Mixed Methods in Landscape Archaeology: An Application to Explore Identity Formation in the Romano-British Period,Shropshire Region

Abstract

This article proposes a novel methodology designed to identify past/ancient identities from archaeological information available in rural landscapes. It consists of a quantitative stage based on network analysis followed by a qualitative analysis based on phenomenology. Such methodology contributes to the field of mixed method research in two ways. First, in contrast to existing approaches that use people’s opinions to collect data, the proposed methodology extracts information from archaeological remains and the landscape; second, it extends the scope of topics considered by the field of mixed method research. The methodology was applied to the study of identity evolution in the Romano-British period in Shropshire. The results illustrate the potential of the proposed methodology to understand the formation/evolution of past identities.

Keywords

mixed method Roman Britain identity formation landscape archaeology

Introduction

The objective of this article is to present a mixed method approach, which uses archaeological material cultural remains in the landscape, and has the potential to identify the heterogeneous identities which evolved in the Romano-British period in Shropshire as a result of a culture clash between different societies. The research consists of a sequential design with an initial quantitative phase based on network theory, followed by a qualitative phase based on phenomenology (for details, see the methodological section below).

The quantitative stage imposes a network structure in the landscape where archaeological sites correspond to nodes, and similar material cultural remains in different sites are assumed to be links. That is, two sites are linked when they share a significant number of material culture indicators. Using standard mathematical algorithms developed in network theory, sites in this network can potentially be separated into clusters which are interpreted as broad identities. This stage also uses Digital Elevation Models (DEM), where clusters are plotted to explore potential relationships between the identified identities and the landscape at the large scale. On the other hand, the qualitative stage is based on phenomenology and was implemented to explore possible relationships between the identified identities and the landscape on a small scale (i.e., the surrounding landscape).

There are two main benefits to integrating both the network approach and phenomenology in landscape archaeology. First, the proposed mixed method allows researchers to collect complementary information from different scales of analysis in terms of the landscape. Second, the proposed methodology has the ability to identify patterns without imposing, ex-ante, hypotheses or associations between material culture that could potentially introduce bias in the results. In contrast to traditional research, the resulting network models are generated endogenously from data on nodes and links without imposing any previous belief or expectations about material culture remains or site distribution.

The proposed methodology was adopted to answer the following research questions: Can different identities be identified from cultural material remains available in the landscape? Is there a relationship between features of the rural landscape and the possible identities detected from such cultural material remains?

To address these questions, the methodology was applied to a case study of the identity formation during the Romano-British period in Shropshire, which was brought about by the Roman invasion of Britain in AD 43 under Emperor Claudius, and was an event that forced native societies to react and adjust to the invaders.

It is important to highlight that the proposed methodology is novel with regard to both mixed methods research and landscape archaeology. How it is novel in each of these fields is covered in the relevant sections below.

Contribution to the Field of Mixed Methods

The proposed methodology contributes to the field of mixed methods research in two main areas. First, it builds on traditional approaches that generally involve participants responding to questionnaires and interviews to collect data; such an approach is clearly not applicable to the field of archaeology. In the proposed mixed method, data and other relevant information are extracted from material culture remains in sites distributed in the landscape as well as the surrounding local landscape. Second, the proposed mixed method extends the scope of topics considered by the field of mixed method research by including the research in landscape archaeology, defined by David and Thomas (2008) as “an archaeology of how people visualized the world and how they engaged with one another across the space, how they chose to manipulate their surroundings or how they were subliminally affected to do things by way of their locational circumstances” (p. 38).

Contribution to Landscape Archaeology

Currently, the research in landscape archaeology is based on two competing approaches. One is a quantitative approach referred to as processual archaeology, which consists of the application of scientific principles to the interpretation and explanation of past changes and culture processes (Renfrew & Bahn, 2000). The other is the phenomenology of Martin Heidegger, which is concerned not only with consciousness, but also with the complex relationships between people and things constituting a human world understood as a structure of intelligibility (Frede, 1993; May, 2023). This phenomenology is used by landscape archaeologists to propose the idea that direct experience with the landscape through embodiment can inform about past evidence from the past (Hamilton et al., 2006; Thomas, 2006; Tilley, 2010).

The proposed methodology contributes to Landscape Archaeology by linking these competing approaches into a mixed method with the purpose of finding past identities and their evolution. In relation to this contribution, it is important to highlight that qualitative and quantitative analyses have already been combined in archaeology since its inception in several investigations by integrating visual description and other qualitative methods with quantitative measurement (see Binford, 1978; Hamilton et al., 2006; Maxwell, 2016; Richards-Rissetto, 2017). However, they have been used to explore areas other than identity formation, which is the focus of the proposed methodology.

As explained in the introduction, this methodology was tested by means of a case study on identity formation during the Romano-British period in Shropshire. To contextualize this case study, a brief literature review of the archaeology of Shropshire in the period is presented in the next section.

The Archaeology of Roman Britain in Shropshire

Traditionally, research on Roman Britain in Shropshire focused on Wroxeter Roman city with excavations dating from the 19th century. However, little was known about the area surrounding this city, apart from a few studies providing cropmark typologies from aerial reconnaissance (Buteux et al., 2000).

A major change in this research trend was the introduction of the Wroxeter Hinterland Project (WHP) in 1994, a long-term research project which aimed to study the relationship between Wroxeter Roman town and its hinterland (Gaffney et al., 2000; Gaffney et al., 2007). The results from this project revealed that Wroxeter actually exercised a strong influence on the surrounding area because it directed communal wealth into the town suggesting the presence of a strong and centralized elite (Buteux et al., 2000). The project also found evidence relating to the wider landscape, based on buffers in GIS around -ley suffixed placenames, suggesting both the existence of woodlands in Roman periods, and that settlements tended to be located in unwooded areas (White, 2022).

While Roman rural settlements remain largely unexplored in Shropshire, some archaeological evidence suggests that over the ensuing centuries, societies in the proximity of urban areas and roadside settlements were influenced by Roman ways to some extent. However, the lack of Roman material culture in areas located away from urban and road settlements suggests lack of interest in adopting Roman ways in the rural population, or lack of resources to afford them (White et al., 2013; White & Wigley, 2018).

Methodology

This section presents the proposed methodology. Because it was designed to identify past identities, it starts by outlining the concept of identity adopted in this research. After that, a formal description of the methodology is provided. Finally, this section finishes by presenting an example to illustrate how the network for the Later Roman period was created. The aim is to show explicitly the computations that are needed to create the network.

The Concept of Identity

The notion of identity used in the current investigation is related to collective identities: identity reflects the role of individuals within a collective group and how they perceive themselves as belonging to certain groups and not others (Jenkins, 2008). According to this concept, social identity is multidimensional and can be seen as a composite of individual identity types. Identity types, in turn, can be linked to material culture by means of specific social behavior or habits (Eckardt, 2014; Swift, 2017).

