Letters from distant lands: Carolingian intellectuals and their network(s)

Letters from distant lands: Carolingian intellectuals and their network(s)

Clare Woods, Classical Studies, Duke University

In the age of Facebook, Twitter, LinkedIn, and other global social networking tools, networks, and ways in which we can (re)construct and analyze them, have become an important focus of research across the disciplines. This is no less true for those of us studying the premodern world, where networks (whether social, political, familial or intellectual) were both the fabric and the driving force of societies.

This post draws on material from my ongoing Carolingian Intellectual Networks project, the broad aim of which is to collect data on and map all the various connections between scholars, teachers and students in the late eighth through ninth centuries. The intellectuals I am studying were located across a large area of Europe comprising what is now France, the Low Countries, Switzerland, western parts of Germany and Austria, and northern Italy. This vast geographical area constituted the Carolingian Empire, a territory united by the military campaigns of Charlemagne in the later eighth century, and holding together, more or less, through the reigns of his successors until the late ninth century. This early medieval period–from about the 770s to the later ninth century–is usually referred to as the Carolingian Renaissance.

As the term “Renaissance” might suggest, the period was distinguished by its rich and sophisticated literary culture. Modern scholars have long been interested in Carolingian intellectuals and in the centres that housed them, usually monasteries or cathedrals, or the courts of aristocrats and kings. [1] To date, studies have tended to focus on individual scholars or individual schools, and typically on the very small number of individuals or centres for which we have the most data. Crucially, these studies are text-based. Maps and diagrams might be used to illustrate or clarify a point, but rarely work to enhance data and provoke further discovery. This post interrogates two recent (and representative) text-based studies of early medieval scholars and intellectual networks. Through visualizing data captured from these studies, I explore the challenges encountered in reconstructing early medieval intellectual networks and demonstrate the limitations of text-based network analyses. As we shall see, these are of necessity selective, offering in consequence only partial or fractured glimpses of the whole. If we are to reconstruct a total intellectual network in all its complexity, we need to start exploring the possibilities offered by new digital tools, preferably tools that allow for dynamic visualizations of our data. In addition, I hope that this post demonstrates the power of the visualization process itself, a process that enables–even requires–us to ask different and larger questions of our data.

What we know about Carolingian intellectuals we know because they wrote–to each other, for each other, occasionally about themselves. Quite possibly the richest source of information about Carolingian intellectuals–their interactions, connections, and networks–are the collections of letters that survive from the period. [2] The two studies I have chosen to interrogate in this post use letters as their primary source for establishing and discussing connections between Carolingian intellectuals.


Letters from the Classroom

The first study is an article by John J. Contreni entitled “The Carolingian School: Letters from the Classroom”. [3] While Contreni did not explicitly set out to reconstruct an intellectual network, his piece is useful for our purposes because of the wide selection of scholars featured in it. From the intellectual scene Contreni sketches, networks emerge.


Diagram 1

Diagram 1, constructed using Wordle, serves to introduce the more than forty individuals whose connections are discussed in Contreni’s paper. By counting the number of times each individual’s name occurs in Contreni’s text, I was able to produce a word cloud that provides a sense of the relative influence and importance of each intellectual, at least for Contreni. This representation of Contreni’s “network” gives an impression of the size of the scholarly crowd, but is most useful for identifying those individuals within it that loom largest in his study: Lupus, Ermenric, Alcuin, Hrabanus, and John Scottus.

Two of these names–John Scottus and Ermenric–achieve prominence by the accident of Contreni’s argument, as it were. Contreni’s article was first published in the proceedings of a conference devoted to the Carolingian philosopher John Scottus, which explains the relatively high number of references to this important scholar. These references to John Scottus are almost entirely confined to the opening and closing paragraphs of Contreni’s article, however. They belong merely to the frame and not the main picture. We know so little about the where and when of John Scottus that it is almost impossible to connect him with any other known Carolingian intellectual. Ermenric is another matter. A monk originally from the monastery of Ellwangen, he is perhaps most famous for one long and rather eccentric letter. [4] Contreni’s discussion of the letter over several pages of his article is the reason Ermenric’s name has made it into the top five. If we filter out John Scottus and Ermenric we are left with three names:

  1. Lupus (c. 805-862), a monk and later abbot of the monastery of Ferrières in north-central France;
  2. Alcuin (†804), an Anglo-Saxon cleric, and one of the most famous and important scholars and teachers of the Carolingian age. He became an adviser to Charlemagne, a teacher at his court, and later abbot of the monastery of Tours.
  3. Hrabanus (c.780-856), a monk at, and then abbot of the monastery of Fulda on the eastern edge of the Carolingian Empire. He was later elected to the archbishopric of Mainz.

