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

The study investigates the spatiality of heresy prosecutions in high and late medieval Europe through a systematic analysis of locational information in inquisition records, focusing on Peter Seila’s 1241–1242 sentencing register from Quercy (650 individuals across nine locations). Coverage of medieval inquisitions was uneven, shaped by political, logistical, and geographic factors, yet scholarly judgments often rest on impressionistic assessments rather than structured data. Toponymic surnames—frequent but ambiguous indicators of spatial association—pose challenges of coverage and interpretation (residence vs. origin vs. familial association). This paper aims to transform such names into structured, geocoded data, quantify naming practices and social context (sex, noble status, sect interaction), and analyze spatial distributions and distances relative to sentencing centres to infer whether toponyms more likely reflect residence or migration. The overarching goals are to refine our understanding of the spatial coverage of Peter’s trials and to contextualize perceived dissidence (heretici vs. valdenses) within Quercy’s human geography, moving beyond institutional narratives to ground-level interactions.

The paper builds on calls for a systematic “geography of heresy” (Duby) and recent spatial approaches to religious cultures (e.g., Slatter 2023; Jakubowski 2023). It engages anthroponymic scholarship on toponymic surnames and their interpretive ambiguities (Emery 1952, 1955; Lopez 1954; Chareille 2008), including issues of inheritance and multiple meanings, alongside regional evidence suggesting surname updating in Languedoc and double toponyms indicating residence vs. origin (Bourin 1995; Mousnier 1995; Billy 1995). For Quercy and Peter Seila’s activities, the study considers Feuchter’s hypothesis that Peter’s sentencing locations align with western bailiwick seats and may represent an integral, systematic inquest (Feuchter 2007), as well as Taylor’s observations about the spread and novelty of valdenses north of the Garonne–Tarn–Aveyron confluence and possible limits in Peter’s penetration of their networks (Taylor 2011, 2022). Comparative context includes the large-scale inquisition of Bernard de Caux (1245–1246). The study situates its contribution against these debates, offering quantitative, spatially explicit analysis to nuance assumptions about coverage, urban–rural bias, and sect-specific patterns.

Source and scope: The dataset comprises 650 sentenced individuals recorded in Peter Seila’s register across nine locations in Quercy (Ascension 1241: Montauban, Moissac; Advent 1241: Gourdon; Lent 1242: Montcuq, Beaucaire, Sauveterre, Montpezat, Almont, Castelnau-Montratier). The source is a 17th-century copy of the 1241–1242 register (BnF, MS Doat 21), edited by Duvernoy (2001). Contextual regional information was used to interpret spatial patterns. Data extraction and structuring: - Toponymic identification: Surnames suggestive of toponyms (often with “de”) were matched to medieval Latin and modern French place-names using Duvernoy’s annotations and toponymic references (e.g., Taylor 2011; Feuchter 2007; Albe 1910; Fénié & Fénié 1998; Cassange & Korsak 2013; Longnon 1874; Perrin & de Font-Réaulx 1972). For each identified toponym: Latin/modern form and coordinates were recorded. Multiple candidates were adjudicated, with the chosen geolocation flagged as ambiguous and justified in notes. Individuals with two toponyms had both recorded in original order. - Apparent toponymic surnames: All surname elements with toponymic appearance were counted, regardless of geocoding success, to quantify coverage gaps and naming practice. - Name construction: The number of major name components (excluding prepositions and honorifics) was counted; Bourin–Chevalier classification was assigned for comparability. - Identifying relatives: 106 individuals (mainly women) were identified via a relative (often ‘wife of’). Relative information was captured in mirrored fields. Where appropriate, relatives’ toponyms were substituted to augment geocoding and toponymic prevalence metrics. - Supporting variables: Sex; sentencing location; sect interaction (H=heretici only, V=valdenses only, B=both, NS=not stated); noble status (y if stated in secondary literature). Summary metrics and counts: - Geocoding success (before relative substitution): 148/650 individuals (22.77%) had ≥1 geocoded toponym; 155 surnames geocoded in total, 30 flagged as ambiguous. - Apparent toponyms (before substitution): 289/650 (44.46%) had ≥1 apparent toponymic element. - After relative substitution: 173/650 (26.62%) individuals linked to a geocoded toponym; 331/650 (50.92%) had ≥1 apparent toponymic element. - Name components: Before substitution, 471 (72.46%) had two components; 44 (6.77%) had three or more; 142 (20.62%) had a single name (many women identified via relatives). After substitution: one-component 51 (7.85%); two components 547 (84.15%); three or more 52 (8.00%). - Socio-cultural: Sex overall: 391 men (60.15%), 259 women (39.85%); nobles 93 (14.31%). Sect interaction: H 359 (55.23%), V 188 (28.92%), B 99 (15.23%), NS 4 (0.62%). Analytical strategy: - Name and social context analysis: Compared prevalence of apparent and geocoded toponyms across sentencing events with distributions of name lengths, sex, noble status, and sect interaction (Tables 1–3; correlations summarized in Table 4). - Mapping: Plotted geocoded toponyms with precedence rules: if an identifying relative had a geocoded toponym, that location was mapped; otherwise, for individuals with two toponyms, the second element (more likely current residence) was mapped. Ambiguities flagged. - Distance analysis: Computed distances between distinct mapped toponyms and associated sentencing centres to infer whether toponyms reflect residence (local clustering) versus migration (elongated distributions). Boxplots used to summarize distance distributions by sentencing centre; ambiguity rates reported to assess robustness. - External comparison: For Montauban, compared toponymic prevalence with two 1243 Peace of Lorris witness lists (citizens of Montauban vs. residents of the surrounding bailiwick) to contextualize urban vs. regional naming practices and assess notarial effects.

