the ALS derived DTM we have tried to estimate the 
area of the barrow more accurately. As a natural 
dolina dominates the southern part of the plot and 
barrows at Cvinger seem to have generally omitted 
this type of landscape, it is most probable that the 
barrow was located in the central or northern part 
of the plot, where the terrain is more even. Also, 
the surface there shows signs of mayor modern 
interventions, which might have completely erased 
all the remains of a former barrow (Fig. 1).
A possible cemetery at Gomivnica was proposed, 
based on the chance find of a bronze arm ring, 
ceramic fragments and cremated remains in 1979,
7
 
followed by field observation (Dular, Križ 2004, 
5
  Most of the barrows already published could also be 
identified on the DTM and in the field. Besides that, the 
already published dimensions are very much comparable 
with the ones obtained by the recent study.
6
  Plot. nr. 3411 k. o. Podturen.
7
  Plot. nr. 3835 k. o. Podturen.
210–211, Fig. 3). The results of the analysis of the 
DTM confirmed a possible heavily ploughed-out 
barrow in the suggested location and identified 
another barrow approx. 70 m to the north
8
 (Fig. 1).
FIELD RESEARCH BETWEEN
2017 AND 2019
Settlement
Considering the settlement, we have researched 
the karstic abyss in the centre of the settlement as 
well as the fortifications, i.e. the rampart and the 
embanked approach path.
Cvingerska jama abyss
In the central part of the settlement, there is 
a vertical karstic abyss, the so-called Cvingerska 
jama, which was mentioned for the first time 
in the 1900 and sketched out by A. Müllner in 
1909 (Pršina 2017, 107–108). The first modern 
research followed in 1982, when the abyss was 
documented by cavers and registered in the cave 
cadastre of Slovenia. Soon after, in 1986–1987, 
the first archaeological research took place, when 
approx. 10 m
3
 of material was removed from the 
area just below the entrance, but no archaeologi-
cal material was retrieved (Dular, Križ 2004, 212; 
Pršina 2017, 109–110).
The latest research campaign of 2016 and 2017 
was executed in cooperation with the Novo mesto 
caving association (Jamarski klub Novo mesto). 
The intention was to clear the continuation of 
the entrance part of the shaft of modern material 
and compare the situation with Müllner’s sketches 
from 1909, which pointed to a wide hall at the 
bottom of the shaft. Large amounts (over 20 m
3
) 
of material were transported out of the shaft. At 
a depth of approx. 10 m, we stopped the removal 
of material as the abyss splits into two shafts. One 
of them allowed penetration into the deeper parts 
of the system without removing the remaining 
debris still covering a part of the entrance shaft. 
Moving forward to the next parts of the abyss and 
exploring if humans entered it before it became the 
8
  The remains of the previously identified barrow are 
most probably stretching over two plots (plot. nr. 3834 and 
3835 k. o. Podturen) as does the newly identified barrow 
(plot. nr. 3829/2 and 3830 k. o. Podturen).

534 ČREŠNAR, MUŠIČ, HORN, VINAZZA, LESKOVAR, HARRIS, BATT, DOLINAR 
focus of the campaign in 2017. The team of cavers 
entered the newly discovered parts of the abyss 
using technical equipment. They finally reached 
a depth of over 51 m (the depth was corrected from 
53 [Pršina 2017] to 51 m: revisional measurement, 
pers. comm. M. Pršina). However, below the existing 
cap of debris, consisting of earth and fallen rock, 
at a depth of approx. 15 m, no signs of human 
presence were observed.
The work consisted of documenting the research 
inside and outside the shaft. All the material re-
moved was inspected and the vast majority of the 
finds; ceramic, bone and metal, were determined 
as modern. Only a few rounded fragments of pre-
historic pottery were found, however, all in layers 
recently washed into the shaft by water. Although 
some debris are still present in the upper part of 
the shaft, which could also include archaeological 
remains, the observations are leading us to an as-
sumption, that the entrance of the abyss was much 
smaller, completely blocked or did not even exist in 
the time of the prehistoric inhabitation. Therefore, 
the Cvingerska jama abyss most probably has no 
archaeological significance.
RAMPART AND WALL
Excavation of Trench 7
Trench 7 cut the rampart in a place recently 
damaged by a forest track. Although the excava-
tions were limited, the whole section of the rampart 
from the interior of the settlement to its foot was 
exposed (Figs. 3; 4). The remains, described below, 
do not represent intact stratigraphy, as the forest 
track partly damaged the layers all the way to the 
limestone bedrock.
The geological base in Trench 7 was formed from 
solid limestone (SE 7026) and an archaeologically 
sterile clayish layer (SE 7027). These were covered 
by several clayish layers (SE 7017, 7018, 7020 
and 7023), two of them (SE 7018 and 7020) also 
containing a moderate number of archaeological 
finds. These layers correspond well with those 
belonging to the first phase of the fortification, 
as recognized by Dular and Križ in their Trench 
1 (Dular, Križ 2004, 215–217, App. 1). The layers 
SE 7017 and 7020 seem also to have been partly 
removed and manipulated during the erection of 
the defence wall (SE 7009, 7010, 7010a), which was 
in part resting also on a clayish layer mixed with 
stones (SE 7009a). The approx. 1.5 m thick wall 
was built with an inner (SE 7009) and an outer 
(SE 7010) face, made of bigger stones, whereas 
the core of the wall was filled with smaller stone 
material (SE 7010a).
On the inner side of Trench 7, the first layer 
above the archaeologically sterile soil (SE 7027) 
was SE 7018, which was partly covered by layer 
SE 7023, the lowest layer, understood as a part 
of the settlement fortification. It was cut by a 
posthole (SE 7021/7022), which could be part of 
the fortification. We must also be aware of a thin 
clayish layer 7024 and a layer with signs of burn-
ing (SE 7019) covering it. It is possible that these 
layers are connected to the destruction of the first 
fortification, as observed in other trenches (Dular, 
Križ 2004, 217–220). It seems that the next set-
tlement layer (SE 7016) was already connected to 
the phase of the dry stone wall, as are also both 
the others (SE 7015, 7013), which are leaning 
on the lower part of its construction (SE 7009a). 
The settlement layers included typical finds such 
as baking lids, pieces of burned wall plaster and 
a spindle whorl. The uppermost layers (SE 7002, 
7008) can be understood as wall ruins.
On the outside of the wall, several sloping lay-
ers were recognized. The lower layers (SE 7012 
and 7014) represent the covering of the base of 
the dry stone wall (SE 7010). The layer SE 7003 
includes bigger stones, probably from the wall 
and can be understood as a layer of ruins of the 
fortification. It can be most probably also con-
nected to the layer SE 7028, excavated at the foot 
of the rampart, whereas the sloping layers SE 7004 
and 7006 can be attributed to the later (modern) 
destruction of the rampart by the forest activities 
inside the forest track.
The unearthed finds from the settlement layers 
did not include any typologically distinctive forms, 
which could be used for more precise dating and 
add to the interpretations already presented about 
the fortification and its chronology (Dular, Križ 
2004, 215–221, 230–232).
Electrical resistivity tomography
(ERT) of the rampart
Trench 7 was one of the first profiles with 
electrical resistivity tomography conducted on a 
prehistoric rampart in Slovenia, where comparative 
results from an ongoing excavation were available. 
In Slovenia, electrical resistivity tomography has 
been used in prehistoric studies only in the recent 

535 Interdisciplinary research of the Early Iron Age iron production centre Cvinger near Dolenjske Toplice ...
Fig. 3: Cvinger near Dolenjske T oplice with immediate cultural landscape on a shaded DTM.
Sl. 3: Cvinger pri Dolenjskih T oplicah s kulturno krajino v neposredni okolici na senčenem DMR.
years; however, the basic principles and some 
possibilities of its applications have already been 
presented (Mušič et al. 2015; Mušič et al. 2018; Horn 
et al. 2018a; 2018b; Horn, Mušič, Črešnar 2019).
In order to examine the structure of the set-
tlement rampart and to allow direct comparison 
with the profile/cross-section of above described 
Trench 7 (Fig. 4), we set a profile (ERT 6) only 
1.5 m away and parallel to the trench (Figs. 3–5). 
We used the dipole-dipole electrode array and an 
electrode spacing of 0.3 m.
Inversion resistivity model ERT 6 shows a very 
good correlation with the profile of Trench 7 (Figs. 
4; 5). Knowing the shape of the solid bedrock and 
thickness of soil changes over very short distances 
in the karst environment and given it was measured 
1.5 m away from the already open trench, we have 
to allow some minor differences in material (and 
thus also resistivity) distribution between the two.
Defence dry stone wall ruins are defined with 
high resistivity values (Fig. 5: [A, A1, A2: 200–1000 
Ωm]) with approx. 4 m in length and up to 1 m 
depth (corresponding to layers SE 7002, 7008, 
7009, 7010) and are separated from the settle-
ment layers below with slightly lower, but still 
high resistivity values (Fig. 5: [B1: 200–400Ωm]) 
of layers SE 7013, 7015, 7016. Behind the rampart 
inside the settlement two resistivity areas can be 

536 ČREŠNAR, MUŠIČ, HORN, VINAZZA, LESKOVAR, HARRIS, BATT, DOLINAR 
recognised – a higher resistivity area A3 (up to 
300 Ωm), most probably corresponding to the 
stone debris and very low resistivity area B2 (~30 
Ωm), which is a continuation of the B1 layer and 
corresponds to the wetter clayey sediment, which 
very likely contains archaeological material. This 
low resistivity anomaly continues deeper, down 
to 2 m below the surface, in the area marked D1. 
The first 2.5 m of the ERT profile length was not 
excavated, so the border between the interpreted 
archaeological layers and geological base (red 
dashed line) in this section is extrapolated from 
the known border in the profile of Trench 7 (Fig. 
4: SE 7016/7018); thus, it could also be deeper 
regarding resistivity distribution (it can continue 
to the area D1), or shallower as well. On the outer 
slope, there are high resistivity areas (A4: 40–350 
Ωm) of redeposited clayey layers mixed with de-
bris belonging to the covering of the base of the 
stone wall (Fig. 4: 7012, 7014) and the wall ruins 
(Fig. 4: SE 7003, 7006, 7028). There is also a low 
resistivity area, belonging to a redeposited clayish 
Fig. 5: Cvinger. 3D model of Trench 7 (a) in comparison with the nearby inversion resistivity 
model ERT 6 (b) (see also Figs. 2; 3; 4).
Sl. 5: Cvinger. 3D model sonde 7 (a) v primerjavi z bližnjim inverznim upornostnim
modelom ERT 6 (b) (glej tudi sl. 2; 3; 4).
Fig. 4: Cvinger. Section of Trench 7, cutting the remains of the rampart.
Sl. 4: Cvinger. Presek sonde 7, izkopane čez ruševine obzidja.

