Toponymic surnames and the spatiality of heresy prosecutions: Peter Seila’s register of sentences from the Quercy region (Languedoc), 1241–1242

Abstract

Our knowledge of the geography of medieval religious dissent and its repression in the Latin West is limited by a lack of systematic study of locational information in inquisition trial records. Spatial analysis of these rich details has the potential to help build a bottom-up picture of interaction between dissidents and inquisitors that moves beyond institutional perspectives. This task is rendered challenging due to the inconsistencies and uncertainties of what inquisitors and their notaries typically recorded about the spatial associations of suspects. Probably the most common indicator of such associations found in inquisition records are toponymic surnames. They present challenges of coverage (not everybody had a toponymic surname) and interpretation (multiple possible meanings). This study attempts to tackle the challenge of interpreting such surnames within the context of the nine sentencing events held by the inquisitor Peter Seila in 1241 and 1242 in the Quercy region of Languedoc: covering 650 sentenced individuals, the register documenting these events is the earliest extant record of an inquisition of such scale. Rather than taking the interpretative challenges of toponymic surnames as reasons to limit ourselves to qualitative analysis, our approach shows the value of rendering and analysing them as structured data. Firstly, we quantify the context of toponymic surnames, placing them against the background of broader name construction practices and other social factors. Secondly, we plot and analyse the geocoded data derived from toponymic surnames with the benefit of this contextualisation, looking especially at the distance of toponyms from their associated sentencing centres, in order to derive narratives that best explain the generality of their meaning. The results allow us to appraise the actual spatial coverage of the nine sentencing events. The first two, centred on the important towns of Montauban and Moissac, seem most likely to have been primarily urban affairs, with little evidence of rural coverage. The remainder, which took place in castra (fortified villages), appear to have covered more of the surrounding countryside. These results geographically contextualise the reports of dissidence conveyed within Peter’s register, and suggest narratives for how Peter optimised his strategy for impact in the face of constraints.

Introduction

Our knowledge of religious dissidence in high and late medieval Europe largely derives from its prosecution by churchmen. The documents they have left behind, predominantly records of inquisition trials, are concentrated in certain regions — Southwestern France (Languedoc), Northern and Central Italy, Germany, England (to a lesser degree), and Spain (from the fifteenth century) — and certain localities within them. This is not just a result of uneven manuscript preservation. Some places simply received more intense attention from heresy prosecutors (Duby 1968, 399; Zbíral and Shaw 2022, 4).

This uneven coverage had various influences. At least at a local level, it is not unreasonable to suppose that prosecuting inquisitors were influenced by the actual spatial distribution of what they perceived as heresy as they followed the leads that they found. But following leads requires the ability and will to detect them. That inquisitors did not investigate all places evenly was not simply a function of where dissidence might be perceived, as the lack of any known investigation in vast swathes of Europe makes clear. The local emergence and subsequent reach of heresy inquisitions was strongly conditioned by various circumstances. The attitudes and capabilities of lay authorities and their relation to church institutions mattered, as did more prosaic factors, such as ease of travel and the resources and time available for investigation. Such factors affected not only whether inquisitors were appointed at all for a particular region, but every stage of their work if they were (Zbíral and Shaw 2022, 4).¹

Nevertheless, our judgements on the coverage of heresy trials usually remain more based on feel than systematic analysis of available data. While some aspects of the total “geography of heresy” — or perhaps, more proximately, geography of inquisition — that the Annales-school historian Georges Duby called for in 1962 are stymied by the relative paucity of coinciding data on demography, topography and infrastructure, records of heresy inquisitions themselves do offer copious locational details, concerning the trial itself, the actions of suspects, and their geographic associations (Duby 1968, 402). As Ruth Slatter and others have recently shown, studying the ground-level spatial interactions that locational details represent can move us beyond overly institutional visions of past religious cultures, even when using sources of institutional origin (Slatter 2023, 5–6; Jakubowski 2023). That such spatial studies of heresy trials are lacking is partly due to the laboriousness of collecting these details as structured data. Moreover, even taking just one category of information found within inquisition records — for instance, the places where those prosecuted lived, which will be the primary focus of this study —, there are significant interpretative challenges. The notaries working with inquisitors offered different levels of information about where people came from, if they provided any at all. Quite frequently, much of the extant locational information derives from toponymic surnames, typically in the form of “Someone of somewhere” (for instance, “Ricardis Fabrissa del Pruliaco”). Such toponymic surnames could signify many different things: a current residence, a personal origin, a family origin or association, among other possibilities (Emery 1952; Emery 1955; Lopez 1954; Chareille 2008, 178–180). In relation to an inquisitorial trial, they tell us something about personal mobility, but what kind exactly — the inquisitor’s reach to where suspects resided, or the migrations of suspects or their ancestors towards trial centres — is not immediately clear.

Such issues of relating people to places present significant obstacles both for the transformation of geographic information found in the records into structured data and for systematic analysis: it is thus understandable that historians have tended to treat these details only in qualitative discussions. But the difficulties can also be looked at from a more constructive aspect. Treating the locational information in the registers systematically forces us to evaluate the amount of missing data and the elements of uncertainty more precisely. Careful quantitative diagnostic work around these uncertainties may produce important interpretive knowledge in and of itself: understanding what we do not know has the potential to tell us much about the production of the information we do know. This understanding in turn helps us to interpret such spatial data in their generality, even if individual examples retain their uncertainty.

This case study takes up the challenge of capturing, plotting, and analysing information concerning the spatial coverage of inquisition trials at a relatively local level, focusing on a specific register: the record of those penances meted out to 650 individuals by the inquisitor Peter Seila (also spelled Sellan, Sella, Cellan) in nine locations across the Quercy region of Languedoc between 1241 and 1242.² While a document of significant historical interest — the earliest inquisition register to cover hundreds of defendants — it also embodies the difficulties mentioned above: crucially, the great majority of the information on the locational ties of the sentenced, beyond their place of trial, comes from toponymic surnames. The analyses presented here show how handling such information and its uncertainties in a formal manner can open opportunities. Firstly, the transformation of names into structured data allows us to quantify the context of toponymic elements, not only measuring their prevalence among the sentenced at each of the sentencing events, but also placing this in the context of broader name construction practices and other social factors. Secondly, plotting geocoded toponymic surnames with the benefit of this context and analysing their distance from the trial locations (a proxy for travel time and difficulty) enables us to refine our understanding of what kind of mobility they speak to. This approach produces new reflections on the source, the local societies affected by religious dissidence, and, above all, Peter Seila’s variable reach in his prosecutions across Quercy.

