508
Documenta Praehistorica XLVII (2020)
Introduction
In pre- and protohistory the spheres of influence
and functionalities of places are usually estimated
and reconstructed based on a theoretical framework
joined with spatial data. Whilst assumptions and
approaches differ, this method predominantly con-
nects a set of ideas of how economic and political
power as well as cultic centres are reflected within
the archaeological and environmental record. These
theories are usually based on the work of Georg
Simmel (1903), who argued that social structures
and power relations are projected to space. Further,
the works of Walter Christaller concerning central
places (Christaller 1933[2006]) and Eike Gringmuth-
Dallmer’s medieval centrality criteria (Gringmuth-
Are we creating our past|
Exploring theory building, geospatial statistics, and
the reconstruction of the function of fortified
Urnfield culture settlements
Chiara G. M. Girotto
Goethe University Frankfurt, Frankfurt, DE
chiara.girotto@web.de
ABSTRACT – Urnfield Culture hilltop settlements are often associated with a predominant function
in the settlement pattern. This study challenged the idea of centrality by means of density estimates
and spatial inhomogeneous explanatory statistics. Reflecting on the differences in spatial trends and
material culture, no conclusive evidence for a consolidation of power, economic, or cultic domi-
nance was observed. The dataset strongly points towards the inapplicability of commonly used para-
metric and/or homogenous spatial algorithms in archaeology. Tracer variables as well as the metho-
dological and theoretical limitations are critically reviewed and a methodological framework to
increase the reproducibility and reusability of archaeological research is proposed.
IZVLE∞EK – Vi∏inska naselja kulture ∫arnih grobi∏≠ pogosto povezujemo z vodilnim polo∫ajem v vzor-
cu poselitve. V pri≠ujo≠i ∏tudiji izpodbijamo idejo o centralnosti z uporabo ocen gostote poselitve in
prostorskih nehomogenih pojasnjevalnih statisti≠nih podatkov. Pri opazovanju razlik v prostorskih us-
meritvah in materialni kulturi nismo na∏li nobenega prepri≠ljivega dokaza o utrjevanju mo≠i ter o
gospodarski ali kultni prevladi. Podatkovna baza ka∫e na neuporabnost parametri≠nih in/ali homo-
genih prostorskih algoritmov, ki se obi≠ajno uporabljajo v arheologiji. Nudimo kriti≠en razmislek o
sledilnih spremenljivkah ter o metodolo∏kih in teoreti≠nih omejitvah, hkrati pa ponudimo metodolo∏-
ki okvir, s pomo≠jo katerega lahko pove≠amo ponovljivost in ponovno uporabnost arheolo∏kih raziskav.
KEY WORDS – social theory; methodology; central place; geospatial statistics; Urnfield culture
KLJU∞NE BESEDE – dru∫bena teorija; metodologija; centralni prostor; geoprostorski statisti≠ni
podatki; kultura ∫arnih grobi∏≠
Ali ustvarjamo svojo preteklost| Raziskovanje gradnje teorij,
geoprostorskih statisti;nih podatkov in rekonstrukcije funkcije
utrjenih naselij kulture /arnih grobi[;
DOI> 10.4312\dp.47.29
Are we creating our past| Exploring theory building, geospatial statistics, and the reconstruction of the function of fortified Urnfield ...
509
While treating measurement as a state of uncertain-
ty reduction this paper aims to investigate function-
al differences in the material record of Urnfield cul-
ture fortified hilltop settlements and lowland set-
tlements in the Central Swabian Alb by comparing
them among themselves and to other features, such
as graves or hoards, based on an eclectic approach –
both in method and theory. The reflection on simi-
larities and differences in spatial location and cat-
egorical functional groups, as well as a critical asses-
sment of the variables’ symbolism and statistical
methods, will complement the interpretation of for-
tified hill-top settlements and their function. Finally,
it will provide ideas for a framework to overcome
current problems in the application of geospatial al-
gorithms.
Material and methods
The data used was originally published by Rainer
Kreutle (2007). As he collected the majority of the
data himself during a relatively short period of time
(1986/1987; Kreutle 2007.15) a low intra-observer
error rate, especially concerning colour and temper-
ing descriptions, is to be expected. Therefore sites
should be comparable to each other and differences
do not arise based on different description styles,
terminologies or categories. However, as the land-
scape of the Swabian Alb is predominantly charac-
terized by valley systems, reaching up to 1000m asl
in the central parts, the archaeological knowledge is
influenced by local collectors and modern settlement
areas. Nonetheless, a statistical correlation between
construction activity and the discovery of sites was
not observed.
The dataset consists of 283 sites of seven types (i.e.
hilltop settlement, lowland settlement, cave finds,
graves, hoards, single finds, and objects of unknown
contexts). Although the central part of the Swabian
Alb is not favourable to agriculture, due to its climate
and lack of water resources, 27 hilltop and 37 low-
land settlements were observed in the area, produc-
ing 1023 and 432 recorded pottery units, respecti-
vely. A total of 2088 ceramic objects, including 296
units originating from 61 burial sites were included.
In general, pottery from settlement sites is coarser
and more broken than those recovered from graves.
Most vessels are from settlement sites and served as
urns, the most characteristic finds are bowls with
creased sides and beakers. In order to simplify the
data and include as many objects as possible, the pot-
tery was categorised as beakers, bowls, cups, pans,
pots, and special shapes.
Dallmer 1999) have greatly influenced the interpre-
tation of prehistoric hilltop settlements as places of
centrality (e.g., Schauer 1993.62; Stegmaier 2017.
265), dynastic seats (Winghart 1999.532), castles
(Gersbach 2006.96–97), and even of proto-urban
character (Ostermeier 2012.143). The adapted cen-
trality criteria are usually described as the presence
of fortifications, political power, specialised crafts-
manship, and religious centres (e.g., Posluschny
2010.362). There is no unchallenged set of tracers,
distinctive variables, and objective interpretative
value. However, their existence is necessary – as any-
thing in existence is somehow reflected on the sys-
tem – and therefore is theoretically measurable (Mc-
Call 1939.15; Thorndike 1918.16). Most common-
ly, the presence of political power is substituted by
the nearby presence of extraordinary rich graves,
characterised by ‘class-specific’ objects (Jockenhö-
vel 1990.224). These are associated with drinking
and hospitality, especially if made out of bronze,
along with weapons, wagons, riding accessories,
amulets, objects made of precious metal, rich orna-
mental décor, and extraordinary mortuary construc-
tions (Bockisch-Bräuer 1999; Clausing 1999; Fal-
kenstein 2005; Fischer 1997; Jockenhövel 1971;
Knöpke 2009; Kossack 1974). The connection of
elites, economic power, and specialised craftsman-
ship is assumed to be indicated by the presence of
advanced metallurgy as well as high-quality objects
(Jockenhövel 1990.227). There is no archaeological
evidence for the assumption that supposedly less pre-
stigious craftsmanship, such as pottery or textile pro-
duction, was solely conducted in lowland settlements
(Jockenhövel 1994.25). Metal hoards are often asso-
ciated with economic dominance in metal trade, but
also with religious functions (Falkenstein, Oster-
meier 2015.21; Roymans, Kortlang 1999.27–28).
