Transport, Interaction, and Connectivity

2.1 Means of Transport

Pedestrian transport must be assumed as it leaves little evidence in the archaeological record beyond the presence of shoes. Wagons and the invention of the wheel is an archaeological relevant issue. Wagons are a substantial simplification of traffic and hence, it is no surprise, that the wheel was invented independently in different locations (Kaiser, 2010). The extraordinary importance of wagons is reflected with their role in funerary practice and rites of some cultures, which reflect their importance in everyday life as well (Harbison, 1969; Pare, 1988, 1991). Richly decorated four-wheeled ritual wagons and two-wheeled chariots were used in the Iron Age of Europe and the Mediterranean. Ritual wagons refer to a sphere of ritual and spiritual transport, mostly in the context with religious transitions. Egypt or Celtic chariots (Furger-Günti, 1991; Harbison, 1969; Veldmeijer, 2017) are rather associated with combat. Wagons require draft animals, which are sometimes present in graves. Equestrians also need to be considered (Bliujienė, 2009; Koch, 1999). Horses play an important role for equestrian nomads (Alexeev, 2001).

Ships and boats representing waterborne transport are known from rock art, the decoration of bronze objects, pottery vases, etc., from different times and regions (Kaul, 1998) as well as from shipwrecks (Johnstone, 2013). The most successful type is the logboat used in Europe from the Mesolithic to the twentieth century (Hirte, 1987; McGrail, 1978). Plank boats provide more technological details and more flexibility for shaping the hull. The Hjortspring boat from the Pre-Roman Iron Age and the Nydam boat from the Roman Iron Age are examples of a stepwise technological development and show the transition from the canoe to the rowing boat (Rieck, 1994; Rosenberg, 1937). Both combat vessels represent a strong military component. Specific technical solutions enable us to identify boat building traditions such as ships with sewn planks from the Mediterranean (Kahanov & Pomey, 2004). The strict functional requirements of ships and boats, which are life-threatening if not met, led to a relatively stable typological development of watercraft, where arbitrary attempts are rather rare. Each typological change is due to functional, economic, or social reasons. Social aspects are seen with the late Nordic vessels and the early cogs (Crumlin-Pedersen, 1997; Englert, 2015; Fliedner & Pohl-Weber, 1972). The Nordic ships are built in a traditional way and have larger capacity, while the cogs break with tradition and introduce entirely different people to shipbuilding and navigation. Shipbuilding and naval seafaring become the domain of medieval citizens. The wide range of topics covered with ships and wagons allows us to gain many information on both, transport and society. The means of transport tells us something about transport velocity, capacity, technical skills, and the social integration of transport.

2.2 Roads, Routes, and Networks

The aspect of roads, routes, and networks corresponds with the main topic of this volume, and includes material remains of roads as well as the reconstruction of routes based on indirect evidence (Margary, 1973; Pitz, 1962; Rathmann, 2003; Schneider, 1848; Tilburg, 2007; Vermeulen & Antrop, 2001). These empirical route models result in a network that comprises connections between places with specific features such as length, effort of movement, building costs, suitability for different means of transport, and technical realisation. Three levels are demonstrated in both the empirical and the theoretical network models (Nakoinz & Knitter, 2016, p. 171).

At the local level, networks represent the exact location of pathways. In this case, nodes represent turning points of the pathways and, rarely, occupied places. The edges of the network, the exact pathways are tightly connected to environment in this level. Slope, soil type, wetness, and means of transport determine the location of the pathways. Local level transport networks only make sense as geographical networks mapped in a geographical space. A network always connects nodes with straight (or arbitrary curved if there is no geographical meaning) lines. A least cost path connection would produce a curve between two sites by creating many artificial nodes that do not have a specific meaning beside they are placed on the pathway. Hence, these nodes do not make any sense without the geographical location.

Regional level networks do not focus on the exact location of the connection but on the question of which sites are connected to other sites. Both, geographical and non-geographical network representations can make sense. On the regional level, the nodes are settlements at the starting and end points of interaction processes. The edges are any kind of relationship or interaction. In this case, the edges are either interpreted as roads, pathways, or as measure of the actual interaction. Although the environment has influence on the edges, they are not directly related to the environment in the network model.

The supra-regional level networks represent dominant and abstract interaction patterns. On this level, the nodes do not include certain settlements but rather whole regions represented by a random point inside the region. Accordingly, the edges are not certain connections or roads but rather general routes and communication axes. The supra-regional level represents the general structure of the transport system.