These three men are, to medievalists, the “usual suspects” when we talk about scholarship and education in the Carolingian Renaissance. Their prominence in the Wordle diagram of Contreni’s article is not surprising.

Diagram 2

Diagram 2

Diagram 2, created using Gephi, an open-source visualization platform, better illustrates emergent networks in Contreni’s piece. It shows more clearly how Lupus, Alcuin and Hrabanus achieved their prominence in the word cloud. Diagram 2 is based purely on the letters used by Contreni to explore relationships between scholars. Each edge in the diagram represents a letter.

Visualizing Contreni’s data in this way is interesting because it suggests relative sizes of network: Lupus enjoys the largest, then Alcuin, then Hrabanus. It is also striking that these three scholars and their networks appear not to intersect with each other, or with anyone else featured in the diagram. A curious result because, as every historian of the early middle ages–and certainly Contreni–knows, Alcuin taught Hrabanus who in turn taught Lupus, yet these particular connections aren’t evidenced in the letters selected by Contreni for his article. Instead, when we visualize Contreni’s data, separate egocentric networks emerge around each letter writer. This, I would argue, is typical of the way historians tend to approach early medieval networks. The reality is far more complex.

I will return to the problem of selectivity in text-based network studies later. First, let us consider two important factors that tend to get taken for granted, downplayed, or even overlooked in network studies: the location of the individuals in question, and the dates when they were active.



Most of Contreni’s scholars can be geographically located because we know the institutions with which they were affiliated. The letters themselves more often than not begin with an address mentioning the name, rank and affiliation of both the sender and recipient.

Diagram 3

Diagram 3

Diagram 3 locates Contreni’s scholars on a map of Europe (custom constructed using Tableau). [5]

The names in red are the senders of the letters studied by Contreni; those in green the recipients. Arrows linking individuals show the direction of correspondence. Alcuin as abbot of Tours, Lupus as abbot of Ferrières, and Hrabanus as abbot of Fulda are rendered in larger letters. Less easy to read in this rendition are those places emerging as important focal points in the Carolingian intellectual network. Notice the cluster of scholars at Fulda, for example: Hrabanus, Rudolf, Ercanbert and Meginhard; the smaller cluster at Corbie of Ratramnus and Hildemar.

But the act of placing an individual on a map involves in some cases guesswork, in most cases, choice. Some correspondents–the gaggle of students associated with Alcuin, for example–are impossible to place. They float on this map, unmoored from any particular institution. In contrast, many of the scholars represented here were associated at one time or another with several places. The Carolingian educational system allowed gifted young monks and clerics to travel and study with masters outside their own institutions, for example. This is how Hrabanus came to be a student of Alcuin, traveling first to Charlemagne’s court in Aachen, then making a second trip to Tours before returning to Fulda (see Diagram 4 below). It is how Lupus came to be a student of Hrabanus, traveling from Ferrières to Fulda where he remained for about seven years before returning to his home monastery. From Ferrières Lupus made additional trips to Aachen, to the court of Charlemagne’s successor, the Emperor Louis the Pious. Monks like Lupus traveled. Not all of the letters composed by Lupus and represented on this map were written at Ferrières. Some–those to Einhard whom I’ve placed in Seligenstadt–were written while he was Hrabanus’s student at Fulda. The map has simplified a situation that was more dynamic and complex, but in doing so it reflects the common practice of labeling individuals according to the place with which they had the closest (or longest) association: hence medievalists routinely refer to Lupus as “Lupus of Ferrières”.

In reality most of the men on this map had multiple associations: Gottschalk “of Orbais”, for example, started out his career as a monk at Fulda, and can be counted among Hrabanus’s many students. But Orbais is only one out of many places that hosted him over the course of his troubled life. Christian “of Stavelot” began his career as a monk at Corbie, as did Odo, later Bishop of Beauvais. Arn, Bishop of Salzburg, and probably one of Alcuin’s closest friends (to judge from the large number of letters Alcuin wrote to Arn) was also abbot of St. Amand, a monastery much closer to Tours than Salzburg. If we are to do justice to the networks these men forged and maintained, we ought to be able to take account of their multiple associations. Text-based scholarship, for clarity’s sake, must needs fall back on familiar labels (rather like the modern practice of using surnames). Digital representations of networks need to find more dynamic solutions.



