Mapping Spheres of Influence on Medieval Iberia's Religious Frontier via Viewshed Analysis and Cost-Distance Analysis
Since the nineteenth century, Iberia's transition from an Islamic Caliphate into a handful of medieval Christian Kingdoms (often called the Reconquista) has been mapped using two principal schemas. The most common is a sequential comic-strip showing the moments of greatest change over nearly six hundred years. The second schema compresses this comic strip into a single map with a handful of dated horizontal lines representing the southward progress of Christian forces into Muslim-controlled territory. Not unlike maps of "Manifest Destiny" in North American history, medieval Iberia's frontier zones have been depicted as edges rather than the wide, permeable spaces that they appear to be in the historical and archaeological record. Given how easily these sharp edges of territorial partitioning can mislead viewers of historical maps, it is imperative that historical geographers do more than simply gesture at the shape and extent of frontier space. New maps of medieval Iberia (and similar fluctuating frontiers) should visualize how spheres of influence were projected onto frontier landscapes by rival groups, and how those spheres overlapped.
This article seeks to visualize medieval Iberia's wide, shifting, and contentious frontier by running GIS analyses on a data set of fortresses and settlements occupied by Christian military-religious orders, Christian nobles, and Andalusian and North African Muslims. By itself, the ADIMO data-set (Architectural Database of Iberian Military Orders) reveals that fortresses invariably represented the forward-most positions on the religious frontier during the Reconquista, but these sites are only the epicenters for spheres of influence that projected over the frontier landscape. In order to understand how a switch in fortress occupation from Christian to Muslim or vice-versa influenced the surrounding landscape, the ADIMO project asks which areas of territory can be seen or quickly traveled to by the new occupants. In GIS terms, these questions were answered using viewshed and cost-distance analysis. Put simply, the ability to see a wide swath of territory on the frontier made it easier to secure it from invasion, and because vision is reciprocal, highly visible hilltop fortresses could effectively stamp the landscape below them as under Christian or Muslim influence. Similarly, cost-distance analysis visualizes where the threat of violence by a fortress-garrison was most acute by drawing the boundaries of single-day travel from each site. Finally, when layered as a historical GIS on an even field with major events of the Reconquista, viewshed and cost-distance analysis reveal strategic patterns where even sites with very meager remains – either physically or in the historical record – often held highly strategic, central positions in the definition of the frontier.
Geographer William Shepherd's historical atlas of 1911 contains a quintessential map of the stages of the Christian (re)conquest of Iberia (Figure 1). Like many versions of this map in successive historical atlases and surveys, Shepherd's map begins with the "height" of Islamic dominance of the Iberian Peninsula in 910 CE, and ends just before the conquest of the Nasrid Kingdom of Granada by the recently united Christian kingdoms of Castile and Aragon in 1492. The first three quadrants on the double-page spread of Shepherd's map represent frozen [End Page 66] moments from 910, 1030 and 1150 CE, but the map in the lower right corner depicts "Spain 1212-1492." The result is a problematic visualization that privileges the moments of greatest change at the expense of the far more common patterns of incremental change and territorial fluctuation on the frontier. Unfortunately, the temporal and spatial "jumping" in this map has become a dominant method of visualizing the Reconquista. Rather than depicting wide, porous borders between religious groups, and overlapping spheres of influence as they are written in modern works of history and art history, most historic maps of the Reconquista continue to depict the frontier as a thin, winding horizontal line that periodically jumps south.
These maps partially succeed at visualizing a long period of history in a succinct way, yet they also misrepresent the contested borders of medieval Iberia. The frontier between Christian and Muslim-held territories may have compressed to the width of a river at various times over seven hundred years of "conflict and coexistence" on the peninsula, but they never resembled the hard lines of nationalist boundaries.1 Unfortunately, most attempts to depict a mixing of groups on the frontier – often through alternating lines or hash marks – are not drawn at a resolution that allows them to be scrutinized. The frontier in these maps appears to be a perfectly heterogeneous zone where Christians and Muslims both resided. Most importantly for this study, most historical [End Page 67] maps of the Reconquista fail to spatially and temporally locate the Christian military-religious orders, whose existence was defined by their proximity to the frontier. In response, I created a geospatial database called ADIMO (The Architectural Database of Iberian Military Orders) to bring the historical evidence of these groups back to the literal and figurative forefront of the Christian Reconquest of Iberia.
This article proposes methods for depicting the porous, overlapping spheres of influence on the frontier that are more precise than previous visualizations of this topic – especially those that include medieval "border-lines." I agree with fellow medievalists like Ronnie Ellenblum who successfully argue that the words "border" and "borderline" are not synonymous, and further, that despite their associations with castles in the middle ages, linear definitions of territorial sovereignty were almost nonexistent during this period.2 Many historians of medieval Iberia have preferred to define territory in cultural terms or, as Foucault did, by making it synonymous with sovereignty. I agree with Stuart Elden that Foucault's definition of sovereignty as "holding onto conquered territory" is simplistic, and more importantly, impossible to visualize.3 We know that the military orders were sovereign over units of land that had some degree of nearness to their fortresses, and we could claim that this nearness made these spaces more Christian than Islamic, but how do we draw these spheres of influence? The answer that has proved most satisfying for me is to focus on "centers" rather than borderlines, and to run analyses that visualize which parts of the landscape could be surveilled or physically policed by garrisons occupying each center or fortified site. I have also attempted to test the hypothesis that "chains of fortresses" were designed to accomplish the same result as modern borderlines. For this portion of the project, rather than using complex network tools in GIS, I have focused on connecting only those sites that were intervisible with each other or sites whose garrisons could travel to other sites in a single day on horseback.