Source material: background, questions, and challenges

Peter Seila’s register of penances (1241–1242) and the geography of Quercy

Peter Seila, a key figure in the early Dominican order, was one of the first inquisitors appointed by the papacy in 1233 to root out heresy from Languedoc in the aftermath of the Albigensian Crusade (Feuchter 2007, 257–284; Dossat 1959, 122; Pelhisson 1994, 44). The register of the sentences he handed out in the Quercy region (1241–1242), which survives in a seventeenth century copy, is the earliest trial document with large-scale coverage to emerge from the work of the Languedocian inquisition, coming at the end of a period of challenge. Due to opposition and obstinacy centred on Raymond VII, count of Toulouse, the inquisitors had been temporarily ejected from Toulouse in 1235, and Gregory IX had suspended their activities between 1238 and 1241 (Feuchter 2007, 292; Dossat 1959, 131–145). Despite such setbacks, Peter did not shy from delving quite deeply into the local society of Quercy, a county lying just to the north of the Toulousain. His register documents the sentencing of 650 people, 391 men and 259 women from a wide variety of social backgrounds, who were deemed to have supported and/or believed dissident ministers. Primarily, we find those who interacted with the heretici (the “heretics”, ascetic religious specialists in what later scholarship has often called Catharism) and/or the valdenses (the Waldensian brothers and sisters, named after the founder of the movement, Valdès) (Duvernoy 2001, 25; Taylor 2011, 7–11).

Quercy was not a geographically cohesive region at the time of Peter’s work there, even if its broad boundaries were recognised by contemporaries. The part to the east of the episcopal seat of Cahors, an area which Peter did not investigate at all, was sparsely populated, owing to high elevation and poor soils. The remainder of the region, meanwhile, was home to unevenly settled societies, often divided by forests, whose horizons were typically still quite local. Easy transportation ran only along the east-west axes of the major regional tributaries of the Garonne watershed: the Lot, Tarn and Aveyron (Taylor 2011, 51–61). While our knowledge of the region’s infrastructure and human geography in this period remains fragmentary, Peter’s inquisition nevertheless provides an intriguing case for systematic spatial analysis due to the register’s organisation. The text separately documents sentencing events that occurred in nine locations: Montauban and Moissac around Ascension Sunday 1241; Gourdon during Advent 1241; Montcuq, Beaucaire, Sauveterre, Montpezat, Almont, and Castelnau-Montratier during Lent 1242 (see Fig. 1).³ The locations were not all alike, nor were the scales of sentencing events associated with them. The southernmost, Montauban, was a major town (population estimated at around 8000 in 1300), situated at the confluence of the Tarn and Tescou rivers; it was one of Quercy’s two major commercial centres, alongside Cahors (Taylor 2011, 72–73; Feuchter 2007, 111–163; Bairoch et al. 1988, 27). The sentencing event there was the largest (256 sentenced individuals). Around 30 km west and also situated on the Tarn, Moissac (99 sentenced) was probably a somewhat smaller town in this period (population estimated at around 8000 by 1480), but one of significant religious prestige, having grown up around an important Cluniac abbey (Taylor 2011, 53; Fayolle-Bouillon 2001, 190). The remaining trial centres were significantly smaller — we lack population estimates for them — albeit not without local importance. The northernmost location, Gourdon (143 sentenced), was a well-developed castrum — in Languedoc, the term typically denoted a fortified village more than a free-standing castle (Taylor 2011, 74–76) — whose lords held significant political reach within the north-western parts of Quercy (Taylor 2011, 55–8, 71). The other six locations — Montcuq (84), Beaucaire (7), Sauveterre (5), Montpezat (22), Almont (23), and Castelnau-Montratier (11) —, smaller castra characteristic of the central Vaux de Quercy region (between the Lot, Tarn and Aveyron rivers), all fell somewhat under the influence of Gourdon (Taylor 2011, 53–55). Overall in 1241–1242, Peter thus appears to have shifted his focus from larger urban centres to much smaller settlements. While we do not know whether he visited every sentencing location, he would almost certainly have had to move from the south (Ascension 1241) to the north (Advent 1241) and then to the centre of Quercy (Lent 1242) to be within summoning range of those he sentenced. The spatial division of the register thus opens the possibility not only of studying the local variability of perceived dissidence and its repression under the same inquisitor, but to do so in the context of individual mobility rather than assumed institutional coverage.

Fig. 1: Overview of Peter Seila’s sentencing events; size of sentencing event icon is proportional to number sentenced.

Historical Quercy boundaries derived from the map of Jean Duvernoy (2001, 8).

Theorisations and unresolved questions

Theorisations concerning the spatiality of heresy and its repression in Quercy have already been made at a general level on the basis of Peter’s register. It has not escaped historians that the relative frequency with which people interacted with heretici vis-à-vis valdenses varies significantly between the sentencing locations. We thus have the impression of relatively widespread penetration of heretici across the part of Quercy covered by Peter, with valdenses found more in specific areas, above all around Montauban and, to a lesser degree, Montcuq and Gourdon (Duvernoy 2001, 26–27; Taylor 2011, 124–139, 161–164; Feuchter 2007, 226–231; Barmby 2017, 70–96). Claire Taylor (2022, 55–56) has suggested that the appearance of valdenses north of the confluence of the Garonne, Tarn and Aveyron was relatively new — the result of dissidents fleeing north from the Toulousain following the establishment of French royal officials there in 1229 in the aftermath of the Albigensian Crusade — whereas their larger presence at Montauban, in southern Quercy, was longer established. Beyond this, however, we know relatively little, despite obvious questions. The primary religious target of the Albigensian Crusade had in fact been the heretici, and those of their number who had survived would have been well-known as fugitives throughout Languedoc, including Quercy: were these dissident ministers more likely to interact with those who lived away from bigger urban centres? To what extent were the valdenses — still a somewhat secondary target — filling areas or types of areas in Quercy that the fugitive heretici might have abandoned? Any understanding of the spatial distribution of dissidence derived from the source, however, must be framed by whatever knowledge can be derived of the spatial variation in investigation and prosecution. Taylor, for instance, notes that while only two people who had contact with valdenses were sentenced in Moissac, these few mentions suggest that their ministers were many in number, which could mean that Peter did not penetrate the networks associated with the valdenses too deeply in the area (Taylor 2011, 130).