In considering an individual’s multiple associations, I have already introduced the notion of change over time. The problem with the diagrams so far is not only that they restrict individuals to one particular place, they also fix these scholars at a particular moment in time, at that moment in their careers when they composed or received the letter selected by Contreni for discussion in his paper. Alcuin is the teacher here. His students remain fixed in this scholarly network as students, despite the fact that some went on to enjoy distinguished careers: Fridugisus, for example, after a stint at Charlemagne’s court, became abbot of Tours after Alcuin’s death. Ermenric is captured here as monk of Ellwangen in correspondence with Grimald, but Ermenric, his eccentric letter notwithstanding, went on to become Bishop of Passau, a diocese in lower Bavaria, Germany. The status and scholarly role of these individuals evolved over time, generating new sets of connections that aren’t recorded, or even hinted at in these network diagrams.

But our diagrams of Contreni’s data run into far deeper temporal problems. As is perhaps already clear, Contreni was not concerned to map scholarly connections and exchanges over time, but rather to discuss some of the features and major themes in Carolingian education more generally. The language and content of the letters were what mattered most to Contreni, not the dates when they were written. If we were to make a time-line of his article according to the date of the letter he discusses at any given moment, where the Y axis is the date of the letter, and the X axis the page in his article where Contreni uses that particular letter, the resulting graph would run a jagged course of peaks and troughs. Diagram 3 reflects Contreni’s use of his sources, but the following video adds a temporal dimension to his data. Date range is indicated at the centre of each slide; individuals writing and receiving letters are again coloured red (writers) or green (recipients).

The display focuses our attention on the connections discussed by Contreni as these emerge decade by decade. But this more dynamic visualization of his data need not efface the other individuals alive (or likely to have been alive) and active at the same time. I have retained these scholars in the maps, albeit greyed out. Additionally, with data rendered in a time-line, we can achieve more precision in locating those individuals known to have moved around over the course of their careers.

Viewing Contreni’s data in this way begs a couple of questions. The connections he features can’t have been the only intellectual connections made: what are the other people, the greyed-out intellectuals, doing? And can’t we add more scholars to these maps?

As already noted, Contreni never intended in this relatively short paper to provide an exhaustive treatment of his topic, but rather to give a flavour of the relationships, themes, and educational strategies evident in the letters he sampled. One of the challenges of representing a network in a short written piece is that one must be selective. But in fact the length of the study makes little difference when it comes to selectivity as my second case study demonstrates.



The second study I’d like to consider is a recent and lengthy volume devoted to the topic of medieval intellectual networks. [6] Its author, Sita Steckel, investigated scholarly networks from about 800 to the twelfth century, devoting around 500 pages to Carolingian intellectual networks. As with Contreni’s paper, letters between individuals were one of her primary sources, as were letter-prefaces in which scholars dedicate their books to one another; she also used personal poetry and biographies (or vitae) of individuals, where these exist. The quantity of written materials available is large, so Steckel was strategic: the networks she focused on for the Carolingian period are those of just three scholars, our usual suspects: Alcuin, Hrabanus, and Lupus.

For the remainder of this post, I will focus on Steckel’s treatment of Hrabanus’s network. Diagrams 4 and 5 represent data from two different sections of Steckel’s book.

Diagram 4

Diagram 4

Diagram 5

Diagram 5

Diagram 4 visualizes the data from a section concerned with the earliest phase of Hrabanus’s career. It reveals the connections Hrabanus made through travel to Charlemagne’s court (Alcuin, and the important scholar and bishop Theodulf), and Tours (further study with Alcuin).

Diagram 5 displays the different sets of connections forged and/or maintained through letters by Hrabanus in his capacity as scholar and exegete, from 822 when he was elected abbot of Fulda to his death in 856. It helps us answer, at least in part, the questions raised by the time-line view of Contreni’s data, namely what else was going on? From the 820s until his death as Archbishop of Mainz, Hrabanus wrote many more than three letters!

But even Steckel must be selective. Diagram 6 represents Hrabanus’s entire network of correspondents — as complete a picture as we can gain of it, at any rate, from the letters that have survived. [7]

Diagram 6

Diagram 6

The circled names are the only ones featured by Steckel in the sections discussed above. The names in purple circles are all members of the Carolingian royal family, for whom Hrabanus wrote and/or dedicated treatises and biblical commentaries. It is tempting to suggest that these letters, and the connections to which they attest, feature prominently in Steckel’s discussion because they have been the recent focus of scholarly attention. [8] The eleven green-circled names are all monks, abbots or bishops. Just as many monks and bishops, and a layman too (Count Eberhard of Friuli), all of whom were also the recipients of Hrabanus’s books or treatises, are omitted from the network Steckel reconstructs.