The ADIMO project began by locating the physical manifestations of the contested border – fortresses and fortified settlements that were built and occupied by Andalusian and North African Muslims, the Christian military orders, and Christian nobles – and ended with GIS analyses that radiate out from these sites. The first analysis, called Viewshed Analysis, "paints" geographical space according to which gridded units of land could be seen by a viewer at the top of a hilltop fortress. The second analysis, called Cost Distance, visualizes which units of this same landscape could be traveled to by a fortress garrison on horseback within a single day (cost-distance). When combined, these analyses reveal which areas of the frontier landscape could be surveilled and/or physically attacked or defended, and by whom. They also begin to explain why so many hilltop fortresses were constructed on the frontier in the first place, while disrupting the common, false assumption that the best-preserved sites also had the most strategic significance. By making it possible to include sites that survive as little more than place names, dates of occupation and bare foundations, the ADIMO project offers a higher-resolution image of the frontier than past mapping efforts. In addition, the patterns that emerge in ADIMO's interactive maps – i.e. clusters of intervisible sites, remote isolated outposts that survive despite being surrounded by rivals, or suspiciously large spatial or temporal gaps in the data – create more opportunities for further research than one scholar could pursue. Like many complex forms of data visualization, these analyses require more than textual definitions and examples in order to be translated by someone other than their creator. Consequently, this article will offer 3D views of the maps and in-depth historical context when possible. In summary, this article will describe the inspiration for this approach, outline the methods used, and unpack the complex patterns that the Historical Geographic Information System (HGIS) project creates. [End Page 68]
ADIMO, the Architectural Database of Iberian Military Orders, began as an expansion of the topic of my MA thesis on the thirteenth-century fortress-monastery of Calatrava la Nueva in central Spain.4 After finishing this case study, my dissertation research began with three assumptions, two of which were verified, and one important one that was quickly disproven. The first assumption was that Christian military-religious orders were the vanguard of the Reconquest, occupying the edge of the frontier. I had little doubt that identifying where the military orders were at various times during the twelfth, thirteenth, and fourteenth centuries would also visualize the fortified frontier during those times, and the ADIMO database offered confirmation at an early stage of the project. The idea that the military orders occupied the "front lines" of the Reconquista was first put forward in foundational documents of the military orders. For example, the following passage from the Rule of the order of Santiago inspired the title of my dissertation A Wall of The Faithful: Spatial Analysis of Military Order Architecture on Medieval Iberia's Religious Frontier.5 The rule reads:
With the kings in such disagreement, a multitude of Saracens came from beyond the seas to lay waste the lands of the Christians and to destroy the Church of God. The aforesaid knights, inspired by the grace of the Holy Spirit and seeing the great peril that threatened the Christians – unless it could be checked – imprinted on their chests the cross in the shape of a sword with the ensign and invocation of the Blessed James to stop the hostile advance of the enemies of Christ, defend the Church, and expose themselves as a wall of the faithful to the fury of the infidels.6
This description of the order of Santiago, while metaphorical, held literal connotations as well. Even before it was visualized in a GIS, the data entry stage for the ADIMO project confirmed that the military orders were often granted castles in the most advanced positions on the frontier between the Christian Kingdoms of Castile, Aragon, and Portugal and the Muslim-controlled territories held by the Almohads, Nasrids and others. Preliminary efforts to include occupation dates in my maps of military order fortresses revealed that the orders were often ahead of the major jumps of territorial expansion by Iberia's Christian kingdoms.
The second assumption was that, while the military orders did control hundreds of castles and other fortified sites on the frontier, there were not so many as to make mapping them impossible. This assumption was formed as a counterpoint to a very specific line from a 1984 article on the subject of the fortresses of the military orders by Alan Forey titled "The Military Orders and the Spanish Reconquest in the Twelfth and Thirteenth Centuries."7 This article represented the first, and to date only, effort outside of my dissertation to directly assess the role of the military orders in the formation of the frontier across the entire peninsula. In short, Forey concluded that it was "impossible" to create a survey of all of the frontier fortresses that had come into the possession of Iberia's military orders. Citing an imprecise "high" number of sites, and the tendency for sites to change or share place-names, Forey was unwilling to offer a conclusion about the spatial distribution of the military orders on the frontier other than that there were "regions [where] particular orders tended to predominate."8 My initial reaction to this statement was that while seemingly innocuous, it actually explained why a vast majority of maps of the military orders focused on small regional and temporal units, and why peninsula-wide historical maps of medieval Iberia still drew the frontier as a single wavy line. As a result, I pinned my dissertation research to the conviction that a systematic approach would make it possible to locate hundreds of sites and lead to a more complete image of frontier fortress strategy. I was also encouraged by [End Page 69] the thought that Forey did not have access to readily available satellite data and GIS technology in 1984 when he claimed a survey would be impossible. The survey I produced relied heavily on satellite imagery, and early in the data-entry stage I began to notice that the ADIMO database was producing denser, higher-resolution spatial distributions of the military orders in Iberia. The point data of the ADIMO database began to back up my initial assumption that there was a deliberate strategy to the location and occupation of frontier fortresses, but it was not until I analyzed the points via viewshed and cost distance tools that the most interesting patterns appeared.
My third assumption was that the military orders created enough surviving architecture to trace their decisions morphologically. This idea very quickly proved incorrect. In truth, Calatrava la Nueva, and other extant "fortresses-monasteries" built by the military orders in Iberia are rare survivals of exceptional buildings. The vast majority of the sites they occupied are little more than half-walls or outlines of cisterns today. By virtue of being built on a shifting frontier, fortresses could become irrelevant and abandoned soon after they were conquered due to the new location and direction of the frontier. Consequently, as I collected data for the first iteration of ADIMO, fields that were meant to record the number of castles that contained cloisters, churches, or other substantial alterations by the military orders were largely filled with null values. As an architectural history project focusing on the way the military orders manifested their identity through architecture, ADIMO seemed hollow – especially after I came to the conclusion that the orders were better described as occupiers than as builders of frontier fortresses. Fortunately, after shifting the design of the database to focus on data that was available for most sites regardless of the degree of survival, I made an important discovery: buildings made no sense as the base unit of the database. ADIMO was better at identifying occupation events. Put simply, these events determined which military order or frontier group occupied each fortress or walled settlement, as well as the beginning and end dates for their occupation of the site. Consequently, unique survivals such as Calatrava la Nueva could exist in the same GIS, and with equal weight as the much more common sites that were little more than place names, locations, and dates of occupation.