537 Interdisciplinary research of the Early Iron Age iron production centre Cvinger near Dolenjske Toplice ...
layer, probably also influenced by recent activities 
around the forest track (Fig. 4: SE 7004). A low to 
medium resistivity layer (C: 70–200 Ωm) beneath 
the defence wall ruins corresponds to archaeologi-
cally clayey sediment (Fig. 4: SE 7017, 7020) and 
archaeologically undisturbed layer (Fig. 4: 7027). 
The high resistivity anomaly (E: ~400 Ωm) reflects 
the solid limestone bedrock (Fig. 4: SE 7026). V ery 
low resistivity anomalies in geologic base (D1 and 
D2: ~30 Ωm) indicate the presence of narrow steep 
depressions in the obviously severely weathered 
limestone bedrock, filled with clayey sediments 
with higher moisture content.
EMBANKED APPROACH PATH
The embanked approach path leading to the 
hillfort’s main entrance (Figs. 1; 3) is a structure 
with no appropriate comparisons in the region. 
The only similar structure is the much shorter 
(20 m) simple linear embanked approach path 
at the entrance to the Vinkov vrh hillfort, which 
lies not far away toward the north above the Krka 
valley (Dular, Tecco Hvala 2007, 183–184, 341, 
Fig. 104; 263).
Our primary aim was to investigate the feature, 
recognized in its whole complexity only with the 
analysis of the DTM. With a trench we cut the 
second transverse feature, which closes the path 
and probably represents an important point of this 
fortification element (Fig. 3).
Excavation of Trench 6
(Figs. 6; 7)
The geological base in Trench 6 was formed from 
solid limestone (SE 6016 = 6009), which presented 
also the basis on which the dry-wall was built. Only 
under the main outer wall face we have recognized 
the foundation of the wall (SE 6021), whereas the 
rest of the wall was placed directly on the lime-
stone bedrock. The dry-wall construction, with a 
width of approx. 4 m, had an outer face (SE 6006) 
and an inner face (SE 6010), as well as additional 
intermediate lines made of bigger stones (SE 6012 
and 6015). The intermediate areas were filled with 
loose smaller stones and soil (SE 6011, 6013 and 
6019). It has to be emphasized, that only the outer 
face of the wall could be followed in several layers 
of construction, whereas the inner face and both 
of the intermediate lines were preserved only in 
one complete layer. The outer face also included 
a possible niche in its eastern part.
The earliest feature on the inner side of the wall 
was a layer with pieces of iron slag and animal 
bones (SE 6018) filling up a natural depression in 
the limestone bedrock (SE 6016 = 6009). Most of 
the excavated area behind the wall was covered by 
a succession of rather thin layers; the clayish layer 
SE 6008, including a larger amount of charcoal, 
and the layer SE 6004 above it, yielding atypical 
pottery fragments. Both layers can be understood 
as remains of former open surfaces on the inner 
side of the wall.
The excavation of Trench 6 did not yield enough 
evidence to show the reconstruction of the wall. 
The one possible niche, recognized in the outer 
face of the wall, has parallels in the Stična type 
fortifications, not however at Cvinger (Dular, Križ 
2004, 216–217). Also interesting are the interme-
diate lines of bigger stones. They were preserved 
only in one layer and have no suitable parallels; 
therefore, we cannot draw any conclusions whether 
they were only building elements influencing the 
stability of this very wide dry stone construction, 
or they also indicating its original form. Besides 
that, the amount of stone discovered at the location 
is very limited; therefore, a high stone structure is 
not very probable and a stone/wood construction 
is more likely.
IRON-SMELTING AREA
AT BRANŽEVEC
Magnetic survey
W e have applied magnetic survey (magnetometer 
and magnetic susceptibility) (Fig. 2), as it was suc-
cessfully used for identifying iron-smelting sites 
at this location and beyond (Mušič, Orengo 1998, 
157–186; Abrahamsen et al. 1998, 61–70; Powell et 
al. 2002, 651–665; Vyncke et al. 2014, 109–113). 
It is the most appropriate geophysical technique 
to identify strong magnetic anomalies generated 
by Early Iron Age furnaces with or without slag 
pits and other archaeological remains associated 
with iron-production activities, e.g. waste mate-
rial deposits. The magnetic responses of different 
iron slag forms always produce a wide range of 
magnetic anomalies. The exceptional variability 
of magnitudes and shapes of magnetic anomalies 
may be basically connected to different portions 
of differently magnetic iron mineral types. Besides 

538 ČREŠNAR, MUŠIČ, HORN, VINAZZA, LESKOVAR, HARRIS, BATT, DOLINAR 
Fig. 6: Cvinger. Plan of Trench 6, cutting across the second transverse wall of the embanked approach path to the hillfort 
(see also Fig. 2).
Sl. 6: Cvinger. Načrt sonde 6 čez drugi prečni zid utrjene pristopne poti do gradišča (glej tudi sl. 2).
Fig. 7: 3D model of Trench 6, cutting across the second transverse wall of the embanked approach path to the Cvinger 
hillfort (see also Fig. 3).
Sl. 7: 3D model sonde 6 čez drugi prečni zid nasipa za pristopno pot do gradišča Cvinger (glej tudi sl. 3).
6 7

539 Interdisciplinary research of the Early Iron Age iron production centre Cvinger near Dolenjske Toplice ...
Fig. 8: Branževec. Results on first magnetic prospection obtained by Fluxgate gradiometer Geoscan FM36 (see 
Mušič, Orengo 1998) after applying Gaussian filter in colour scale display using Histogram equalization. Dynamics: 
-62.31/+74.48 nT/m.
Sl. 8: Branževec. Rezultati prve magnetne prospekcije, pridobljeni s pretočnim gradiometrom Geoscan FM36 (glej Mu-
šič, Orengo 1998) po uporabi Gaussovega filtra v barvnem prikazu z uporabo histogramskega izenačevanja. Razpon 
prikazanih vrednosti: -62,31 /+74,48 nT/m.
the iron mineral composition, depth, dimensions 
and geometry of the iron-production remains can 
also significantly contribute to the magnitude and 
shape of anomalies. Strong magnetic anomalies from 
waste slag deposits frequently blur the magnetic 
responses of individual furnaces. Therefore, only 
the case sensitive, adapted examination can assure 
reliable identification of individual structures 
(Powell et al. 2002, 651–665).
At Branževec a geophysical survey using a 
total field magnetometer (Geometrics G-858) in 
(pseudo)gradient mode was undertaken. Magnetic 
measurements in this configuration generally am-
plifies the weak magnetic anomalies of induced 
magnetization at shallow depths in favour of 
long-wave magnetic anomalies caused mostly by 
the magnetic susceptibility differences in karstic 
limestone bedrock (Fig. 8). As this magnetometer 
consists of two separate sensors, it allows the 
observation of magnetic field readings separately 
on the top and the bottom sensors. The analyses 
of single sensor data presents a more refined ap-
proach for specific archaeological contexts such 
as iron-production sites (Tabbagh 2003, 75–81). 

540 ČREŠNAR, MUŠIČ, HORN, VINAZZA, LESKOVAR, HARRIS, BATT, DOLINAR 
Such measurement procedures are normally sup-
ported by diurnal magnetic field changes correct- 
ions using a base magnetometer. If single sensor 
measurements are acquired in a short period of 
time when the diurnal magnetic field changes are 
rather small, a base magnetometer is not neces-
sary. For smaller diurnal variation corrections, 
adequate processing steps are suggested (Tabbagh 
2003, 75–81). In our case, the diurnal variations 
were surprisingly small. Therefore, no additional 
corrections were required, and processing steps 
resembled those used for gradient mode (Fig. 9). 
The magnetometer used attained a resolution of 
0.1–0.2 nT/m in measuring the total magnetic field 
density, with an acquisition at a rate of 5 Hz along 
the 0.25–0.5 m spaced transects. The height of the 
bottom sensor above the ground was approx. 0.3 
m and the distance between vertically positioned 
sensors was 0.7 m. The readings were interpolated 
to a sample interval of 0.25 m in both directions 
using the cubic “spline approximation to the sinc 
function” (Engels, Stark, Vogt 1988, 225–236) and 
smoothed by Gaussian filter (Nixon, Aguado 2008). 
The magnetic maps were created using Histogram 
equalization adjustment (see for instance: Acharya, 
Ray 2005). In this way, the total magnetic field 
readings can be better distributed on the histogram 
to enable insight into weaker, discrete magnetic 
9 10