Broader theorisations concerning the spatial reach of Peter’s trials remain more conjectural. No documentary evidence survives of the preceding investigation and trial, beyond the fact that the register itself makes two vague references to a grace period of uncertain length and operation, in which people could come forward to confess.⁴ Jörg Feuchter offers the most comprehensive attempt to fill in the gaps that the register leaves. He posits that the 1241–1242 sentence register caps a chain of trial events leading back into the mid-1230s, pointing to Guillaume de Pelhissons’s Chronica, which states that Peter first visited Moissac in 1234/1235, some unstated locations in the region in 1235, and Montauban in 1236 (Pelhisson 1994, 56–58, 70, 90–92; Feuchter 2007, 292–304). In line with this image of long, deliberate action, Feuchter also takes the view that Peter was systematic in his approach to spatial coverage, noting that the nine trial locations accord almost precisely with the seats of the nine western bailiwicks of Quercy under the aegis of the count of Toulouse (Feuchter 2007, 294–295). The suggestion here appears to be that Peter attempted to cover an area, rather than just specific places. For Feuchter (2007, 296), Peter’s register represents “the complete document of an inquisition in which no local sections are missing … [Peter] systematically covered an area in a way that research has so far only assumed for the ‘great’ inquisition of Bernard de Caux that took place in 1245/1246 in the Toulousain Lauragais and its environs” (our translation).⁵

Feuchter’s observation relates to an institutional correlation. The coverage of the trials, however, was ultimately conditioned by individual movements of Peter and his suspects: should we actually believe that he covered the area of these bailiwicks completely and evenly? Building on her observation concerning the seemingly less vigorous pursuit of supporters of the valdenses around Moissac, Claire Taylor notes that “even such a vast inquest was not thorough” (Taylor 2011, 130). The dramatic variations in numbers between the trials might convey a similar impression. Population size differences between the bailiwicks must surely account for some of that: it is perhaps no surprise to see that Montauban, with its major urban population centre, was the place of the largest sentencing event, nor that the relatively proximate sentencing events in the bailiwicks governed by very small castra in the Vaux de Quercy typically feature smaller numbers. We might also attribute some variation to the levels of dissidence Peter actually perceived within the bailiwicks. But, even considering such possibilities, the wide differences in the scale of sentencing events jars with the idea that Peter’s register of sentences covered both the trial centres and the surrounding bailiwicks consistently. Might such variations suggest that the document represents only a partial result (with others perhaps sentenced by Peter at a different time and recorded elsewhere)? Or, in line with the often hampered nature of the first decade of the Languedocian inquisition, might the register be the product of a more selective procedure which found different reaches in different places?

The challenge of toponymic surnames

To gain greater definition on these interrelated issues — the spatial distribution of dissidence in its different forms and the reach of Peter’s ambitious investigation — it is necessary to go beyond comparing the nine local sentencing events of Peter’s inquisition at a general level. Systematically capturing, plotting, and analysing the known spatial ties of those tried in this register provides a constructive way forward.

As already briefly described, this analytical process is not without challenges. Peter Seila’s register does not provide information on the locational ties of the sentenced at all consistently and, crucially, what is provided is uncertain on multiple levels. Some uncertainties are general to the transformation of textual information from medieval sources into spatial data: these concern the identification of placenames and geocoding. Most centrally in this regard, associating medieval Latin placenames with known locations is not always possible with certainty or even at all (see “Identified Toponyms” section).

A more challenging type of uncertainty, however, concerns the actual association between people and places. In Peter’s register, almost all the information on where people might come from derives from toponymic surnames. These not only cover only a portion of the sentenced (see “Identified Toponyms” and “Apparent Toponymic Surnames” sections), but represent a thorny interpretative challenge. To take one example, how should we read a name such as “Arnaldus Guillelmi de La Viga”, a man sentenced at Gourdon? His toponymic surname literally means “of La Viga” or “from La Viga”. In this case, the toponym is at least readily identifiable with the village of La Vigan, some 5 km east of Gourdon. But what was Arnaldus’s actual association with the place? In the context of a medieval West where surnames of any sort only became common from the eleventh century (Bourin and Chevalier 1990, 7), toponymic surnames have often been read as reflecting personal origin, if not necessarily residence, since people could potentially move during their lifetimes. As a result, such surnames have been seen as useful evidence in studies of medieval migration to towns and cities (Chareille 2008, 178–179; Lopez 1954, 10). But, it has also been noted that the choice of toponymic surname sometimes reflected less direct associations and, most crucially, that surname inheritance was a potential feature from an early stage: any association might be generations out of date (Emery 1952, 45, 47, 49–50; Chareille 2008, 179–180). Medieval civil jurists themselves occasionally touched upon the fragility of the link between toponymic surnames and real geographic associations (Kezdar 1973, 123). Moreover, we must also acknowledge the potential influence of those who recorded the names within official interactions: one might suppose that officials would have preferred toponymic surnames reflecting current or recent associations, but how often they intervened to alter the personal or inherited choices of those whose names they recorded and with what exact purpose is difficult to reconstruct.

In the context of Peter’s register, however, there are reasons for optimism about the interpretability of toponymic surnames. The most challenging element of the association uncertainty is heritability, since it has the potential to sever any real connection between person and place. In Languedoc, however, while such surname inheritance is noted from shortly after surnames became common there in the eleventh century, it was far from the rule: surnames, and in particular toponymic surnames, were often updated (Bourin 1995, 193–194, 196–198; Mousnier 1995, 165; Billy 1995, 182). One study of the Toulousain region, covering the eleventh to fourteenth centuries, shows that it was not uncommon to find people with two toponymic surnames (and we find them too in Peter Seila’s register), the latter usually being a more up-to-date indication of residence, the former a personal or familial origin (Billy 1995, 180). While very limited, the other evidence concerning residence provided by Peter’s register only appears to contradict the association suggested by surname in a solitary instance: Ramunda de Mazerac’s surname refers to Saint-Jean de Mazerac, around 15 km east of Montpezat where she was sentenced, but the text also identifies her as a prioress of Lalécune, a priory of Augustinian canonesses around 7 km northwest of Montpezat.⁶ In four out of the nine other instances in which residence appears clearly stated within the source, toponymic surnames provide essentially the same information or reference a very proximate location. Bertrandus, a lord of Gourdon, was resident in the castle there.⁷ Guillelmus de Labarta (Labarthe) saw heretics in his estate at Labarthe.⁸ Gaillardus de Godor (Goudou) had a house at La-Bastide Murat, but this is just a few kilometres from Goudou.⁹ Guillelmus de Santo Genesio (Saint-Génies, around 3 km east of Montcuq) was described as a “lord of Laborde”, where he had an estate, but, similarly, this Laborde (Laborde-Haute according to Jean Duvernoy [2001, 158]), is little more than a kilometre west of Saint-Génies.¹⁰ Regarding the other five cases, three surnames are non-toponymic — Petrus Peregrini, who lived in Gourdon and was sentenced there; Ademarius Civada, who lived in Villeneuve, which Duvernoy (2001, 59) describes as a quarter of Gourdon, where he was also sentenced; Stephanus Sobressen, a tenant of a certain mansus del Poi (“Mas del Pech”), an unidentifiable estate, sentenced at Almont — and thus offer no clue.¹¹ The other two appear toponymic (Ramunda, wife of “de Bosolens”, whose husband had a house in Montauban where she was sentenced; Hugo de Portu, said to have had a house in Corbarieu, around 8 km south of Montauban where he was also sentenced) but are not geolocatable and thus offer no confirmed contradiction: it may even be that “de Portu” (“of the harbour”) references part of Corbarieu, which was on the river Tarn, and “Bosolens” somewhere in Montauban or its very close environs.¹²