After populating ADIMO with roughly eleven hundred occupation events joined to more than six hundred walled settlements and fortresses, the next steps required identifying analyses that would allow even the most meager architectural survivals to "radiate" influence over the broader frontier landscape and link with other sites. These choices were driven by research into what medieval builders considered the purpose of their castles and what effect they had on the frontier landscape. This question of motivation for castle construction is contentious in the sub-field of castle-studies, especially for historians of medieval English castles. The traditional argument that castles existed to either exert power or communicate status (occasionally referred to as the "war or status debate") is not particularly applicable to Iberia's religious frontier.9 In addition to Iberia's less feudal character, the conflict over territory was longer and at times (but not always) more ideological for the military orders, their Christian patrons, and their Muslim rivals on the peninsula than elsewhere in Europe. There are very few statements from medieval writers that directly relate how castles were used on a religious frontier, but much can be gained from the few that have been identified. One of Forey's best discoveries in this regard was his quotation of a 1240 commentary on the construction of the Templar-constructed castle of Safed in northern Israel, in which the medieval author claimed this new castle:
would provide defense and security against the Saracens and be like a shield for the Christians as far as Acre; it would also be a strong and formidable point of [End Page 70] attack… and through the construction of the said castle the sultan would lose a lot of money and much support and service from the men who would be subject to the stronghold and from their lands. He would also in his own territories lose villages, agricultural land, pasture and other rights, because his men would not dare to work the land for fear of the castle.10
According to Forey, the principal benefits of castle building on a Christian-Muslim frontier were all present within this single quotation. Castles were a strategic base to monitor and defend territory from future attack, they could be an offensive weapon for future incursions, and most interestingly, they dominated the landscape by frightening Muslims who wished to work on land that was within a castle's sphere of influence.11 Despite the often extreme cost of their construction or capture, and despite their ubiquity on the landscape of medieval Europe, direct contemporary statements about a castle's purpose or effect such as this are relatively rare. Almost two decades after Forey's 1984 article, Adrian Boas summarized the uses for crusader and military order fortresses in much the same way, but with an added element of surveillance, and minus their psychological effect:
Frontier castles were generally no more than isolated, fortified, forward positions which could house large garrisons, contain stores of weapons, food and equipment, and serve as lookout positions and as refuges if necessary. None the less, these were significant roles which could be decisive in the success or failure of the field army.12
Neither Forey nor Boas were wrong in focusing on these uses for frontier fortifications, yet what has been particularly difficult to define has been the shape and size of each fortress' sphere of influence. It is my view that the size and shape of a castle's sphere of influence can be visualized via GIS analyses, and that viewshed and cost-distance analysis are particularly useful. These advanced GIS tools offer new opportunities for pattern detection that earlier scholars did not have access to, and they allow centers to take precedence over borders as definers of the frontier.
The first analysis – viewshed – is relatively simple to define, yet often difficult to interpret. Simply put, a viewshed is a 2D representation of which parts of a 3D landscape could be seen from a given location. As with most tools in GIS software, the results of viewshed analysis are entirely dependent on the data that are set to limit it, such as a digital elevation model (DEM), vegetation data, or weather. As tempting as it is to consider adding more and more layers of data, no model can account for all of the factors affecting vision at any particular moment in time, especially in the past. At the scale of the entire Iberian Peninsula, there is no way to know how tall, or even where the trees were in the twelfth century, much less how many were chopped down over the course of two centuries. Consequently, in the interest of keeping all the ADIMO viewsheds consistent, regardless of the likelihood of deforestation or tree cover around the sites, the most significant shaper of viewshed, the topography of the surrounding landscape, was used as the critical variable. The DEM breaks a given landscape into square units – for instance twenty square meters – that have heights associated with them. The 2.5D data is captured by satellite, LIDAR, or aerial photography and represents changes in height visually via an (often) monochromatic color-scale with lightest pixels representing the highest peaks of a landscape.13
Viewsheds processed from a single viewer location yield rasterized, binary images that ask each pixel location if the space could be seen by the viewer. Composite viewsheds use multiple [End Page 71] viewer locations to offer degrees of visibility for different units of landscape. The ADIMO project iterated single viewsheds for every occupation event in the database, then transformed the positive values into vectors for a very specific purpose – to attach dates, castle names and affiliations to each viewshed. The viewsheds were styled according to their affiliation, and can be queried according to the dates of occupation. In the newest visualizations of this data, the viewshed polygons are displayed as slightly transparent. As a result, the maps also reveal how areas of the landscape could be visible from more than one location, and more than one group. Querying the data for a particular year offers a snapshot of which groups occupied or constructed fortresses in Iberia in that year, and what these groups could see from these locations. For instance, Figure 2 shows the viewsheds projected from significant sites around southern La Mancha shortly after the order of Templars occupied the Almohad-built castle known as Calatrava la Vieja.