While this concept of social behavior as a mediating variable is useful to link material culture to collective identity, it has the disadvantage that past social behavior/habits cannot be directly observed. Nonetheless, in recent research, some archaeologists have related lifestyles and activities to material culture suggesting that the former may be considered as proxies of behavior; this is the focus that has been adopted in this investigation (Eckardt, 2014; Millett, 2016).

In this extension, collective identity can potentially be identified by linking lifestyles and activities to existing material culture. In considering these ideas, twelve broad categories related to lifestyles/activities have been identified from the literature (see Allason-Jones, 2011; Brindle, 2018; Eckardt, 2014; Ferris, 2012; Herrle, 2008; Millett, 2016; Pearce, 2016; Smith et al., 2018): building lifestyle, household lifestyle, kitchen activities and food preparation, transport, literacy, security, defense, recreation, personal appearance, personal grooming, religion, and economic activities.

In considering these lifestyles/activities, two sites are said to belong to the same collective identity when both of them have similar material culture indicators reflecting similar lifestyles/activities.

The Proposed Mixed Method Research

The proposed mixed method follows a sequential design consisting of the following phases.

Phase 1: Network Analysis

As outlined in the introduction, the network approach is used to analyze quantitative data with the objective of identifying possible social identities. In this approach, nodes correspond to archaeological settlements during the Roman period in Shropshire and weighted links (i.e., connections between sites with values between zero and one) correspond to similar material culture indicators (i.e., if two sites are connected by a link with a value of 0.7, then they share 70% of the material cultural indicators considered in the research). Alongside this, the modularity class indicator (a measure of the structure of a network that was designed to indicate the extent to which a network could be divided into modules or clusters) was used to identify groups or clusters of sites, and the strength indicator was employed to identify dominant sites in the network. These indicators are described as follows.

The modularity class indicator is used to determine clusters of relatively similar nodes that in the current research would represent broad identity classes. Clusters that are estimated by Gephi software (i.e., the software that was adopted in the current investigation) use the Louvain optimization algorithm consisting of an iterative algorithm that is based on two equations (Agarwal, 2012). This approach involves estimating a modularity value for each node by means of one of these equations under the assumption that each node belongs to a single class. After that, gains in modularity values are explored by means of the second equation by assuming that determined nodes belong to the same class. If there is an increase in the value of modularity in these nodes, then it is assumed that they belong to the same class. In contrast, if modularity values decrease, then the opposite conclusion is drawn. This process is repeated iteratively until no gain in modularity values is possible.

A refinement of existing modularity algorithms was introduced by Lambiotte et al. (2008) which consists of a dynamic approach that uses a resolution parameter to determine the number of clusters in a network. It is based on the idea that the size of stable clusters increases with time, and time can be considered in this context as a resolution parameter. Thus, by selecting a specific value of the parameter, a number of clusters can be obtained depending on the structure of the network.

The strength indicator, on the other hand, is a measure of the involvement of a node in the network and is estimated by adding the weight of its links. A node with a high strength value means that it is highly connected to other nodes in terms of the number of links and the weight in these links (Opsahl et al., 2010).

To facilitate the understanding of the strength and modularity class indicators, consider the following example in Figure 1.

This particular network can be described in terms of strength and modularity class. For example, the strength of node i is 1.4 (i.e., the sum of the weight of its links: 0.3 + 0.4 + 0.7). In this example node k is the node with the highest strength (i.e., a value of 2.3) implying that this node is relevant and central in the network because it directly and indirectly bridges a number of nodes. In contrast, node m is the node with the lowest strength (i.e., a value of 0.2) implying that the structure of the network would not strongly be affected if this node was deleted. Regarding modularity class, this network is composed of two clusters. One of them is composed of nodes i, j, k, l, and m. The second cluster is formed of nodes u, v, and w. As illustrated in this example, it is possible to characterize this network both in terms of groups of nodes that belong to the same cluster, and nodes that have an important role in bridging different areas of the network.

To create links between nodes, an indicator of similarity is proposed. It consists of a number of indicators related to material culture items that are outlined in the Rural Settlement of Roman Britain Online Resource of the Archaeological Data Service (available at https://archaeologydataservice.ac.uk/archives/view/romangl/). An inspection of this database led to the conclusion that some indicators are not useful for the network analysis because this information is fragmented across several sites. In considering this limitation, 41 indicators were employed to estimate weighted links across sites. Following the criterion described above for the concept of identity, these indicators were classified according to the twelve categories reflecting lifestyles/activities (see Table A1).

To assess the possible evolution of these identities, two networks were created: one for the earlier Roman Period (i.e., from 43AD to 250AD) and one for the later Roman period (i.e., from 250AD to 450AD). Differences in these networks were considered as the basis to explore identity evolution.

Finally, the sites in the clusters were plotted in Digital Elevation Models (i.e., maps that contain information about topography, land elevation, rivers, and ancient Roman roads) with the purpose of identifying possible relationships between the identified identities and the landscape at the large scale.

Phase 2: Phenomenology

The analysis was complemented with a phenomenological approach to landscape archaeology with the purpose of determining whether some of the features of the resulting network model are related to features of the surrounding landscape (i.e., at the small scale). Phenomenology was applied on selected key sites that were identified as relevant according to the results obtained from the network analysis in Phase 1.

According to the phenomenological approach, direct experience with the landscape can inform about past interpretations because the landscape and humans are intimately related in a bidirectional or reciprocal relationship that can be understood through embodiment. That is, landscapes provide affordance for living as well as a number of constraints that not only influence social activities, but also individuals’ perceptions of the world (Tilley, 2010).

In considering this view, affordance is seen in this article not only as available resources that facilitate economic, social, and symbolic expressions, but also as resources that facilitate the development of traditions and skills contributing, in this way, to the formation of social identities over long periods of time. For this research, therefore, affordance is analyzed by means of two indicators: physical affordability and visibility affordability. The former was introduced to determine whether the visited sites had available natural resources in the proximity. Visibility affordability, on the other hand, was considered as a type of affordance that inform about the ability to be seen from the surrounding landscape. Visibility affordability is relevant because there is evidence that high-status settlements in Roman times were concerned about being seen for commercial/marketing purposes (Bain & Evans, 2011). It is also argued that other high-status settlements such as villas were located in areas of high visibility enabling them to send messages of status, wealth, and power (Purcell, 1995).

Five principles of phenomenology research in landscape archaeology proposed by Tilley (2008) were considered for Phase 2: embodiment (the phenomenologist’s body was used as the primary tool of research in the selected sites); places and paths (each visit included a search for possible places and paths, geological and topographic features, as well as archaeological remains); sensory experience (local visibility was considered for the current investigation); poetics and the body (phenomenology was carried out in the surrounding landscape of each selected site because of one’s own bodily limitations); and nature and culture (sensory affordance and constraints were analyzed in prominent features in the surrounding landscape of each selected site). These principles were selected because the activities involved in them are physically feasible for a researcher to undertake.