Information from external sources on the sentenced, while only forthcoming for the more elevated individuals, shows a similar pattern. A number of the nobles from the Gourdon region identified by Claire Taylor had toponyms associated with their seat of power.¹³ In the case of those with double toponymic surnames, such as the members of the “de Engolesme de Milhac” branch of the “de Engolesme” (Goulême) family, it is the second toponym (Milhac, around 8 km north of Gourdon) that provides indication of the location most currently associated with them, following the pattern witnessed in the aforementioned Toulousain evidence (Taylor 2011, 213). We know that several families tried at Montauban — the Carbonnel, the Engelbaudi, the Folcautz/Folcaldi, the de la Mota — owned properties in the heart of town and several others alongside these — the d’Aussac, Geraldi, Lauteri, de Sapiac, de Castilho — provided consuls for the town (Feuchter 2007, 193–202). But most of these had surnames that are either non-toponymic, non-geolocatable (de Castilho), or reference a suburb (de Sapiac). In the case of the noble de la Mota family, meanwhile, there is no evidence that they had entirely ended their residential association with the seat most probably (in Feuchter’s view [2007, 199]) suggested by their surname — Château de la Mothe, around 20 km to the southeast of Montauban near La Salvetat —¹⁴ even if all or some of them also lived in Montauban.

There are thus some good grounds for using locational data derived from toponymic surnames as an indicative proxy for the coverage of Peter Seila’s trials. While individual data points must be treated as potentially multivalent, their generality deserves analysis as an indicator of the spatial distribution of the sentenced. Their arrangement in relation to the trial locations may in turn allow us to weigh the probability of interpretative narratives and place bounds on our uncertainties.

Nevertheless, the aforementioned challenges mean that strong contextualisation of toponymic surname data is critical to grounding any potential insights. As mentioned, toponymic surnames were not the only type of surname; surnames based on a personal name (for instance, “Bernardi”) also commonly appear, as do, more rarely, those with roots in an occupation (such as “Textor” [“Weaver”]). Moreover, even where a surname is clearly toponymic, it may not be geocodable. Understanding the proportion of those sentenced who possessed toponymic surnames, and within this, the proportion whose surnames can actually be geocoded, helps us to understand the amount and nature of missing locational data and thus to avoid misinterpretation of the available information.

More positively, these proportions, their relationship to wider patterns of name construction and other socio-cultural factors, and the variance of these things across the sentencing events may suggest something concerning local society, the inquisition process, and/or the interaction between the two. As a starting point, the prevalence of toponymic surnames must be seen in the context of broader anthroponymic preferences, above all the number of name elements typically used. The relative prevalence of nobility at the different sentencing events should also be taken into account, since it has been noted in other research on French localities that toponymic surnames were particularly common among (if far from the exclusive preserve of) nobles in this period (Barthélemy 1990, 38–41; Mousnier 2002, 72–76; Bourin 1995, 192–193). But one might also wonder about the influence of local geography itself — or, more strictly, the local geography covered by Peter’s inquisition — on naming preferences. Might toponymic surnames have been more prevalent in predominantly rural areas, perhaps due to heightened relevance as an identifier among such populations? Here, seeing how patterns of toponymic surname prevalence across the sentencing events align with the spatial distribution of mapped toponyms may be revealing for what those surnames typically mean.

Transforming toponymic surnames into structured data

Using data principally derived from toponymic surnames to study the spatiality of heresy prosecutions represented in Peter Seila’s register thus presents challenges but also opportunities that might be exploited. In order to unlock potential insights into both the dispersal of dissidence in Quercy and the reach of Peter’s investigation, we need not only to plot and analyse toponymic surname data spatially, but also to quantify the context from which the data derive and to explore these findings in conversation with each other. Our dataset (n = 650) thus represents toponymic surnames as structured and geocoded data. It also captures data on name construction as well as other pertinent details concerning the sentenced.¹⁵

Identified toponyms

During the initial data collection, any surname that looked like it might be toponymic — most commonly preceded by “de” — was carefully checked for matches with the Latin names of nearby locations, assisted by the annotations of Duvernoy (2001) in his edition of the register and other resources on local toponyms (e.g., Taylor 2011; Feuchter 2007; Albe 1910, 283–288; Fénié and Fénié 1998; Cassange and Korsak 2013. Longnon 1874; Perrin and de Font-Réaulx 1972). Where a match was found, we recorded both the Latin and modern French name of the location, as well as latitude and longitude coordinates. Where the person had two geocodable toponymic surnames (such as “de Engolesme de Milhac”), we captured both within the dataset, retaining the order they appear in the text (“Identified Toponym 1” and “Identified Toponym 2” column sets). If there were multiple candidate locations for the same toponym, we coded what we judged the most plausible geolocation but recorded it as ambiguous (with a “y” in the “Ambiguous?” columns) and wrote a textual note (“Notes” column) concerning our choice. Overall, 148 (22.77%) of the 650 sentenced had at least one geocoded surname (prior to the addition of information from identifying relatives, as discussed below, which increased this number), with 155 surnames (including second surnames) geocoded in total; of these geocoded surnames, 30 were ambiguous.