The second form of analysis used to study the site-landscape relationship is called cost-distance analysis. In simple terms, the hundreds of time-stamped, fuzzy-edged polygons that surround all the ADIMO sites are impressionistic visualizations of each site's possible religious/military influence on the landscape, rather than an image of land ownership for each site (Figure 3). Like viewshed analysis, cost-distance analysis is a form of agent-based modeling (ABM) – a category of analysis that employs hypothetical agents to act upon the dataset according to rules set up by the map maker. Unlike viewshed analysis, however, the agents are not static features. [End Page 72] The cost-distance analysis used in ADIMO simulated an agent – specifically a rider on horseback – and set the agent to walk 32 kilometers (roughly a day's travel) in all possible directions from the starting point of each of the 621 sites in ADIMO. Without taking into account the cost of traveling up increasingly steep slopes, the cost-distance would look like a perfect circle with a radius of 32 kilometers and a center at one of the sites in ADIMO. However, by reclassifying a slope-map from the DEM so that each degree of slope larger than eight degrees costs an increasing number of available 32 cost-kilometers, the shape of the limits that a person on horseback can travel in 32 cost-kilometers tends to follow the valleys that are closest to the frontier settlement or castle. The cost-map was further reclassified using a layer of the largest Roman roads in Iberia in order to lessen the cost of traveling directly on these routes.14
Figure 4 shows the cost-distance polygon estimating which areas of the landscape could be reached within one day's travel on horseback from the fortress-monastery of Montesa. As with all agent-based modeling, the results of cost-distance analysis are highly dependent on the number and weight of variables added to the model. Vegetation, rivers, roads, political boundaries, tolls, seasons and weather can all influence the route taken to reach a particular destination or prevent travel entirely. Adding these variables is beyond the scope of this work and would require an entire team of researchers to build a model that would be useful at the scale of the Iberian Peninsula.15 Nonetheless, the current model – which again uses the slope of the landscape as the primary determinant for accessibility – suggests which parts of the surrounding landscape were most likely to be under the influence of a given fortress or city garrison. In more concrete terms, the shape of the cost-distance polygon shows where a fortress garrison would be able to send a mounted soldier if a local insurrection, tax collection, or some other immediate (within a [End Page 73]
[End Page 74] day) policing need arose. Metaphorically, the cost-distance polygon maps the extent of a castle's immediate ability to physically exert its influence over others. There are, of course some trade-offs for transforming the raster data into vectors that can be attached to occupation events. The most significant drawback is that the cost-distance polygons become monochromatic boundaries. Thus, any sense of "fuzziness" or ambiguity – represented by a color ramp of pixels showing the spaces within the cost-distance boundary that were easiest to travel over – are lost. Conversely, table-data such as dates and names of fortress occupants cannot be attached to pixelated rasters. Taking the various advantages and disadvantages of raster and vector data into account, I decided to convert the cost-distance boundaries into vectors so that I could query the data temporally.
Even without additional variables affecting travel cost, the cost-distance model in Figure 4 can at least eliminate those parts of the landscape that were impossible to access in a single day's ride. Moreover, when compared with viewshed analysis, we can see how the direction of greatest visibility may not have necessarily been the direction or extent of greatest influence on the surrounding landscape. In these cases, hilltop or ridge sites could have been chosen because they had an unobstructed view in the direction of a perceived threat, such as a mountain pass, but the landscape made it easier to travel to low lying areas in the opposite direction of this mountain pass. This characteristic could be interpreted as a site with a less aggressive posture, depending on who was residing in the site at a particular time. In other cases, when a fortress was sited against the side of a large mountain chain, cost-distance offers a relatively clear, visual impression of a site's "orientation" (Figure 5).
Compared to visibility, landscape accessibility is a much more difficult motivation to measure. The ability to ride out from a fortress and reach a specific location within a single day's travel is reflective of that fortress's ability to enforce its power through the threat of violence. It is especially useful in the case of the ADIMO database because so little is known about the locations of smaller, agrarian frontier settlements across the peninsula. Christopher Gerrard's survey of Christian and Muslim hamlets and farmsteads around the fortified town of Ambel in north-west Aragon is a rare archaeological study of medieval settlements, but his localized approach does suggest that Christian and Muslim frontier space could have been organized topographically.16 Gerrard's observation that Christian settlements tended to be sited in the foothills with Muslim hamlets below them supports the idea that the major fortified settlements were designed to exert power over irrigated valleys.17 Because Ambel was periodically controlled by the order of Templars, the site and this occupation fit into the ADIMO data set, and the viewsheds results support Gerrard's claims even though Ambel is not one of the more dramatically sited possessions of the Templars. Ambel's fortress sits at the edge of a rise just above a valley that is actively cultivated today, and its viewshed is more expansive than one might expect based on the site's position on a low foothill (Figure 6).
Visual dominance of the landscape, as Gerrard suggested, is one way that a fortress or settlement could express power over the landscape. Another, more concrete method for expressing power in medieval Iberia was to demonstrate an armed presence in the area through regular reconnaissance on horseback. Thinking in terms of cost-distance and viewshed, it makes sense that a militarized, minority Christian population would want to secure sites that were uphill and visually elevated from the majority Muslim populations that still lived in and cultivated land on the frontier. By visualizing the spatial extents of day-long movements that begin at each fortress, the cost-distance polygons reveal the shape of valleys under the fortress' influence. Christian and Muslim settlers within the polygon could reach the castle if an external threat appeared in the area, and the fortress could regularly send armed soldiers to these areas to collect taxes.
Intervisibility and Accessibility
Viewshed and cost-distance analysis are also capable of illustrating more specific relationships between sites. These visualizations establish connectivity via two elements: nodes (or to use Ellenblum's term: "centers") and edges. In the ADIMO project, each occupation event, including the location, date, and affiliation is represented as a node. Edges – the lines [End Page 75] between nodes – are formed by querying the viewshed and cost-distance analyses to determine if a relationship exists between nodes like those in Figure 4. Unlike far more complex forms of network analysis, the nodes in this study do not have gravitational pull based on the number of edges connected to them or the size of the garrisons at each site. For the purpose of this study, the edges reveal where bottlenecks in visibility or accessibility occurred in the frontier landscape. For instance, sites like the castle of Ferral at the top of Figure 3 are nearly invisible in the historical and archaeological record, yet this meager tower stands out very clearly in the ADIMO data set. Also in Figure 3, the dashed white lines symbolize sites that were intervisible with each other, the solid black lines reveal sites between which people could (potentially) travel within a single day on horseback, and alternating black and white dashes symbolize intervisible sites between which people could (potentially) travel in a day.