The presence of a somehow stratified society in con-
trol of the aforementioned aspects has been sum-
marised by various authors (e.g., Bockisch-Bräuer
1999; Clausing 1998; Falkenstein 2005; Knöpke
2009; Kristiansen 1982; 1998; Rowlands 1998;
Sperber 1999; Tomedi 1999; Wirth 1999). Usually
it is connected to the presence of a warrior-based
elite, as the few extraordinary rich Urnfield culture
graves have a higher proportion of weapons (Knöp-
ke 2009.16). This has been named an ‘aristocratic
warrior elite’ (Kristiansen, Larsson 2005.218), class
of warlords (‘Kriegerherrenschicht’; Stary 1980.
64), and sword bearing aristocracy (‘Schwertträge-
radel’; Sperber 1999.643–644). For the latter, ter-
ritories of 4–6km were reconstructed based on the
distances of graves (Sperber 1999.629–635).
Chiara G. M. Girotto
510
Furthermore, 1089 metal objects (tools, jewellery,
weapons, indicators of metallurgy, indicators of
other craftsmanship, harnesses or wagon parts, spe-
cial objects and unidentified items) were included in
the analysis. The collection represents a typical Ur-
nfield culture assemblage. Most of the items were
recovered from graves (581) and hoards (232). The
majority consisted of knives, needles, and various
kinds of rings and bracelets, whilst tools (sickles and
axes) predominantly originate from settlements or
hoards. It is notable that most weapons, especially
spears, were single finds or parts of grave inventories.
The data is supposed to serve as an example dataset
of Urnfield culture sites, to not only allow for func-
tional reconstruction, but also a critical assessment
of the statistical algorithms.
Quadrant counts on a 10km pattern and kernel den-
sity estimates based on opti-
mised likelihood cross vali-
dation bandwidth with edge
correction were calculated to
compare the spatial intensity
distributions of the different
site types. Theoretical territo-
ries were constructed based
on Dirichlet tessellations and
Delaunay triangulations. Spa-
tial distance trends were vi-
sualised by Stienen diagrams.
Correlations of site types and
specific objects were analysed
through an inhomogeneous
L- function and evaluated
through Monte Carlo enve-
lopes (19 runs). The similari-
ty of site assemblages was
graphically compared. All sta-
tistical analyses were based
on the method and algorithm
recommendations by Adrian
Baddeley et al. (2015) and
computed in R with the rele-
vant packages (Arnold 2018;
Bivand et al. 2013a; 2013b;
2018; Bivand, Lewin-Koh
2017; Bowman et al. 2007;
Pebesma, Bivand 2005; R
Core Team 2016; Urbaneck
2013; Wickham 2009; Wick-
ham et al. 2017; 2018a;
2018b; Wickham, Henry
2018).
Results
Spatial data
The data displayed varying spatial intensities, not
only presenting regional trends but also differing
among site types (Fig. 1). The trend corresponds to
the central plateau of the Swabian Alb and its asso-
ciated passes.
Dirichlet tessellation centred on hilltop settlements
and its associated Delaunay triangulation indicate
slight hexagonal trends (Fig. 2a), and Stienen dis-
tance measures indicate preferable spheres of influ-
ence of around 6km (Fig. 2b).
The explanatory power of the association of all site
types was correlation stationary (T = 0.06, p = 0.001)
and had a constant scale of spatial interaction (T =
< 0.01, p = 0.001). This suggested weak explanatory
Fig. 1. Distribution of sites and their associated density estimates based
on a Gaussian kernel with optimised bandwidth (likelihood cross valida-
tion) with edge correction. (a) s = 6.2km, (b) s = 7.1km, (c) s = 4.2km
(d) s = 9.0km. Digital terrain model by the University of Heidelberg,
based on an SRTM dataset projected on EPSG:5683, Gauss-Kruger zone 3.
One step on the x-axis corresponds to 20km. The symbols on each map
correspond to the same site types on every other map in the paper: a tri-
angle = hilltop settlement, dot = lowland settlement, square = burial sites,
and diamond = hoards.
Are we creating our past| Exploring theory building, geospatial statistics, and the reconstruction of the function of fortified Urnfield ...
511
Archaeological data
As no significant indicators for interaction could be
generated from the spatial data, the artefact assem-
blages were analysed for their similarity. No signi-
ficant differences in the pottery firing technique
were observed (Girotto 2018). The differences in
the proportion of drinking vessels were insignifi-
cant (Fig. 4).
Discussion
An extensive discussion of all possible elements and
factors of the dataset is not only beyond the scope
of this study, it is also impossible based on the cur-
rent state of knowledge. The presented results are
a regional study and are foremost a tactile explana-
tory vehicle to highlight the problems of settlement
interpretation in prehistory. The inability to detect
why certain settlement sites are exceptional within
a certain cluster is most often based on qualitative
approaches and theoretical interpretation. A notable
exception being Oliver Nakoinz (2013a), as the lack
of statistical significance of explorative analyses re-
quires either a descriptive approach or holistic mo-
dels. However, most interpretations are based on
the a priori selection of ‘prestigious’ factors, such as
the universal perception of gold as of utmost value
(e.g., Gersbach 2006.97) or their rarity is ‘proof’ of
their special functions in society. For this study, the
main theoretical contribution to select the tradition-
al markers is provided by Albrecht Jockenhövel (Jo-
ckenhövel 1990.220): “Fortified settlements pro-
tected large and small settlement clusters. They can
be considered their economic, power and possibly
cultic centre” (author’s own translation).
Hilltop settlements as centres of protection
and power
The pattern of the hilltop settlements is reminiscent
of the passes allowing passage through the central
plateau of the Swabian Alb (Fig. 2a), thus possibly
indicating gateway functions (e.g., Nakoinz 2013b.
96). Nonetheless it should not be considered as evi-
dence for the control of trade or routes as their lo-
cation is heavily influenced by the landscape. None
of the sites offers causal archaeological evidence to
assume more than a correlation of site type and to-
pology.
This is further implied as in a only few cases were
lowland settlements observed close to hilltop set-
tlements. Nearest-neighbour distances of the hilltop
settlements (Fig. 2b) suggest a preferred distance of
approx. 6km on the Alb, however there was no cor-
Fig. 2. (a) Dirichlet tessellation (black line) with
associated Delaunay triangulation (dotted line)
of the hilltop settlements. Lowland settlements,
burial sites and hoards are shown in dark grey.
(b) Stienen diagram of the hilltop settlements
with the other site types shown in dark purple.
Urnfield culture dated fortifications highlighted
in yellow, burials with swords in green.
power, and Monte Carlo envelope validation of the
centred inhomogeneous L-function indicated evi-
dence against an inhomogeneous Poisson point pro-
cess (Fig. 3).
Chiara G. M. Girotto
512
relation with Urnfield culture
dated fortifications. Therefore
no conclusive evidence could
be gained concerning their
protective function for settle-
ment clusters.
Whilst graves occur at higher
rates within the Swabian Alb,
the location of graves of sword
bearing individuals is interest-
ing. Comparatively often such
a grave is located at the bor-
der of a Stienen circle (Fig.