Apparent toponymic surnames

As expected, not all surnames that appear toponymic were readily geocodable. Nevertheless, it is necessary to know the prevalence of all apparent toponymic surnames both to understand the level of geographic data missing due to identification failure and to grasp the overall place of toponymic surnames within local naming practices. We thus present an “Apparent Toponymic Surnames” value for each person in our dataset, counting surname elements with a toponymic appearance whether they are geocoded or not: thus “de Portu” (non-geolocatable) = 1, “de Engolesme de Milhac” (both toponyms geolocatable) = 2. Overall, 289 (44.46%) of the 650 sentenced had at least one apparent toponymic surname (prior to the addition of information from identifying relatives, as discussed below).

Major name components

In order to understand how toponymic surnames relate to overall name construction, we recorded a “Major Name Components” value for each sentenced person. This counts the number of components in a name, excluding prepositions and honorifics (in Languedoc, most commonly “En” for men and “Na” for women). Overall, 471 individuals (72.46%) of the 650 sentenced had a name consisting of two major components, while 44 (6.77%) had three or more. 142 (20.62%) had just a single name; 93 of these (92 women, 1 man), however, are identified in the text by their relation to a surnamed relative (see below). Similarly, one woman (0.15%) had no personal name, but was simply labelled as “the wife of Arnaldus Tabart”.¹⁶

Beyond this, we have also classified each person’s name in line with the schema of Monique Bourin and Bernard Chevalier (1990, 11–12) used throughout the important Études d’anthroponymie médiévale series (“Bourin-Chevalier Classification” column). While this does not feature in our analysis, it classifies other name constructions that are not in our focus and renders our dataset more accessible to scholars of medieval European anthroponymy.

Identifying relatives

106 sentenced individuals (98 women, 8 men) are identified with reference to a relative, almost always male (100 cases): in 89 cases, it is a “wife of” relation. While the men with identifying relatives usually have their own surnames (7 cases, in 6 the same as the relative), the women almost always lack a surname of their own (93 cases). Identifying relatives can thus provide important substitute data, and their details are included in the “Identifying Relative” columns of our dataset, which mirror those concerning the naming of the sentenced person. Substituting data from identifying relatives, a further 25 individuals (all women) can be linked to a geocoded surname, taking the total to 173 (26.62%) of the 650 sentenced; a further 42 (all women) have at least one apparent toponymic surname, taking the total to 331 (50.92% of total sentenced). Inferring surnames from such substitutions also reduces the number of individuals with only one name component to 51 (7.85% of total sentenced); 547 have two name components (84.15%), while 52 (8%) have three or more.

Supporting data

In order to analyse toponymic surnames against their socio-cultural background and the context of Peter’s trial, our dataset also contains other details concerning the sentenced individuals. Where they were sentenced (“Sentencing Location” column) and their sex (“Sex” column) were obvious necessities. Given our interest in how dissident sect alignment might correlate with toponymic surname locations, the dataset also shows (in the “Sect Interaction” column) whether these individuals interacted only with heretici (“H”; 359 individuals, or 55.23% of the total 650 sentenced), only with valdenses (“V”, 188 individuals, 28.92%), both sects (“B”, 99 individuals, 15.23%), or whether their interactivity is not clearly stated (“NS”, 4 individuals, 0.62%). We also provide an indication of the noble status of the sentenced (if known) given its potential influence on the possession of toponymic surnames. Here we were not led by the use of the aforementioned Languedocian honorifics (“En”, “Na”), since these could also be applied to non-nobles (Biller et al. 2011, 125), but rather relied on existing research concerning the sentenced individuals and their families (esp. Albe 1910, 283–288; Taylor 2011; Feuchter 2007). While some nobles have surely been missed or misidentified despite the best efforts of researchers (in part due to the somewhat unclear boundaries of noble status), the data in the “Noble” column (“y” if nobility has been stated by a historian, otherwise blank) allows for an indicative comparison of noble prevalence across the sentencing events and for the identification of noble toponymic surname associations in our mapping. Overall, 93 (14.31%) of the 650 sentenced were identified as noble.

Analyses and discussion

Analysis of the geography of Peter Seila’s heresy prosecutions requires a two-fold approach, with as much effort given to understanding the context of toponymic surname data as to spatial analysis. Our first step is thus to quantify that context across the different local sentencing events, presenting tables that allow us to see the data derived from toponymic surnames amid the anthroponymic practice and other socio-cultural features of those sentenced. The second is to map and analyse the geocoded toponymic surname data with the benefit of that context. Understanding the distance of plotted toponyms from the different sentencing centres allows us better to theorise what sort of mobility — the migration of suspects / their ancestors or the reach of the inquisitor — they most likely represent, and thus to interpret the spatial distribution of dissidence that Peter perceived and the coverage of his investigations.

Name construction analysis

Tables 1–3 provide a local breakdown of toponymic surname prevalence in the context of broader naming practice (Table 1 takes the names of the sentenced in isolation, Table 2 includes identifying relative substitutions) and other socio-cultural details concerning the sentenced (Table 3). Here we have chosen to combine the five sentencing events with the fewest sentenced people — Beaucaire (7), Sauveterre (5), Montpezat (22), Almont (23), and Castelnau-Montratier (10) — to produce a total (68) more commensurate with the others. This is also justified by their proximity within the Vaux de Quercy and occurrence within the same period (Lent 1242).

Table 1 Toponymic surname prevalence and name construction: sentenced individuals in isolation.

Table 2 Toponymic surname prevalence and name construction: with identifying relative substitutions.

Table 3 Other socio-cultural details.

To begin with broader naming practice, it is clear that the model of a first name and a single surname (that is, two components, not counting prepositions and honorifics) is dominant. Table 2, which uses the data of identifying relatives, where stated, to infer surnames for those sentenced who otherwise lack them, shows that this model appears quite consistent (c. 85% prevalence across all sentencing events), suggesting that this reflected the standard naming practice within Quercy and/or among Peter’s notaries, who may well have been locals too.

Table 1 (sentenced people without identifying relative substitutions) shows less consistency, as one might expect, given the higher prevalence at some sentencing events (especially Moissac) of single-named individuals identified by someone else’s name. As mentioned, these single-named individuals are almost always women: before identifying relative substitutions, women have a mean name length of 1.54 components vs. 2.07 for men, whereas after such substitutions the gap is much smaller (1.89 vs. 2.08). An interesting outlier here is Montcuq, which shows relatively little difference between the name length figures in Tables 1 and 2. There, the percentage of sentenced individuals with a single name is just 11.90% even before identifying relative substitutions, despite Montcuq having the highest percentage of female sentenced of all sentencing events (see Table 3): this is because women here have a mean name length of 1.78 components (vs. 2.12 for men in the region) before substitutions (which reduce the gap a little further to 1.90 vs. 2.12). This seems less likely to represent a difference in cultural practices towards identifying women in and around Montcuq — the geographically proximate “All Other” sentencing events do not produce similar data — than an inquisitorial and/or notarial decision connected with the comparatively high prevalence of women sentenced there (for instance, saving time and writing materials by appending surnames to women directly).