The lines of intervisibility and mutual accessibility in the ADIMO maps act as a direct, empirical test of the popular description of medieval frontiers as "chains of castles." This phrase is very popular in studies of ancient, medieval, or early-modern borders, such as the Roman limes, the Christian Reconquest of Iberia, the Norman conquest of the British Isles, or the western limits of the Ottoman Empire.18 The temptation to describe castles as existing in a "chain" likely stems from modern conceptions of territorial borders, but this problematic term is usually accepted without comment. One question I ask in this study is whether connections between castles can validate the common linear definition of borders in historical maps. In truth, it seems there were [End Page 76] cases where "chain" seems to be an appropriate descriptor of the spatial distribution of certain castles during the Reconquista, such as the ones shown in Figure 4. In the mid-fourteenth century castles like Ayora, Almansa, Caudete and Biar could be described as forming a linear border between Murcia and Valencia – two former Almohad Taifas that had recently been conquered by the Christian kingdoms of Castile and Aragon, respectively. The single line connecting each node appears to make this border seem even more rigid than the more complex networks of connectivity elsewhere in the peninsula, yet in fact, likely the opposite was true. Medieval garrisons did not assume that the space "behind" their castles was safe from attack – the direction of security or insecurity was not that clear, and sites that were visually or spatially connected to allies in multiple directions had a much clearer sense of sovereignty over the landscape than sites that could only count on one or two allies in opposite directions.
GIS Patterns & Historical Context
Overlapping spheres of influence, as they are visualized in ADIMO, are abstractly observable even with minimal historical context. For instance, already by the early-twelfth century, the military orders in Iberia were developing a pattern where their headquarters were built very near the frontier, yet they also added watchtowers in more advanced positions that were intervisible with a specific "targeted" Muslim city. This pattern can be seen with the large Templar fortress of Monzon, its hilltop tower of Chalamera, and the Muslim-held city of Fraga in 1118 CE (Figure 7). A wider pattern was the tendency for Christian-Muslim intervisibility between castles to be a short-lived condition. In most cases, when rival castles were left "staring" [End Page 77] at each other after a shift on the frontier, Muslim groups abandoned these sites in favor of ones that lay behind natural barriers to vision. While this simple trend is worthy of pursuit on its own, much more complex patterns emerge when the GIS is observed with previous familiarity
[End Page 78]
of the historical context of the Christian Reconquest. In this section, I will describe the events surrounding an especially recognizable pattern in ADIMO's viewshed and connectivity data as a way of advocating for future deep dives into historical GIS patterns. The case study below should also help clarify the importance of contextualizing visual secondary sources like the ADIMO GIS.
The period that most clearly challenges the teleological image displayed in traditional historic maps of the Christian Reconquest were the years leading up to the battles of Alarcos in 1195 CE and the conclusion of the battle of Las Navas de Tolosa in 1212 CE. During this seventeen-year period, the spatial structure of the Christian-Muslim frontier fluctuated dramatically, and the fortresses occupied by the military orders and their Almohad rivals were central to the outcome. The pivotal sites in this episode were the city of Toledo – the frontier capital of Castile in the twelfth and early-thirteenth centuries – the castle of Calatrava la Vieja, and the fortress of Salvatierra. Toledo had been a Roman foundation and was sited in a very defensible position in a bend of the steep-sided Tagus River (Figure 8). Nonetheless, it was vulnerable to raids in the mid-twelfth century because it was essentially blind due to a natural ridge in the south – which is easily observable in the viewshed analysis in Figure 9. The original response to these raids was to conquer the Almohad castle of Qal'at Rabah (Latinized as Calatrava) and turn it into a Christian outpost and buffer by donating it to the Templars in 1147. Upon hearing that the castle would face the full force of a new wave of Almohad invasions form North Africa in 1155, the Templars abandoned the castle, forcing king Sancho III of Castile to promise the castle to anyone willing to defend it. The first native Iberian military order of Calatrava was formed by taking this castle as its namesake, and with the assistance of a Christian army that was raised in 1158, it managed to survive their first years and begin to expand in the following decade. By 1194, a wide swath of territory south of Toledo was under the surveillance of the order of Calatrava thanks to their occupation of several hilltop fortifications south of the Tagus River (Figure 10). [End Page 79]
Unfortunately for Castile and the order of Calatrava, forty years of slow gains in the "Campo de Calatrava" were wiped out after the Almohad victory at the battle of Alarcos (Figure 11). The Order of Calatrava was nearly destroyed after this battle, so much so that they did not have enough members left to garrison any of its castles south of Toledo. After abandoning these castles to the Almohads the order proved even further that it considered proximity to the enemy essential to its identity when the remnants of the Order of Calatrava unexpectedly selected and conquered the fortress of Salvatierra for its new frontier base of operations in 1198 (Figure 12). This fortress became so essential to the Order's renewed purpose that the order chose Salvatierra as its new namesake – calling itself the Order of Salvatierra while harassing the Almohads from this location (Figure 13). The order's effectiveness in this regard is supported by the fact that the Almohad Caliph al-Nāsir again invaded from Morocco with the specific purpose of removing the order from this castle in 1210. The Order of Salvatierra eventually caved under the pressure of a fifty-one day long siege, but it was allowed to leave with their possessions and march to Toledo as a condition of the order's surrender.19
One of the most important aspects of the siege of Salvatierra, besides eliciting a call to crusade in Iberia by Pope Innocent III, was the reaction of the Caliph to the surrender. Despite failing to destroy the garrison itself, the Caliph relished the victory over Salvatierra and remarked that he had "[c]ut off the right hand of the King of Castile."20 The Caliph's complaints about this castle also help explain why viewshed analysis immediately came to mind as a useful form of analysis for the ADIMO project:
Salvatierra had fallen into the traps of the adorers of the cross and the presence of a bell on its church-tower was an insult to the Muslims, who, to the four points
[End Page 80]
[End Page 81] of the compass around this place, heard the muezzins glorify God and call them to prayer; it was a watch-tower rising against the sky in the bare plain,… an observatory which spied on us. This castle gave the Muslims no peace because the Christians made it the base for all their raids and organized it so that it was a sort of key guaranteeing the security of their strongholds and towns.21