541 Interdisciplinary research of the Early Iron Age iron production centre Cvinger near Dolenjske Toplice ...
Fig. 9: Branževec. Results on magnetic prospection applying total field magnetometer Geometrics G-858 in gradient 
mode and single sensors without any corrections of diurnal magnetic field changes. Gradient and total magnetic field 
data after applying Gaussian filter and colour scale display using Histogram equalization. Dynamics: –268.68 /+104.84 
nT/m (gradient); 47577/47954 nT (bottom sensor); 47736/47870 nT (top sensor).
Sl. 9: Branževec. Rezultati magnetne prospekcije z uporabo magnetometra totalnega magnetnega polja Geometrics G-858 
v gradientnem načinu in s posameznimi senzorji brez popravkov dnevnih variacij Zemljinega magnetnega polja. Podatki 
o gradientnem in totalnem magnetnem polju po uporabi Gaussovega filtra v barvnem prikazu z uporabo histogram-
skega izenačevanja. Razpon prikazanih vrednosti: –268,68 /+104,84 nT/m (gradient); 47577/47954 nT (spodnji senzor); 
47736/47870 nT (zgornji senzor).
Fig. 10: Branževec. Comparison of the magnetic method results obtained by caesium magnetometer Geometrics G-858 in 
gradient mode and single sensors with the excavated remains of furnaces (Fig. 12). According to theoretical background 
on magnetic method, furnaces are located approx. half the distance between the highest positive and lowest negative 
parts of magnetic anomalies. Mean values of magnetic susceptibility: 5.54 × 10
-3
SI (furnace); 15.44 × 10
-3
SI (slag); 0.07 
× 10
-3
SI (stone); 1.14 × 10
-3
SI (archaeologically undisturbed soil).
Sl. 10: Branževec. Primerjava rezultatov magnetne metode, pridobljenih s cezijevim magnetometrom Geometrics G-858 
v gradientnem načinu in s posameznimi senzorji z izkopanimi ostanki peči (sl. 12). Glede na teoretično ozadje mag-
netne metode se peči nahajajo približno na polovici razdalje med najvišjimi pozitivnimi in najnižjimi negativnimi deli 
magnetnih anomalij. Srednje vrednosti magnetne susceptibilnosti: 5,54 × 10
-3
SI (peč); 15,44 × 10
-3
SI (žlindra); 0,07 × 
10
-3
SI (kamen); 1,14 × 10
-3
SI (arheološko intaktna tla).
anomalies. Basically, this histogram manipulation 
allows for areas of lower contrast to gain a higher 
contrast. Histogram equalization accomplishes 
this by effectively spreading out the most frequent 
intensity values (Acharya, Ray 2005). Results of a 
previous magnetic survey using Fluxgate gradiom-
eter Geoscan FM36 (Mušič, Orengo 1998, 157–186) 
also contributed additional information obtained 
from Histogram equalization. Namely, the height 
of gradiometer above the ground was only approx. 
0.1 m with a strong influence of slag fragments 
beneath the topsoil. By making relatively weaker 
magnetic responses of high frequency visible it 
becomes evident which parts of an iron-smelting 
complex are covered by slag fragments and other 
strongly magnetic waste material in relatively larger 
quantities (Fig. 8) in comparison with caesium 
magnetometer results (see for instance: Dirix et 
al. 2013, 233–247).
The comparison of the magnetometry survey 
obtained by caesium magnetometer and the ex-
cavated remains (Fig. 12) of the furnaces shows, 
in general, a good spatial correlation between 
archaeological objects and magnetic anomalies 
(Fig. 10). According to the theoretical background 
on magnetometry, furnaces are located approx. 
half the distance between the highest positive and 
lowest negative parts of magnetic anomalies (see 
e.g.: Telford, Geldart, Sheriff 1990). Even on the 
high-resolution results of gradient measurements 
(min=–117 nT/m; max=+299 nT/m; sd=35), we 
cannot clearly identify the individual furnaces, 
because the magnetic anomalies of adjacent fur-
naces at short distances overlap. In some situations, 
younger furnaces have been dug into older ones, 
but they also differ in the depths at which they 
occur. It is therefore obvious that the situation on 
the iron-smelting area is even more complex than 
it was discernible from magnetic method results. 
From these results it can be concluded that the total 
number of furnaces is much larger than the number 
estimated solely from magnetometer survey. This 
problem can be at least partly solved by applying 
3D magnetic modelling as it was described in 
the previous publication on the magnetic survey 
at Branževec (Mušič, Orengo 1998). Magnetic 
gradient and bottom sensor (min=47647 nT; 
max=48143 nT; sd=54) give quite similar results 
but single sensor results are in general clearer be-
cause they incorporate less high-frequency noise, 
which is an inherent characteristic of surveys in 
gradient mode. While gradient mode and bottom 
sensor enable high resolution and recognition of 
magnetic anomalies, generated by small clusters of 
furnaces, top sensor (min=47730 nT; max=47880 
nT; sd=24) gives insight into the areas with the 
strongest magnetic response. The latter can be the 
consequence of very well preserved furnace remains 
and/or large quantities of deposited slag (Fig. 10).
Electrical resistivity tomography survey
To be able to further investigate the type of 
archaeological sediments in the iron-smelting 
area (furnaces, bigger waste depositions or else) 
electrical resistivity tomography (ERT) was also 
employed.

542 ČREŠNAR, MUŠIČ, HORN, VINAZZA, LESKOVAR, HARRIS, BATT, DOLINAR 
For the possible recognition of individual fur-
nace remains, estimation of the total thickness 
of archaeologically disturbed layers and depth to 
the solid bedrock (limestone) we have measured 
five ERT profiles at the southern part of the iron-
smelting area as shown on Figs. 2 and 11. Accord-
ing to the previous research (Mušič, Orengo 1998; 
Dular, Križ 2004) the furnace remains discovered 
at the northern part of the iron-smelting area lay 
0.5–0.6 m below the topsoil (mixed with humus, 
iron slag and burnt clay). Their radii reached be-
tween 0.5–0.9 m and heights of 0.3–0.4 m (below 
topsoil). According to the recent excavation at the 
western part of the iron-smelting area (Fig. 12) 
furnace remains lie approx. 0.4 m below the top-
soil, are 0.6–0.7 m wide and are up to 0.4 m deep.
Considering this information a 0.5 m electrode 
spacing was applied with a 48-electrode system, 
resulting in the profiles lengths of 23.5 m and depth 
reach of approx. 4 m with the applied Wenner-
beta electrode configuration. The latter is in fact 
a reduced version of the dipole-dipole electrode 
array (e.g. Loke 2013), with better sensitivity to 
the lateral changes in subsurface resistivity, which 
Fig. 11: Branževec. Magnetic anomaly map (gradient mode) with shown positions of the ERT lines 1–5 (a), combined 
with inversion resistivity models ERT 1–5 shown in 3D view (b) and with their more detailed explanations (c).
Sl. 11: Branževec. Karta magnetnih anomalij (gradientni način) s prikazanimi položaji linij ERT 1–5 (a) v kombinaciji 
z inverznimi modeli upornosti ERT 1–5 v 3D pogledu (b) in z njihovimi natančnejšimi razlagami (c).

543 Interdisciplinary research of the Early Iron Age iron production centre Cvinger near Dolenjske Toplice ...
makes it better in resolving vertical artefacts such 
as furnaces and depressions in limestone bedrock. 
Its data acquisition time is 3–4-times faster than 
for dipole-dipole at the expense of lower resolution 
and lower depth reach. After adding topography 
information, all the pseudo sections were inverted 
using the focused inversion (Portniaguine, Zhdanov, 
1999; Zond geophysical software, 2016
9
).
We have located the five profiles (ERT 1–5) 
in the midst of the highest magnetic anomalies 
in the southern part of the iron-smelting area 
(Fig. 11a). Based on the resistivity distribution 
on inversion models ERT 1–5 (Fig. 11c) we can 
define a high resistivity topsoil layer (dashed black 
line) at an approximate depth 0.5 m. Below that 
line, multiple small oval-shaped high resistivity 
anomaly disturbances are defined (black arrows) 
9
  ZONDRES2D – Program for two-dimensional in-
terpretation of data obrained by resistivity and induced 
polarization methods. (n.d.). Zond geophysical software. 
– Saint-Petersburg 2001–2016.
down to depths up to approx. 1 m. These might 
reflect the remains of furnaces. While profiles 
ERT 1 and ERT 2 show no significant disturbances 
below this depth, on ERT 3 and ERT 4 medium 
to high resistivity anomalies are also present up 
to the depth of approx. 2 m (red arrows), which 
can indicate archaeologically disturbed layers, i.e. 
stone debris or burnt clay deposits.
Deeper, medium resistivity anomalies on ERT 5 
can be probably attributed to the unrealistic dis-
tortion of the shallower anomalies (black arrows), 
which are common with Wenner-beta electrode 
configuration. Low resistivity areas (D) can be 
interpreted as clayey sediments, and most probably 
can be considered as archaeologically undisturbed 
below 1 m depth on profiles ERT 1 and ERT 2 
and 2 m on ERT 3 and ERT 4. High to medium 
resistivity areas (E) represent the solid limestone 
bedrock on ERT 1 and weathered limestone blocks 
on ERT 2–5.
Excavation of Trench 8
The results of the recent caesium magnetometer 
study (Figs. 8; 9) were used to choose the location 
Fig. 12: Branževec. Trench 8 with the remains of furnaces 
(1–15), excavated in the southern part of the iron-smelting 
area (see Figs. 3; 8–10).
Sl. 12: Branževec. Sonda 8 z ostanki peči (1–15), izkopana 
v južnem delu železarsko-talilniškega območja (glej tudi 
sl. 3; 8–10).
Fig. 13: Branževec. Simple furnace remains without any 
remains of superstructures (F12, F13) (a); and more 
complex remains pointing to the probable use of domed 
furnaces (F1, F3) (b).
Sl. 13: Branževec. Ostanki preprostih peči brez ostankov 
nadgradnje (F12, F13) (a); in kompleksnejši ostanki, ki 
verjetno kažejo na uporabo kupolastih peči (F1, F3) (b).