2b). Whether this is a repre-
sentation of the territories of
sword bearers (‘Schwertträ-
gerterritorien’; Sperber 1999.629–635) remains que-
stionable. The additional association of elites with
wagons and harnesses cannot be supported by the
sites of the Runder Berg, Bad Urach (1) and the
Hackberg, Gomadingen (2). Both present additional
evidence for management of horses, however the
findings from the Runder Berg were recovered from
the settlement and might also be representative of
agriculture. Bronze parts of a harness were found in
a grave near the Hackberg, but the possibly associat-
ed hilltop settlement is supposed to have only had
space for a singular building (Biel 1987.78–79). Only
a necropolis near the Plettenberg (4) had evidence
that could be conclusive. A wagon and sword were
recovered from a double burial, and lowland settle-
ments as well as other graves are present in the Stie-
nen radius of the Plettenberg. However, due to the
insignificance of the explana-
tory analyses no direct causa-
lity can be deduced from the
findings. Theoretical territo-
ries should only be interpret-
ed, if at all, with great caution
as they might be a result of
random fluctuation present
in the sample or the dataset
due to its lack of large scale
excavations.
This study highlights how the
archaeological search for cen-
trality is driven very much by
current understandings of
where it should occur, and
not which underlying proces-
ses might have created it (e.g.,
Green, Perlman 1985.5; Hen-
ning, Lucianu 2000.533; Maise 1996). A terrific
example of this would be the Heuneburg, Herbertin-
gen Hundersingen (5). Based on early works (e.g.,
Kurz 2012.449–450, Gersbach, 2006.96–97) and
the data published in Kreutle (2007) a much less pro-
minent position in the Urnfield settlement structure
was proposed. However, current research attributes
almost urban characteristics to the site (Stegmaier
2017.264). This illustrates not only the importance
of large scale excavations, but also how much an a
priori idea of the characteristics of a central place
influence the interpretation.
Hilltop settlements as centres of economy and
hospitality
No causality for the location of the different site
types could be established. However, artefact analy-
Fig. 3. Centred inhomogeneous L-function and Monte Carlo envelope of
all recorded sites indicating divergence from an inhomogeneous Poisson
process. Range r in metres.
Fig. 4. Violin plot of the proportion of drinking vessels among the site
types.
Are we creating our past| Exploring theory building, geospatial statistics, and the reconstruction of the function of fortified Urnfield ...
513
sis can still offer insights with regard to their func-
tional differences. The data was characterised in its
original publication (Kreutle 2007) using the tradi-
tional approach, and therefore this paper will focus
on two specific aspects: Indications for the control of
craftsmanship and metallurgy, and the connections
of elites and hospitality.
As mentioned before the control of advanced metal-
lurgy is generally assumed to be a feature of (forti-
fied) hilltop settlements, but there is no conclusive
evidence for this. However, this idea perfectly fits in
a hierarchical and evolutionary mindset with regard
to Bronze Age societies. Metal is often considered a
driving factor of individualisation, increasing wealth,
complexity, and the consolidation of power (e.g.,
Kuijpers 2012.418). Further, it requires that peo-
ple valued metal objects more than any other kind.
In the study region no grave associated with metal-
lurgical craftsmanship has yet been discovered, how-
ever these are usually associated with slightly richer
graves than the average (Nessel 2012). In general
their heterogeneity is more likely a remnant of the
organizational and functional aspects of the related
person in a social system, and not an indicator of a
high social status.
The indicators of metallurgy are distributed among
hilltop and lowland settlements as well as hoards
(Fig. 5).
The presence of the finds suggest areas of general
increased productivity, as around the Runder Berg
(1). Interesting is the optimized bandwidth of c.
37km, which indicates that its distribution is a su-
perregional phenomenon that cannot be analysed
on the small scale of the study region. This comple-
ments archaeological ideas on metal production
during the Urnfield period, and highlights once more
the problem of insignificant point process associa-
tions.
General increased productivity and economic domi-
nance is often connected to elites participating in su-
perregional exchanges to acquire prestigious objects
and resources, like metal. However, these function-
al aspects require the theoretical base of the pres-
ence of an elite or otherwise dominant group of peo-
ple. Usually a connection of
warriors, elites, and hospital-
ity is postulated. This is based
on the idea of the organisa-
tion of power in decentralised
communities where feasting
plays a major role in the con-
solidation of alliances, as well
as in the display of power and
economic dominance (Diet-
ler 2001.77). Following the
idea of hypothesis falsifica-
tion and measurable outputs
of significant variables to this
connection, the increased pre-
sence of sword bearing indi-
viduals should also generate
higher proportions of drink-
ing vessels at hilltop sites.
However, the plot mainly dis-
tinguishes burials and settle-
ment sites (Fig. 4). The in-
creased rates in graves indi-
cate a focus on drinking dur-
ing the funerary ritual or for
the afterlife, regardless of
whether the pots or their
contents were originally con-
sidered as burial gifts. The
slightly higher portion of
Fig. 5. Density estimates of metallurgical indicators (ingots, metallurgi-
cal by-products, moulds, and associated tools) and distribution of the hill-
top settlements, lowland settlements, and hoards. Based on a Gaussian
kernel with optimized bandwidth (likelihood cross validation) with edge
correction, s = 37.6km. Digital terrain model as in Figure 1.
Chiara G. M. Girotto
514
dinking vessels at lowland settlements is insignifi-
cant, as all confidence intervals overlap. The data-
set did not strengthen the argument of hilltop set-
tlements as places of increased hospitality or feast-
ing. However, the pottery assemblages across all
contexts were very variable. Potentially slight
changes in proportion were hidden due to the high
variance. The connection of sword bearing individ-
uals and hilltop settlements could not be reinforced.
Maybe the often quoted “[...] all civilisations owe
their origin to warriors” (Keegan 1993, vi) should
be questioned as too often “elite positions are taken
as a given where their existence needs to be pro-
ven from contextual information” (Kienlin 2012.
21).
Summary
The analyses do not comprehensively support the
idea of hilltop settlements as early central places,
seats of the elites, and economic centres. Even the
presence of an elite group, due or because of the
occurrence of hilltop settlements could not be con-
clusively evidenced. The strongest connections are
sword bearing individuals at the borders of Stienen
nearest-neighbour circles which might indicate some
territorial differentiation.
A priori selection of high status indicators merely
projects modern assumptions on the society of the
Urnfield culture. Therefore, the presented maps do
not necessarily reflect aspects of a past social reality
but might also represent the underlying research
concepts. It allows the observer to critically evalu-
ate commonly used indicators and methods, as any
meaningful variable should produce a significant re-
sult. However, if the original populations’ ideas dif-
fered significantly, it is possible that the currently
tested variables simply do not have explanatory po-
wer, because they were not relevant in the past.
Implications for future research – Are we nar-
rating our past?
Statistical limitations
Aside from theoretical considerations many of to-
day’s limitations are critically linked to the archaeo-
logical geostatistical approach. A complex system,
like human interaction, social organization, and the
symbolism of places, cannot be seen as one solvable
by simple spatial statistics (e.g., Hacıgüzeller 2012.