Empirical local level networks can be developed using archaeological road remains, such as hollow ways and paved roads. The research on Roman roads started rather early and is still an important topic (Becker, 1880; Blöck, 2014; Davies, 2002; Margary, 1973). The reasons are diverse. The most important one is the fact that paved Roman roads are more likely to be discovered than roads without this kind of substantial material structure. Furthermore, the dense pattern of Roman archaeological evidence provides a rich context for interpreting the roads.

Nonetheless, paths without a fixed structure are also the subject of archaeological research (Herzog, 2016). In particular, hollow ways can be observed, but since they can be remains from different time periods and are not suitable for scientific dating or typological dating, their chronology is hard to determine. This is completely different for “Moorwege,” wooden paths built to cross bogs, because they offer much better opportunities for dating since they are often preserved in the wet environment (Endlich & Lässig, 2007; Hayen, 1958; Heumüller, 2018). Furthermore, the effort to build these constructions points to highly relevant connections. Similarly, paved street sections can indicate areas of special significance, as is the case at Borremose, a fortified settlement of the pre-Roman Iron Age in Denmark (Martens, 2010).

We also have to consider bridges and transport infrastructure such as fords (Lehnemann, Urz, & Meiborg, 2021) because they indicate roads or pathways. A disputed topic is the reconstruction of pathways using grave mounds. Müller (1904) not only observed that Bronze Age burial mounds form strange linear patterns in Denmark which can be interpreted as ancient pathways but also reconstructed the course of these paths more than 100 years ago. Similar research has been conducted in Schleswig-Holstein (Hinz, 1951; Marschalleck, 1964). This approach was accepted until Renfrew (1973) observed that megalithic tombs are often located at assumed territorial borders. For the Danish monuments, Klassen (2014) demonstrated that the Bronze Age lines frequently aim for fords, and their interpretation as paths is convincing. It is also possible that lines of monuments are both borders and pathways (Faupel, 2021; Faupel & Nakoinz, 2018; Nakoinz & Knitter, 2016). Although roads connect settlements, not all observed road segments are necessarily aiming at settlements. An example are stubways, which extend a network by running between frequently relocated settlement sites and the road system. They enable a stable road system that can coincide with borders.

Accepting the idea that monuments can indicate the location of paths, the next question is how to extract the network from the points. Müller (1904) used an intuitive approach that led to convincing results, but it is not reproducible. A quantitative and reproducible approach is to use neighbourhood graphs (Katajainen & Nevalainen, 1986; Nakoinz & Knitter, 2016, pp. 173–174) for pattern recognition. This approach is sensitive for gaps in the data and assumes that the pathways run exactly along the monuments (Figure 1).

Figure 1

Neighbourhood graphs of Bronze Age burial mounds in Dänischer Wohld (red: relative-neighbour-graph; grey Delaunay-graph (Nakoinz, 2012, Figure 2).

The disadvantages of the neighbourhood graphs are overcome by the density ridge approach (Faupel, 2021; Faupel & Nakoinz, 2018; Nakoinz & Knitter, 2016). The kernel density estimation of the point density of monuments can be used to extract the ridges of the density “mountains,” indicating the lines of the highest point density. This approach was further developed by Franziska Faupel (Engelbogen) who applied this concept to Iron Age cemeteries and analysed the Iron Age transport system in southwest and west Germany and Alsace (Faupel, 2021) (Figure 2).

Figure 2

Density ridge method (Nakoinz & Knitter, 2016, Figure 9.9).

The empirical models show how the ancient pathway network had been. Usually, it is hard to understand why it is this way. To better understand the empirical models, theoretical models can be applied. Even if archaeological data are used, the outcome of the theoretical models is not directly related to the past. They show how the pathway system would look if some specific principles and parameter were applied. If the theoretical model does not fit the empirical one and the implementation of the model is correct, we can infer that the parameter and settings used in the theoretical model did not play a dominant role in the past. Otherwise, they must be considered for the archaeological interpretation.

A frequently used theoretical pathway model is the least cost path model (e.g. Herzog, 2016), which calculates the route that involves the least effort, taking into account the terrain, slope, and obstacles. The parameter used for this model differs according to preferences. For example, walking has fewer restrictions concerning the slope compared to driving a wagon. However, the standard approach simply calculates the optimal route. The second-best route differing slightly by small degree may run elsewhere. To overcome this problem, probabilistic approaches are used, which results in corridors rather than an exact pathway (e.g. Lewis, 2021).