Many of this article's themes are covered by this short passage. First, the Almohad Caliph believed that the castle of Salvatierra was a signpost for the Christian religion that also signaled the close proximity of a rival religion through the use of a church bell. Second, Al-Nasir responded to the form of Salvatierra in much the same way that modern photographers have – noting the way it stood out against the sky when seen from below. Third, the geographic location, physical site, and form of the castle were all in the service of surveillance for the Christians. He described Salvatierra as a "watch-tower" and "an observatory which spied on us." Caliph Al-Nasir also called Salvatierra a base for sorties into Muslim territory. There can be no doubt that frontier fortresses like Salvatierra – garrisoned by dedicated, military-religious knights – were both the practical and symbolic material of the metaphorical "wall of the faithful." Moreover, the ability to view the landscape – and be seen doing so from an imposing position above it – was clearly understood as central to physical and psychological warfare on the frontier. In this context, it is not surprising that the Caliph considered the capture of this largely ruined and meager fortress a major victory, or that a crusade was called to protect western Christendom after its surrender. The Caliph's description of Salvatierra also seems less hyperbolic after observing how isolated, [End Page 82] yet prominently observable this site appears to be in the viewshed analysis image in Figure 12. If anything, the boldness of the order of Calatrava's decision to occupy Salvatierra, rather than re-grouping at a more northern site is exaggerated by a visual comparison of Figures 11 and 12.
The final castle in this short narrative, Calatrava la Nueva was built shortly after a final pendulum swing in favor of Christian forces at the battle of Las Navas de Tolosa in 1212 CE. After this battle on the southern slope of the Sierra Morena Mountains, all of the fortresses that had been lost by the order of Calatrava after the battle of Alarcos seventeen years earlier were returned to them, with the exception of Salvatierra (Figure 14). The donations included their former namesake and original headquarters, which was soon to be known as Calatrava la Vieja. Word of the definitive Christian victory at Las Navas immediately spread throughout Europe, yet due to unforeseen factors like the death of Alphonso VIII of Castile and a famine that struck La Mancha in 1213-14, the "door" to Andalucía was not immediately opened. As a result, the order of Calatrava chose a symbolic yet conservative site for their new headquarters. Rather than build on the furthest edge of the frontier near the battlefield of Las Navas de Tolosa, they decided on a site directly across from the castle of Salvatierra – which remained in Almohad hands for several years after Las Navas de Tolosa. The viewshed from the new headquarters, (aptly named Calatrava la Nueva) when combined with on-site photography, supports the argument that this castle was oriented toward the south so it could closely surveil an important pass through the Sierra Morena mountains (Figures 15 and 16). The symbolic value of casting a shadow over Salvatierra, (the fortress that had ensured the order's survival following the loss at Alarcos) was likely a factor [End Page 83]
[End Page 84] in the order's choice of site as well (Figure 17). Calatrava la Nueva was larger than any previous military order fortress in Iberia at this time. Besides being very large, the fortress also contained a church, cloister, chapter house and dormitories to support the Cistercian monks integrated into the order (Figures 18 and 19). This military-monastic complex was designed to project permanence in the landscape, but it also revealed a historic awareness that the frontier was anything but a rigid line. In spatial terms, the order of Calatrava knew that its sphere of influence could fluctuate in the landscape due to changes in fortress-occupations around it, and that the southernmost outpost of Castilian territory did not define the religious border between Christianity and Islam. Consequently, the order's new headquarters was architecturally prominent yet sited behind a topographical and visual barrier that had always separated La Mancha from Andalucia.
The above section demonstrates that the ADIMO data set can provide spatial context to narratives that are already known, but more importantly, ADIMO also reveals patterns that run parallel or counter to the dominant narrative surrounding the best-preserved sites. Even within the Campo de Calatrava region in La Mancha, the ADIMO data set contains sites that barely register in the historical or archaeological record, yet appear to have highly strategic roles when visualized in the GIS. As I mentioned briefly above, one of the best examples of this pattern is the castle of Ferral. If not for the enthusiasm of hikers who took geotagged photos of this site and uploaded them to Google Earth, this site would have been very difficult to locate on a satellite map (Figure 20). By noticing the small cluster of image icons near an area where I thought Ferral should be and comparing these images to what was visible in the satellite data, I was able to place a relatively precise pushpin in the location of this mostly ruined fortress. After running viewshed analysis and querying surrounding sites for intervisibility, the ADIMO data-set revealed that this site, and the equally meager castle of Vilches were sited in remote locations with extremely expansive viewsheds of the Guadalquivir valley. The visual bottleneck produced by Ferral is [End Page 85]
[End Page 86] clearly defined by lines of intervisibility with six other sites in the data set, including the recently captured Andalucian city of Baeza (Figure 21). In the years immediately following Las Navas de Tolosa, the fortress of Ferral appears to have been in a highly strategic position. It is my hope that the online ADIMO map can help other scholars discover additional patterns, lead to more in-depth research into other overlooked sites, or even act as a model for castle research in other locations.22
Conclusion: From Viewshed and Cost Distance to Spheres of Influence
Having established that visibility and accessibility factored into the choice of site for fortresses and settlements on the Christian-Muslim frontier, the next step is to transform viewshed and cost-distance analyses into measures of power and influence in the landscape. These ideas can be bridged simply via practical consideration of the needs of fortress occupiers, but there are also theoretical considerations. To begin, occupying a castle and looking out toward the landscape were not merely practical necessities for frontier security. According to Foucault, medieval "territory" was created by an agent's ability to expand as well as hold what had already been gained.23 Stuart Elden argues that this definition of medieval territory is evidence of Foucault's vague notion of sovereignty in this period because it projects a modern sense of territory as a quantified space. I share the opinion of Elden and Ronnie Ellenblum when I argue that religious boundaries were never defined linearly, or geometrically in this period, and that we must look beyond connected centers or fortress chains for our definition of medieval frontiers. Each center should have its own measurable sphere of influence that acts on the landscape. The alternative model described here foregoes popular variables like taxation or grazing rights as ways of measuring the influence of castles on the landscape in favor of models that have more psychological associations. For [End Page 87] instance, in a medieval frontier zone, the effect of being able to see a landscape from a position in which one felt secure lent a measure of at least influence, if not ownership, of that landscape. The phrase "king of the mountain" seems an apt description of the experience of standing at the top of a tower at a hilltop site, but again, the valley below was anything but a modern, measured territory. Obviously, the medieval viewer wasn't "ruler of all they could perceive" by virtue of being able to see it, yet the Reconquest of Iberia was partially driven by this phenomenological experience. Economic, political and religious dominance of the frontier meant control of fertile valleys in lower elevations. By observing the mills, cultivated land, animal herds, fences and other forms of infrastructure before them, the occupant of a hilltop castle was able to understand the value of the land, and target it as a future possession.