544 ČREŠNAR, MUŠIČ, HORN, VINAZZA, LESKOVAR, HARRIS, BATT, DOLINAR 
of the trial Trench 8 (Figs. 1–3). It was located in 
the southern part of the iron-smelting area, ap-
prox. 70 m away from trench excavated by Križ 
in the northern part, which yielded remains of 
12 slag-pit furnaces (Dular, Križ 2004, 228–229, 
Figs. 3; 4; 36). Our basic aims were to investigate 
whether other parts of the smelting area show 
similarly dense occurrence of furnace remains, if 
they belong to the same type and if the remains 
are similarly poorly preserved. Besides that, we 
wanted to get the best possible data for their cor-
relation with the contemporary magnetometer 
survey, which was planned to be used for a more 
relevant interpretation of the whole smelting area. 
Therefore, a location was chosen that showed vari-
ous ranges of magnetic anomalies and could offer 
possible interpretations for many of the observed 
anomalies.
Although the excavated remains are not yet 
fully studied, some initial conclusions can be 
drawn. The excavation of the test trench (3 × 4 m) 
yielded remains of possibly 15 furnaces (Fig. 12: 
F1–15) and a great amounts of smelting remains, 
including limonite iron ore, roasted ore, partly 
reduced ore and various types of slag (Črešnar, 
Burja, Vinazza 2017, Fig. 4). The excavated fur-
naces can be generally divided into two types. 
The simpler ones (F4, F9–10, F12–15) were only 
lightly burned round or oval pits (Fig. 13a), filled 
with a mixture of earth, charcoal as well as pieces 
of burned clay and slag (mostly <10 cm). A more 
complex type (F1–3, F5–8, F11) presents round 
or oval furnaces, with heavily burned thick walls 
(5–10 cm) (Fig. 13b). They were filled with large 
pieces (<50 cm) of burned clay and slag as well as 
charcoal. Furthermore, the complex remains most 
probably present furnaces in which the slag was 
not trapped for the entire time of the smelting 
process but was at least partially removed. This 
assumption is supported by the strongly fired 
furnace walls, which also suggest that the furnaces 
were used multiple times.
When searching for the form of the furnaces, 
most of them lack any parts of the superstructure. 
Nevertheless, furnaces F1, F3 and F6 were preserved 
enough to interpret them as domed and not shaft 
type (cf. Cleere 1972, Fig. 11) as proposed previ-
ously. However, not all the burned clay material 
from the furnace remains have been studied to the 
detail that would allow us to draw final conclusions 
on the forms of the furnaces.
The furnaces were clustered in four distinct 
groups with an empty central space or corridor. 
Furnaces in the groups were built one after the 
other, and partly reused the existing empty space of 
the former furnace. Such a case is the group in the 
northeast where furnaces F5 and F3 were without 
a doubt cut when furnace 1 was built and the slag 
from the furnace ran into the emptied and reused 
bottom part of the deeper furnace F3 (Fig. 13b).
Fig. 14: Branževec. Sample positions (CVG 1, samples 1001–1036) from furnace base deposit (F1: SE 8024) and the clay 
lining (F1: SE 8022) of Furnace 1 (a). Sample positions (CVG 2, samples 2001–2014, 2017–2024) from furnace base 
deposit (F2: SE 8023) of Furnace 2 (b).
Sl. 14: Branževec. Položaji vzorcev (CVG 1, vzorci 1001–1036) iz spodnje plasti (F1: SE 8024) in stene peči (F1: SE 8022) 
peči 1 (a). Položaji vzorcev (CVG 2, vzorci 2001–2014, 2017–2024) iz spodnje plasti peči 2 (F2: SE 8023) (b).

545 Interdisciplinary research of the Early Iron Age iron production centre Cvinger near Dolenjske Toplice ...
ARCHAEOMAGNETIC DATING OF THE 
IRON-SMELTING FURNACES
Archaeomagnetic dating can be a suitable dat-
ing technique when iron-bearing clay material has 
been fired to above c. 400°C and has remained in 
situ. The unearthed iron-smelting furnaces exca-
vated in Trench 8 at Branževec iron-smelting area 
(Fig. 12) offered the opportunity to employ this 
dating technique. Thus, archaeomagnetic studies 
were carried out on two iron-smelting furnaces in 
Trench 8. The main reason for the analysis was the 
dating of the furnaces, as it might add important 
knowledge about the use of the iron-smelting area. 
The site is lacking in any datable pottery or metal 
finds, and the radiocarbon dating of charcoal was 
affected by the Hallstatt plateau and has therefore 
a very broad time-span (Dular, Križ 2004, fn. 40).
Archaeomagnetic dating utilises knowledge of 
past changes in the Earth’s magnetic field to date 
archaeologically fired materials such as fired clays 
and ceramics (McIntosh, Catanzariti 2006; Pavón-
Carrasco et al. 2015). The fundamental principles 
of the method rely on two phenomena. Firstly, 
the geomagnetic field changes significantly on 
archaeologically relevant timescales of decades 
and centuries, and secondly, iron-bearing oxides 
within soils and clays under certain conditions can 
record and retain a magnetic remanence, which 
reflects the past geomagnetic field from the time 
of the last firing. Archaeomagnetic dating is well-
established in Europe (Hervé, Chauvin, Lanos 
2013; Batt et al. 2017) and regularly utilised for 
dating fired archaeological structures, i.e. furnaces, 
kilns or hearths (Aidona et al. 2018; Casas et al. 
2018). Palaeosecular variation curves (PSVC) 
mainly cover country-specific areas and facilitate 
archaeomagnetic dating for those countries where 
there is enough data.
Fig. 15: Branževec. Archaeomagnetic directional data from the tube samples taken from the base deposits of the Furnace 
1 (CVG1) (a) and Furnace 2 (CVG2) (b). The full symbols represent normal inclination values and outlined symbols 
represent reversed inclination values.
Sl. 15: Branževec. Arheomagnetni podatki usmeritev remanentnega magnetnega polja iz cevastih vzorcev, odvzetih iz 
spodnjih plasti peči 1 (CVG1) (a) in peči 2 (CVG2) (b). Polni simboli predstavljajo normalne vrednosti inklinacije, 
prazni pa obratne vrednosti inklinacije.
Fig. 16: Branževec. Archaeomagnetic directional data from 
the button samples taken from the clay lining of furnace 
1 (CVG 1).
Sl. 16: Branževec. Arheomagnetni podatki usmeritev 
remanentnega magnetnega polja iz gumbastih vzorcev, 
odvzetih iz glinene obloge peči 1 (CVG 1).

546 ČREŠNAR, MUŠIČ, HORN, VINAZZA, LESKOVAR, HARRIS, BATT, DOLINAR 
There were three aims, which this analysis 
sought to investigate. Firstly, if the material was 
fired in situ and to a high enough temperature to 
acquire a thermoremanent magnetisation (TRM). 
Secondly, to ascertain if there was a difference 
between the samples taken from the base deposits 
of the furnaces and the furnace wall samples. Fi-
nally, determine if a date for the last firing event 
could be established through calibration with the 
nearest applicable PSVC.
Across the two furnaces (Figs. 12; 13b) a total 
of 58 samples were taken and orientated with a 
north seeking magnetic compass (Fig. 14). The 
first furnace (F1: sample CVG 1) had some of 
the clay lining preserved (SE 8022) which was 
sampled using the button method (10 samples). 
The remaining samples were taken from the 
base deposits of the two furnaces using the tube 
method (48 samples: 26 from SE 8024 and 22 
from SE 8023).
10
 The sampling protocols from 
Clark, Tarling, Noël (1988) and Trapanese, Batt, 
Schnepp (2008) were followed.
10
  The full archaeomagnetic analysis of these two fur-
naces can be found in: S. E. Harris, C. M. Batt, M. Črešnar, 
M. Vinazza, Dating Early Iron Age sites in the SE Alpine 
region: first archaeomagnetic data for Slovenia. Physics of 
the Earth and Planetary Interiors (in preparation).
Fig. 17: Branževec. Summary of the calibration data for the clay lining of furnace 1 (CVG 1) using the Austrian PSVC 
(Schnepp, Lanos 2006) and the Matlab programme developed by Pavón-Carrasco et al. (2011). – Top row: master secular 
variation curves for the observation site (red bold curves with red error bands) of the declination (left), inclination (right) 
and the undated archaeomagnetic direction (blue line). – Middle row: the individual probability density functions for 
the declination (left), inclination (right). The green lines indicate the 95% probability threshold. – Bottom row: map with 
locations of the research area (red circle) and of the master secular variation curve (blue star) (left); combined prob-
ability density (declination and inclination) marked with the green line of 95% probability (right) and archaeomagnetic 
dating 95% probability ranges (far right).
Sl. 17: Branževec. Povzetek kalibracijskih podatkov za steno peči 1 (CVG 1) z uporabo avstrijske krivulje preteklih 
časovnih sprememb usmeritve magnetnega polja (PSVC) (Schnepp, Lanos 2006) in programa Matlab, ki so ga razvili 
Pavón-Carrasco et al. (2011). – Zgornja vrsta: temeljna krivulja preteklih časovnih sprememb usmeritev magnetnega 
polja za območje raziskav (rdeči krepki krivulji z rdečimi pasovi, ki prikazujejo interval pričakovanih odstopanj) za 
deklinacijo (levo), inklinacijo (desno) in časovno neopredeljena usmeritev (modra črta). – Srednja vrsta: funkcija gostote 
verjetnosti za deklinacijo (levo) in inklinacijo (desno). Zeleni črti označujeta 95 % prag verjetnosti. – Spodnja vrsta: 
zemljevid s prikazom območja raziskav (rdeč krog) in lokacije temeljne krivulje preteklih časovnih sprememb usmeritev 
magnetnega polja za območje raziskav (modra zvezda) (levo), združena gostota verjetnosti za deklinacijo in inklinacijo, z 
zeleno črto, ki označuje 95 % prag verjetnosti (desno), in razponi datacij 95 % verjetnosti, pridobljeni z arheomagnetnim 
datiranjem (skrajno desno).

