Hostname: page-component-cd9895bd7-hc48f Total loading time: 0 Render date: 2024-12-23T19:30:03.553Z Has data issue: false hasContentIssue false

2nd c. CE defenses around small towns in Roman Britain structured by road network connectivity

Published online by Cambridge University Press:  07 February 2022

Joseph Lewis*
Affiliation:
Department of Archaeology, University of Cambridge
Rights & Permissions [Opens in a new window]

Abstract

The large-scale provision of defenses around small towns in Roman Britain during the 2nd c. CE is without parallel in the Roman Empire. Although the relationship between defended small towns and the Roman road network has been noted previously, provincial-level patterns remain to be explored. Using network analysis and spatial inference methods, this paper shows that defended small towns in the 2nd c. are on average better integrated within the road network – and located on road segments important for controlling the flow of information – than small towns at random. This research suggests that the fortification of small towns in the 2nd c. was structured by the connectivity of the Roman road network and associated with the functioning of the cursus publicus.

Type
Note
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in any medium, provided the original work is properly cited.
Copyright
Copyright © The Author(s), 2022. Published by Cambridge University Press

The provision of defenses around towns in Roman Britain has attracted interest since the 1930s,Footnote 1 with a well-established general outline. in the 1st c. CE, the provision of defenses was limited to a few public towns (coloniae, municipia, civitas capitals).Footnote 2 In the 2nd c. CE, a large number of earthwork defenses were erected in public towns, as well as in settlements subordinate to civitas capitals in political and administrative status (henceforth, small towns).Footnote 3 In the 3rd c. CE, more small towns received walls, and finally, in the 4th c. CE, external towers were added to existing walls.Footnote 4

Although it should be noted that the issue of defining a “small town” continues to be debated,Footnote 5 with the selection and subdivision commonly based on characteristics such as function or origin,Footnote 6 small towns differ from civitas capitals in several ways: they often lacked official administrative status, planned rectangular street grids, and key public facilities such as a forum and bathhouses.Footnote 7 Although the presence of small towns is common to all provinces of the Roman Empire,Footnote 8 the large-scale provision of defenses around small towns in Roman Britain – whether during or after the 2nd c. – is without parallel.Footnote 9 Fortification of small towns in neighboring provinces such as Gaul, Germany, and Raetia remained rare, with defenses often limited to settlements with civitas status.Footnote 10 Due to this, Esmond-Cleary suggests that small towns from the 2nd c. onward played a more significant role within the civitates and Roman Britain more generally compared to other provinces.Footnote 11

Traditionally, the motivation to construct defenses around towns has been driven by two hypotheses: (1) a centralized government response to an actual or perceived threat,Footnote 12 or (2) a symbol of civic pride initiated by local authorities.Footnote 13 Although civic pride and display is considered a plausible explanation for defenses around higher-status towns (coloniae, municipia, civitas capitals),Footnote 14 the significant number of small towns defended in the 2nd c. remains more difficult to explain.Footnote 15 For example, the defense of small towns has been related to specific periods of historical upheaval,Footnote 16 such as the usurpation of Clodius Albinus (r. 193–197 CE). However, the use of “external threat” as the explanatory model has become less tenable because defenses show chronological phasing, as well as the more general issue of fitting archaeological evidence into historical narratives.Footnote 17 Furthermore, although the motivation of civic pride has been used as an explanation for the provision of defenses around small towns,Footnote 18 Esmond-Cleary notes that this is “no more than a possibility and does little to explain the fortification of the ‘small towns.”Footnote 19

In contrast, the defense of small towns has been linked to important installations, such as the mansiones and mutationes that were integral for the development and functioning of the Roman provincial transportation system (cursus publicus).Footnote 20 Through the cursus publicus, government communication was made possible – which was essential for securing and maintaining control throughout the empire.Footnote 21 Nonetheless, this led Smith and Fulford to ask: “if the common factor linking defended settlements on the major roads was that they supported the cursus publicus […], why were not all such settlements provided with defences?”Footnote 22 With this question in mind, this research will use network analysis and spatial inference methods to identify provincial-level patterns in the relationship between small towns fortified in the 2nd c. and the connectivity of the Roman road network.

Specifically, two testable hypotheses will be investigated:

  1. (1) Defended small towns were well integrated within the road network for the efficient transfer of information.