246; Herzog 2009). Questions concerning social in-
teractions and structures are cognitive features, often
subjective, seemingly irrational, and symbolic. They
are, by their very nature, post-processual but our use
of ostensibly objective environmental and spatial
data rightfully faces the same critique Daniel Miller
and Christopher Tilley made of the processual theo-
ry school in 1984: “A belief by some in mathemati-
sation as the goal of archaeology; the attempt to
reduce past social systems to a suitable equation”
(Miller, Tilley 1984.3). Furthermore, it implies the
“reduction of the analysis of social change to the
elucidation of external factors impinging on the
social system” (Miller, Tilley 1984.3). However, any
of the classical statistical approaches is reasonable
if it can actually answer the posed research ques-
tion. Following the concept of maximum parsimo-
ny the least complex method with the least amount
of required a priori interpretation should always
be chosen. Creating a null hypothesis that can be
investigated is often difficult, as most archaeological
questions ask why certain events happen/assem-
blages occur as they desire causal explanations. How-
ever, statistical methods investigate correlations, and
causality can only be derived from interpretation or
made likely through modelling. The most prominent
difference in today’s research is the use of other sci-
entific methods and theories in a more holistic way
(Müller-Scheeßel 1998.265). If one chooses not to
model a complex system and/or non-linear dynam-
ics, the algorithms and methods, even if they are
spatial and deal with complex phenomena like in
epidemiology and ecology, cannot be simply adapt-
ed for archaeological purposes. Among the most cri-
tical differences are the ones required for a scien-
tifically acceptable research design, such as a lack
of reproducibility or a controlled environment. Most
important of all though are the questions as to
whether archaeological data can be considered a
representative sample of the original population, or
if any past dataset is truly independent (Buccellati
2017.344). Further, most algorithms require certain
assumptions, like homogeneity and stationarity. In
return they themselves impose specific behaviours
on the founding populations, like specific probabil-
ity distributions and parametric behaviour. Whilst
usually not even addressed at all in archaeological
papers, inhomogeneous functions, Bayesian proba-
bility approaches, and the validation of explanatory
analysis allow for a sounder intepretatory basis than
homogeneous estimators and Gaussian or Poisson
distributions for point processes. Overall the inho-
mogeneous counterparts of well-known methods,
like the K-function, tend to perform much better.
Even the few requirements of inhomogeneous me-
thods are often not met in archaeological applica-
tions, as illustrated in the example data. This also
holds true for the disciplines that originally devel-
Are we creating our past| Exploring theory building, geospatial statistics, and the reconstruction of the function of fortified Urnfield ...
515
oped these methods, as Jose A. F. Diniz-Filho et al.
(2007.850) neatly summarized: “[w]hen multiple
assumptions are not being met, as in the case of
virtually all geographical analyses, can a result
from any single method (whether spatial or non-
spatial) be claimed to be better? [...] If different
methods themselves are unstable and generate
conflicting results in real data, it makes no sense
to claim that any particular method is always su-
perior to any other.”
Whilst one should not believe that “anything could
be recovered from the archaeological record, if
you only searched hard enough” (Chippindale
1987.515), new approaches and methods have high-
lighted the potential of statistical algorithms and
models. Besides promising results in the field of
agent-based modelling and exploring concepts of
non-linear dynamics in archaeology, the inclusion of
mathematical chaos, sensitivity analysis and ascen-
dance (e.g., Arias-González, Morand 2006; Deman
et al. 2016) as well as the reconstruction of central-
ity based on parabolic fractal distributions (Hen-
ning, Lucianu 2000; Müller-Scheeßel 2007) have
moved and will continue to move the discipline for-
ward.
Tracer variables
The discussion presented above offered highlighted
some of the most common factors and methods used
to investigate settlement hierarchy. Alongside a ra-
ther classical methodological approach this serves
as a basis to evaluate such a scientific approach.
Conceptual and theoretical assumptions merged
with spatial and archaeological data form – directly
or indirectly – the basis for the reconstruction of
societal structures in illiterate societies. Apart from
complex self-learning algorithms, like neural net-
works or chaotic deterministic autocatalytic cycles,
factors of interest need to be chosen and weighted.
Whether that be the association of swords as symbol
of rank or power (e.g., Egg 1986.203; Kristiansen,
Larsson 2005.213; Roymans, Kortlang 1999.310–
311), the association of elites and fortification
(Jockenhövel 1990), or hoards as a reflection of cul-
tic practices (e.g., Hansen 1994.309). The connec-
tion of elites and metallurgy is presumably based
on a similar context. Promising results for the iden-
tification of prestigious artefacts were presented by
Robert Schumann (2015), who not only differenti-
ated between status and prestige but also added
much to the debate on the explanatory power of
artefacts in relation to different assemblages. There-
fore, whether classical statistical methods or more
complex modelling will provide new ideas or are
even useful depends largely on the research ques-
tion. Standard approaches, e.g., those exploring a
certain set of factors, mainly focus on equilibrium-
related methods, while many questions in social
archaeology are actually about processes. Whilst
these approaches certainly have their place in re-
search they are not ideal to investigate societal struc-
tures, because their emergence is a process, con-
stantly fluctuating between entropy and equilibrium.
These cycles (e.g., Zimmermann 2012) are the dri-
ving process – the behavioural patterns and events –
that once generated the material culture record.
It is obvious how a traditional understanding of so-
ciety as a dichotomous, static phenomenon restricts
the interpretative power of the material record. A
society can be decentralised, have cooperative sys-
tems of power, or a high level of horizontal stratifi-
cation. Even the presence of multiple vertical orga-
nisational structures in close regional proximity or
the domain of an archaeological culture cannot and
should not be ruled out (e.g., McIntosh 1999.1). To
develop new pathways in the interpretation of pre-
historic societies the assumptions of ideal, static
types of societal organisation should be left aside in
favour of processes and strategies associated with
the constitution and maintenance of political po-
wer (Blanton et al. 1996.2; Kienlin 2012). Usually
they can be described as cooperative and exclusion-
ary patterns of behaviour in variable combinations,
and do not require evolutionary logic or follow a
systemic logic (Bossen 2006.89; Jung 2011; 2006).
Similarly it is important to acknowledge that their
material indicators and symbolism and meaning can-
not be easily identified. However, they form an in-
tegral part of the supposedly subjective statistical
analysis. Nonetheless the results do not speak for
themselves, but are rather an outcome of our own
limited culturally specific symbolic understanding,
as a symbol “connotes a certain meaning as se-
mantic unit of an ideological code [..., which] pre-
vents us from seeing the semantic systems in the
totality of their mutual relationships” (Eco 1970.
553–554).