547 Interdisciplinary research of the Early Iron Age iron production centre Cvinger near Dolenjske Toplice ...
Analysis of all the archaeomagnetic samples 
at the University of Bradford’s Archaeomagnetic 
Dating Laboratory showed large discrepancies 
between the base deposit samples and the samples 
originating from the clay lining. Despite the base 
deposit samples containing sufficient magnetic 
minerals to record a TRM the two sample sets from 
the floor of the two furnaces exhibited large scatter 
in the data set (Fig. 15). The presence of reversed 
inclination values (open triangle symbols in Fig. 
15) is not a possible record of the geomagnetic 
field for this geographical location in the Holo-
cene. This in conjunction with the large α
95
 values 
(14.0° and 15.4° respectively which are well above 
the recommended value of 5.0° (Hervé, Chauvin, 
Lanos 2013), show that the base deposits are in 
our case not suitable and do not retain a stable 
remanence of the past geomagnetic field. The 
samples were not considered for further analysis 
due to the clear likelihood that these deposits were 
not fired sufficiently.
The archaeomagnetic samples from the clay-lining 
of the first sampled furnace (F1, sample CVG 1) 
displayed a set of data with higher precision (Fig. 
16). The archaeomagnetic analysis showed that the 
material from the clay lining had been fired to a 
high enough temperature to reset the archaeomag-
netic signal. The archaeological direction obtained 
from the clay linings is a record of the geomagnetic 
field from the last time the furnace cooled down, 
i.e. the last time it was used.
While Slovenia does not have its own calibration 
curve the nearest applicable PSVC is the Austrian 
dataset (Schnepp, Lanos 2006) and the introduc-
tion of errors by comparing over such a distance 
has shown to be minimal (Casas, Incoronato 
2007). The calibrated archaeomagnetic date range 
at 95% confidence which is most in concordance 
with the archaeological evidence is 640–370 BC 
for furnace CVG 1 (Fig. 17). The additional date 
ranges are due to the geomagnetic field displaying 
the same behaviour at more than one point in the 
past (Batt 1997).
The second furnace (F2, samples CVG2) failed 
to provide a date. The archaeomagnetic analyses 
of the base deposit material from both features 
showed large scatter in the data sets. There are 
various possible explanations for that. Firstly, it 
could suggest that the material sampled was not 
fired to sufficiently high temperatures. Secondly, 
the deposits on the base of the furnace could 
have been moved/mixed after the cooling down, 
when later furnaces were erected or might be as 
a result of post-depositional processes; it was 
noted that roots had heavily bioturbated some 
of the contexts.
CONCLUSIONS
This short overview of research at Cvinger near 
Dolenjske Toplice complex in the recent years is 
not a summary of all the results of our work, for 
several analyses are still in progress. However, it 
summarizes some of new findings. On the one 
side, the novelties are concerned with the whole 
archaeological complex and on the other side with 
the methods used and combined in the presented 
interdisciplinary approach (Fig. 2).
The first step of the research was the analysis 
of the ALS data. Therewith we gained new data 
about all parts of the Cvinger complex. Firstly, 
we could recognize a levelling/shallow depres-
sion around the settlement fortification (Figs. 1; 
3), which can be most probably understood as 
signs of material acquisition (soil, stone) for the 
erection of the rampart. Furthermore, an approx. 
180 m long embanked approach path, leading to 
the settlement from the iron-smelting area on the 
saddle called Branževec was recognized.
The observation of the terrain texture helped us 
also to delineate the perimeter of the iron-smelting 
area, which measures around 0.6 hectares. Also, 
additional knowledge about the barrow cemeteries 
was gained (Figs. 1; 3). A more accurate position 
of the previously excavated and levelled barrow 
at Dolgi deli was proposed, a new barrow was 
recognized in the fields at Gomivnica and five 
possible additional barrows were recognized at 
the main barrow group at Branževec.
Cvinger is also an important archaeological site 
for the application of multi-method geophysics 
and the development of new research approaches. 
Although other methods were used, the focus here 
lies in the magnetometry and electrical resistivity 
tomography (ERT). Two case studies presented are 
important beyond the site of Cvinger as they open 
new possibilities for researching on the one hand 
fortifications of hillforts and on the other fortified 
sites as well as iron-smelting areas.
As shown above, the ERT method based on 
resistivity distribution inside ramparts can identify 
approximate dimensions and depths of defence dry 
stoner wall ruins and other features as embank-
ments etc. (Fig. 5b). However, as also demonstrated, 
due to the absence of the electrical contrast in the 

548 ČREŠNAR, MUŠIČ, HORN, VINAZZA, LESKOVAR, HARRIS, BATT, DOLINAR 
same type of material we cannot (for now) further 
differentiate between the in-situ wall remains and 
the collapsed parts of a defence wall. Although not 
discussed further, other parts of the rampart at 
Cvinger and elsewhere were also studied and the 
ERT method is more than comparable with the 
results of previous excavations. The huge advantage 
is however, it is non-destructive, replicable and 
much less time consuming than archaeological 
excavations.
New information also came from the research 
of the iron-smelting area on the Branževec saddle. 
Geophysical survey using a total field magnetom-
eter (Geometrics G-858) (Fig. 9) was performed at 
selected parts of the previously surveyed area by 
the Fluxgate gradiometer (Geoscan FM36) (Fig. 8) 
(Mušič, Orengo 1998, 157–186). It was conducted 
in the (pseudo)gradient mode with the separately 
displayed single sensor measurements that ensured 
more refined magnetograms. The more reliable 
evidence about furnaces and clusters of furnaces 
was also ensured also by a greater distance of the 
lower sensor from the surface, as this reduced the 
effect of the slag on the surface and the shallow-
ness below it.
Magnetic gradient and bottom sensor gave 
similar results but single sensor results are in 
general clearer because of less high-frequency 
noise. Gradient and bottom sensor measurements 
enable high resolution and recognition of magnetic 
anomalies, generated by small clusters of furnaces, 
while the top sensor gives insight into the areas 
with the strongest magnetic response.
The comparison of the magnetic survey results 
obtained by caesium magnetometer and the exca-
vated furnaces (Fig. 10) shows, in general, good 
spatial correlation but almost without clearly 
identified individual furnaces because the magnetic 
anomalies of adjacent furnaces, which were often 
even built one over the other (Fig. 13), at short 
distances overlap. Therefore, the total number of 
furnaces is much larger than it can be estimated 
from magnetic method results.
We have shown that the remains of furnaces 
can also be investigated by ERT. They are mani-
fested on ERT inversion models as small (up to 1 
m deep), oval-shape, medium to high resistivity 
anomalies, below the approx. 0.5 m thick high 
resistivity surface layer (Fig. 11), which is consist-
ent with the results of archaeological excavations.
The combination of these methods is highly 
important for investigations of further probable 
iron-smelting areas, which have already been dis-
covered, but are known only by single furnaces or 
simply finds of iron-smelting waste.
Archaeological excavations of trial trenches 
at Cvinger were in all cases embedded into the 
interdisciplinary approach. Therefore their results 
are important not only for the rather limited ar-
eas of the trenches, but broader areas, researched 
with other methods (ALS, geophysics) as well as 
for verification of methodology in this interdis-
ciplinary approach.
One of the archaeological trenches (Trench 6) 
was positioned on the second transverse rampart, 
which is a part of the embanked approach path, 
recognized on the ALS data (Fig. 3). The excavation 
yielded remains of a dry stone wall with a width of 
approx. 4 m, which is unique fortification element 
in the broader region (Figs. 4–7).
Important data also comes from the Trench 
8, excavated in the iron-smelting area, following 
the results of geophysical investigations. Remains 
of 15 furnaces were unearthed with a variety of 
styles representing simple and complex stratigra-
phy, often cutting earlier furnaces (Figs. 12; 13). 
Furthermore, the complex remains most prob-
ably represent furnaces in which the slag was not 
trapped for the entire time of the smelting process 
but was at least partially removed. In addition, 
some better preserved remains point to the use 
of domed furnaces and not (only) shaft furnaces. 
That is an important new discovery, which will 
have to be investigated further and compared to 
other already researched furnace remains.
One of the furnaces (F 1) was also success-
fully sampled for archaeomagnetic dating (Figs. 
14–16), which was performed at the University of 
Bradford’s Archaeomagnetic Dating Laboratory. 
The calibrated archaeomagnetic date range at 
95% confidence, which is most in concordance 
with the archaeological evidence, is 640–370 BC 
(Fig. 17). That is important as the smelting area 
is lacking in any datable pottery or metal finds, 
and the radiocarbon dating of charcoal was af-
fected by the Hallstatt plateau and has therefore 
a very broad chronological timespan (Dular, Križ 
2004, fn. 40).
The successful archaeomagnetic study of one 
of the iron-smelting furnaces from Cvinger shows 
that it is possible to obtain a date for its last use 
and that the precision of the date range is an 
improvement on that possible from radiocarbon 
dating. Future archaeomagnetic studies within 
Slovenia, particularly in the 1st millennium BC, 
have the potential to introduce a new method to 