  2. (2) Defended small towns were located on roads that were important for controlling the flow of information across the road network.

Materials and methods

The data and code are available at https://github.com/josephlewis/Defended_Small_Towns.

Roman road network

The Roman road network was downloaded from Bishop.Footnote 23 Based on Roman roads identified by Ivan Margary,Footnote 24 this dataset represents the most comprehensive Roman Britain road network currently available. It should be noted, however, that knowledge of the Roman road network is currently incomplete,Footnote 25 with the region north of London showing a greater concentration of roads.Footnote 26 Although this reflects local interest in identifying Roman roadsFootnote 27 rather than the true distribution of Roman roads in Roman Britain, the impact of this on the analysis is outside the scope of this paper and will therefore not be further explored.

To ensure that the Roman road network as currently understood was interconnected, and thus sufficient for network analysis, small gaps were manually fixed by connecting road ends via straight lines (Fig. 1a). The choice to manually join roads – predominately to urban centers – over other automated methods such as connecting all road ends within 500 mFootnote 28 or least-cost path analysisFootnote 29 aimed to minimize the introduction of additional uncertainty.Footnote 30 The road network was converted to a network graph, with road vertices acting as nodes and road segments as edges (Fig. 1a).

Fig. 1. (a) Roman roads in Roman Britain using Bishop (Reference Bishop2014) (grey), with small gaps filled by present author (black) and (b) network graph representation of the Roman road network with vertices acting as nodes and road segments as edges.

Small towns

Data on 90 small towns with sufficient information on the date of defenses was collated from Millett,Footnote 31 using Smith and Fulford for more accurate chronologies when possible.Footnote 32 Four small towns (Brough-on-Fosse, Dorn, Dropshot, and Sandy Lane) were removed from the analysis due to uncertainty in the dates of fortification.Footnote 33 Nonetheless, the large number of small towns included within the analysis aims to sufficiently represent small towns as a category. The 90 small towns were filtered to those within 1 km from the Roman road network, ensuring that the small towns were deemed to be sufficiently integrated within it. The remaining 79 small towns were used within the analysis, with 22 (28%) defended in the 2nd c. The small towns were subsequently attached to the Roman road-network graph to become nodes (Fig. 2).

Fig. 2. Roman roads with 2nd-c. defended and nondefended small towns as nodes. (Map by J. Lewis.)

Network analysis and spatial inference

The application of network analysis to archaeological research has rapidly increased.Footnote 34 By representing data as a network graph – with nodes signifying points, and edges representing the connections between them – the fundamental characteristics of a network structure at a node and network level can be analytically described.Footnote 35 For example, Pau de Soto assessed the accessibility of the Iberian Peninsula using the Roman road network,Footnote 36 Orengo and Livarda analyzed the circulation of goods in relation to Roman Britain road network properties,Footnote 37 and Brookes and Huynh assessed the correlation in town PageRank values to their status in Roman and early medieval England.Footnote 38

Although less common in archaeology, the network representation of archaeological data can also be used in the formal testing of hypotheses.Footnote 39 More specifically, the significance of a spatial pattern – in this case, whether defended small towns were well integrated within the road network and located on Roman roads important for controlling the flow of information – can be quantified by randomly reshuffling the characteristic of interest while keeping all other aspects of the data not relevant to the hypothesis constant.Footnote 40 Through this, the underlying process of interest and whether it resulted in the spatial pattern can be evaluated. More simply put, how likely is it that the observed spatial pattern of defended small towns has arisen from a random process given the underlying spatial distribution of small towns?Footnote 41

Network analysis measures

NODAL EFFICIENCY

The efficiency of defended small towns for communication was measured via nodal efficiency, which is the inverse of the average length of the minimum path length between a given node and all other nodes in the network (Fig. 3).Footnote 42 Nodes with high nodal efficiency are well integrated within the overall network and indicate a higher capability of transferring information efficiently to all other nodes.Footnote 43 The nodal efficiency of all small towns was calculated, with the nodal efficiency of defended small towns mean averaged.

Fig. 3. Nodal efficiency of defended (black border) and nondefended small towns (no border). (Map by J. Lewis.)