• Overall coverage and spatial bounds: Nearly all mapped toponyms fall within western Quercy, supporting Feuchter’s delimitation of Peter’s activity to the western bailiwicks, with few long-distance associations. - Urban vs. rural contrasts: • Montauban (n=256): Very low apparent toponymic prevalence (35.16% with relative substitutions) and geocoding success (14.45%); low noble presence (8.59%). Mapped toponyms show elongated distance distributions, often far from the centre, indicating inward migration typical of urban poles rather than local residence. Peace of Lorris comparison: citizen list (1243) shows similar low apparent toponymic prevalence (~39.55%) vs. higher prevalence in regional bailiwick list (~65.52%), reinforcing an urban signature and limited extra-urban trial coverage. • Moissac (n=99): Apparent toponymic prevalence relatively high (56.57%) but geocoded very low (14.14%); nobles relatively scarce (9.09%). Identifiable toponyms mainly indicate distant origins; many apparent toponyms could reflect urban micro-toponyms or features, suggesting an urban focus with limited rural coverage. • Gourdon (n=143): Apparent (58.04%) and geocoded (42.66%) toponymic prevalence high; nobles frequent (25.17%). Dense, evenly distributed local toponyms around the centre, many corresponding to noble seigneuries; indicates substantial coverage of surrounding countryside. • Montcuq (n=84): Highest apparent (71.43%) and geocoded (47.62%) toponymic prevalence; nobles 22.62%; women 48.81%. Strong local clustering of mapped toponyms implies rural coverage; aligns with suggestions of substantial peasant/labourer presence. • Other castra (Beaucaire, Sauveterre, Montpezat, Almont, Castelnau-Montratier combined, n=68): Apparent 61.76%, geocoded 30.88%; patterns similar to Gourdon with local clustering indicating rural reach. - Ambiguity impact: Ambiguous geolocations remained modest (Montauban 27.03%; Moissac 21.43%; Gourdon 18.03%; Montcuq 7.50%; Others 23.81%), not altering overall spatial interpretations. - Naming practice and social correlates: The two-component name model dominates (~84% after substitutions). Longer names (≥3 components) correlate with higher toponymic presence; nobles more often have toponymic surnames (e.g., 89.25% apparent after substitutions vs. 44.52% for non-nobles) and longer names (17.20% vs. 3.97% with ≥3 components). Sex has limited effect on toponymic prevalence after substitutions. - Sect interactions: Heretici-only interactants show slightly higher toponymic prevalence; nobles are disproportionately among heretici-only (18.66%) versus valdenses-only (7.45%) or both (12.12%). In Montauban, mapped individuals overrepresent heretici-only (43.24% mapped vs. 30.08% overall), driven by urban elites (e.g., de la Mota, Sapiac), indicating a broader association of heretici with higher-status groups, not solely rural nobility. - Coverage heterogeneity: The first two sentencing events (Montauban, Moissac) primarily captured urban populations with limited rural reach. Later events at castra covered surrounding countryside more comprehensively, especially around Montcuq and Gourdon.

The findings clarify how to interpret toponymic surnames and what they reveal about Peter Seila’s activities. Distance distributions and clustering show that in major towns (Montauban, Moissac) toponyms predominantly index migration (origins elsewhere), implying trials focused on urban residents with little rural coverage. In contrast, at castra (Gourdon, Montcuq, others), dense local clustering supports interpreting toponyms as current residence proxies, revealing broader countryside reach. These patterns resolve competing narratives: while institutional alignment with western bailiwicks exists, coverage was uneven and shaped by settlement hierarchy, logistics, and strategic considerations. The sect interaction patterns, correlated with noble and elite presence, suggest heretici were more associated with higher-status networks (urban and rural), whereas valdenses appear more diffuse. The analysis reframes Peter’s inquisition not as uniformly systematic area coverage but as a sequence of localized engagements optimized for impact and feasibility: initial urban spectacles, followed by rural investigations around smaller centres. This advances the “geography of inquisition” by anchoring it in individual mobility and practical constraints rather than solely institutional jurisdictions.

By transforming toponymic surnames and their socio-cultural context into structured, geocoded data, this study delineates the spatial coverage of Peter Seila’s 1241–1242 trials. Montauban and Moissac show predominantly urban, migration-marked toponyms and minimal rural reach, whereas Gourdon, Montcuq, and nearby castra display dense local toponym clusters indicating broader countryside coverage. The results nuance prior claims of systematic territorial coverage, revealing heterogeneous reaches shaped by settlement size, resources, and strategic aims. Methodologically, the work demonstrates that despite individual uncertainties, toponymic surnames can yield robust spatial insights when contextualized, quantified, and analyzed via distance-based reasoning. Future research could extend this framework to other inquisitions (e.g., Bernard de Caux), integrate transport/topographic models and demographic proxies to refine travel cost inferences, and link micro-toponyms and urban toponymy to improve geocoding in towns.

• Data source constraints: The register is a 17th-century copy; preceding investigative processes are undocumented, limiting reconstruction of full procedural context. - Toponymic ambiguity: Surnames may reflect residence, personal/familial origin, or other associations; inheritance and notarial choices complicate interpretation. - Geocoding uncertainty: Medieval Latin toponyms may be unidentifiable or ambiguous; only 26.62% of individuals could be linked to a geocoded toponym after substitutions; urban micro-toponyms likely under-identified. - Coverage bias: Analyses pertain to sentenced individuals only and a specific temporal window (1241–1242); spatial inferences may not generalize beyond the western Quercy bailiwicks investigated. - Social classification: Noble status identification relies on secondary literature and may be incomplete; name-length correlations do not establish causality. - Mapping choices: Priority rules (e.g., mapping second toponym) and distance as a proxy for travel effort, while justified, remain simplifying assumptions.