549 Interdisciplinary research of the Early Iron Age iron production centre Cvinger near Dolenjske Toplice ...
ABRAHAMSEN et al. 1998 = Abrahamsen, N., U. Koppelt, 
B. H. Jacobsen, T. Smekalova, O. Voss 1998, Iron Age 
slags at Snorup (Denmark): magnetic prospecting, 
modelling, reconstruction and dating. − Geologica 
Carpathica 49, 61–70.
ACHARYA, T ., A. K. RAY 2005, Image Processing: Principles 
and Applications. − Wiley-Interscience.
AIDONA et al. 2018 = Aidona, E., G. S. Polymeris, P. 
Camps, D. Kondopoulou, N. Ioannidis, K. Raptis 2018, 
Archaeomagnetic versus luminescence methods: the 
case of an Early Byzantine ceramic workshop in Thes-
saloniki, Greece. − Archaeological and Anthropological 
Sciences 10/4, 725−741.
BATT, C. M. 1997, The British Archaeomagnetic Calibra-
tion Curve: An Objective Treatment. − Archaeometry 
39/1, 153−168.
BATT et al. 2017 = Batt, C. M., M. C. Brown, S.-J. Clel-
land, M. Korte, P. Linford, Z. Outram 2017, Advances 
in archaeomagnetic dating in Britain: New data, new 
approaches and a new calibration curve. − Journal of 
Archaeological Science 85, Suppl. C, 66−82.
CASAS, L., A. INCORONATO 2007, Distribution analysis 
of errors due to relocation of geomagnetic data using 
the ‘Conversion via Pole’ (CVP) method: implications 
on archaeomagnetic data. − Geophysical Journal Inter-
national 169/2, 448−454.
CASAS et al. 2018 = Casas, L., C. Auguet, G. Cantoni, J. L. 
Vilar, N. Guasch, M. Prevosti 2018, Using archaeoma-
gnetism to improve the dating of three sites in Catalonia 
(NE Spain). − Journal of Cultural Heritage 31, 152−161.
CLARK, A. J., D. H. TARLING, M. NOËL 1988, Deve-
lopments in archaeomagnetic dating in Britain. − Journal 
of Archaeological Science 15/6, 645−667.
CLEERE, H. F . 1972, The classification of early iron smelt-
ing furnaces. – Antiquaries Journal 52/2, 8–23.
ČREŠNAR, M., J. BURJA, M. VINAZZA 2017, Nove 
arheološke raziskave na Cvingerju pri Dolenjskih 
Toplicah in njihov doprinos k poznavanju železarstva 
v jugovzhodni Sloveniji v starejši železni dobi. – Arheo 
34, 79‒93.
DIRIX ET AL. 2013 = DIRIX, K., P. MUCHEZ, P. DE-
GRYSE, B. MUŠIČ, J. POBLOME 2013, Integrating 
multi-element geochemical and magnetic survey on 
ancient Sagalassos (Southwest Turkey): anthropogenic 
versus natural anomalies. − Archaeological Prospection 
20, 233−247.
aid the dating of sites or contexts lacking other 
datable material in this region.
Although not crucial for understanding of the 
site, one of the main actors of a series of different 
myths, was the karstic abyss, called Cvingerska 
jama, located in the centre of the hillfort (Fig. 3). 
Our research showed that the entrance of the abyss 
was much smaller, completely blocked or did not 
even exist in the time of the prehistoric inhabita-
tion of the settlement. Therefore, the Cvingerska 
jama abyss most probably has no archaeological 
significance.
Although this article only presents an overview 
of the research conducted at Cvinger, it has revealed 
the potential that Cvinger and its surroundings 
still hold for future research. Having this said, we 
must not forget about other sites in its vicinity, e.g. 
Dolenje Gradišče (Fig. 1), and potential sites in the 
lowlands around Cvinger, where several prehistoric 
sites have been identified, however none of them 
was systematically studied.
11
11
  In an area of less than 2.5 km away from Cvinger 
there are six known or potential archaeological sites 
entered into the database RKD – Register of immovable 
heritage of Slovenia (RKD = Register kulturne dediščine, 
Ministrstvo za kulturo RS) [https://gisportal.gov.si/portal/
apps/webappviewer/index.html?id=df5b0c8a300145fda41
7eda6b0c2b52b].
Acknowledgements
The authors thank first and foremost the colleagues who 
were collaborating with us on the research of the Cvinger 
complex, namely Luka Pukšič, Dejan Udovč, Miha Mihelič, 
Manca Omahen Gruškovnjak, Igor Medarić, Filip Matijević, 
Petra Basar and Jernej Umek as well as Bryan Hanks with 
his team of the University of Pittsburgh (ScienceProHerit-
age project, ARRS, BI-US/18-19-088). We would like to 
thank also colleagues who have previously researched the 
Cvinger complex and shared their knowledge us, namely 
Janez Dular, Borut Križ and Philip Mason. Furthermore, we 
would like to thank regional conservators, who inspected 
our work, Tina Britovšek and Uroš Bavec from the ZVKDS, 
OE Novo mesto. Besides that do our thanks go also to 
Marko Pršina (ZVKDS) for his enthusiastic support, the 
Novo mesto Caving club and the Municipality of Dolenjske 
Toplice as well as to all the other colleagues, who partici-
pated at our work, and very importantly our students from 
the Oddelek za arheologijo, Filozofska fakulteta, Univerza 
v Ljubljani, and Erasmus students many of whom spent 
their fieldwork practicum at Cvinger. Helen McCreary is 
also thanked for measuring some of the archaeomagnetic 
samples at the University of Bradford. The authors would 
also like to thank the reviewers for their comments, which 
we feel have helped us to improve the paper.

550 ČREŠNAR, MUŠIČ, HORN, VINAZZA, LESKOVAR, HARRIS, BATT, DOLINAR 
DULAR, J., B. KRIŽ 2004, Železnodobno naselje na Cvingerju 
pri Dolenjskih Toplicah / Eisenzeitliche Siedlung auf 
dem Cvinger bei Dolenjske T oplice. – Arheološki vestnik 
55, 207–250.
DULAR, J., S. TECCO HV ALA 2007, South-Eastern Slovenia 
in the Early Iron Age. Settlement – economy –society / 
Jugovzhodna Slovenija v starejši železni dobi. Poselitev – 
gospodarstvo – družba. – Opera Instituti Archaeologici 
Sloveniae 12 [DOI: 10.3986/9789612545239].
ENGELS, W ., E. L. STARK, L. VOGT 1988, On the applica-
tion of an optimal spline sampling theorem. − Signal 
Processing 14/3, 225–236.
HERVÉ, G., A. CHAUVIN, P . LANOS 2013, Geomagnetic 
field variations in W estern Europe from 1500BC to 200AD. 
Part I: Directional secular variation curve. − Physics of 
the Earth and Planetary Interiors 218, Suppl.t C, 1−13.
HORN, B., B. MUŠIČ, M. ČREŠNAR 2019, Innovative 
Approaches for Understanding Early Iron Age Forti-
fications. Emphasize on 2D Subsurface Models in the 
Light of Electrical Resistivity Tomography. – In / V: T. 
Tkalčec, T . Sekelj Ivančan, S. Krznar, J. Belaj (eds. / ur.), 
Fortifications, defence systems, structures and features in 
the past. Proceedings of the 4th International Scientific 
Conference on Mediaeval Archaeology of the Institute of 
Archaeology Zagreb, 7th−9th June 2017 / Fortifikacije, 
obrambeni sustavi i strukture u prošlosti, 45–54, Zagreb.
HORN et al. 2018a = Horn, B., B. Mušič, I. Medarić, D. 
Mlekuž 2018, Geophysical and archaeological research 
of Eneolithic pile dwelling settlement Gornje Mostišče: 
emphasis on electrical resistivity tomography method. 
– In / V: I. Miloglav (ed. / ur.), 6th International Scien-
tific Conference Methodology & Archaeometry, Zagreb, 
6th−7th December, 31, Zagreb.
HORN et al. 2018b = Horn, B., B. Mušič, M. Črešnar, P . Basar 
2018, Geofizikalne raziskave v kraškem okolju: rezultati 
električne upornostne tomografije in nizkofrekvenčne 
elektromagnetne metode na primeru utrjene naselbine 
Gradišnica pri Dolenjem Gradišču. − Arheo 35, 7–31.
LOKE, M. H. 2013, Tutorial: 2-D and 3-D electrical ima-
ging surveys. − Geotomo Software Malaysia, July, 127 
[http://seisweb.oma.be/bibadmin/uploads/pdf/COUR-
SENOTES.pdf].
MASON, P., D. MLEKUŽ 2016, Negotiating space in the 
Early iron Age landscape of south-eastern Slovenia: the 
case of Veliki Vinji vrh. − In / V: I. Armit H. Potrebica, 
M. Črešnar, P. Mason, L. Buster (eds. / ur.), Cultural 
encounters in Iron Age Europe, Archaeolingua, Series 
Minor 38, 95‒120.
McINTOSH, G., G. CATANZARITI 2006, An Introduction to 
Archaeomagnetic Dating. − Geochronometria 25, 11−18.
MLEKUŽ, D., M. ČREŠNAR 2014, Landscape and identity 
politics of the Poštela hillfort. – In / V: S. Tecco Hvala 
(ed. / ur.), Studia Praehistorica in Honorem Janez Dular,  Opera Instituti Archaeologici Sloveniae 30, 197−213.
MUŠIČ, B., L. ORENGO 1998, Magnetometrične raziskave 
železnodobnega talilnega kompleksa na Cvingerju pri 
Meniški vasi / Magnetic Investigation of the Iron Age 
Iron-Smelting Complex at Cvinger near Meniška vas. 
− Arheološki vestnik 49, 157−186.
MUŠIČ et al. 2015 = Mušič, B., M. Vinazza, M. Črešnar, 
I. Medarić 2015, Integrirane neinvazivne raziskave in 
terensko preverjanje: izkušnje s prazgodovinskih najdišč 
severovzhodne Slovenije / Integrated non-invasive rese-
arch and ground truthing: experiences from prehistoric 
sites in North-Eastern Slovenia. − Arheo 32, 37–64.
MUŠIČ et al. 2018 = Mušič, B., M. Črešnar, I. Medarić, 
B. Horn 2018, Neinvazivne raziskave gomil, pomni-
kov starejše železne dobe pod Poštelo pri Mariboru / 
Non-invasive research of barrows, monuments of the 
Early Iron Age below Poštela near Maribor. – In / V: 
M. Črešnar, M. Vinazza (eds. / ur.), Srečanja in vplivi 
v raziskovanju bronaste in železne dobe na Slovenskem. 
Zbornik prispevkov v čast Bibi T eržan, 317–334, Ljubljana 
[https://e-knjige.ff.uni-lj.si/znanstvena-zalozba/catalog/
view/113/206/2790-1].
NIXON, M. S., S. A. AGUADO 2008
2
, Feature extraction 
and image processing. – Amsterdam (DOI: https://doi.
org/10.4312/9789610600855).
PAVÓN-CARRASCO et al. 2011 = Pavón-Carrasco, F. J., 
J. Rodríguez-González, M. L. Osete, J. M. Torta 2011, 
A Matlab tool for archaeomagnetic dating. − Journal 
of Archaeological Science 38/2, 408−419.
PAVÓN-CARRASCO et al. 2015 = Pavón-Carrasco, F. J., 
M. L. Osete, S. A. Campuzano, G. Mcintosh, F. Martín-
-Hernández 2015, Recent Developments in Archaeoma-
gnetism: The Story of the Earth›s Past Magnetic Field. 
− In / V: L. Eppelbaum (ed. / ur.), New Developments 
in Palaeomagnetism, 99−158, New York.
PORTNIAGUINE, O. N., M. S. ZHDANOV 1999, Focusing 
geophysical inversion images. − Geophysics 64, 874–887.
POWELL et al. 2002 = Powell, A. J., J. G. Mcdonnell, C. M. 
Batt, R. W . Vernon 2002, An assessment of the magne-
tic response of an iron-smelting site. − Archaeometry 
44/4, 651−665.
PRŠINA, M. 2017, Razkrita skrivnost jame na Cvingerju. 
‒ Dolenjski kras 7, 105‒117.
SCHNEPP , E., P . LANOS 2006, A preliminary secular va-
riation reference curve for archaeomagnetic dating in 
Austria. − Geophysical Journal International 166/1, 91−96.
TABBAGH, J. 2003, Total field magnetic prospection: are 
vertical gradiometer measurements preferable to single 
sensor survey? − Archaeological Prospection 10, 75−81.
TELFORD, W . M., L. P . GELDART, R. E. SHERIFF 1990
2
, 
Applied geophysics. – Cambridge.
TERŽAN, B. 1976, Certoška fibula / Die Certosafibel. – 
Arheološki vestnik 27, 317‒536.
TRAPANESE, A., C. M. BATT , E. SCHNEPP 2008, Sampling 
methods in archaeomagnetic dating. A comparison 
using case studies from W örterberg, Eisenerz and Gams 
Valley (Austria). − Physics and Chemistry of the Earth, 
Parts A/B/C 33/6−7, 414−426.
VYNCKE et al. 2014 = Vyncke, K., B. Mušič, P. Degryse, 
M. W aelkens 2014, The metal production at Düzen T epe 
(Southwest T urkey): an archaeological and archaeometric 
study. − Open Journal of Archaeometry 2/1, 109−113.