Humanity, the prime example of a complex non-lin-
ear system, thrives on its two main characteristics:
emergence over scale and self-organisation over
time. People interact with their environment, and
their spontaneous interactions create unforeseen
changes that amplify over time in new structures
and social systems (Lam 1998.37–38). Therefore,
especially with equilibrium-based approaches, the
Chiara G. M. Girotto
516
choice of tracer variables rooted in the perceived
universality of modern day values indeed creates a
past. In this case, any model is not a tool to heuris-
tically understand processes, context, and possibil-
ities but rather, as Alfred Tarski (1933) stated, a
realisation of a theory. However, in prehistoric ar-
chaeology, this cannot be the way forward to re-
construct society, as it restricts its greatest asset: the
view on the longue durée (Braudel 1977) of social
change. Archaeology might be “the discipline with
the theory and practice for the recovery of unob-
servable hominid behavior patterns from indirect
traces in bad samples” (Clarke 1973.17), but emb-
racing new theories and concepts such as complex,
non-linear models will bring the discipline forward
to a more holistic understanding of objects and their
value for the people that used them. Gaining these
insights and the inclusion of variables will reduce
the amount of narration fed into the reconstruction
of past societies.
A way forward
Synthesising the remarks on the chosen tracer vari-
ables it becomes obvious that is not only the choice
of methods which influences the result. The findings
can also be interpreted in different ways, because
science is never truly objective, and will always be
influenced by the world and culture we live in today.
Further, any statistical finding is only a statement of
probability how with regard to well the data used
fits a hypothesis, never proof of reality. Therefore,
the results should much more be seen as a reduction
of uncertainty. As in information theory, new know-
ledge is produced by the exclusion of unlikely events
or processes, following the idea of falsification. Fur-
ther, the reconstruction of function clearly requires
us to accept that, as in the Bayesian school of stati-
stics, this uncertainty is a feature of the observer
and the data itself (see Bolstad 2007). The hope that
one day the patterns left by past societies can be
understood comes from the fact that things that
exist have some quality, and that this can be mea-
sured (McCall 1939.15; Thorndike 1918.16). The
search for such meaningful variables should con-
tinue in the current eclectic school of thought (e.g.,
Chippindale 1987.515; Pearce 2011.87), choosing
the most suitable methods, approaches, and algori-
thms. Creativity and new ideas should not be hin-
dered by a lack of implementation of inhomoge-
neous functions or Bayesian functionalities in easily
accessible statistical applications, but rather should
aim to produce new methods and concepts to imple-
ment them.
Considering the limits imposed on the interpretabi-
lity of archaeological investigations, both reprodu-
cibility and a documented approach should be the
prime criteria of any methodological or statistical
paper. Whilst readily applied in modelling and more
complex strategies, this is often overlooked in more
classical archaeological approaches.
It is therefore proposed to include an adapted ver-
sion of the ODD protocol, devised by Volker Grimm
et al. (2006), as a supplement to any statistical ap-
proach in archaeology (Fig. 6).
Concisely stating the project overview, design con-
cepts, and details will force the researcher to not
only arrive at a statistically testable hypothesis, but
also specify what the prime research questions are,
by which factors those are supposedly influenced
and why these specific factors were chosen. Fur-
thermore, as clearly outlined in the previous sec-
tions it is of great importance to state and discuss
not only the theoretical background and framework
of the researcher and model, but also the data itself.
Whilst the theoretical background of the model
might already impose prior assumptions, it is critical
to address the requirements and assumptions of the
Fig. 6. Proposed general procedure to increase the
reproducibility and reusability of statistical re-
search and interpretation in archaeology.
Are we creating our past| Exploring theory building, geospatial statistics, and the reconstruction of the function of fortified Urnfield ...
517
algorithms as this will further influence the out-
come. In particular because it is common that they
are not met (Diniz-Filho et al. 2007), they need to
be accounted for in the interpretation. Furthermore,
any results should be tested not only for statistical
significance, but for reliability and logic as well.
Statistical results can never prove a hypothesis, but
if the underlying factors and methods are clearly
outlined any result can not only be reproduced and
reused by other researchers, but itself be considered
a reduction of uncertainty and an exclusion of the
unlikely.
Conclusion
This paper presented a statistical analysis and its
critical evaluation concerning the function of Urn-
field culture hilltop settlements as power, economic
and cultic centres. For the study region, the Central
Swabian Alb, the hilltop settlements were approxi-
mately regularly spaced at around 4km linear dis-
tance and followed the natural topology of the re-
gion. Density and distance measures were the foun-
dation of the study, as the requirements for most
explanatory functions were not met. Solely based
on descriptive statistics and theoretical considera-
tions, the spheres of influence of hilltop settlements
sometimes had graves with swords and rarely char-
iots at their borders. If one is willing to accept the
semantic attribution of sword bearing individuals to
territories and power, there are indications of the
power functions of hilltop settlements. However, no
reliable indications connected to economic and cul-
tic dominance were observed.
The study highlighted the importance of critically
discussing commonly used statistical algorithms and
how they influence the results of research. Depend-
ing on the choice of factors the researcher is indeed
narrating a past in accordance with perceived ‘uni-
versal’ ideas of value or status. Although archaeolo-
gical data does not meet the requirements for inho-
mogeneous, non-parametric point processes, this
suggests that these pathways should be explored
deeper and probability functions as well as the val-
idation of explanatory analyses should not remain
scarce phenomena in geospatial archaeological ap-
plications, in order to better understand intangible
variables. It also suggests that researchers should
use non-linear methods to reconstruct society more
holistically. Further, it means that researcher should
adhere to a methodological framework to increase
the reproducibility and reusability of the research
they conduct.
This research was part of a Master’s thesis submitted
at the Goethe University Frankfurt, Institute for Ar-
chaeological Sciences and associated with the LOEWE
focal point “Prehistoric Conflict Research – Bronze
Age Fortifications between Taunus and Carpathian
Mountains”. My thanks extends to my supervisors R.
Krause and M. Jung, as well as L. Linde, O. Nakoinz,
H. C. W. Price, and L. Schubert for interesting discus-
sions and ideas, as well as T. Rose for his continuous
support. The valuable comments of T. Evans and R.
Rivers on an early version of this paper as well as the
remarks of two anonymous reviewers are also grate-
fully acknowledged. All remaining errors are my own.
ACKNOWLEDGEMENTS
References
∴
Arias-González J. E., Morand S. 2006. Trophic functioning
with parasites: A new insight for ecosystem analysis. Ma-
rine Ecology Progress Series 320: 43–53.
https://doi.org/10.3354/meps320043
Arnold J. B. 2018. ggthemes: Extra themes, scales and
geoms for ‘ggplot2’. R package version 3.5.0.
Baddeley A., Rubak E., and Turner R. 2015. Spatial Point
Patterns: Methodology and Applications with R. CRC
Press. Boca Raton. London, New York.
Biel J. 1987. Vorgeschichtliche Höhensiedlungen in Süd-
württemberg-Hohenzollern. In parts similar to doctoral
thesis, University of Tübingen. Theiss. Stuttgart.
Bivand R., Keitt T., and Rowlingson B. 2018. rdgal: Bin-
dings for the geospatial data abstraction libary. R pac-
kage version 1.2–20.
Bivand R., Lewin-Koh N. 2017. maptools: Tools for read-
ing and handling spatial objects. R package version 0.9–2.
Bivand R., Pebesma E. J., and Gómez-Rubio V. 2013. Ap-
plied spatial data analysis with R. 2nd ed. Springer. New
York.