According to Bailey (2005), a radius of 5 km is believed to be the maximum radius of daily exploitation for agriculture, and a radius of 3 km of more intensive economic activity around a site. After visiting some sites, it was found that embodiment engagement with the landscape is difficult for these radii because of the physical effort and distance involved. In considering this problem, it was found that direct embodiment engagement is feasible for a radius of 1 km around the site. A larger radius of 3 km was also considered, but the features contained at that distance were only analyzed visually and with the use of a vehicle.

It is important to clarify that there are a number of processual researchers who criticize the use of phenomenology in landscape archaeology arguing that it may reflect only accidental associations indicative of hyper-interpretations of archaeological evidence (Brück, 2005; Fleming, 2006; Hamilton et al., 2006). The current research acknowledges these limitations and accepts that the results obtained in this investigation should be considered with caution. Nonetheless, the results were reviewed by two experts in the field.

The Sampling Strategy

The Rural Settlement of Roman Britain Online Resource of the Archaeological Data Service contains a list of rural settlements with archaeological data related to the Roman period in Shropshire reported in published academic works as well as grey literature. 21 sites with reliable information in the database were considered in the research (see Table A2 in the appendix). Other sites were omitted because their archaeological information is very fragmented. The selected sites were classified as either Early or Later Roman by identifying the main period of occupation from primary sources. This made it possible to construct cultural networks for both periods.

For Phase 2 (i.e., phenomenology), a small number of cases in each cluster were selected using the ego-network approach which involves the selection of key nodes of the network (Edwards & Crossley, 2009). These were studied intensively using a purposive sampling technique referred to as Diverse Cases technique (Teddlie & Yu, 2007). This technique consists of achieving the maximum variance along relevant dimensions. In the current research, the relevant dimension is the node strength indicator because it indicates the degree of connectivity of a site in the network. Following this idea, sites with high, medium, and low connectivity were selected for each cluster.

Example: How the Network for the Later Roman Period was Created

To facilitate understanding of the network analysis, some of the data and calculations made in this research for the network representing the later Roman period are described as follows. The Rural Settlement of Roman Britain Online Resource of the Archaeological Data Service contains two types of data: (a) data informing about the presence of some material culture remains (e.g., masonry, tiled roof, and inscriptions) and (b) data that inform about the quantity of a determine material culture remain (e.g. coins, finger rings, and agricultural tools). The first group of data was processed as binary variables with a value equal to one if the archaeological remain is present, and zero otherwise. On the other hand, the data that contains quantified remains were represented as categorical variables using the interquartile criterium. For example, if a site with a determined number of coins belongs to the first quartile, then this site is assigned a number one to that quartile and zero to the others. These quantities were divided by the area under excavation (in hectares) to make sites comparable (Brindle, 2017). The distribution of coins and brooches was also considered to capture the most frequent types of these remains in the sites. Sites with similar distribution were considered as being similar with respect to this criterion. Using this information, it was possible to create weighted links between sites. As an example, consider the Later Roman sites at Tern Bridge, Wroxeter, and Yarchester in Table 1:

Table 1.

Similarities and Differences Between the Sites at Tern Birdge and Yarchester.

Material culture	Tern Bridge	Yarchester	Similarity
Coins distribution	Group 1	Group 2
Coins quantity	Quartile 1	Quartile 3
Building material: Masonry	No	Yes
Tiled roof	No	Yes
Painted plaster wall	No	Yes
Structural worked stone	Yes	Yes	Yes
Ceramic building material	No	No	Yes
Window glass	No	No	Yes
Tessellated floor mosaic	No	Yes
Hypocaust	No	Yes
Corndrier	No	No	Yes
Storage	No	No	Yes
Bathhouse	No	No	Yes
Evidence metal-work	No	No	Yes
Iron slag	No	No	Yes
Evidence bone-work	No	No	Yes
Hobnails	No	No	Yes
Bracelets quantity	Quartile 1	Quartile 1	Yes
Finger rings	Quartile 1	Quartile 1	Yes
Hairpins	Quartile 1	Quartile 1	Yes
Other dress accessories	Quartile 1	Quartile 1	Yes
Brooch distribution/type	Group 1	Group 1	Yes
Brooch quantity	Quartile 1	Quartile 1	Yes
Toilet cosmetic	Quartile 1	Quartile 1	Yes
Military fitting weapons	Quartile 1	Quartile 1	Yes
Inscriptions	No	No	Yes
Writing equipment	Quartile 1	Quartile 1	Yes
Equine transport	Quartile 1	Quartile 1	Yes
Religious objects	Quartile 1	Quartile 1	Yes
Knives tools	Quartile 1	Quartile 1	Yes
Agricultural tools	Quartile 1	Quartile 1	Yes
Security objects	Quartile 1	Quartile 1	Yes
Recreation objects	Quartile 1	Quartile 1	Yes
Weighing objects	Quartile 1	Quartile 1	Yes
Spinning	Quartile 1	Quartile 1	Yes
Weaving	Quartile 1	Quartile 1	Yes
Sewing	Quartile 1	Quartile 1	Yes
Quernstones	Quartile 4	Quartile 1
Millstones	No	No	Yes
Glass objects	Quartile 1	Quartile 1	Yes
Vessel glass	Quartile 1	Quartile 1	Yes
			Total = 33

The first column of this table presents the 41 indicators of material culture considered in this research. The second and third columns show the categories to which these sites belong with respect to these indicators. Finally, the last column shows the similarities between the two sites in terms of either common archaeological remains or remains that are not present in both sites. According to this column, the sites at Tern Bridge and Yarchester have 33 similarities from the 41 indicators. That means that these sites have a weighted link between each other equal to 0.80 (i.e., 33/41) implying 80% of similarity.

Using this approach, links between all the sites in the sample were created. For the case of the Later Roman period, these links are shown in the following table.

The weighted links presented in this table are the information that is needed to run the Gephi software. Once this information is introduced, the software estimates the modularity class indicator using the mathematical algorithms described above. Using this approach, two clusters were identified by the software which, according to the current research, represent different identities. One of these clusters is formed of the sites at Meole Brace, Whitley Chapel, Lea Cross, Yarchester, and Whitley Grange; and the other cluster by the sites at Tern Bridge, Nesscliff Hill, Ellesmere Road, Sheinton, Bridgwalton Quarry, and Chilton Farm (a formal analysis of these clusters is provided in the results section).

For the strength indicator of a particular site, on the other hand, all the weighted links of this site must be added. The strength indicator for each site is calculated using the information in Table 2 and the results are shown in Table 3.

Table 2.

Weighted Links Between the Later Roman Sites.