EDGE BETWEENNESS

The importance of a road segment for controlling the flow of traffic was measured via edge betweenness, which is the number of shortest paths in the graph that pass through a given edge.Footnote 44 Edges with high edge betweenness therefore control the flow of information and act as “bridges” between multiple parts of the networkFootnote 45 (Fig. 4). The edge betweenness of all road segments was calculated, with the edge betweenness of roads with defended small towns situated along them mean averaged.

Fig. 4. Edge betweenness of each road segment in the road network with defended small towns. (Map by J. Lewis.)

CREATING THE NULL HYPOTHESIS

The null hypothesis of mean nodal efficiency and mean edge betweenness was calculated by randomizing the status of small towns – that is, randomly relabeling whether a small town was “defended” or “nondefended” (Fig. 5). The randomization in the status of the small towns reflects that of a random process,Footnote 46 with the rejection of the null hypothesis signifying that the observed pattern of defended small towns is unlikely to have occurred by chance. A one-sided Monte Carlo hypothesis test was used to statistically compare the mean nodal efficiency and mean edge betweenness of the observed pattern of defended small towns against 999 simulations where the “defended” or “nondefended” status of small towns was randomized (p < 0.05). Although a relatively small number of simulations are required when using Monte Carlo hypothesis testing, with 99 simulations often sufficient,Footnote 47 999 simulations results in a better estimate of the p-value.Footnote 48 Under a Monte Carlo hypothesis test procedure,Footnote 49 N values of the test statistic (e.g., mean nodal efficiency) from simulations under the null hypothesis (where status is randomized) are compared to the observed statistic (where status is not randomized), with the p-value obtained by counting the number m of simulated test statistics at least as extreme as the observed test statistic (p-value = [m + 1] / [N + 1]).Footnote 50

Fig. 5. Roman road network with observed distribution of small towns (a) and examples after randomly shuffling the “defended” status of small towns (b, c, and d). (Maps by J. Lewis.)

Results

Network measures of defended small towns

The mean nodal efficiency of the defended small towns is greater than nondefended small towns, with the defended small towns showing less variance in nodal efficiency values (Fig. 6a). Furthermore, the mean edge betweenness of road segments with defended small towns is greater than road segments with nondefended small towns (Fig. 6b).

Fig. 6. (a) Nodal efficiency of defended and nondefended small towns and (b) edge betweenness of defended and nondefended small towns.

Using the Monte Carlo hypothesis test procedure, the mean nodal efficiency of the 22 defended small towns is statistically significant (p < 0.05) compared to the 999 randomized simulations (Fig. 7a). This means that the defended small towns on average possess a higher capability of transferring information to other nodes and are therefore more integrated within the road network than small towns chosen at random. Similarly, the mean edge betweenness of road segments with defended small towns is statistically significant (p < 0.05) compared to the randomized simulations (Fig. 7b), showing that defended small towns were on average located on roads that had more control over the flow of information while also acting as bridges to other parts of the Roman province.

Fig. 7. (a) Distribution of mean nodal efficiency from randomized simulations and the true mean nodal efficiency of defended small towns (black dashed line) and (b) distribution of mean edge betweenness from randomized simulations and true mean edge betweenness of defended small towns (black dashed line). (Simulations by J. Lewis.)

Discussion and conclusion

Although the relationship between defended small towns and the Roman road network has been noted previously,Footnote 51 this research has shown that the connectivity of the Roman road network is associated with the fortification of small towns in the 2nd c. CE. Defended small towns in the 2nd c. on average not only possessed higher nodal efficiency but were located on roads with higher values of edge betweenness than nondefended small towns. Furthermore, the rejection of the null hypotheses shows that the small towns defended in the 2nd c. were on average better integrated within the road network than small towns at random, and they were situated along road segments that were important for controlling the flow of information across the road network. Although the reason for fortifying small towns was complex and likely not due to a single cause,Footnote 52 the provincial-level pattern of defended small towns in the 2nd c. having high connectivity while also being well integrated within the road network echoes the fundamental role defended small towns played in the communication of information via the cursus publicus.Footnote 53 This suggests that, as noted by Millett,Footnote 54 although civic competition between communities might have driven the provision of defenses, government officials were keen to ensure that the defenses also protected administrative infrastructure related to the functioning of the cursus publicus.