One further obstacle to a complete view of Hrabanus’s network in Steckel’s study is the fact that she presents his network in pieces, laying out fragments of it for us over the course of several sections (as reflected in diagrams 4 and 5 above). A larger, more general observation about her book as a whole is that she trains a spotlight on one scholarly network at a time. In her analyses of Hrabanus’s network, we catch no hint or glimpse of ways in which his network might have overlapped with someone else’s. The same is true of her treatments of Alcuin and Lupus. To a large extent, we might blame this on the nature of the source evidence. The large letter collections that survive for Hrabanus, Alcuin, and Lupus keep our attention firmly fixed on the man at the centre of the correspondence, like a spider dominating its own web. What emerges from Steckel’s study of Carolingian intellectual connections–highly detailed, nuanced and rich though it undoubtedly is–is a series of fragmented egocentric networks. No attempt is made to merge those individual datasets into a more dynamic picture of Carolingian intellectual life.

The diagrams presented in this post represent an early stage of my Carolingian Intellectual Networks project. In many ways they aren’t an end product but an important part of the process of working out how medievalists currently do networks, and what needs to change. In conclusion, I would argue that if we are to reconstruct networks of early medieval intellectuals, but especially if our goal is a total intellectual network, text-based studies are not the most appropriate medium. We need to build digital tools that can handle large data sets and allow more quantitative analysis of our data in the aggregate. We need tools that can merge multiple networks, generated from different types of sources, not just the literary/textual evidence of the sort I’ve showcased here, but material and physical evidence also, like the manuscript copies of the books that are the subject of so many of Hrabanus’s letters, for example. It is even possible now, using tools like ORBIS, to plot the routes our scholars might have taken as they traveled from St. Amand to Salzburg, or from Tours to Aachen. Only by merging and layering all these different kinds of network will we come close to understanding the complex reality of Carolingian intellectual life.


Note: All images and the embedded presentation above were produced, and are owned, by the post’s author.

[1] Most intellectuals for which any evidence survives were men. We know of only one female writer, Dhuoda, who composed a handbook for her son in the early 840s. Women, usually royal women, often nuns, were the recipients of books and treatises by men, including some of the men featured in this post. Women were also active as scribes, copying manuscripts. For a useful brief survey of female intellectual activity see John J. Contreni, “The Carolingian renaissance: education and literary culture”, in Rosamond McKitterick (ed), The New Cambridge Medieval History (Cambridge: Cambridge University Press, 1995), 717-20.

[2] Nearly all extant early medieval letters have been edited, mostly in the late 19th century, as part of a series entitled Monumenta Germaniae Historica (Historical Records of Germany). The series contains primary sources for the history of post-classical, pre-modern Germany (broadly interpreted), and the letters appear in six volumes, all now accessible online here. At least one Carolingian letter collection, that of Lupus of Ferrières, has been edited again more recently, by L. Levillain, Correspondance (Paris: H. Champion, 1927-35), and by P. K. Marshall, Servati Lupi Epistolae (Leipzig: Teubner, 1984). Lupus’s letters are also available in English translation: Graydon W. Regenos, The Letters of Lupus of Ferrières (The Hague: Martinus Nijhoff, 1966).

[3] John J. Contreni, “The Carolingian School: Letters from the Classroom” in Giovanni Scoto nel suo tempo. L’organizzazione del sapere in età carolingia. Atti del XXIV Convegno storico internazionale, Todi 11-14 ottobre, 1987 (Spoleto: Centro Italiano di Studi sull’Alto Medioevo, 1989), 81-111.

[4] Addressed to Grimald of St. Gall: Ernst Dümmler (ed) MGH, Epistolae V, Karolini Aevi III (Berlin, 1899), 536-79; available online here.

[5] I would like to thank Eric Monson for his assistance in preparing the maps for this post, and for all his advice with the visualization process.

[6] Sita Steckel, Kulturen des Lehrens im Früh- und Hochmittelalter. Autorität, Wissenskonzepte und Netzwerke von Gelehrten (Köln: Bölau Verlag, 2011).

[7] For Hrabanus’s entire correspondence see Ernst Dümmler (ed) MGH, Epistolae V, Karolini Aevi III (Berlin, 1899), 379-516; available online here. For a useful explanation of why (and how) certain medieval letters survive while others don’t, see Mary Garrison, “‘Send More Socks’: On Mentality and the Preservation Context of Medieval Letters”, in Marco Mostert (ed), New Approaches to Medieval Communication. With an Introduction by Michael Clanchy = Utrecht Studies in Medieval Literacy 1 (Turnhout: Brepols, 1999), 69-100.

[8] See Mayke De Jong, “The empire as ecclesia: Hrabanus Maurus and biblical historia for rulers”, in Yitzhak Hen and Matthew Innes (edd) The Uses of the Past in the Early Middle Ages, (Cambridge: Cambridge University Press, 2000), 191-226.

One Response

Leave a Reply

Your email address will not be published. Required fields are marked *