	Meole Brace	Whitley Chapel	Tern Bridge	Lea Cross	Nesscliff Hill	Yarchester	Whitley Grange	Ellesmere Road	Sheinton	Bridgwalton Quarry	Chilton Farm
Meole Brace	—
Whitley Chapel	0.51	—
Tern Bridge	0.46	0.93	—
Lea Cross	0.49	0.85	0.78	—
Nesscliff Hill	0.44	0.85	0.88	0.73	—
Yarchester	0.44	0.88	0.80	0.83	0.80	—
Whitley Grange	0.49	0.68	0.63	0.66	0.61	0.76	—
Ellesmere Road	0.44	0.93	0.95	0.78	0.93	0.80	0.61	—
Sheinton	0.44	0.88	0.90	0.73	0.88	0.76	0.56	0.95	—
Bridgwalton Quarry	0.46	0.90	0.93	0.76	0.90	0.78	0.59	0.98	0.93	—
Chilton Farm	0.44	0.88	0.90	0.73	0.95	0.83	0.61	0.95	0.90	0.93	—

Table 3.

Strength Indicator for the Sites in the Later Roman Period.

Meole Brace	Whitley Chapel	Tern Bridge	Lea Cross	Nesscliff Hill	Yarchester	Whitley Grange	Ellesmere Road	Sheinton	Bridgwalton Quarry	Chilton Farm
4.61	8.29	8.17	7.34	7.98	7.68	6.20	8.32	7.93	8.15	8.12

As can be seen in this table, the site with the highest strength indicator value is Ellesmere Road, and this means that this is the most integrated site in the network. In contrast, Meole Brace has the lowest value suggesting that this site has features that are not common in the network.

The strength indicator was also employed to select the sites in each cluster that were to be visited for phenomenological analysis (i.e., the ego-network approach described above). This is the connection between the quantitative and qualitative phases. That is, for each cluster three sites were selected: the one with the highest strength indicator among the sites in the cluster; the one with the lowest indicator value, and a site with a value in between. For example, the selected sites in one of the clusters are Lea Cross with the highest strength indicator (i.e., 7.34) in that cluster; Meole Brace with the lowest indicator in this cluster (i.e., 4.63); and Whitley Grange with an intermediate value (i.e., 6.22). This criterion was useful to ensure that the explored sites have reasonable variance between them in terms of material culture remains, making it possible to identify some patterns.

The results given in this section have been presented for illustrative purposes. However, a deeper description of these results is presented in the results section below.

Results and Discussion

This section presents the results from Phase 1 and 2 of the current investigation considering first the Earlier Roman period and then the Later Roman period.

Results for the Earlier Roman Period

This part presents the results from the quantitative and qualitative analyses for the Early Roman period.

Results from the Network Analysis

Ten sites of the 21 in the sample were classified as Earlier Roman settlements. The resulting network model that was created with these sites contains only a single cluster (Cluster 1) suggesting the existence of a single identity in this area during this period. This model is shown in Figure 2 (numbers indicate the ID of the sites as presented in Table A2, and letters in brackets the type of sites: nucleated settlement (N); farms (F); industry (I); and villa (V)).

According to this figure, all the sites in the cluster occupy areas of relatively low elevation (i.e., green areas). Some of them are not far from more elevated hilly areas (i.e., red areas). What these sites have in common, is that they are located in the proximity of Roman roads and rivers suggesting that these sites were highly connected through road networks at that time, and also took advantage of available watercourses.

In terms of material culture, the cluster includes a small percentage of sites containing high status building remains, and there is also evidence of metal work. However, a significant number of the material culture indicators considered in the research are not present or they exist in very small quantities which is a marked regional characteristic. This may account for the algorithm being unable to extract different clusters. This can also be inferred from the information presented in Tables 4, 5, and 6.

Table 4.

Weighted Links Between the Sites in Cluster 1.

Site (ID in brackets)	Duncote Farm	Craven arms	Hay Farm	Heath Road	Sharp-stones Hill Site E	Ismore Coppice Wroxeter	Linley hall	Oak street head	Redhill, telford
Duncote Farm (30003(F))
Craven Arms (30004(F))	0.93
Hay Farm (30005(F))	0.93	0.95
Heath Road (30006(N))	0.88	0.95	0.90
Sharpstones Hill Site E (30012(F))	0.88	0.88	0.93	0.83
Ismore Coppice (30016(I))	0.90	0.93	0.98	0.88	0.90
Linley hall (30020(I))	0.83	0.85	0.90	0.80	0.83	0.88
Oak Street Head (30026(F))	0.90	0.93	0.98	0.93	0.90	0.95	0.88
Redhill, Telford (30028(N))	0.76	0.78	0.73	0.73	0.71	0.71	0.73	0.71
Alkmund Park (30030(V))	0.61	0.61	0.66	0.56	0.61	0.68	0.76	0.63	0.59

Table 5.

Strength Indicator Values of Cluster 1 (Higher and Lower Values in Grey).

Table 6.

Material Culture in Cluster 1 (Grey Indicates Similarities Across Sites, and Empty Cells Means That the Material Culture Remain is Not Present in a Site).

According to Table 4, the sites of this network present some degree of homogeneity with similarity ranging from 56% to 98%. As can be seen in Table 6, this is explained by a common feature of the network in general which is the lack of a number of archaeological remains. In spite of this, there is also variance across sites as can be inferred from the strength indicator (see Table 5). In fact, the most integrated site is the one at Whitley Chapel and this is because this site does not have material culture remains at all. In contrast, the site at Alkmund Park is the least integrated as this is the site with the highest number of material culture remains implying that this site contains remains that are not common in this network.

Results from the Phenomenological Analysis

The results from the network analysis revealed that Shropshire in the Earlier Roman period was characterized by a single cluster (i.e., Cluster 1). This cluster was explored by means of phenomenological analysis considering three heterogeneous sites in terms of the strength indicator (SI) value: Whitley Chapel (SI = 7.96); Redhill, Telford (SI = 6.45); and Alkmund Park, Berwick (SI = 5.71).

Although the selected sites belong to the same cluster, they are by no means identical. They differ in terms of both settlement type and specific material culture remains identified. In particular, the site at Whitley Chapel is a farm with few material culture remains, evidence that has been used to suggest a low status native dwelling (Howlett, 1992). The site at Redhill, on the other hand, corresponds to a nucleated settlement that is richer in material culture. Finally, the site at Alkmund Park is a villa richer in material culture than the previous sites, as indicated by the even lower value, suggesting a higher status settlement in this region.