Despite the evidence of relatively consistent cultural and (for the most part) notarial practices concerning name length, we do find some noteworthy divergences in the prevalence of toponymic surnames across the sentencing events. If we might note a generally similar level of apparent toponymic surnames (regardless of whether they could be geocoded) at Gourdon, Moissac, and All Other (around 55–60% after identifying relative substitutions, as per Table 2), Montcuq (71.43%) is elevated from this level, while Montauban (35.16%) is a clear outlier in its low prevalence. The geocoded toponymic surname figures are still more inconsistent: this is not unexpected, given the vagaries of geocoding on the basis of centuries-old Latin labels. The result for Moissac here (14.14%, after identifying relative substitutions), however, is particularly noteworthy, since this represents a very low level of successful identifications in comparison to the other sentencing events: it is markedly lower than the figures for Gourdon (42.66%) and All Other (30.88%) despite their similar prevalence of apparent toponymic surnames. On the other hand, the very high percentage for Montcuq (47.62%) and converse for Montauban (14.45%) follows the same pattern as apparent toponymic surnames.

Table 4 shows how sex, name length, noble status and sect interactivity relate to the prevalence of toponymic surnames across Peter’s investigations. Once identifying relative substitutions are factored in, sex appears to make relatively little difference to the likelihood of having either an apparent or geocoded toponymic surname. Longer than usual names (three or more name components) more commonly coincide with the presence of at least one apparent toponymic surname (77.27%, sentenced individuals in isolation; 76.92%, with identifying relative substitutions) than the typical two-name model (54.14%, sentenced individuals in isolation; 53.02%, with identifying relative substitutions). It is not clear, however, whether preference for a longer name stands more as a function than a cause of preference for a toponymic surname. Furthermore, this factor also relates to another, nobility (17.20% of identified nobles have names of three or more components vs. 3.97% of others), for which the relationship with toponymic surname prevalence is still clearer. As expected, those with known noble status have a greatly increased prevalence of apparent toponymic surnames (70.97%, sentenced individuals in isolation; 89.25%, with identifying relative substitutions) in comparison to others (40.04%, sentenced individuals in isolation; 44.52%, with identifying relative substitutions). With both these factors (name length and noble status), similar disparities can be seen in the prevalence of geolocatable toponymic surnames. While there might appear to be a small association between sect association and toponymic surname prevalence — heretici-only interactants having somewhat higher prevalence of both apparent and geocoded toponymic surnames than other categories — this can also be related back to nobility. Nobles appear to have been much more prevalent among heretici-only interactants (18.66% noble) than valdenses-only (7.45% noble) or both sects interactants (12.12% noble), a noteworthy finding in the context of heresy studies.¹⁷

Table 4 Correlation between toponymic surname prevalence and other variables.

Comparing Tables 1–2 with Table 3, the disparities in nobility stand out as having the strongest alignment with toponymic surname prevalence: Montauban, for instance, which has a very low prevalence of toponymic surnames, also has a relatively low prevalence of nobility (8.59%), especially when compared with Montcuq (22.62%) and Gourdon (25.17%), which have much higher prevalences of toponymic surnames. On the other hand, Moissac has the second lowest prevalence of identified nobles (9.09%), but has a relatively high prevalence of apparent toponymic surnames, even if relatively few are geocodable. Moreover, it is also clear that low prevalence of nobility cannot on its own account for Montauban’s very low toponymic surname prevalence, especially at the “apparent” level. The relative paucity of nobles there, however, may be suggestive of a wider influencing factor related to the spaces occupied by those sentenced at Montauban, by far the largest town of any of the sentencing centres: were they simply a more urban group, living predominantly in the same town, for whom toponymic surnames were perhaps less relevant as an identifier?

External evidence suggests this to be a very plausible narrative. It is instructive to compare the Montauban data to toponymic surname data derived from two external documents: the witness lists for the Peace of Lorris (1243, between King Louis IX of France and Count Raymond VIII of Toulouse) concerning 1) citizens of Montauban (134 individuals—this list is published by Feuchter [2007, 499–502]) and 2) residents of the surrounding bailiwick of Montauban (29 individuals [Teulet 1866, 501, n. 3056; see also Feuchter 2007, 180–181]). The first list, confined to just the urban population of Montauban itself, provides very similar figures to Peter Seila’s Montauban trial: just 39.55% feature apparent toponymic surnames (taking into account identifying relatives). The second, related to the surrounding region, shows what we have seen to be a more typical level of apparent toponymic surname prevalence: 65.52% (taking into account identifying relatives). Interestingly, it does not seem particularly elevated beyond the typical level we have seen in Peter’s trial despite this list statedly containing nobles and gentry alone. As Feuchter has identified, the first list does have a partial overlap with Seila’s sentenced (58 of peace witnesses from the citizenry also were among the 256 sentenced at Montauban, as marked in the “Notes” field of our dataset), which doubtless influences the similarity of the figures to some degree. But the overlap figures are also telling in another way: the regional witness list shows no overlap at all with those sentenced by Peter. Beyond this, it is pertinent to note that the scribe for the citizenry witness list was not the same as Peter’s, since there are many orthographic differences in the shared names:¹⁸ the low toponymic surname prevalence among those sentenced by Peter at Montauban thus does not appear the result of notarial idiosyncrasy but rather of local social practices. Overall, the evidence from Montauban builds to a picture that Peter Seila covered the town but relatively little beyond and, moreover, that low toponymic surname prevalence might be relatable to the urban character of its population.

