Our new study published in Digital Scholarship in the Humanities explores the extent to which information gathered from deponent testimonies affected an inquisitor’s investigation. By applying a network and temporal analysis to the register of Peter Zwicker’s 1393–94 inquisition in Stettin, we found that the sequence of interrogations largely followed a pre-planned structure, yet deponents’ incriminations of others were selectively utilised to find new suspects. This suggests that the late medieval inquisition was not an isolated series of trials but a more sensitive information-processing system.
How did medieval inquisitors gather the names of suspected heretics? Inquisitors’ question lists typically asked a deponent to reveal the names of all heretics known to her or him. This points to the process of testimony-driven snowballing: depositions generated new suspects. On the other hand, both the inquisitors’ manuals and some procedural documents emphasise the role of pre-trial information gathering. When the actual interrogations began, the inquisitor would proceed based on a preprepared list of suspects.
Both snowballing and pre-process information gathering have been attested in the previous scholarship. What has remained unclear is how to test the presence and relationship of these two practices at the process-level: did inquisitors actually use information produced in interrogations to shape the course of an investigation, or were interrogations primarily an administrative component of a pre-planned process?
A recent study from the DISSINET (Masaryk University) and PERSECUTIO (University of Turku [link: https://sites.utu.fi/persecutio/]) projects approaches this question from a different angle. Rather than treating inquisitorial records only as legal texts or cultural testimonies, it analyses a late medieval inquisition as an information-processing system: a process that collected, stored, and reused information while making decisions over time.
The case study focuses on the large anti-Waldensian inquisition conducted in 1393–1394 in Stettin (today’s Szczecin), led by the inquisitor Peter Zwicker. 195 depositions of the ca. 450 original have survived, mentioning more than 1000 individuals. Using network and temporal analysis of surviving trial records, the study reconstructs how names entered the system, when they reappeared, and whether they triggered new interrogations.
This makes it possible to observe feedback between testimony and decision-making and to test how strong that feedback really was.
The results show that the inquisition did not unfold as a purely reactive process driven by ongoing testimony. Instead, the sequence of interrogations followed a largely pre-planned structure, strongly shaped by spatial planning based on the geography of the region. Individuals living closer to Stettin were typically summoned earlier, while more distant communities appeared later in the process.
This spatial ordering strongly suggests that the inquisitor began with a substantial amount of prior information and organised the investigation accordingly, rather than discovering suspects step by step through testimony alone.
At the same time, this spatial pattern was far from absolute. The order of individual summonses frequently departed from it due to local resistance, voluntary depositions, or other contingent events. It is precisely the accumulation of many such deviations that would have obscured the pattern of inquisitorial investigation in smaller or purely qualitative samples.
At the same time, the investigation was not rigid. New information did feed back into decision-making—but selectively. General reputational accusations (fama of heresy) were systematically used to identify and summon new suspects, whereas detailed confessions describing concrete heretical interactions with other suspects mainly served as legal evidence against the person being interrogated.
By combining network structure with the temporal order of interrogations, the analysis also evaluates the extent of testimony-driven “snowballing”. The results show that large-scale snowballing was limited. Most individuals were not summoned simply because they had been named by others, and apparent follow-up effects often disappear once the massive loss of trial records is taken into account. This challenges the assumption of the campaign as a primarily empirical, bottom-up investigation driven by accumulating testimony.
Methodologically, the key contribution of the study lies in treating the inquisition as a system of information processing, rather than as a collection of isolated trials. By tracing how information circulated within that system and by explicitly modelling what is lost from the record, it becomes possible to reconstruct strategic decision-making by inquisitors even from fragmentary sources.
The result is a more precise picture of inquisitorial practice: neither a blind bureaucratic ritual nor a fully reactive investigative machine, but a hybrid system combining pre-planned structure with limited, targeted responsiveness to new information.
The full article, How inquisitive was medieval inquisition? A network-analytical approach to information flow in the trials for Brandenburg-Pomeranian Waldensians (late 14th c.), is published open-access in the Digital Scholarship in the Humanities.
We have published the DISSICON dataset (which stands for DISSINET Conceptual Network), offering an ontology for analysing medieval inquisition records and other trial documents. It formalizes actions, concepts, and their relationships to support computational analyses of historical material.