The results from the phenomenological research revealed additional possible patterns in terms of similarities and differences between these sites. In terms of similarities, it was found, as expected for people depending on agricultural production, that the sites are located in areas with easy to moderate access to farmlands and local natural resources. In relation to differences, on the other hand, local visibility seems to be linked to the strength indicator and status of the sites in this cluster. That is, the low status site (i.e., Whitley Chapel) with low material culture (i.e., higher values of SI) is found in an area of restricted local visibility. In contrast, the higher status site (Alkmund Park) contains several material culture remains associated with different lifestyle types suggesting greater Roman influence (Brindle, 2018). This site is highly visible from much of the surrounding landscape. Finally, the site at Redhill occupies an intermediate position in relation to these considerations. A possible explanation for this apparent relationship is that people in higher status sites wanted to demonstrate this status by placing the settlement in areas with significant local visibility. This need for signalling status may have served as a way to reflect wealth and control over the surrounding landscape, or to control the work of slaves/tenants. A similar conclusion was drawn by Bain and Evans (2011) for the site at Sharpstones Hill Site E/Meole Brace who point out that display was an important factor in the location of the site.

Results From Phases 1 and 2 for the Later Roman period

This subsection presents the results from the quantitative and qualitative analyses for the Later Roman period.

Results from the Network Analysis

Eleven sites were found suitable to develop a network for this period. The resulting model is composed of two clusters (i.e., Clusters 2 and 3). They are presented in Figure 3.

A similar pattern to that described in Figure 2 was found in Figure 3. However, sites in Cluster 2 (in green), composed mainly of villas, seem to be more connected through the road network than the sites in Cluster 3 (in red), formed mainly of farms. This higher connectivity may be related to the fact that Cluster 2 contains sites that are richer in material culture, suggesting the need for more connectivity for commercial purposes. This is explored in more detail in the phenomenological analysis below.

Cluster 2 is generally rich in several material culture indicators considered in this investigation suggesting a set of relatively wealthy settlements. In contrast, Cluster 3 is relatively poor in material culture, although it contains some sites with evidence of material culture such as brooches, quernstones and hobnails, among others. This is consistent with existing evidence revealing lack of Roman material culture in areas away from urban areas suggesting lack of interest in adopting Roman ways amongst the rural population (White & Wigley, 2018). A deeper exploration of these clusters is presented as follows.

(a) Cluster 2: The main characteristics of this cluster are summarized in Tables 7, 8, and 9.

Table 7.

Weighted Links Between the Sites in Cluster 2.

Site (ID in brackets)	Meole Brace	Whitley Chapel	Lea cross, Pontesbury	Yarchester
Meole Brace (30001 and 30007(N))
Whitley Chapel (30015(F))	0.51
Lea Cross, Pontesbury (30018(V))	0.49	0.85
Yarchester (30021(V))	0.44	0.88	0.83
Whitley Grange (30022(V))	0.49	0.68	0.66	0.76

Table 8.

Strength Indicator Values of the Sites in Cluster 2 (Higher and Lower Values in Grey).

Table 9.

Material Culture in Cluster 2 (Grey Indicates Similarities Across Sites, and Empty Cells Means That the Material Culture Remain is Not Present in a Site).

While this cluster is composed mainly of villas, it is relatively heterogeneous in material culture ranging from 44% to 88% in terms of similarity between the sites (Table 7). This is explained by the fact that there are several archaeological remains that are not shared by all the sites (Table 9). In fact, the only common remains are some associated with high-status building (e.g., masonry, tiled roof, and structural worked stone).

Regarding the strength indicator (Table 8), the site with the highest value for this indicator is the site at Whitley Chapel. Its higher connectivity with the network is explained by the observed lack of material culture remains other than building material. On the other hand, the site with the lowest strength indicator is the site at Meole Brace. Its low connectivity is due to the fact that this site contains remains that are rare in the network (see Table 9).

(b) Cluster 3: Information relevant to this cluster is summarized in Tables 10, 11, and 12.

Table 10.

Weighted Links Between the Sites in Cluster 3.

Site (ID in brackets)	Tern Bridge, Wroxeter	Nesscliff Hill	Ellesmere Road, Shrewsbury	Sheinton	Bridgwalton Quarry
Tern Bridge, Wroxeter (30017(I))
Nesscliff Hill (30019(F))	0.88
Ellesmere Road, Shrewsbury (30023(F))	0.95	0.93
Sheinton (30027(F))	0.90	0.88	0.95
Bridgwalton Quarry (30029(F))	0.93	0.90	0.98	0.93
Chilton Farm, Atcham (30031(F))	0.90	0.93	0.95	0.90	0.93

Table 11.

Strength Indicator Values of the Sites in Cluster 3 (Higher and Lower Values in Grey).

Table 12.

Material Culture in Cluster 3 (Grey Indicates Similarities Across Sites, and Empty Cells Mean That the Material Culture Remains are Not Present in a Site).

This is a very homogeneous cluster with similarities between 85% and 98% (Table 10). This homogeneity is explained by a general lack of material culture in all the sites in the cluster (Table 12). In relation to the strength indicator (Table 11), the site at Ellesmere Road has the highest value, and this high connectivity within the network is because this site does not contain any of the material culture considered in this study. In contrast, the site at Sheinton has the lowest connectivity and this is because this site contains brooches which are uncommon in this network.

Results from the Phenomenology Analysis

The clusters identified in the network analysis were studied from the phenomenological approach by means of the ego-network approach (see the methodology). For Cluster 2, the selected sites are: Meole Brace (SI = 4.63); Lea Cross, Pontesbury (SI = 7.34); and Whitley Grange (SI = 6.22). For Cluster 3, the selected sites correspond to: Sheinton (SI = 7.93); Bridgwalton Quarry, Morville, Bridgnorth (SI = 8.16); and Chilton Farm, Atcham (SI = 9.10). The results from the phenomenological approach are presented as follows.

The sites at Meole Brace, Lea Cross, and Whitley Grange of Cluster 2 have significant similarities in terms of the surrounding landscape. First, all of them are wealthy settlements in areas with a significant expanse of flat land. Second, all of them had some access to natural resources such as watercourses and woodland resources. Finally, these sites can be seen from much of the surrounding landscape suggesting that people of these settlements wanted to be seen. In the case of the site at Meole Brace, this high degree of visibility may be explained by the services and marketing activities that they offered to travelers to and from Wroxeter (Evans et al., 1999). In the case of the sites at Lea Cross and Whitley Grange, on the other hand, high visibility may reflect signal of ownership, landscape control, slave control, and/or status. Signalling status is particularly a possibility for Whitley Grange which, according to current evidence, seems to have been used as a hunting lodge by wealthy elites, a practice that was common in the later Roman period (White, 2022).

In spite of these similarities, there are also some differences that are captured by the strength indicator as discussed by the network analysis for this period.

The sites in Cluster 3 are also located in areas with access to farmlands and natural resources such as forests and watercourses/rivers. However, two of the sites in this cluster, namely, Sheinton and Bridgwalton Quarry, differ from those in Cluster 2 in that there is significantly less available farmland in the surrounding landscape. This may explain why few material cultural remains have been identified in these sites. That is, people in these settlements were probably less able to generate agricultural surplus for trade activities and used farmlands mainly for self-subsistence. In addition, visibility beyond the area considered for phenomenological analysis in the sites at Sheinton and Bridgwalton Quarry is prevented by local features such as hills and lands with irregular slopes. This suggests that people in these settlements were less concerned about signalling status and control over the landscape.