551 Interdisciplinarne raziskave železarskega središča Cvinger pri Dolenjskih Toplicah ...
V prispevku je predstavljen zgoščen pregled 
rezultatov interdisciplinarnih raziskav, izvedenih v 
zadnjih letih na arheološkem kompleksnem najdišču 
Cvinger pri Dolenjskih Toplicah. Novosti na eni 
strani dopolnjujejo dosedanja arheološka spoznanja 
o Cvingerju (Dular, Križ 2004), ki ga z naborom že 
uveljavljenih sodobnih metod lahko raziskujemo 
celoviteje tudi na neinvaziven oz. nizko invanziven 
način. Opravljene raziskave so lahko spodbuda za 
nadaljnje uvajanje interdisciplinarnega pristopa na 
kompleksnih utrjenih prazgodovinskih najdiščih 
dolenjskega krasa. Poudarek je predvsem na upo-
rabi nedestruktivnih metod, tj. zračnem laserskem 
skeniranju in različnih geofizikalnih meritvah, a 
so seveda v vseh primerih ključnega pomena za 
razumevanje rezultatov teh metod zlasti rezultati 
arheoloških izkopavanj (sl. 2). V prispevku so 
predstavljeni rezultati (arheo)magnetne datacije 
talilnih peči, izkopanih na železarsko-talilniškem 
območju na ledini Branževec pod gradiščem.
Z analizo podatkov zračnega laserskega ske-
niranja (ZLS) smo pridobili nova spoznanja o 
notranji ureditvi naselbine s terasiranjem, poteku 
obrambnih struktur in manjših pregrad ter o vr-
tačah, umetno oblikovanih izravnavah in seveda 
številnih gomilah. Pri tem kaže poudariti, da smo 
izstopajoče površinske oblike na prikazih ZLS 
v vseh primerih preverili tudi na terenu. Ob že 
znanih terasah v notranjosti naselja in okopu, ki 
v celoti obdaja naselje, smo na njegovi zunanji 
strani prepoznali izravnavo terena, ki na jugoza-
hodni strani najdišča prehaja v plitek jarek (sl. 1; 
3). Gre za nezvezno, na več mestih prekinjeno in 
različno široko negativno topografsko obliko, ki ji 
ne pripisujemo obrambne funkcije, verjetneje gre 
za posledico obsežnega odstranjevanja zemljenega 
in kamninskega materiala za gradnjo okopa in 
zidu okoli naselja.
Gradišče je nekoč imelo dva vhoda, severnega in 
južnega, vsi drugi današnji prehodi skozi okop so 
nastali v novejšem času. Glavno vlogo pripisujemo 
južnemu vhodu, ker smo pred njim v obdelanih 
podatkih ZLS prvič v celoti zaznali kompleksno, na 
obeh straneh utrjeno dohodno pot v dolžini okoli 
180 m in širini 4–5 m. Ta pot vodi od železarsko-
-talilniškega območja na Branževcu v naselje in 
na tej razdalji večkrat zavije, ob tem pa se na dveh 
mestih nanjo navezujeta prečna okopa (sl. 1; 3).
Uspelo nam je natančneje zamejiti tudi železar-
sko-talilniško območje na podlagi očitnih razlik 
v teksturi na prikazih ZLS. Zavzema okoli 0,6 ha 
veliko območje, ki je na severnem delu zamejeno 
še z nizkim nasipom.
Pregledi ZLS območij vseh treh grobišč prav tako 
prinašajo nova spoznanja (sl. 1; 3). Izkopano in 
izravnano gomilo na Dolgih delih smo poskušali 
nekoliko natančneje locirati. Na njivskih površinah 
pri Gomivnici, kjer je bila doslej znana le ena go-
mila, smo v njeni neposredni bližini prepoznali še 
eno. Na območju osrednjega gomilnega grobišča na 
Branževcu pa smo nekoliko natančneje locirali že 
veliko prej dokumentirane in opredeljene gomile 
in jim dodali še pet drugih, ki smo jih zaznali v 
okviru te raziskave (glej tudi Dular, Križ 2004, 
208–212).
Analize podatkov ZLS in ogled najdišča smo 
na razmeroma obsežnih območjih dopolnili z 
geofizikalnimi raziskavami (sl. 2). Rezultate teh 
raziskav smo upoštevali pri izboru lokacij za 
testne sonde. Raziskave na Cvingerju so tako 
primer interdisciplinarnega pristopa za utrjena 
arheološkega najdišča na dolenjskem krasu, kjer 
so geofizikalne meritve že zaradi kompleksnosti 
naravnega okolja izjemno redke. Čeprav so bile 
že prej uporabljene različne metode, je bil to-
krat poudarek na magnetni metodi in električni 
upornostni tomografiji (ERT). V prispevku pred-
stavljena primera obravnavata uporabnost teh 
metod pri ugotavljanju arheoloških ostankov in 
nista pomembna zgolj za Cvinger, temveč širše, 
saj uvajata nove rešitve za raziskave obrambnih 
struktur (sl. 4; 5) in železarskih kompleksov tudi 
drugje v podobnih okoljskih kontekstih (sl. 8–11).
Z izkopom testne sonde 7 prek glavnega okopa 
okoli naselbine (sl. 3; 4) smo preverjali primerljivost 
rezultatov izkopavanja na tem delu z razmeroma 
kompleksno stratigrafijo z rezultati, pridobljenimi 
z metodo ERT. Meritve s to metodo, ki temelji 
na porazdelitvi upornosti do določene globine, 
smo opravili na odseku, oddaljenem 1,5 m od 
Interdisciplinarne raziskave železarskega središča
Cvinger pri Dolenjskih Toplicah
iz starejše železne dobe
Povzetek

552 ČREŠNAR, MUŠIČ, HORN, VINAZZA, LESKOVAR, HARRIS, BATT, DOLINAR 
roba arheološke sonde. Prepoznali smo približne 
dimenzije in globine ruševin obrambnega zidu, 
kulturne plasti in nasutja pred obzidjem, globino 
matične geološke podlage itn. (sl. 5). Vendar pa 
za zdaj še ne moremo razlikovati med ostalinami 
zidu in situ in ruševino zidu zaradi odsotnosti 
električnega kontrasta pri isti vrsti materiala.
Z metodo ERT so bili preiskani še drugi deli 
okopa na Cvingerju in utrdbene strukture na ne-
katerih drugih najdiščih, ki pa jih na tem mestu 
nismo vključili v razpravo, a so rezultati tudi v teh 
primerih dobro primerljivi z izsledki izkopavanj 
(Horn, Mušič, Črešnar 2019). Velika prednost 
metode ERT je v tem, da je povsem nedestruk-
tivna, ponovljiva in precej manj zamudna, kot so 
arheološka izkopavanja. Ta metoda lahko v priho-
dnje močno olajša raziskave utrdbenih struktur in 
pripomore k primerni izbiri lokacij za arheološka 
izkopavanja obrambnih struktur.
Geofizikalne metode so imele pomembno vlogo 
tudi v raziskavi železarsko-talilniškega območja 
na sedlu Branževec. Na izbranih delih območja, 
ki je že bilo raziskano s pretočnim gradiometrom 
(Geoscan FM36) (Mušič, Orengo 1998, 157–186), 
so bile izvedene geofizikalne meritve z magnetome-
trom totalnega polja (Geometrics G-858) (sl. 8–10). 
Glede na zastavljene arheološke cilje so meritve, 
izvedene v (psevdo)gradientnem načinu, in ana-
lize rezultatov meritev na spodnjem in zgornjem 
senzorju dale preglednejše magnetograme (sl. 9; 
10). K bolj izpovednim podatkom o posameznih 
pečeh oz. skupinah peči je prispevala razmeroma 
velika oddaljenost spodnjega senzorja od površine 
(pribl. 30 cm), kar je zmanjšalo magnetni učinek 
žlindre na površini in plitvino pod njo. Magne-
tni gradient in meritve na spodnjem senzorju so 
dali glede opredeljevanja peči podobne rezultate, 
vendar so ti na splošno jasnejši zaradi manjšega 
deleža visokofrekvenčnega šuma meritev s po-
sameznimi senzorji blizu površja. Gradientne 
meritve in meritve na spodnjem senzorju zago-
tavljajo dobro ločljivost in s tem prepoznavanje 
magnetnih anomalij na mestih posameznih peči 
oz. majhnih skupin peči, medtem ko magnetne 
meritve na zgornjem senzorju omogočajo boljši 
pregled nad območji z najmočnejšimi magnetnimi 
anomalijami (sl. 9; 10).
Primerjava rezultatov magnetne metode, pri-
dobljenih s cezijevim magnetometrom, in izko-
panimi pečmi kaže na splošno dobro prostorsko 
korelacijo (sl. 10), vendar pa posamezne peči niso 
jasno opredeljene. Razlog je v tem, da ležijo peči 
na tem območju zelo blizu druga drugi in se tako 
magnetne anomalije sosednjih peči prekrivajo ali 
pa so mlajše peči vkopane v ostaline starejših in 
se njihovi magnetni učinki seštevajo (sl. 9; 10; 
12). Zaradi navedenih spoznanj lahko upravičeno 
zaključimo, da je skupno število peči mnogo večje, 
kot je mogoče oceniti le na podlagi na magneto-
gramih vidnih izrazitih magnetnih anomalij.
Preizkusno smo metalurško območje ob tem 
preiskali še z metodo ERT. Na inverznih modelih 
ERT-profilov smo peči prepoznali kot majhne, do 
1 m globoke ovalne anomalije s srednjo do visoko 
upornostjo pod pribl. 0,5 m debelo površinsko 
plastjo z visoko upornostjo (sl. 11). To je skladno 
z rezultati arheoloških izkopavanj.
Kombinacija obeh metod je dala pomembne 
rezultate, ki so lahko ključni za nadaljnje raziska-
ve podobnih železarsko-talilniških območij tudi 
drugje. Številna potencialna najdišča so namreč 
že bila ugotovljena, a doslej z njih večinoma 
poznamo le posamezne peči ali najdbe odpadnih 
produktov železarstva.
Arheološka izkopavanja na Cvingerju so bila v 
vseh primerih vključena v interdisciplinarni pristop 
in pomenijo del celote, kjer uporabljene metode 
dopolnjujejo druga drugo. Rezultati izkopavanj torej 
niso pomembni le za razlago arheoloških vsebin na 
izkopanih območjih, temveč pomenijo izhodišče 
za razlago veliko širšega prostora, od koder so na 
voljo zgolj rezultati neinvazivnih metod. Izsledki 
izkopavanj tako prispevajo ključne podatke za 
razlago rezultatov nedestruktivnih metod.
Arheološka sonda 6 je tako pripomogla k razja-
snitvi dela utrjene pristopne poti, prepoznane na 
prikazih ZLS (sl. 3). Umeščena je bila na enega od 
prečnih okopov. Izkop je razkril ostanke približno 
4 m širokega kamnitega zidu. Čeprav je bil slabše 
ohranjen, je bilo mogoče jasno prepoznati njegovo 
zunanje in notranje lice, grajeno iz večjih kamnov, 
ob tem pa sta bili odkriti še dve liniji večjih ka-
mnov v notranjosti zidu. Vmesni prostori so bili 
zapolnjeni z manjšimi kamni (sl. 6; 7).
Mesto arheološke sonde 8 je bilo na železarsko-
talilniškem območju izbrano na podlagi analize 
rezultatov geofizikalnih meritev. Z njo smo poskušali 
zajeti karseda različne oblike in jakosti magnetnih 
anomalij. Odkrili smo ostanke petnajstih peči, 
ki so se razlikovale po obliki in razporejenosti. 
Ene so stale bolj osamljeno, ob jami za žlindro 
pri njih tudi nismo odkrili ostankov nadgradnje. 
Druge so imele bolj zapleteno zgradbo in so bile 
pogosto vkopane druga v drugo (sl. 12; 13). Pri 
večini smo ugotovili, da ne gre za tip peči, v katerih 
je bila žlindra ves čas taljenja ujeta v jamici pod 

