Skip to main content

Book of Fortresses

The Book of Fortresses project seeks to spatially reconstruct an exceptional architectural source called the Livro das Fortalezas, which contains 120 perspective drawings and plans of 55 fortresses on Portugal’s border with Spain in 1510.

Published onMar 02, 2022
Book of Fortresses


The Book of Fortresses project seeks to spatially reconstruct an exceptional architectural source called the Livro das Fortalezas, which contains 120 perspective drawings and plans of 55 fortresses on Portugal’s border with Spain in 1510. GIS and 3D modeling techniques are used to translate the book into a comprehensible spatial source for modern viewers.

The Livro das Fortalezas, which contains 120 perspectival drawings and 51 precisely measured, but roughly drawn plans of 55 fortresses on Portugal’s border with Spain, represents a very early (1510 CE) example of an early modern “city atlas.” This rare snapshot of a premodern, fortified border provides a wealth of information, but it still remains spatially confusing for modern viewers. The primary goal of this digital project is to free the book from its original codex form and reassemble its drawings in a single spatial system. GIS and 3D modeling techniques make it possible to study how each place in the book was rendered using a mixture of observational drawing and cartographic principles.

The Book of Fortresses

The Book of Fortresses project is a spatial study of a unique codex of drawings of castles and fortified towns on the Kingdom of Portugal’s border with Spain in 1509-1510. This image shows the first true “map” of Portugal by Alvares Seco, which was created more than half a century after the Book of Fortresses. The red icons indicate sites that appear in the book.

A description of the Book of Fortresses by historian Alfredo Pimenta in 1944.

This quote served as an important inspiration for the spatial approach taken by the Book of Fortresses digital project. In short, it remains true that the Book of Fortresses has not yet been properly explored. The primary reason is because flipping through its pages remains spatially confusing, especially for a modern viewer. Duarte de Armas’ mixture of multiple vantage points in each image, his individual, non-mathematical approach to perspective drawing, and his simultaneously observational, and cartographic approaches to drawing these sites causes them to feel both attractive and disorienting. 

Brief overview of the source.

Brief overview of the source.

The Livro Das Fortalezas is quite literally a bound series of drawings and plans of fortresses. The original, and best preserved copy of the book, known as “Codex A” is held at the Torre do Tombo; the National Archives of Portugal in Lisbon. This digital project is deeply indebted to the publication of two facsimiles of the book that were published in 1997 and 2015. A copy of the 1997 facsimile is presented as part of this exhibition. 

A brief overview of the book’s creator.

The Book of Fortresses was drawn by a squire at the court of Manuel I of Portugal who went by “Duarte de Armas.” Very little is known about him, but it was recorded that he trained in cartography in Lisbon, and based on his name “de Armas” it is believed that he also served as a military engineer or architect. De Armas was certainly not shy about expressing his role on the project, as he drew himself (on horseback) with his servant (survey-poles / spears in hand) in 19 of his 120 perspective drawings.

A geospatial visualization of the Book of Fortresses.

The Book of Fortresses is an extremely rare, visual source of information about military architecture, early cartography, and the nature of medieval and early modern borders. It is a remarkably complete source that tells us as much about the appearance of these border towns and castles (many of which have been drastically altered or destroyed during the 18th and 19th centuries) as it does about early 16th century modes of assessing and describing the state of national borders.

Pairs of photographs taken from vantage points that approximate those used by Duarte de Armas when he created his two perspective drawings of Mértola, Portugal.

One of the first spatial investigations taken by the Book of Fortresses project was to travel to the first sites on Duarte de Armas’ itinerary and photograph the sites using a variety of lenses and vantage points. This initial effort began to reveal patterns in the artist’s drawings, such as his exaggeration of the vertical plane in order to capture as much information on his pages as possible.

A summary of the first three digital approaches taken by the Book of Fortresses project. 

Because there is no single system that is capable of capturing Duarte de Armas’ rendering of multiple scales of space in his book, a variety of digital environments have been used to try to disassemble the original codex and reconstruct it in ways that allow viewers to experience the book interactively. The first two environments were a CAD (computer aided design) program that allowed the modelers to combine de Armas’ precisely measured but imprecisely drawn plan drawings with his perspective drawings, and GIS (Geographic Information Systems) which enabled the team to spatially contextualize the drawings as 3D “Billboard” objects, and run additional analyses on the level of connectedness between each site. 

A summary of the second set of digital approaches taken by the Book of Fortresses project.