The site at Chilton Farm was included as part of Cluster 3 by the network framework because few material cultural remains were identified in this site. However, its surrounding landscape is more consistent with the landscape of the high-status sites of Cluster 2. That is, there is a significant amount of flat farmland which could produce agricultural surplus, and the site can be seen from a long section of the landscape along the River Severn suggesting that these people wanted to be seen from commercial routes associated with the river. The discrepancy between lack of material culture and the opportunity offered by the local landscape suggests that rich material culture in this site has not been identified yet, probably because only two small trenches were excavated. Actually, five Samian sherds (i.e., a pottery that was adopted by high-status people in Roma times) have been found in this site reinforcing the idea that it may have been a high-status settlement.

In summary, the phenomenological analysis for the Later Roman period revealed a similar pattern to the analysis for the Earlier Roman period, but between the two clusters in the network for this period rather than sites within the clusters: high-status settlements of Cluster 2 tend to be placed in areas of good visibility; and low-status settlements of Cluster 3 in areas of low visibility.

Conclusions

The objective of this article is to present a mixed method approach which was designed to identify heterogeneous past/ancient identities and their evolution as a result of culture clash between different societies using archaeological remains in the landscape. It consists of a quantitative phase based on network analysis that was designed with the purposes of identifying clusters of sites reflecting possible identities, and possible relationships between these identities with the landscape at a large scale. It is followed by a qualitative phase based on phenomenological analysis that was designed to get a better understanding of the relationship between selected sites and their surrounding landscape or landscape at the small scale. This methodology was applied to explore the case of the Roman invasion of Britain in the Shropshire region.

The results obtained in this investigation provide evidence that the proposed methodology is effective. That is, analysing archaeological information present in the rural landscape using the network approach not only permitted the identification of a possible identity in the Early Roman period, but also its evolution into two different identities in the Later Roman period: one of them was low-status with low level of wealth; and the other identity developed in wealthy sites with different degrees of Roman influence.

In relation to the landscape at a large scale, the results revealed that sites of the high-status identity were placed near Roman roads suggesting the need to be integrated into the commercial network in the area. Regarding the surrounding landscape, on the other hand, one of the key results from the phenomenological analysis for the Early Roman period, is that high-status sites with more abundant material culture are in places with good local visibility. This tendency was also detected in the Later Roman period. That is, sites in the richer cluster tended to be located in areas of good visibility suggesting that they wanted to be seen with the purpose of signalling status as a way to reflect wealth and control over the surrounding landscape, or as a way of controlling slaves/tenants who needed to be supervised. In contrast, the sites in the poorer cluster were placed in places with lower visibility. This result illustrates the benefit of using phenomenology in the proposed methodology: it informs about specific attributes of the landscape that are linked to specific identities. These results suggest a polarization that is consistent with the nativist model (i.e. a polarization between Roman and native identities. For a discussion, see Webster, 2001; Hingley, 2005) because there is a clear difference in material culture between the two clusters, and differences in relation to the surrounding landscape.

The results obtained from the proposed methodology are consistent with the findings of alternative approaches implying that the proposed mixed method is robust. Consider for example the Wroxeter Hinterland Project. According to Gaffney et al. (2007), the Cornovii people in Wroxeter’s surrounding countryside seems to have rejected Roman ways given the absence of material culture used by people living in urban areas. In contrast, Wroxeter seems to have been the place of the Cornovian elite who adopted Roman ways through the manipulation of pre-existing social and power relationships. This rural-urban and native-Roman lifestyle duality is consistent with the results obtained from the proposed methodology.

It is important to highlight the results obtained from this research must be considered with caution because some limitations might have introduced potential biases. First, the network analysis omitted some relevant material culture indicators because this information is fragmentary in several sites. Second, the information for Shropshire is relatively poor and it was only possible to obtain samples from a small number of sites. It is important to clarify that the possible biases listed above reflect mainly limitations of data rather than limitations of the methodology itself. Consequently, more robust results may be obtained when more reliable data is available in the future.

Footnotes

Acknowledgments

The author wishes to thank Dr. Alison MacDonald and Dr. Alex Smith for their useful suggestions at different stages of the research, and to the anonymous reviewers and editors for their observations that improved the manuscript significantly.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Daniel E. May

Appendix

Table A1.

Criteria to Classify Material Culture Indicators Into Lifestyle/Activity Types.

Material culture	Justification
Coins distribution and quantity.	Economic activities (commerce and trade), payment for services and household lifestyle because coins may reflect wealth.
Building material, tiled roof, painted plaster wall, structural worked stone, ceramic building material (CBM), window glass, tessellated floor mosaic, and hypocaust.	Building lifestyle related to building material, architectural elaboration and Roman influence of house architecture.
Corndrier (apparatus for drying solid materials).	Economic activity associated with crop processing.
Storage.	Economic activity related to cereal processing.
Bathhouse (baths used for recreation and relax).	Personal grooming and recreation because bathhouses were used for personal hygiene, relaxation and social relationship.
Evidence metal-work, iron slag (residuals of iron processing).	Economic activity related to metal production.
Evidence bone-work.	Economic activity because bone was used as raw material for the production of personal items, dressing, tools and recreation objects.
Hobnails (shoe tacks, a type of clinching nail).	Personal appearance because hobnails were used in boots.
Bracelets, brooches, finger rings, hairpins, other dress accessories.	Personal appearance because these objects were used as personal adornment.
Toiletry cosmetics (e.g., cosmetic spoons, grinders, and palettes).	Items used for personal grooming.
Military fitting weapons.	Items related to defense.
Inscriptions, writing equipment.	Intellectual activity reflected as literacy.
Equine transport.	Animals used for transport and military needs.
Religious objects.	Objects that appear to have been specifically manufactured for religious use.
Knife tools.	Used for domestic tasks and food preparation, and economic activity such as leather working and non-cereal crops harvesting.
Agricultural tools.	Economic activity related to food production.
Security objects.	Items used for security such as locks and keys.
Recreation objects.	Objects used for recreational activities (e.g. gaming counters).
Weighing objects.	Economic activity because they correspond to tools used for weighing goods during commercial transactions.
Fiber processing, weaving, spinning, sewing.	Economic activity related to textile production.
Quernstones (stone tools for hand-grinding different materials), millstones (stones used in gristmills, for grinding wheat or other grains).	Economic activity and food preparation because they were used for food production and grain processing.
Glass objects.	Personal appearance because these objects were normally used as personal adornment items.
Vessel glass.	Economic activity related to glass production and trade in glass, and domestic lifestyle (vessel glass was used for storage and serving food and drinks).