Nevertheless, it is important to be aware of limitations. First, the nodal efficiency and edge betweenness are based on the edges present within the road network. Although the impact of missing road segments – as well as filling in gaps using straight lines – is assumed to be minimal, it highlights the need for a more complete understanding of the road network in Roman Britain.Footnote 55 Second, this analysis ignores the possibility of maritime and river navigation, resulting in potentially undervaluing the nodal efficiency and edge betweenness for small towns and road segments near rivers and the coast. Although this effect is assumed to be small due to the majority of defended small towns being in the central belt of Roman Britain, future research is needed to incorporate multiple types of transportation when conducting provincial-level analyses.Footnote 56 Last, given that the present analysis focused exclusively on small towns defended in the 2nd c., there is a need to determine whether and how much the road network continued to have an influence on which small towns were fortified in later periods.

Despite the limitations, it has been shown that the connectivity of the Roman road network played a fundamental role in which small towns were defended during the 2nd c. The provision of defenses around small towns and the association with the protection of the administrative infrastructure reflects the importance of the cursus publicus in the establishment and maintenance of Roman rule within the Roman province of Britannia.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Footnotes

6 E.g., Burnham Reference Burnham1986; Burnham and Wacher Reference Burnham and Wacher1990.

8 De Ligt and Bintliff Reference De and Bintliff2020 for an overview.

13 Millett Reference Millett1990, 139–40.

16 E.g., Frere Reference Frere1991, 241; Wacher Reference Wacher1962.

18 Millett Reference Millett1990, 137–40.

22 Smith and Fulford Reference Smith and Fulford2019, 18.

24 Margary Reference Margary1973, 170.

25 Estimated to be 40% at best by the Roman Roads Research Association 2021.

27 E.g., The Viatores 1964.

28 E.g., Brookes and Huynh Reference Brookes and Huynh2018.

29 E.g., Orengo and Livarda Reference Orengo and Livarda2016.

30 E.g., Lewis Reference Lewis2021.

31 Millett Reference Millett1990, 154–56.

32 Smith and Fulford Reference Smith and Fulford2019.

33 Smith and Fulford Reference Smith and Fulford2019.

34 Brughmans and Peeples Reference Brughmans and Peeples2017.

37 Orengo and Livarda Reference Orengo and Livarda2016.

38 Brookes and Huynh Reference Brookes and Huynh2018.

39 Östborn and Gerding Reference Östborn and Gerding2014.

41 Modified from Fotheringham and Brunsdon Reference Fotheringham and Brunsdon2004.

42 Latora and Marchiori Reference Latora and Marchiori2001.

43 Latora and Marchiori Reference Latora and Marchiori2001; Latora and Marchiori Reference Latora and Marchiori2003.

46 Fotheringham and Brunsdon Reference Fotheringham and Brunsdon2004; Östborn and Gerding Reference Östborn and Gerding2014.

48 Gentle Reference Gentle2002, 55–57.

50 North, Curtis, and Sham Reference North, Curtis and Sham2002; Turner and Jeffs Reference Turner and Jeffs2017.

51 e.g. Burnham and Wacher Reference Burnham and Wacher1990, 235–78

54 Millett Reference Millett1990, 140.

55 E.g., Roman Roads Research Association 2021

56 E.g., Carreras and de Soto Reference Carreras and de Soto2013; Carreras, de Soto, and Muñoz Reference Carreras, de Soto and Muñoz2019