Blanton R. E., Feinman G. M., Kowalewski S. A., and Pere-
grine P. N. 1996. A Dual-Processual Theory for the Evolu-
tion of Mesoamerican Civilization. Current Anthropology
37: 1–14. https://doi.org/10.1086/204471
Chiara G. M. Girotto
518
Bockisch-Bräuer C. 1999. Zur Aussagekraft von Gräbern
bei der Rekonstruktion sozialer Strukturen: Überlegungen
am Beispiel der Spätbronze- und Urnenfelderzeit in Nord-
bayern. In Eliten in der Bronzezeit: Ergebnisse zweier
Kolloquien in Mainz und Athen. Römisch-Germanisches
Zentralmuseum. Verlag des Römisch-Germanischen Zen-
tralmuseums. Mainz: 533–564.
Bolstad W. M. 2007. Introduction to Bayesian statistics.
2nd edition. Wiley. Hoboken.
Bossen C. 2006. War as practice, power and processor: A
framework for the analysis of war and social structural
change. In T. Otto, H. Thrane, and H. Vandkilde (eds.),
Warfare and society: Archaeological and social anthro-
pological perspectives. Aarhus Univ. Press. Aarhus: 89–
102.
Bowman A., Crawford E., Alexander G., and Bowman R. J.
2007. rpanel: Simple interactive controls for R functions
using the tcltk package. Journal of Statistical Software
17: 1–18.
Braudel F. 1997. Geschichte und Sozialwissenschaften.
Die longue durée. In M. Bloch, C. Honegger (eds.), Schrift
und Materie der Geschichte. Vorschläge zur systemati-
schen Aneignung historischer Prozesse. Edition Suhr-
kamp 814. Suhrkamp. Frankfurt: 47–85.
Buccellati G. 2017. A critique of archaeological reason:
Structural, digital and philosophical aspects of the exca-
vated record. Cambridge University Press. Cambridge.
Chippindale C. 1987. Comment to Earle, T. K., and Preu-
cel, R. W. 1987, Processual archaeology and the radical
critique. Current Anthropology 28: 501–512.
http://www.jstor.org/stable/2743487
Christaller W. 1933 [2006]. Die zentralen Orte in Süd-
deutschland: Eine ökonomisch-geographische Untersu-
chung über die Gesetzmäßigkeit der Verbreitung und
Entwicklung der Siedlungen mit städtischen Funktionen.
2nd edition. Wissenschaftliche Buchgesellschaft. Darmstadt.
Clarke D. L. 1973. Archaeology. The loss of innocence.
Antiquity 47: 6–18.
https://doi.org/10.1017/S0003598X0003461X
Clausing C. 1998. Zur Bedeutung der Bronze als Anzeiger
urnenfelderzeitlicher Sozialstrukturen. In C. Mordant, M.
Pernot M. (eds.), L’atelier du bronzier en Europe du 20e
au 8e siècle avant notre ère: Production, circulation et
consommation du bronze. The Comité des travaux his-
toriques et scientifiques (CTHS). Paris: 309–321.
1999. Untersuchungen zur Sozialstruktur in der Urnen-
felderzeit Mitteleuropas. In Römisch-Germanisches Zen-
tralmuseum (ed.), Eliten in der Bronzezeit: Ergebnisse
zweier Kolloquien in Mainz und Athen. Verlag des Rö-
misch-Germanischen Zentralmuseums. Mainz: 319–420.
Deman G., Konakli K., Sudret B., Kerrou J., Perrochet P.,
and Benabderrahmane H. 2016. Using sparse polynomi-
al chaos expansions for the global sensitivity analysis of
groundwater lifetime expectancy in a multi-layered hydro-
geological model. Reliability Engineering & System Sa-
fety 147: 156–169. 
https://doi.org/10.1016/j.ress.2015.11.005
Dietler M. 2001. Theorizing the feast: Rituals of consump-
tion, commensal politics, and power in African contexts.
In M. Dietler, B. Hayden (eds.). Feasts: Archaeological and
ethnographic perspectives on food, politics, and power.
Smithsonian Institution Press. Washington, DC: 54– 114.
Diniz-Filho J. A., Hawkins B. A., Bini L. M., De Marco P. Jr.,
and Blackburn T. M. 2007. Are spatial regression methods
a panacea or a Pandora’s box?: A reply to Beale et al.
(2007). Ecography 30: 848–851.
Eco U. 1970. Codes and ideology. In B. Visentini, C. Oli-
vetti (eds.), Linguaggi nella società e nella tecnica: Con-
vergo promosso dalla C. Olivetti, per il centenario della
nascita di Camillo Olivetti, Museo Nazionale della Sci-
enza e della Tecnica Milano. Edizioni di Comunità. Mi-
lano: 545–557.
Egg M. 1986. Zum “Fürstengrab” von Radkersburg (Süd-
steiermark). Jahrbuch des Römisch-Germanischen Zen-
tralmuseums 33: 199–214.
Falkenstein F. 2005. Aspekte von Alter und Geschlecht im
Bestattungsbrauchtum der nordalpinen Bronzezeit. In J.
Müller (ed.), Alter und Geschlecht in ur- und frühge-
schichtlichen Gesellschaften: Tagung Bamberg, 20.–21.
Februar 2004. R. Habelt. Bonn: 73–90.
Falkenstein F., Ostermeier N. 2015. Höhensiedlungen in
Süddeutschland: Zentralorte, Kultzentren oder Handels-
plätze. Archäologie in Deutschland 4: 20–21.
Fischer C. 1997. Innovation und Tradition in der Mittel-
und Spätbronzezeit: Gräber und Siedlungen in Neffen-
bach, Fällanden, Dietikon, Pfäffikon und Erlenbach. Fo-
torotar. Zürich.
Gersbach E. 2006. Die Heuneburg bei Hundersingen,
Gemeinde Herbertingen: Eine Wehrsiedlung, Burg der
Bronze- und frühen Urnenfelderzeit und ihre Stellung
im Siedlungsgefüge an der oberen Donau. Theiss. Stut-
tgart.
Girotto C. G. M. 2018. … und was macht sie besonders?:
Ein statistischer Ansatz zur Rolle befestigter Höhensied-
Are we creating our past| Exploring theory building, geospatial statistics, and the reconstruction of the function of fortified Urnfield ...
519
lungen innerhalb des urnenfelderzeitlichen Siedlungs-
gefüges der zentralen Schwäbischen Alb. Master thesis.
Goethe University Frankfurt. Frankfurt.
Green S. W., Perlman S. M. 1985. Frontiers, boundaries,
and open social spaces. In S. W. Green, S. M. Perlman
(eds.), The archaeology of frontiers and boundaries.
Academic Press. Orlando: 1–30.
Grimm V. and 27 co-authors. 2006. A standard protocol
for describing individual-based and agent-based models.
Ecological Modelling 198: 115–126.
https://doi.org/10.1016/j.ecolmodel.2006.04.023
Gringmuth-Dallmer E. 1999. Methodische Überlegungen
zur Erforschung zentraler Orte in ur- und frühgeschichtli-
cher Zeit. In S. Moździoch (ed.), Centrum i zaplecze we
wczesnośredniowiecznej Europie środkowej. Werk. War-
saw: 9–20.