Table A2.

Sites Included in the Sample and References (Database ID in Brackets).

Site	References
Meole Brace (30001) and Duncote Farm (30003)	Ellis, P. (1991) A5/A49 Shrewsbury bypass archaeological project 1989-1990. Site narratives and post excavation research design. Birmingham: Birmingham University Field Archaeology Unit https://doi.org/10.5284/1026960
Meole Brace (30001) and Duncote Farm (30003)	Ellis, P. et al. (1994) Excavations in the Wroxeter hinterland 1988-1990: The archaeology of the Shrewsbury bypass, Shropshire Archaeology and History LXIX, 1-119.
Meole Brace, Pulley (30007)	Evans, C.J., Jenks, W.E. and White, R.H. (1999). Romano-British kilns at Meole Brace (Pulley), Shropshire, Shropshire Archaeology and History LXXIV, 1-27
Craven Arms (30004)	Ferris, I. (1991). An Archaeological Evaluation at Craven Arms, Shropshire, 1991. Birmingham: Birmingham University Field Archaeology Unit. Report No. 0157 https://doi.org/10.5284/1026962
Hay Farm, Eardington (30005)	Howlett, C. (1992) An archaeological evaluation at Hay Farm, Eardington, Shropshire. Oxford: Tempus Reparatum https://doi.org/10.5284/1026963
Hay Farm, Eardington (30005)	Hunn, J.R. (2000). Three Iron Age and Roman-British rural settlements on English gravels settlements British Archaeological Reports (British Series) 312 Archaeopress.
Heath Road, near Whitchurch (30006)	Hannaford, H. (1991) An Excavation at the Roman Roadside Settlement at Heath Road, near Whitchurch, 1991. Gifford and Partners https://doi.org/10.5284/1026965
Sharpstones Hill Site E/Meole Brace (30012)	Barker, P.A., Haldon, R. and Jenks, W.E. (1991). Excavations at Sharpstones Hill near Shrewsbury, 1965-71, Shropshire Archaeology and History LXXXVI, 15-57.
	Bain, K. (2007). Meole Brace,Shrewsbury. Archaeological Investigations 2005-6. Post-excavation Assessment. University of Birmingham: Birmingham Archaeology. Report No. 1283 https://doi.org/10.5284/1004573
	Bain, K. and Evans, C.J. (2011). A late Iron Age and early Romano-British enclosure at Meole Brace, Shrewsbury, Transactions of the Shropshire Archaeological and Historical Society 86, 121-151.
Whitley Chapel/Weir Meadow (30015)	Toms, G. (1985). Whitley Chapel and Weir Meadow - A Romano-British site near Shrewsbury, Shropshire Archaeology and History LXIV, 13-16.
Ismore Coppice, Wroxeter (30016)	Houghton, A.W.J. (1966). A Roman tilery and brickfield at Ismore Coppice, Wroxeter, Shropshire Archaeology and History 57, 7-12.
Tern Bridge, Wroxeter (30017)	Houghton, A.W.J. (1966) A Roman pottery factory near Wroxeter, Shropshire Archaeology and History 57, 101-111.
Lea Cross, Pontesbury (30018)	Houghton, A.W.J. (1961). A note on excavations at the Roman villa at Lea Cross during 1956-57, Shropshire Archaeology and History 56, 26-7.
Nesscliff Hill (30019)	Hume, C.R. and Jones, G.W. (1961). Excavations at Nesscliff Hill, Shropshire Archaeology and History 56, 129-32.
Linley hall (30020)	Page, W. (1908). No report. Information in The Victoria history of Shropshire: volume one, p. 257-8.
Yarchester (30021)	Information from Shropshire HER
Whitley Grange (30022), Alkmund Park (30030) and Chilton Farm, Atcham (30031)	Gaffney, V., White, R., Bevan, L. and Goodchild, H. (2007). Wroxeter, the Cornovii, and the urban process: final report on the Wroxeter Hinterland Project 1994-1997. Vol. 1, Researching the hinterland Journal of Roman Archaeology Supplementary series no. 68 Journal of Roman Archaeology.
Ellesmere Road, Shrewsbury (30023)	Anon (1996) Ellesmere Road, Shrewsbury, Shropshire. Archaeological Excavation. Oxford: Oxford Archaeological Unit https://doi.org/10.5284/1026969
Oak Street Head, Belle Vue, Shrewsbury (30026)	Hannaford, H. (1992). An Archaeological Evaluation at Oak Street Head, Belle Vue, Shrewsbury, Shropshire. Archaeology Service Shropshire County Council https://doi.org/10.5284/1026971
Sheinton (30027)	Hannaford, H. (2006). Archaeological Investigations at Sheinton, Shropshire, 2006. Archaeology Service Shropshire County Council. Report Number 244 https://doi.org/10.5284/1026976
Redhill, Telford (30028)	Andrew Mann. (2011). Archaeological evaluation and watching brief at Redhill Reservoir, Telford, Shropshire. Worcester: Worcestershire Archaeology. report 1886, revision 2 https://doi.org/10.5284/1019166
Redhill, Telford (30028)	Browne, D.M. and Boon, G.C. (2004). Excavation at Redhill, Lilleshall, Shropshire: an interim Report, Shropshire Archaeology and History LXXVII, 1-9.
Bridgwalton Quarry, Morville, Bridgnorth (30029)	Wood, M. (2013). An archaeological Evaluation by trial trenching: Proposed extension to Bridgwalton Quarry, Morville, Bridgnorth, Shropshire. Allen Archaeology Ltd. Report Number AAL2013056 https://doi.org/10.5284/1026979

References

Agarwal

(2012). Networks in nature: Dynamics, evolution, and modularity Doctoral thesis, University of Oxford. Oxford University Research Archive ORA. https://ora.ox.ac.uk/?utf8=%E2%9C%93&q=Networks+in+Nature%3A+Dynamics%2C+Evolution%2C+and+Modularity&search_field=all_fields

Allason-Jones

(2011). Artefacts in Roman Britain. Cambridge University Press.

Bailey

(2005). Site catchment analysis. In Renfrew

Bahn

(Eds.), Archaeology: The key concepts (pp. 172–176). Routledge.

Bain

Evans

C. J.

(2011). A late iron age and early Romano-British enclosure at Meole Brace, Shrewsbury. Transactions of the Shropshire Archaeological and Historical Society, LXXXVI, 121–151.

Binford

(1978). Nunamiut ethnoarcheology. Academic Press.

Brindle

(2017). Coins and markets in the countryside. In Allen

Lodwick

Brindle

Fulford

Smith

(Eds.), The rural economy of roman Britain, new visions of the countryside of Roman Britain, Vol. 2 (pp. 230–265). , Britannia Monograph 29.