What about the other sentencing locations? Montcuq was, as a settlement, at the opposite end of the scale to Montauban: a small castrum, of less stature than Gourdon in this regard. It is thus tempting to relate its very high prevalence of apparent and geocodable toponymic surnames to a particularly rural set of sentenced individuals dispersed beyond the sentencing centre, for whom such surnames were more relevant identifiers. That it should stand out somewhat from the other sentencing centres in the Vaux de Quercy, also small castra, may be partly attributable to the higher prevalence of nobility tried there. Moissac also deserves some consideration within this same discussion. While the prevalence of surnames that appear toponymic is relatively high, similar to Gourdon and the “All Other” castra of the Vaux de Quercy, the proportion that can be geocoded is comparatively small: as a result, the prevalence of identifiable toponyms is very low, comparable to Montauban. This latter fact cannot simply be dismissed as a result of the vagaries of the geocoding process, for Moissac was the most significant town among the sentencing centres after Montauban. Moreover, as Edmond Albe had previously noted and as borne out in Table 3, the prevalence of nobility in the trial also appears to have been at a relatively low level, which might also seem fitting with a more urban set of sentenced individuals (Albe 1910, 287). It should thus not be ruled out that the seeming obscurity of the apparent toponyms witnessed in the surnames of Moissac is a consequence of their pertaining to smaller-scale urban features and/or places swallowed up by the growth of the town around its famous abbey.

Spatial analysis

To get a better sense of what toponymic surnames relate to, we must look at the spatial distribution of the geocoded locations. Our maps plot these in the following order of preference. If an identifying relative possesses a geocoded toponymic surname, we plot the location derived from their name, since in every case where this occurs, the sentenced person themselves lacks one; no identifying relative possesses more than one such surname, so here no choice needs to be made concerning which to plot. In the absence of an identifying relative, we turn to the sentenced individual themselves, prioritising any second toponymic surname over the first (thus “Milhac” in “de Engolesme de Milhac”), based on the view that the second is more likely to denote residence (Billy 1995, 180). The location we have plotted is expressed in the “Mapped Toponym” columns of our dataset.

Figure 2 shows that almost all toponyms are inside Quercy, and concentrated in its west, roughly around the western bailiwicks that Feuchter suggests Peter actually covered: long-distance associations of individuals appear rare. We thus have a strong suggestion that those sentenced were generally locals by personal and/or family background. Within this general concentration, however, we see divergent distribution patterns of mapped toponyms around the sentencing centres. Around Montcuq and Gourdon especially, the locations are distributed quite evenly and densely in the area immediately around the trial centres. Meanwhile, around Montauban and Moissac, the two largest settlements among the sentencing centres, the geocoded locations appear to have a much less even and more widespread distribution, reaching far beyond the immediate surroundings. While that sparseness is partially relatable to the relatively low prevalence of geocoded toponymic surnames among those sentenced in these places, it is notable that those plotted very often seem quite distant from the trial centre.

Fig. 2

Geocoded toponymic surnames of those sentenced by Peter Seila, showing geolocation ambiguities.

These patterns among the identifiable toponyms and their divergences do not appear to be significantly influenced by geolocation uncertainties (see Fig. 2). Ambiguities do not affect more than around a quarter of mapped toponyms anywhere (Montauban, 27.03%; Moissac, 21.43%; Gourdon, 18.03%; Montcuq, 7.50%; All Other, 23.81%), and thus do not much reduce our confidence in interpreting their general spread. One potential exception to this is found with the Montauban trial, where we have a cluster of members of the de la Mota family. The identification of “La Mota” with the Château de la Mothe, around 20 km southeast of Montauban, is made on the basis of Feuchter’s survey of evidence concerning the family, but another identification, with the village of Lamothe-Capdeville, around 7 km north of Montauban, is also possible (Feuchter 2007, 199; Taylor 2011, 73).

Looking at the distribution of distances between the sentencing centres and the distinct mapped toponyms drawn from the surnames of those sentenced at them (see Fig. 3) limits the influence of such family clusters and shows more clearly the greater dispersion of toponyms around the larger settlements. Even in the absence of other data concerning local geography, these distances can be taken as a proxy for time and effort required in travel.¹⁹ Longer and more difficult journeys are more likely to represent either personal or ancestral migration over time than inquisitorial reach and mobility in the moment. On the one hand, at Moissac and Montauban, the elongated distributions speak to a greater reach of inward migration, as one would expect of important urban centres. At these two sentencing centres, interpreting toponym distribution as relating more to residence than to personal/familial origin is implausible: many points are very distant or nearer to other sentencing centres. On the other hand, the more local concentrations around the castra (Gourdon, Montcuq, All Other) that hosted a sentencing event appear more likely to be related to residence than migration (even if it is also reasonable to assume that the migrational reach of these settlements was not as long as that of the larger towns). To hold otherwise would be to suppose that these castra had more of a migrational draw on their local countryside than Moissac and Montauban. With Gourdon, we can see that a good portion of the local spread is associated with nobility (see Fig. 4): nobles account for 49.18% of mapped individuals in this subregion. Their toponymic surnames typically represent their seigneuries, likely to entail some sort of residential association. With Montcuq, nobles account for fewer of the mapped toponyms (35%) despite a similar overall nobility prevalence to Gourdon (see Table 3), but the local toponyms around this castrum form an even more defined area surrounding the settlement, leaving little local countryside uncovered. Such a distribution seems in line with Edmond Albe’s suggestion that those sentenced there contained a good number of local peasants and labourers (Albe 1910, 284).

Fig. 3: Distances between distinct mapped toponyms and associated sentencing events.

The central rectangle in each boxplot represents the common range of these distances (25th–75th percentile). The line inside the rectangle denotes the median distance. The “whiskers” extending from the box indicate variability outside the middle 50%, while significant outliers are plotted as individual points.

Fig. 4

Geocoded toponymic surnames: known nobles vs. others.

In sum, this analysis refines our suggestions from the name analysis. We see nothing to contradict the suspicion that the great majority of those tried at Montauban were its townspeople, since the mapped toponyms associated with this sentencing event appear primarily related to migration. The idea that the very high prevalence of toponymic surnames at Montcuq may in fact suggest the rural character of a good portion of those sentenced there receives support from the particular density and coverage of mapped toponyms in its immediate surroundings. Gourdon and the other castra in the Vaux de Quercy also feature relatively high prevalence of toponymic surnames and relatively high density of mapped surnames around the trial centres; we can thus interpret that those sentenced there were probably also relatively rural. As for Moissac, the fact that its small number of geocodable toponymic surnames appear mostly suggestive of migration strengthens the case that those sentenced there were in fact quite urban, despite a relatively high prevalence of apparent toponymic surnames.

What does our spatial interpretation of toponymic surnames tell us about the spread of dissidence and reach of inquisition in the region? On the former, a map differentiating between the toponymic surnames of those who interacted with the heretici, valdenses or both (see Fig. 5) reflects what was already known from the broad differences between those sentenced at different trial centres: we see more valdenses-only or both sects contacts around Montauban, Montcuq, and Gourdon than elsewhere. We do not see, however, the kind of concentrations of multiple distinct toponyms (that is, multi-family concentrations) associated with contacts of a particular sect that would allow interpretation of differential local spread at a finer level than this.