Hacıgüzeller P. 2012. GIS, critique, representation and be-
yond: The analysis of settlement patterns. Journal of So-
cial Archaeology 12: 245–263.
https://doi.org/10.1177/1469605312439139
Hansen S. 1994. Studien zur Metalldeponierung wäh-
rend der älteren Urnenfelderzeit zwischen Rhônetal
und Karpartenbecken. Universitätsforschungen zur Prä-
historischen Archäologie 21. Bonn.
Henning H., Lucianu C. 2000. Zipf, Christaller, Gräberfel-
der: Sind latente Besiedlungsstrukturen der Hallstattzeit
aus der Verteilung der Nekropolen ersichtlich? Archäo-
logisches Korrespondenzblatt 30: 527–548.
Herzog I. 2009. Analyse von Siedlungsterritorien auf der
Basis mathematischer Modelle. In D. Krausse, O. Nakoinz
(eds.), Kulturraum und Territorialität: Archäologische
Theorien, Methoden und Fallbeispiele; Kolloquium des
DFG-SPP 1171, Esslingen 17.–18. Januar 2007. Leidorf.
Rahden: 71–86.
Jockenhövel A. 1971. Die Rasiermesser in Mitteleuropa:
Süddeutschland, Tschechoslowakei, Österreich, Schweiz.
Prähistorische Bronzefunde. Beck. München.
1990. Bronzezeitlicher Burgenbau in Mitteleuropa: Un-
tersuchungen zur Struktur frühmetallzeitlicher Gesel-
lschaften. In T. Bader (ed.), Orientalisch-ägäische Ein-
flüsse in der europäischen Bronzezeit: Ergebnisse
eines Kolloquiums vom 16.–19.10.1985 in Mainz. R.
Habelt. Bonn: 209–228.
1994. Schutz und Repräsentation: Burgenbau: Eine
Neuerung im Siedlungswesen. In A. Jockenhövel, W. Ku-
bach (eds.), Bronzezeit in Deutschland. Theiss. Stut-
tgart: 22–26.
Jung M. 2006. Zur Logik archäologischer Deutung: Inter-
pretation, Modellbildung und Theorieentwicklung am
Fallbeispiel des späthallstattzeitlichen “Fürstengrabes”
von Eberdingen-Hochdorf, Kr. Ludwigsburg. Universitäts-
forschungen zur prähistorischen Archäologie 138. Habelt.
Bonn.
2011. Der “Big Man”: Die Verselbstständigung eines
theoretischen Konstrukts und ihre Adaption in der Ar-
chäologie. Das Altertum 56: 187–204.
Keegan J. 1993. A history of warfare. Plimico. London.
Kienlin T. L. 2012. Beyond elites: An introduction. In T.
L. Kienlin, A. Zimmermann (eds.), Beyond elites: Alterna-
tives to hierarchical systems in modelling social forma-
tions. International conference at Ruhr-Universität Bo-
chum, Germany. Habelt. Bonn: 15–32.
Knöpke S. 2009. Der urnenfelderzeitliche Männerfried-
hof von Neckarsulm. Theiss. Stuttgart.
Kossack G. 1974. Prunkgräber: Bemerkungen zu Eigen-
schaften und Aussagewert. In G. Kossack, G. Ulbert (eds.),
Studien zur vor- und frühgeschichtlichen Archäologie:
Festschrift für Joachim Werner zum 65. Geburtstag.
Beck. München: 3–33.
Kreutle R. 2007. Die Urnenfelderkultur zwischen
Schwarzwald und Iller: Südliches Württemberg, Hohen-
zollern und südöstliches Baden. Verlag Dr. Faustus. Bü-
chenbach.
Kristiansen K. 1982. The formation of tribal systems in
Later European Prehistory: Northern Europe, 4000–500
BC. In C. Renfrew, M. Rowlands, and B. A. Seagraves
(eds.), Theory and explanation in archaeology: The
Southampton Conference. Academic Press. New York:
241–280.
1998. The emergence of the European World System in
the Bronze Age. In K. Kristiansen, M. Rowlands M. (eds.),
Social transformations in archaeology: Global and
local perspectives. Routledge. London: 287–323.
Kristiansen K., Larsson T. B. 2005. The rise of Bronze Age
society: Travels, transmissions and transformations.
Cambridge University Press. Cambridge.
Kuijpers M. G. H. 2012. Towards a deeper understanding
of metalworking technology. In T. L. Kienlin, A. Zimmer-
mann (eds.), Beyond elites: Alternatives to hierarchical
systems in modelling social formations. International
conference at Ruhr-Universität Bochum. Germany. R. Ha-
belt. Bonn: 413–422.
Chiara G. M. Girotto
520
Lam L. 1998. Nonlinear Physics for Beginners. Fractals,
Chaos, Solitons, Pattern Formation, Cellular Automata
and Complex Systems. World Scientific. Singapur.
Maise C. 1996. Höhensiedlungen als Normalform hallstat-
tzeitlicher Siedlungen? Archäologisches Korrespondenz-
blatt 26: 65–74.
McCall W. A. 1939. Measurement: A revision of ‘How to
measure in education’. Macmillian Company. New York.
McIntosh S. K. 1999. Pathways to complexity: An African
perspective. In S. K. McIntosh (ed.), Beyond chiefdoms:
Pathways to complexity in Africa. Cambridge University
Press. Cambridge: 1–30.
Miller D., Tilley C. Y. 1984. Ideology, power, and prehi-
story: An introduction. In C. Y. Tilley, D. Miller (eds.),
Ideology, power, and prehistory. Cambridge University
Press. Cambridge: 1–15.
Müller-Scheeßel N. 1998. Archaeology is nothing if it is
not critique: Zum Archäologieverständnis von Michael
Shanks und Christopher Tilley. In M. K. H. Eggert (ed.),
Theorie in der Archäologie: Zur englischsprachigen Dis-
kussion. Waxmann. Münster, Berlin: 243–271.
2007. Weitere Überlegungen zu den latenten Besied-
lungsstrukturen der Hallstattzeit Süddeutschlands. Ar-
chäologisches Korrespondenzblatt 37: 57–66.
Nakoinz O. 2013a. Archäologische Kulturgeographie der
ältereisenzeitlichen Zentralorte Süd-westdeutschlands.
Universitätsforschungen zur prähistorischen Archäolo-
gie. R. Habelt. Bonn.
2013b. Räumliche Interaktionsmodelle. Prähistorische
Zeitschrift 88: 226–257.
Nessel B. 2012. Metallurgen im Grab: Überlegungen zur
sozialen Einstufung handwerklicher Spezialisten. In T. L.
Kienlin, A. Zimmermann (eds.), Beyond elites: Alter-
natives to hierarchical systems in modelling social for-
mations. International conference at Ruhr-Universität Bo-
chum, Germany. R. Habelt. Bonn: 423–432.