It quickly became clear that additional 3D data was required to help align De Armas’ perspective drawings in a geographic system. This need was alleviated through several “photogrammetry” campaigns in 2018 and 2019 that sought to capture dense 3D data of the extant architecture at the best preserved sites through a process which combines hundreds of 2D photos into dense 3D data of each site. Because photogrammetry requires photos from many overlapping vantage points, this process led to several important conclusions about the ways de Armas drew each site. For instance, it appears that the artist drew each view from up close as well as far away and stitched these observations together into images that represented architectural as well as landscape-scale features simultaneously. The team also developed a relational database and a new procedural modeling system to further analyze the drawings without always focusing on comparing the drawings to a modern “basemap.”

Conversely, this process also led to the conclusion that the comparison between the “real” sites and those drawn by the artist was inadvertently causing the project to become too invested in the “accuracy” of the drawings. Consequently, in 2020, Dr. Triplett teamed with Dr. Phil Stern (Duke, History Dept.) on a new branch of the project called the “Sandcastle Workflow”

The Book of Fortresses Database.

 A view of the relational database that drives an analysis of the iconography, serves as a project management system, and acts as the back-end database for the project website:

Spatial patterns in the book’s iconography.

The iconographical analysis has allowed the Book of Fortresses team to run a series of “multivariate clustering” analyses on the book. This view shows two of the more interesting patterns in the book that are anything but clear for a viewer that flips through its pages. The correlation between new town charters that were established during the years that Duarte de Armas was traveling to these sites, and the artist’s inclusion of symbols of justice and punishment in his drawings of those sites was one of the more stark patterns that emerged from this data.

Viewshed analysis of the Alentejo region.

The Book of Fortresses project is also a study about borders, and the way castles acted as “markers” of sovereign space at a time when our modern concept of linear boundaries between states had not yet emerged through advances in cartography. This image reveals how visible each castle was in the surrounding landscape of the Portuguese-Spanish border through a GIS tool called “viewshed analysis”.” Because vision is reciprocal, the blue colored spaces in this map also show what portions of the landscape could be surveilled by the castles and towns in the book. This analysis allowed the team to better understand possible gaps, or spaces of perceived vulnerability in the border, as well as the areas where visual overlap between sites was emphasized. 

Comparative GIS layers in the Book of Fortresses project.

Another concept that is implied in the book, yet difficult to see in its codex form, is the way the fortresses form a “chain” of visually connected sites. These maps reveal that most of Duarte de Armas' claims of intervisibility between sites (drawn as iconographic castles in the background of some images) occurred nearest to historical paths of invasion from Spanish staging areas in Extremadura toward Lisbon. 

An analysis of the paths that Duarte de Armas might have taken between sites based on his short written itinerary. 

The only purely textual portion of the Book of Fortresses is Duarte de armas’ short itinerary outlining the sequence of sites he visited, the distance between them, and the quality of the roads at each stage. This image shows how closely a modern wayfinding algorithm, (which is set to avoid modern highways) matches the distances that Duarte de Armas claimed between each site. In most cases, the paths closely match with de Armas’ estimates, and it appears that he also took more direct paths by mule than some of the switchback roads that are designed for modern cars.

Brief conclusions and patterns in the Book of Fortresses.

The Book of Fortresses project is a “deep dive” into a single source. Consequently, one of the defining principles of the project was to extract as much information from the source directly. Structuring the information from the source (spatially, categorically, and visually) has allowed us to build a recombined view of the book from many different vantage points. This image outlines only a few of the discoveries made through our digital methods.

Student Credits.

In 2020, The Book of Fortresses Project branched out to develop a workflow that could be applied toward other examples of pre-modern “city views” that are often referred to as “chorography.” This branch of the project, called “The Sandcastle Workflow” was funded by the National Endowment for the Humanities’ Office of Digital Humanities and has developed methods for translating similar perspectival views into navigable 3D environments through a process called “procedural modeling.” Both of these projects have had many students involved in the content creation, annotation and analysis stages, and we are very grateful for their contributions.

Project Creator or Project Team Name

 Edward Triplett: Duke University, Art, Art History and Visual Studies

Institutional affiliation/s

Duke University, Art Art History and Visual Studies

The Office of Digital Humanities at the National Endowment for the Humanities

Years active

2017 - present


Chorography, Mapping, Castles, Fortresses, GIS, 3D Visualization, Perspective

To learn more

Book of Fortresses website

The Sandcastle Workflow

No comments here
Why not start the discussion?