Fig. 5

Geocoded toponymic surnames: dissident sect associations.

Considering, however, the mapped individuals alongside the unmapped produces some points of interest. As seen in Table 4, heretici-only contacts have a marginally higher prevalence of geocoded toponymic surnames than the valdenses-only and “both” categories, but among those sentenced at Montauban the difference appears much more striking. The representation of the latter category plotted on Fig. 5 is disproportionately large in comparison to overall figures for this trial: whereas heretici-only interactants account for just 30.08% of those sentenced at Montauban (vs. 46.09% valdenses-only and 23.44% both; see Table 3), they represent 16 out of 37 (43.24%) of those mapped. This seems to be affected by the strong positive association between nobility (a class which, as seen, seem particularly associated with heretici) and the possession of toponymic surnames: 17 of the 37 mapped were nobles, and of these 12 were associated only with heretici. We cannot, however, conclude that heretici-only interaction around Montauban was also a more extra-urban phenomenon, even if both nobility and the possession of toponymic surnames might typically seem more associated with rural environments. We have observed that, for the Montauban sentenced, toponymic surnames seem more generally to indicate inward migration than residence; and as previously mentioned, the noble, primarily heretici-associated de la Mota family (5 geocoded individuals, 4 heretici-only interactants), are known to have had a house within the town, whatever residences they still held outside it. Meanwhile, the toponym of the noble Sapiac family (9 geocoded individuals, 6 heretici-only interactants) ties them to a suburban location, just across the Tarn from Montauban itself. Both the de la Mota and the Sapiac families had members who served as consuls of Montauban and can thus also be said to have belonged to an urban elite that included many non-nobles (Feuchter 2007, 193–202). As Feuchter has also observed (2007, 246–7), this class of townspeople likewise appear to have leant towards the heretici: of the 58 men among the sentenced who also served Montauban as peace witnesses in 1243, just 18 (31.03%) were valdenses-only contacts, as opposed to 100 of the other 198 (50.51%) tried there.²⁰ The Montauban data thus suggests a more general association between the heretici and those of higher social rank that was not specifically related to nobility or the countryside.

While we thus do not gain much new knowledge either of the local spatial associations of the heretici or valdenses or of any general locational biases in their operation, we must acknowledge that the limits of Peter’s coverage colour our perception of the distribution of dissidence in the region. Taylor’s view that Peter simply failed to penetrate valdenses networks in the Moissac region is open to some doubt on the basis of the sect interaction statistics presented in Table 3 — even if Peter primarily targeted heretici networks there, one would still expect to find some who interacted with both sects (as we do in every other region) if valdenses were a major presence within the area he covered — but the general point concerning the existence of a spotlight effect is undeniable, especially at a spatial level. While our data may not tell us too much more of the map of heresy in Quercy, they do tell us much about the area covered by the investigation. Our analyses confirm Feuchter’s suggestion that the western bailiwicks of Quercy officially delimited Peter’s work, but they also suggest that the inquisitor’s coverage of this area was highly uneven. We have little evidence to suggest that Peter covered much more than the urban populations at Montauban and Moissac, while elsewhere, away from these large towns and especially at Montcuq, he seemingly covered more of the surroundings.

We can also suggest influences on the spotlight that Peter shone. One can imagine how this uneven coverage might have occurred quite organically. If Peter investigated a larger town and found many suspects there, he might have had less time, resources, and perhaps even inclination (having already achieved tangible results) to investigate the surroundings; and where he focused on smaller castra, it is understandable that he would have needed to investigate the surrounding countryside to find suspects in good numbers. Around Montcuq, it seems most likely that Peter pushed hard to cover the countryside; the numbers sentenced at Beaucaire, Sauveterre, Montpezat, Almont and Castelnau-Montratier seem commensurate with what might happen if he were not able to be as thorough in this. Looking at the way Peter ordered his sentencing circuit in 1241–2 — starting in bailiwicks dominated by large urban centres, moving to those with smaller ones — he was perhaps also thinking of how best to use his limited time and resources to make the most dramatic impression on the locals: by first sentencing large numbers of townspeople, he could make a strong initial impact, saving the practical challenges of handling dispersed rural populations for later.

Overall, these narratives support Feuchter’s suggestion that the document represents the integral results of a set of trials, but offer no further suggestion over whether the initial investigations / trials were contemporaneous with final sentencing. On the question of how systematic Peter’s spatial coverage was, however, we can offer significant nuance. Peter covered different bailiwicks in different ways; if he had any system, be it conscious or unconscious, it appears to have been driven by practicalities and/or the need to make a strong impact, rather than aiming at even coverage.

Conclusion

The study of toponymic surnames in the register of Peter Seila adds much to our understanding of dissidence and repression in Quercy. Above all, by analysing both the socio-cultural context of such surnames and the distances between the locations they reference and the sentencing centres, we have been able to define better the areas Peter actually covered: the earliest sentencing events at Montauban and Moissac suggest a strong focus on urban populations, while those that follow had greater reach into the countryside. This spatially frames the reporting of heresy contained within the register, and also helps us to understand Peter’s process from a less institutional and more individual perspective. Our results are suggestive of an inquisitor who did not, indeed almost certainly could not, cover every area evenly, but made understandable, perhaps even strategic, decisions grounded in the geography of the region: his sentencing events focused first on townspeople in the large urban centres of southern Quercy before tackling more rural societies around Gourdon and in Vaux de Quercy. While just a case study, it is one that helps us understand better the spatial compromises that must have been inherent in many medieval heresy investigations, which, even where better institutionally supported and resourced than this early inquisition in Quercy, relied heavily on the efforts of small numbers of individuals.

More broadly, this study suggests paths forward in the spatial interpretation of medieval toponymic surnames. The uncertainties inherent within them at an individual level mean that they cannot be used simplistically to represent a specific type of locational tie. Our research has indeed suggested that in different places, their generality should be interpreted differently. But it is also instructive that such a nuanced interpretation was achievable, building from the transformation of toponymic surnames and their socio-cultural context into structured data. The subsequent exploration of this data via the weighing of explanatory narratives has allowed us to go far beyond what would have been achievable through qualitative means alone. The presence of significant uncertainties in historical source materials thus should not always be seen as a reason to avoid structuring or quantifying information, but rather a pressing reason to do so.