References

Barnard, G. A. 1963. “Comment on ‘The spectral analysis of point processes’ by M. S. Bartlett,” Journal of the Royal Statistical Society: Series B 25: 294.Google Scholar
Besag, J., and Diggle, P. J.. 1977. “Simple Monte Carlo tests for spatial pattern.” Applied Statistics 26: 327.CrossRefGoogle Scholar
Bishop, M. C. 2014. The Secret History of the Roman Roads of Britain. Barnsley, South Yorkshire: Pen & Sword Military.Google Scholar
Black, E. W. 1995. Cursus publicus: The Infrastructure of Government in Roman Britain. London: BAR.CrossRefGoogle Scholar
Booth, P. 1998. “Defining small towns in Roman Britain.” JRA 11: 613–23.Google Scholar
Brookes, S., and Huynh, H. N.. 2018. “Transport networks and towns in Roman and early medieval England: An application of PageRank to archaeological questions.” JAS: Reports 17: 477–90.Google Scholar
Brughmans, T. 2010. “Connecting the dots: Towards archaeological network analysis.” OJA 29: 277303.Google Scholar
Brughmans, T., and Peeples, M. A.. 2017. “Trends in archaeological network research: A bibliometric analysis.” Journal of Historical Network Research 1: 124.Google Scholar
Burnham, B. C. 1986. “The origins of Romano-British small towns.” OJA 5: 185203.Google Scholar
Burnham, B. C., and Wacher, J. S.. 1990. The Small Towns of Roman Britain. Berkeley: University of California Press.Google Scholar
Carreras, C., and de Soto, P.. 2013. “The Roman transport network: A precedent for the integration of the European mobility.” Historical Methods 46: 117–33.CrossRefGoogle Scholar
Carreras, C., de Soto, P., and Muñoz, A.. 2019. “Land transport in mountainous regions in the Roman Empire: Network analysis in the case of the Alps and Pyrenees.” JAS: Reports 25: 280–93.Google Scholar
De, Ligt, and Bintliff, J. L. (eds.). 2020. Regional Urban Systems in the Roman World, 150 BCE–250 CE. Leiden: Brill.Google Scholar
de Soto, P. 2019. “Network analysis to model and analyse Roman transport and mobility.” In Finding the Limits of the Limes: Modelling Demography, Economy and Transport on the Edge of the Roman Empire, eds. Verhagen, P., Joyce, J., and Groenhuijzen, M. R., 271–89. Cham: Springer International.Google Scholar
Esmond-Cleary, A. S. 2003. “Civil defences in the west under the high empire.” In The Archaeology of Roman Towns: Studies in Honour of John Wacher, ed. Wilson, P., 7285. Oxford: Oxbow.Google Scholar
Esmond-Cleary, A. S. 2007. “Fortificación en la Britannia Romana: ¿Defensa militar o monumento cívico?” In Murallas de ciudades en el occidente del Impero Romano: Lucus Augusti como paradigma, eds. Rodríguez Colmenero, A. and de Llanza, I. Rodá, 155–65. Lugo: Museo provincial de Lugo.Google Scholar
Fotheringham, A. S., and Brunsdon, C.. 2004. “Some thoughts on inference in the analysis of spatial data.” International Journal of Geographical Information Science 18: 447–57.CrossRefGoogle Scholar
Freeman, L. C. 1978. “Centrality in social networks: Conceptual clarification.” Social Networks 1: 215–39.CrossRefGoogle Scholar
Frere, S. S. 1984a. “British urban defences in earthwork.” Britannia 15: 63.CrossRefGoogle Scholar
Frere, S. S. 1984b. “The early development of the cities of Roman Britain.” Revue Archéologique de Picardie 34: 1117.CrossRefGoogle Scholar
Frere, S. S. 1991. Britannia: A History of Roman Britain. 3rd ed. London: Pimlico.Google Scholar
Gentle, J. E. 2002. Elements of Computational Statistics. New York: Springer.Google Scholar
Good, P. I. 2005. Permutation, Parametric and Bootstrap Tests of Hypotheses. 3rd ed. New York: Springer.Google Scholar
Gorenflo, L. J., and Bell, T. L.. 1991. “Network analysis and the study of past regional organisation.” In Ancient Road Networks and Settlement Hierarchies in the New World, ed. Trombold, C. D., 8098. Cambridge: Cambridge University Press.Google Scholar
Haynes, I. 2002. “Britain's first information revolution: The Roman army and the transformation of economic life,” in The Roman Army and the Economy, ed. Erdkamp, P., 111–26. Amsterdam: Gieben.CrossRefGoogle Scholar
Hobley, B. 1983. “Roman urban defences: A review of research in Britain.” In Roman Urban Defences in the West, eds. Maloney, J. and Hobley, B., 7784. London: Council for British Archaeology.Google Scholar