553 Interdisciplinarne raziskave železarskega središča Cvinger pri Dolenjskih Toplicah ...
pečjo, temveč da je vsaj delno iztekla oz. je bila 
odstranjena. Poleg tega nekateri bolje ohranjeni 
ostanki kažejo na uporabo kupolastih, in ne (le) 
jaškastih peči (sl. 13), kot se je domnevalo doslej 
(Dular, Križ 2004, 228–230).
Na eni od peči (peč 1) je bila uporabljena 
magnetna datacijska metoda, ki se je izkazala za 
uspešno (sl. 14–17). Analiza je bila izvedena v 
laboratoriju za arheomagnetno datiranje Univerze 
v Bradfordu. Umerjena datacija zadnje uporabe 
peči, ki je s 95-odstotno zanesljivostjo opredeljena 
v čas 640–370 pr. n. št., se ujema s predhodnimi 
arheološkimi dognanji (sl. 17). Uspešnost te me-
tode za datacijo je še toliko pomembnejša, ker na 
območju taljenja ni bilo kronološko indikativnih 
keramičnih ali kovinskih najdb, uporabnost 
radiokarbonskega datiranja oglja pa je zaradi t. 
i. halštatskega platoja, v katerega razpon sodi, 
omejena (Dular, Križ 2004, str. 229, op. 40). To je 
pri nas prva tovrstna raziskava, ki odpira široke 
možnosti za datiranje žgane gline v najrazličnejših 
arheoloških kontekstih (ognjišča, kurišča, ostanki 
pogorišč, različne peči, kovaška ognjišča itn.), 
zlasti če ni na voljo drugih arheoloških ostalin ali 
najdb. Osnovni pogoj za uspešno rabo te datacijske 
metode je, da so vzorčeni arheološki ostanki v 
primarni legi oz. in situ.
Illustrations: Fig. 7 (prepared by: Jernej Umek, PJP d. o. o.).
Slikovno gradivo: Sl. 7 (izdelava: Jernej Umek, PJP d. o. o.).
The research was in the years 2013–2016 conducted in the framework of the ENTRANS (Encounters and transformations 
in Iron Age Europe) project, which was led by Ian Armit. The project has received funding from the European Union’s 
Seventh Framework Programme for research, technological development and demonstration under grant agreement no 
291827. The project was financially supported by the HERA Joint Research Programme (www.heranet.info) which is co-
funded by AHRC, AKA, BMBF via PT-DLR, DASTI, ETAG, FCT, FNR, FNRS, FWF, FWO, HAZU, IRC, LMT, MHEST, 
NWO, NCN, RANNÍS, RCN, VR and The European Community FP7 2007–2013, under the Socio-economic Sciences 
and Humanities programme. In the following years 2017–2019 the research and other activities were conducted in the 
framework of the Iron-Age-Danube project, led by Marko Mele from the Universalmuseum Joanneum Graz. The project 
was co-financed by the Interreg Danube Transnational Programme.
Arheološke raziskave smo izvajali tudi v sode-
lovanju z jamarji Jamarskega kluba Novo mesto, 
in sicer v kraškem breznu, poimenovanem Cvin-
gerska jama, ki leži sredi naselja (sl. 3). Čeprav 
samo brezno ni ključnega pomena za razumevanje 
najdišča, je bilo za raziskovanje zanimivo zato, ker 
se je okoli njega spletlo kar nekaj zgodb o neodkri-
tih zakladih. Spoznanje, do katerega smo prišli pa 
je, da je bil vhod v brezno v času prazgodovinske 
poselitve bodisi precej manjši ali popolnoma zaprt 
oziroma ga morda sploh ni bilo. V preiskanem 
delu brezna, ki je bilo 51 m globoko,
1
 namreč ni 
bilo arheoloških najdb (Pršina 2017).
Pričujoči članek predstavlja kratek in zgoščen 
pregled dela raziskav na Cvingerju pri Dolenjskih 
T oplicah v letih 2017–2019 in izpostavlja potencial, 
ki ga ima cvingersko gradišče s svojo okolico za 
sedanje in prihodnje raziskave. Pri tem ne kaže 
zanemariti bližnjih najdišč, kot so Dolenje Gradi-
šče tik nad reko Krko (Horn et al. 2018) in druga 
potencialna prazgodovinska najdišča v nižinskem 
delu okoli Cvingerja, od katerih še nobeno ni bilo 
sistematično raziskano.
1
  Revidirana globina jame s 53 m (Pršina 2017) na 51 m 
je posledica revizije meritev; osebna komunikacija M. Pršina, 
Jamarski klub Novo mesto.

554 ČREŠNAR, MUŠIČ, HORN, VINAZZA, LESKOVAR, HARRIS, BATT, DOLINAR 
Matija Črešnar
Univerza v Ljubljani
Filozofska fakulteta
Oddelek za arheologijo
Center za interdisciplinarne raziskave v arheologiji
Aškerčeva 2
SI-1000 Ljubljana
and / in
Zavod za varstvo kulturne dediščine Slovenije
Center za preventivno arheologijo
Poljanska 40
SI-1000 Ljubljana
matija.cresnar@gmail.com 
matija.cresnar@ff.uni-lj.si
https://orcid.org/0000-0002-7856-6384
Branko Mušič
Univerza v Ljubljani
Filozofska fakulteta
Oddelek za arheologijo
Center za interdisciplinarne raziskave v arheologiji
Aškerčeva 2
SI-1000 Ljubljana
and / in
Gearh d. o. o.
Radvanjska cesta 13
SI-2000 Maribor
brankomusic1@yahoo.com
branko.music@ff.uni-lj.si
https://orcid.org/0000-0003-1713-4550
Barbara Horn
Gearh d. o. o.
Radvanjska cesta 13
SI-2000 Maribor
barbarahorn01@gmail.com
Manca Vinazza
Univerza v Ljubljani
Filozofska fakulteta
Oddelek za arheologijo
Center za interdisciplinarne raziskave v arheologiji
Aškerčeva 2
SI-1000 Ljubljana
manca.vinazza@ff.uni-lj.si
Tamara Leskovar
Univerza v Ljubljani
Filozofska fakulteta
Oddelek za arheologijo
Center za interdisciplinarne raziskave v arheologiji
Aškerčeva 2
SI-1000 Ljubljana
tamara.leskovar@ff.uni-lj.si
https://orcid.org/0000-0002-4585-4726
Samuel E. Harris
School of Archaeological and Forensic Sciences,
University of Bradford,
Bradford
West Yorkshire, BD7 1DP
United Kingdom
samh128@hotmail.com
https://orcid.org/0000-0003-3519-1487
Catherine M. Batt
School of Archaeological and Forensic Sciences,
University of Bradford,
Bradford
West Yorkshire, BD7 1DP
United Kingdom
C.M.Batt@bradford.ac.uk
https://orcid.org/0000-0001-8629-2718
Nejc Dolinar
Zavod za varstvo kulturne dediščine Slovenije
Poljanska 40
SI-1000 Ljubljana
dolinarnejc@gmail.com
https://orcid.org/0000-0001-6043-0705