Brindle

(2018). Personal appearance in the countryside of Roman Britain. In Smith

Allen

Brindle

Fulford

Lodwick

Rohnbogner

(Eds.), The rural settlement of Roman Britain, new visions of the countryside of Roman Britain, Vol. 3, pp. 30–65). Britannia Monograph 32.

Brück

(2005). Experience the past? The development of a phenomenological archaeology in British prehistory. Archaeological Dialogues, 12(1), 45–72. https://doi.org/10.1017/S1380203805001583

Buteux

Gaffney

White

van Leusen

(2000). Wroxeter hinterland project and geophysical survey at wroxeter. Archaeological Prospection, 7(2), 69–80. https://doi.org/10.1002/1099-0763(200006)7:2<69::AID-ARP143>3.0.CO;2-R

10.

David

Thomas

(2008). Handbook of landscape archaeology. Left Coast Press.

11.

Eckardt

(2014). Objects and identities: Roman Britain and the North-Western provinces. Oxford University Press.

12.

Edwards

Crossley

(2009). Measures and meanings: Exploring the ego-net of Helen Kirkpatrick Watts, Militant Suffragette. Methodological Innovations Online, 4(1), 37–61. https://doi.org/10.1177/205979910900400

13.

Evans

C. J.

Jenks

W. E.

White

R. H.

(1999). Romano-British kilns at Meole Brace (Pulley), Shropshire (Vol. LXXIV, pp. 1–27). Shropshire Archaeology and History.

14.

Ferris

(2012). Roman Britain through its objects. Amberley.

15.

Fleming

(2006). Post-processual landscape archaeology: A critique. Cambridge Archaeological Journal, 16(3), 267–280. https://doi.org/10.1017/S0959774306000163

16.

Frede

(1993). The question of being: Heidegger’s project. In Guignon

(Ed.), The Cambridge companion to Husserl (pp. 42–69). Cambridge University Press.

17.

Gaffney

C. F.

Gater

J. A.

Linford

Gaffney

V. L.

White

(2000). Large-scale systematic fluxgate Gradiometry at the Roman city of wroxeter. Archaeological Prospection, 7(2), 81–99. https://doi.org/10.1002/1099-0763(200006)7:2<81::AID-ARP145>3.0.CO;2-6

18.

Gaffney

V. L.

White

R. H.

Goodchild

(2007). Wroxeter, the Cornovii and the urban process final report on the Wroxeter Hinterland Project 1994-1997. Volume 1: Researching the Hinterland. Journal of Roman Archaeology Supplementary Series, 68.

19.

Hamilton

Whitehouse

Brown

Combes

Herring

Thomas

M. S.

(2006). Phenomenology in practice: Towards a methodology for a “subjective” approach. European Journal of Archaeology, 9(1), 31–71. https://doi.org/10.1177/1461957107077704

20.

Herrle

(2008). Architecture and identity? In Herrle

Wegerhoff

(Eds.), Architecture and identity (pp. 11–22). Deutsche National bibliothek.

21.

Hingley

(2005). Romanizing global culture: Unity, diversity and empire. Routledge.

22.

Howlett

(1992). An archaeological evaluation at hay farm, eardington, Shropshire. Tempus Reparatum.

23.

Jenkins

(2008). Social identity. Routledge.

24.

Lambiotte

Delvenne

J. C.

Barahona

(2008). Laplacian dynamics and multiscale modular structure in networks. eprint arXiv, Article 0812.1770. https://doi.org/10.48550/arXiv.0812.1770

25.

Maxwell

J. A.

(2016). Expanding the history and range of mixed methods research. Journal of Mixed Methods Research, 10(1), 12–27. https://doi.org/10.1177/1558689815571132

26.

May

D. E.

(2023). The development of landscape archaeology in Britain, present conflicts and possible new directions. Landscape History, 44(2), 109–118. https://doi.org/10.1080/01433768.2023.2284553

27.

Millett

(2016). By small things revealed: Rural settlement and society. In Millett

Revell

Moore

(Eds.), The Oxford handbook of Roman Britain (pp. 699–720). Oxford University Press.

28.

Opsahl

Agneessens

Skvoretz

(2010). Node centrality in weighted networks: Generalizing degree and shortest paths. Social Networks, 32(3), 245–251. https://doi.org/10.1016/j.socnet.2010.03.006

29.

Pearce

(2016). Status and burial. In Millett

Revell

Moore

(Eds.), The oxford handbook of roman Britain (pp. 341–362). Oxford University Press.

30.

Purcell

(1995). The Roman villa and the landscape of production. In Cornell

T. J.

Lomas

(Eds.), Urban society in roman Italy (pp. 151–179).

31.

Renfrew

Bahn

(2000). Archaeology: Theories, methods and practice. Thames and Hudson.

32.

Richards-Rissetto

(2017). What can GIS + 3D mean for landscape archaeology? Journal of Archaeological Science, 84, 10–21. https://doi.org/10.1016/j.jas.2017.05.005.

33.

Smith

Brindle

Fulford

Lodwick

(2018). Lifestyle and the social environment. In Smith

Allen

Brindle

Fulford

Lodwick

Rohnbogner

(Eds.), The Rural Settlement of Roman Britain, New Visions of the Countryside of Roman Britain (Vol. 3, pp. 48–77). Britannia Monograph 32.

34.

Swift

(2017). Roman artefacts and society: Design, behaviour and experience. Oxford University Press.

35.

Teddlie

(2007). Mixed methods sampling: A typology with examples. Journal of Mixed Methods Research, 1(1), 77–100. https://doi.org/10.1177/1558689806292430

36.

Thomas

(2006). Phenomenology and material culture. In Tilley

Keane

Küchler

Rowlands

Spyer

(Eds.), Handbook of material culture (pp. 34–59). Sage.

37.

Tilley

(2008). Phenomenological approaches to landscape archaeology. In David

Thomas

(Eds.), Handbook of landscape archaeology (pp. 271–276). Left Coast Press.

38.

Tilley

(2010). Interpreting landscapes: Geologies, topographies, identities: Explorations in landscape phenomenology 3. Left Coast Press.

39.

Webster

(2001). Creolizing the Roman provinces. American Journal of Archaeology, 105(2), 209–225. https://doi.org/10.2307/507271

40.

White

Wigley

(2018). Shropshire in the Roman period. In White

Hodder

(Eds.), Clash of cultures? The Romano-British period in the West Midlands (pp. 115–135). Oxbow Books.

41.

White

R. H.

(2022). Whitley Grange villa, Shropshire: A hunting lodge and its landscape. In Henig

, et al. (Eds.), Villas, sanctuaries and settlement in the Romano-British countryside (Vol. 95, pp. 81–92). Archaeopress Roman Archaeology.

42.

White

R. H.

Gaffney

V. L.

(2013). Wroxeter: The Cornovii, and the Urban Process. Final Report on the Wroxeter Hinterland Project 1994-1997. Volume 2: Characterising the City. Archaeopress.