Least cost path models can be used to identify the relevant factors determining the location of ancient pathways, allowing different theoretical models to be compared to the empirical one. The similarity of the theoretical and empirical model somehow measures the relevance of the theoretical parameters, allowing a more complete understanding of the pathways and their location (Faupel & Nakoinz, 2018; Nakoinz, 2012).

To summarise, local level networks show how the sites are connected and, in particular, which factors are significant or relevant for the exact location of the connection. This provides insight into the practical requirements, the purpose, and restrictions of the transport system. This level of analysis contributes to questions concerning the minimisation of transport costs, the role of transport costs in general, the relevance of single target sites and connections, and the degree of organisation and development of the transport network. We need to know which sites were connected to fully benefit from these analyses. This is the task of the regional level analysis.

The regional level ignores the precise location of a pathway and asks whether two sites are connected. Furthermore, some general characterisations such as the intensity of the connection can be used as edge weights to characterise the connection on an abstract level. Theoretical network models of the regional level are usually based on simple principles. For instance, connections can be established with sites within a certain distance (buffer approach) or with neighbours of different grades (k-neighbours approach). While the neighbourhood definition based on the ranking of differences to other sites is a mere metrical approach, neighbourhood graphs apply a topological concept. The Delaunay graph or the relative neighbour graph is defined by certain rules for the connections which not only involve distances but also the geometrical configuration of points. The Delaunay graph tests for all subsamples of three points if other points are inside the circle touching the three points. If there are no other points, the connections of the three points forming a three-angle are included in the graph. All other neighbourhood graphs are sub-graphs of the Delaunay graph.

The Delaunay graph represents a model with all required short connections and avoids any long or redundant connections. It can be used to describe a transport system where each settlement is connected to the next one but without highways running between the settlements or specific long-distance connections. At the same time, the many connections of the Delaunay graph represent a system in which it is easy to find a short path between all pairs of points. The relative neighbour graph still provides a network connecting all points but with fewer connections. From the perspective of people who build or maintain roads, this a good solution because connectivity is provided with a minimum of edges. From the traveller’s perspective, a complete graph would be preferred because it offers direct connections to all possible targets. The Delaunay graph is a kind of universal optimisation that considers both perspectives. In addition to these simulations of random graphs, simulations of specific processes can serve as theoretical models.

The minimisation of connections as in the neighbourhood graphs is also preferable from a cognitive perspective. People tend to reuse tracks even when it is not necessary. Empirical studies revealed that new pathways emerge when they provide a saving of distance of about 30% (Helbing, Keltsch, & Molnár, 1997). The theoretical networks help understanding the factors of empirical networks. Which parameters are optimised? Empirical networks need to be known to answer this question.

Empirical regional level networks can be established based on written sources or maps, such as the Tabula Peutingeriana (Albu, 2014). Although results of the local level networks can also be used for this purpose, they are usually too fragmented. Archaeological observations can be used in two ways. First, an object moved from one site to another, identified either typologically or based on scientific analysis, forms a connection between the two sites. It is often only possible to determine the region rather than the exact origin of an artefact, so very few connections can be drawn this way. A more probabilistic approach is to use similarity networks (Habiba, Athenstädt, Mills, & Brandes, 2018; Österborn & Gerding, 2015). In this case, the similarity of archaeological observations is used to deduce a certain degree of exchange between the two sites. The more complex the archaeological observation and the more unlikely a convergence phenomenon, the more reliable is the inferred interaction. The concept of cultural distances (Nakoinz, 2013a) aims to provide a consistent and theory-based approach of this idea; it is discussed further below.

Beyond the visualisation of these networks, network analysis reveals the characteristics of the nodes, edges, or the whole network. Centrality scores are one example of this (Aleskerov, Shvydun, & Meshcheryakova, 2021; Freeman, 1978; Koschützki et al., 2005; Tavassoli, 2018). Many case studies in regional level network analysis apply specific combinations of different methods (Feugnet, Rossi, & Filet, 2017; Filet, 2017; Fulminante, 2020; Knappett, 2013; Leidwanger & Knappett, 2018; Sindbæk, 2007 and other papers in this volume).

The network analysis of the regional level aims to answer the question of which and why sites are connected. A deeper understanding of the system of interaction is gained by neglecting the exact location of the connection. The next step of this abstraction process is to focus on the interrelationship of regions, which is the subject of the supra-regional level networks. This level of analysis is not discussed here as the methods are the same as those of regional level and only the nodes are defined differently.

Roads, routes, and networks show which destinations are considered worthy of being connected, which points are in fact connected, and, furthermore, which parameters have been optimised and are therefore more relevant than others.