Hope, A. C. A. 1968. “A simplified Monte Carlo significance test procedure.” Journal of the Royal Statistical Society: Series B (Methodological ) 30: 582–98.Google Scholar
Kolb, A. 2001. “Transport and communication in the Roman state: The cursus publicus.” In Travel and Geography in the Roman Empire, eds. Adams, C. E. P. and Laurence, R., 95105. London and New York: Routledge.Google Scholar
Latora, V., and Marchiori, M.. 2001. “Efficient behavior of small-world networks.” Physical Revue Letters 87: 198701.CrossRefGoogle ScholarPubMed
Latora, V., and Marchiori, M.. 2003. “Economic small-world behavior in weighted networks.” The European Physical Journal B 32: 249–63.CrossRefGoogle Scholar
Lewis, J. 2021. “Probabilistic modelling for incorporating uncertainty in least cost path results: A postdictive Roman road case study.” Journal of Archaeological Method and Theory 28: 911–24.CrossRefGoogle Scholar
Margary, I. D [The Viatores, pseud.]. 1964. Roman Roads in the South-East Midlands. London: V. Gollancz.Google Scholar
Margary, I. D. 1973. Roman Roads in Britain. 3rd ed. London: John Baker.Google Scholar
Millett, M. 1990. The Romanization of Britain: An Essay in Archaeological Interpretation. Cambridge and New York: Cambridge University Press.Google Scholar
North, B. V., Curtis, D., and Sham, P. C.. 2002. “A note on the calculation of empirical P values from Monte Carlo procedures.” The American Journal of Human Genetics 71: 439–41.CrossRefGoogle ScholarPubMed
Orengo, H. A., and Livarda, A.. 2016. “The seeds of commerce: A network analysis-based approach to the Romano-British transport system.” JAS 66: 2135.Google Scholar
Östborn, P., and Gerding, H.. 2014. “Network analysis of archaeological data: A systematic approach.” JAS 46: 7588.Google Scholar
Roman Roads Research Association. 2021. RRRA Home. http://www.romanroads.org/.Google Scholar
Rust, T. C. 2006. Architecture, Economics, and Identity in Romano-British “Small Towns.” Oxford: John and Erika Hedges.CrossRefGoogle Scholar
Smith, A., and Fulford, M.. 2019. “The defended Vici of Roman Britain: Recent research and new agendas.” Britannia 50: 109–47.CrossRefGoogle Scholar
Todd, M. 1970. “The small towns of Roman Britain.” Britannia 1: 114.CrossRefGoogle Scholar
Turner, R., and Jeffs, C.. 2017. “An extension of Monte Carlo hypothesis tests.” Communications in Statistics – Simulation and Computation 46: 6545–58.CrossRefGoogle Scholar
Verhagen, P., Nuninger, L., and Groenhuijzen, M. R.. 2019. “Modelling of pathways and movement networks in archaeology: An overview of current approaches.” In Finding the Limits of the Limes: Modelling Demography, Economy and Transport on the Edge of the Roman Empire, eds. Verhagen, P., Joyce, J., and Groenhuijzen, M. R., 217–49. Cham: Springer International.CrossRefGoogle Scholar
Wacher, J. 1962. “A survey of Romano-British town defences of the early and middle second century.” ArchJ 119: 103–13.Google Scholar
Wacher, J. S. 1975. The Towns of Roman Britain. Berkeley: University of California Press.Google Scholar
Figure 0

Fig. 1. (a) Roman roads in Roman Britain using Bishop (2014) (grey), with small gaps filled by present author (black) and (b) network graph representation of the Roman road network with vertices acting as nodes and road segments as edges.

Figure 1

Fig. 2. Roman roads with 2nd-c. defended and nondefended small towns as nodes. (Map by J. Lewis.)

Figure 2

Fig. 3. Nodal efficiency of defended (black border) and nondefended small towns (no border). (Map by J. Lewis.)

Figure 3

Fig. 4. Edge betweenness of each road segment in the road network with defended small towns. (Map by J. Lewis.)

Figure 4

Fig. 5. Roman road network with observed distribution of small towns (a) and examples after randomly shuffling the “defended” status of small towns (b, c, and d). (Maps by J. Lewis.)

Figure 5

Fig. 6. (a) Nodal efficiency of defended and nondefended small towns and (b) edge betweenness of defended and nondefended small towns.

Figure 6

Fig. 7. (a) Distribution of mean nodal efficiency from randomized simulations and the true mean nodal efficiency of defended small towns (black dashed line) and (b) distribution of mean edge betweenness from randomized simulations and true mean edge betweenness of defended small towns (black dashed line). (Simulations by J. Lewis.)