Ostermeier N. 2012. Urnenfelderzeitliche Höhensiedlun-
gen in Bayern nördlich der Donau: Topographische,
chronologische und funktionale Aspekte. Universitäts-
forschungen zur prähistorischen Archäologie. R. Habelt.
Bonn.
Pearce M. 2011. Have rumours of the ‘Death of Theory’
been exaggerated? In J. Bintliff, M. Pearce (eds.), The
Death of Archaeological Theory? Oxbow Books. Haver-
town: 80–89.
Pebesma E. J., Bivand R. 2005. sp: Classes and methods
for spatial data. R News 5(2): 9–13.
https://cran.r-project.org/doc/Rnews/
Posluschny A. 2010. “Fürstensitze”, Zentralität und Hin-
terland: ERste Aspekte aus Sicht des Projektes “Fürsten-
sitz” & Umland. In D. Krausse (ed.), “Fürstensitze” und
Zentralorte der frühen Kelten: Abschlusskolloquium des
DFG-Schwerpunktprogramms 1171 in Stuttgart, 12.–15.
Oktober 2009. Theiss. Stuttgart: 359–374.
R Core Team. 2016. R: A language and environment for
statistical computing. R Foundation for Statistical Com-
puting. Vienna.
Rowlands M. 1998. Conceptualising the European Bronze
and Early Iron Ages. In K. Kristiansen, M. Rowlands (eds.),
Social transformations in archaeology: Global and local
perspectives. Routledge. London: 49–69.
Roymans N, Kortlang F. 1999. Urnfield symbolism and so-
cial organisation in the Lower Rhine Region: The Beeg-
den cemetery. In Römisch-Germanisches Zentralmuseum
(ed.), Eliten in der Bronzezeit: Ergebnisse zweier Kol-
loquien in Mainz und Athen. Verlag des Römisch-Germa-
nischen Zentralmuseums. Mainz: 277–318.
Schauer P. 1993. Befestigte Siedlungen der Urnenfelder-
zeit und älteren Eisenzeit in Süddeutschland. In H. Dan-
nheimer (ed.), Das keltische Jahrtausend: [Landesaus-
stellung des Freistaates Bayern, Prähistorische Staats-
sammlung, und der Stadt Rosenheim vom 19. Mai – 1.
November 1993 im Lokschuppen Rosenheim]. von Za-
bern. Mainz am Rhein: 62–74.
Schumann R. 2015. Status und Prestige in der Hallstat-
tkultur. Verlag Marie Leidorf. Rahden.
Simmel G. 1903. Die Grosstädte und das Geistesleben. In
T. Petermann (ed.), Die Grossstadt: Vorträge und Auf-
sätze zur Städteausstellung. Dresden: 185–203.
Sperber L. 1999. Zu den Schwertträgern im westlichen
Kreis der Urnenfelderkultur: Profane und religiöse As-
pekte. In Römisch-Germanisches Zentralmuseum (ed.),
Eliten in der Bronzezeit: Ergebnisse zweier Kolloquien
in Mainz und Athen. Verlag des Römisch-Germanischen
Zentralmuseums. Mainz: 605–659.
Stary P. F. 1980. Das spätbronzezeitliche Häuptlingsgrab
von Hagenau, Kr. Regensburg. In K. Spindler (ed.), Vor-
zeit zwischen Main und Donau: Neue archäologische
Forschungen und Funde aus Franken und Altbayern.
Univ.-Bund Erlangen-Nürnberg. Erlangen: 46–97.
Stegmaier G. 2017. “Akropolis und Suburbium”: Neue Un-
tersuchungen zur bronzezeitlichen Heuneburg bei Herber-
Are we creating our past| Exploring theory building, geospatial statistics, and the reconstruction of the function of fortified Urnfield ...
521
tingen-Hundersingen, Kr. Sigmaringen. In D. Brandherm,
B. Nessel (eds.), Phasenübergänge und Umbrüche im
bronzezeitlichen Europa: Beiträge zur Sitzung der Ar-
beitsgemeinschaft Bronzezeit auf der 80. Jahrestagung
des Nordwestdeutschen Verbandes für Altertumsfor-
schung. R. Habelt. Bonn: 253–269.
Tarski A. 1933. The concept of truth in the languages of
the deductive sciences (Polish). Prace Towarzystwa Nau-
kowego Warszawskiego, Wydzial III Nauk Matematyczno-
Fizycznych 34. Warsaw; reprinted in Zygmunt 1995, pp.
13–172; expanded English translation in Tarski 1983
[1956]: 152–278.
Thorndike E. L. 1918. The nature, purposes, and general
methods of measurement of educational products. Year-
book of the National Society for the Study of Education
17: 16–24.
Tomedi G. 1999. Eliten und Dynastien der späten Urnen-
felderzeit und Hallstattzeit im südostalpenraum. In Rö-
misch-Germanisches Zentralmuseum (ed.), Eliten in der
Bronzezeit: Ergebnisse zweier Kolloquien in Mainz und
Athen. Verlag des Römisch-Germanischen Zentralmuse-
ums. Mainz: 661–681.
University of Heidelberg. SRTM Terrain Model.
http://opendem.info/download_srtm.html
Urbaneck S. 2013. tiff: Read and write TIFF images. R
package version 0.1–5.
Wickham H. 2009. ggplot2: Elegant graphics for data
analysis. R package version 3.5.0.
Wickham H., François R., Henry L., and Müller K. 2018a.
dplyr: A grammar of data manipulation. R package ver-
sion 0.7.5.
Wickham H., Henry L. 2018. tidyr: Easily tidy data with
‘spread’ and ‘gather’ functions. R package version 0.8.1.
Wickham H., Hester J., and Chang W. 2018b. devtools:
Tools to make developing R packages easier. R package
version 1.13.5.
Wickham H., Hester J., and François R. 2017. readr: Read
rectangualr text data. R package version 1.1.1.
Winghart S. 1999. Die Wagengräber von Poing und Hart
a. d. Alz: Evidenz und Ursachen spätbronzezeitlicher Eli-
tenbildung in der Zone nordwärts der Alpen. In Römisch-
Germanisches Zentralmuseum (ed.), Eliten in der Bron-
zezeit: Ergebnisse zweier Kolloquien in Mainz und
Athen. Verlag des Römisch-Germanischen Zentralmuse-
ums. Mainz: 515–532.
Wirth S. 1999. Auf der Suche nach Eliten der späten Bron-
zezeit und der Urnenfelderzeit: Bausteine zum thema aus
dem Altsiedelland am unteren Lech in Bayerisch-Schwa-
ben. In Römisch-Germanisches Zentralmuseum (ed.), Eli-
ten in der Bronzezeit: Ergebnisse zweier Kolloquien in
Mainz und Athen. Verlag des Römisch-Germanischen Zen-
tralmuseums. Mainz: 565–604.
Zimmermann A. 2012. Cultural cycles in Central Europe
between the Neolithic and the Iron Age. In T. L. Kienlin,
A. Zimmermann (eds.), Beyond elites. Alternatives to hie-
rarchical systems in modelling social formations. Inter-
national conference at Ruhr-Universität Bochum, Ger-
many, Bonn 22.–24.10.2009. R. Habelt. Bonn: 137–146.