2.3 Transport Infrastructure

In addition to roads, transport infrastructure comprises bridges (Lehnemann et al., 2021), ferries (Nakoinz, 2005), harbours (Vött, Fischer, & Hadler, 2015), and accommodation. Furthermore, stores or local merchants providing supplies and gear replacement can be assumed. In a time before maps, specialist knowledge must be provided by pilots and guides. Shoals, currents, dynamic fords, and dangerous passes make travelling life-threatening without proper information. Another aspect of security is threats of being robbed or encounters with unfriendly individuals or groups. The pacification of the path as well as fortification of the stations is important but partly intangible aspects of the infrastructure. Hillforts and dykes can be part of the security network as well as used by the threatening party. Archaeology definitely has knowledge gaps concerning transport infrastructure. Well investigated oppida being certainly the target as well as mid-way stations of many transport processes show that our knowledge is still limited (Metzler, 1995). However, the example of the Mont Lassois harbour or lock (Chaume, Cheetham, Komp, Lüth, & Weißl, 2020) shows there is potential for substantial archaeological knowledge. However, this is based on identifiability. A house, however, cannot easily be identified as a travellers’ hostel. If such information were available, it would reveal the degree of organisation of transport, the attitude of residents towards travellers, and conditions of transport.

2.4 Items Transported

The question of which items have been transported is not always a trivial one. According to Jacob-Friesen (1928), people, commodities, and ideas are the common categories of transported items. Although people are always involved, they can be the operators of the transport process or the passenger. Realistic scenarios involve combinations of all three categories, but limited archaeological evidence results in simplified models. However, it depends on which category is dominant. For example, a foreign type of ceramic vessel can be an imported object in which scientific analysis proves the material originated from a remote location. It is also possible that a craftsman moved to a new location and used his or her traditional techniques on local material, indicating the transport of people and ideas. Alternatively, just the idea might have travelled if the object is an imitation by a local craftsman. Obviously, the craftsman has to have been at a place where he could observe the original object or one of the original objects has to be moved into the reach of the local craftsman. From a transport perspective it appears to be relevant, if just people travel or if one template or many products are moved to the new location.

The mode of exchange is also relevant. Is it trade of products, gifts, loot, or just “by-catch”? Trade is defined as exchanging products after an offer to sell. Payment can be assumed to be in the form of money, commodities, or services. Gifts can be viewed as gifts of guests, tips, or bribery, and either reciprocal or not. Loot, unlike trade and gifts, is an involuntary transfer of the good and is one-sided. What I call by-catch are all objects which do not fit into these categories and are at the target location by accident. Objects lost by travellers are an example of this category.

Each category tells a different story, and it is challenging to decide in which category a foreign object belongs. A large number of foreign objects indicate trade, but even that is uncertain since regular gifts can have the same effect; regular trading commodities might be invisible in the archaeological record: for example less durable material such as leather, wood, or human being themselves.

In addition, some items of material culture serve as a kind of currency (Millot-Richard, 2021). For the La Tène culture, semi-finished products play different roles in different contexts, i.e. they can be considered products or currency. The exchange network of these objects is likely to map to the political-economic skeleton that supports the spatial organisation of a region.

There are also technical modes of exchange, such as direct exchange or down-the-line exchange (Renfrew, 1975, 1977). The main factors of the different modes are the moving actors and the successive steps of exchange. Who is travelling: professional merchants, customers, or producers? Are traded commodities passed on from one person to another one or directly from producer to consumer? Is the exchange process intermittent?

Renfrew (1977) compared an empirical fall-off curve with different theoretical distance decay functions to identify the mode of exchange. Fall-off curves represent the quantity of items as a function of the distance from the source. It is assumed that some items “fall off” from the constant stream of items and that the quantity of these items is a measure for the quantity of the remaining stream. Distance decay functions are theoretical functions that show the assumptions “fall off from the stream” or “decay of the remaining amount” for different exchange modes. Although this is a valid scientific approach, in practice, problems may arise due to topography and biases of the archaeological sources, which can distort the fall-off curves.

In conclusion, the items exchanged and the modes of exchange reveal social, cultural, and economic practices. What can be gained by transporting something and how is this put into practice? Assuming that people usually prefer the best solution from their perspective, it can be supposed that the specific transport behaviour was considered ideal from the perspective of perceived opportunities and restrictions. A direct contact might save costs by excluding intermediaries, but this perhaps means a dangerous journey in contrast to down-the-line exchange, just to mention one example. However, humans do not always act in a logical way and may prefer a path for a reason that is unknown. Subjective and objective factors are involved in the decision process.