In 1983, Tadahiko Higuchi wrote a theoretical work on the combination of architecture, landscape and vision that has been a surprising inspiration for the ADIMO project, despite obvious differences in the architecture being studied. His book The Visual and Spatial Structure of Landscapes used the subject of Japanese temples and landscapes to analyze how structures change when they are sited differently or seen from different vantage points.24 This work considered how buildings become more striking to the eye, and thus more "present" in the landscape. Higuchi used the term "isovista" to define the amount of a person's view that is taken up by a particular structure and explained that an easy way to increase the isovista for a given structure was to site it at a high elevation. Taken further, if topographical change is particularly dramatic, where large zones of low territory surround a steep, conical hill capped by a building, that structure will require viewers at middle and even long distances to tilt their head up to view it (Figure 18). The simple act of tilting one's head up, (which Higuchi termed the angle of incidence) is less natural for most viewers, who are inclined to look down while moving through space.25 It is my view (and Higuchi's) that this simple discomfort caused people who worked in the valley to feel even more "beneath" the eyes of those who occupied the higher spaces. This is not an especially unique observation, but Higuchi's attempts to measure this condition were novel. Connections between hilltop siting and architectural expressions of power have been well covered in a variety of fields, but unlike Higuchi, most of this work focuses on the perspective of the projector of power, and other than Higuchi, there are few attempts to actually measure a building's visual dominance in the landscape.26 I see Higuchi's isovista measurement as a phenomenological alternative to the satellite-view of viewshed analysis – it adds degree to what is otherwise a binary of visible/invisible and draws attention to the reciprocal nature of vision between a hypothetical garrison at a hilltop fortress and those working in the valley below. Given that a flag signaling a castle's switch from Christian to Islamic affiliation (or vise-versa) could cause an evacuation of the surrounding area, Higuchi's thoughts on the visual impact of hilltop architecture seem especially applicable.
Similar to Higuchi's isovista, knowing a fortress' "posture" when it was constructed or after it changed hands could be an excellent indicator of how parts of the Christian-Muslim frontier "felt" during the twelfth and thirteenth centuries. Unfortunately, this is a difficult idea to measure. The typical narrative of the Reconquest tends to attribute an aggressive posture to whichever side of the conflict had gained the most recent victory, and momentum is offered as evidence of a grasping, forward-looking policy of territorial expansion. Architecturally, the two sides of the spectrum for aggressive versus defensive fortifications can be demonstrated by two examples. On the defensive side is the Great Wall of China – which was designed with crenellations only on the north side facing the Mongolian threat. The Great Wall was designed strictly to create a static, impermeable border that would never advance. On the "aggressive" end of the spectrum were siege, or "counter-castles" which were built next to fortified targets specifically to offer a strategic advantage in an attack.27 In essence, all fortresses are "defensive" structures, but their orientation [End Page 88] or posture could change dramatically with new occupants, or if landscape-scale factors changed around them. This is why it is so important to symbolize whose gaze is represented in a viewshed or whose garrison had easy access to a valley rather than merely visualizing where or how far these analyses spread. Given that the military orders' possessions more closely resembled a metaphorical "wall of the faithful" than a defensive line of castles, spatial context is essential to understanding the posture of particular fortresses at different times. Consequently, cost distance and viewshed cannot be used in isolation to determine a single fortress' posture or orientation, but they can reveal the overlap of spheres of influence between friendly or rival sites, and this system offers a more realistic image of the character of sovereignty on Iberia's military-religious frontier.
Cost-distance and viewshed analyses are capable of visualizing large, structural concepts like power and influence, and from a historical perspective, they are also capable of visualizing smaller-scale concepts like posture, or the reach of symbolic signaling in the landscape. These concepts, like an interactive GIS, reveal themselves at different scales. One of the goals of the visualizations in this article has been to use 3D GIS to connect high-level patterns in the analyses to the phenomenological experience of standing at the top of a castle, looking up at one, or traveling between them. Similarly, while it is useful to deconstruct linear definitions of borders, and promote a "radiating centers" definition of frontier space, it is equally important to study [End Page 89] what viewshed can reveal about the siting of a single fortress like Calatrava la Nueva (Figure 15), or what cost-distance can tell us about the physical reach and posture of a fortress-monastery like Montesa (Figure 4).
Looking at viewshed analysis and cost-distance analysis with an awareness of what it meant to observe, be observed, or be within true proximity of allies and rivals, I am led to the conclusion that even with the weaknesses of agent-based modeling, these GIS processes visualize spheres of influence in ways strict hermeneutics simply cannot. Granted, it takes several steps to interpret viewshed as a measure of surveillance, spatial anxiety, confidence or posture, just as it takes a good deal of historical context to interpret cost-distance boundaries as images of threatened or defensible space. A healthy dose of skepticism about these methods of analysis is warranted, even necessary for their interpretation, but this should not cause viewers to dismiss their value.
The questions that the ADIMO project sought to answer at the outset were discrete, and the methods of analysis were stripped down to essential questions, such as who could see which parts of the frontier landscape, and when. If the reader accepts how important (and variable) vision and landscape accessibility were in the context of the Christian Reconquest of Iberia, these questions have far more wide-reaching historical implications. Simply put, I argue that knowing what could be seen and traveled to, at a fine-grained temporal and geographical scale, allows forgotten structures to reemerge as historical actors and draws spheres of influence that are far more nuanced and historically appropriate than the boundary lines dividing Christian and Muslim territory in most historical maps (Figure 22).
1. "Conflict and Coexistance" refers to the title of a popular 2002 historical volume edited by Roger Collins and Anthony Goodman, Medieval Spain: Culture, Conflict and Coexistence. (New York: Pelgrave, 2002).
2. Ronnie Ellenblum, "Were There Borders and and Borderlines in the Middle Ages? The Example of the Latin Kingdom of Jerusalem." Medieval Frontiers. Ed. David Abulafia. (2002): 105-6.
3. Stuart Elden "How Should We Do the History of Territory?" Territory, Politics, Governance, 1:1, (2013) 8.
4. Edward Triplett "The Resurrection of the Military order of Calatrava through the Construction of a New Capital" MA Thesis, (University of Virginia, 2009).
5. Edward Triplett, A Wall of the Faithful: Spatial Analysis of Military Order Architecture on Medieval Iberia's Religious Frontier, PhD Diss., (University of Virginia, 2015).
6. Enrique Gallego Blanco, trans., Rule of the Spanish Military Order of St. James 1170-1493, (Leiden NL: Brill Archive, 1971), 2. (emphasis added)
7. Alan J. Forey, "The Military Orders and the Spanish Reconquest in the Twelfth and Thirteenth Centuries," Traditio 40 (1984): 202.
9. Oliver Creighton and Robert Liddiard. "Fighting Yesterday's Battle: Beyond War or Status in Castle Studies." Medieval Archaeology 52,1 (2008): 161-69.
10. R. B. C. Huygens, De Constructione Castri Saphet: Construction et Fonctions d'un Château Fort Franc en Terre Sainte. (Amsterdam: North Holland Publishing Company (1981): 36.
11. Alan J. Forey, The Military Orders; New Studies in Mediaeval History (New York: Palgrave Macmillan, 1991): 59.
12. Adrian Boas, Archaeology of the Military Orders: A Survey of the Urban Centres, Rural Settlements and Castles of the Military Orders in the Latin East (c.1120–1291). (Abingdon UK: Routledge, 2006): 103.
13. The DEM used in this study has a resolution of c. 32 square meters per pixel, and was derived from ASTER. (Advanced Spaceborn Thermal Emission and Reflection Radiometer) – a partnership between NASA and Japan's Ministry of Economy, Trade and Industry. Version 2 of the data was released in 2011.
"ASTER Global Digital Elevation Map." accessed (January 2017) http://asterweb.jpl.nasa.gov/gdem.asp.
15. An excellent example of such a project is the Orbis project created at Stanford University. This project allows a researcher to begin a hypothetical journey from any site in the Roman Empire and end at any other comparable site by ship, horseback, foot, or ox-cart via a network of linear paths identified by roads, sea or river. The model is far more complex than the one proposed here, and also presupposes that there is a set destination, but it is an excellent introduction to cost-distance analysis. In the introduction to the project, the authors succinctly point out that "Cost, not distance is the principal determinant of connectivity." Walter Scheidel, Elijah Meeks, and Carl Grossner, "ORBIS: The Stanford Geospatial Network Model of the Roman World," (June 1, 2012), accessed (January 2017) http://orbis.stanford.edu/.
16. Christopher Gerrard, "Opposing Identity: Muslims, Christians and the Military Orders in Rural Aragon," Medieval Archaeology 43 (2000): 147.
18. For a more detailed historiographical account of the concept of "chains" of medieval (and ancient) fortresses, see: Ronnie Ellenblum, Crusader Castles and Modern Histories. (Cambridge, UK: Cambridge University Press, 2007).
19. Joseph F. O'Callaghan, Reconquest and Crusade in Medieval Spain (Philadelphia: University of Pennsylvania Press, 2003), 97.
20. Ibid. 67.
21. Derek W Lomax, The Reconquest of Spain. (London: Longman, 1978), 122.
23. Elden. 8
24. Tadahiko Higuchi, The Visual and Spatial Structure of Landscapes. (Cambridge, MA: MIT Press, 1983), 41-56.
25. Katherine Ellenberger. "Scales of Visibility at a Chacoan Outlier: The Visual World of People at Kin Klizhin." Thesis. (Binghamton University - State University of New York, 2012), 13-14.
26. Paul Hirst's Space and Power: Politics, War and Architecture. (Cambridge, UK: Polity, 2005) also influenced the ADIMO project through his translations of space and architecture into spheres of influence, and his analysis of militarily contested frontiers. Much like Stuart Elden, the author is primarily concerned with deconstructing ideas of "territory" in modern and pre-modern times and directly engages with Foucault and LeFebvre around the subject of war and the production of space.
27. O. H. Creighton, Castles and Landscapes (London: Continuum, 2002), 57.