Rebuilding a Trial from a Disordered Dossier

The first obstacle is physical. The leaves are not in chronological order. The proceedings that opened the case — the information and witness examinations taken at Trois-Rivières on 19–20 May — sit at the highest folio stamps in the file (fols. 114–119). The interrogations that came later, at Québec on 1–3 June, sit lower (fols. 88–110). Read leaf by leaf, the file runs roughly backward.

Sequence had to be rebuilt from evidence internal to each leaf: the dateline the clerk wrote at the head of every session, and the “Continuation page…” catchwords in the margins that point from one leaf to its true continuation. Those two features — not the binding — are the spine. Plotted together, they produce the map below, and the crossing is the whole problem in one image: as the folios climb, the calendar falls.

Any account that follows the leaves in order tells the story backward. The dates, not the pagination, are the thread.

Every interrogation opens the same way. Before a single question about the crime, the clerk records the prisoner’s name, age, birthplace, condition, and residence — the fixed preamble of French criminal procedure. That boilerplate is the most valuable thing in the file. It captures, under oath and by a disinterested hand, exactly the identifying data a parish register would carry — for three people at once.

The same formula fixes the whole household. Jacques gives his age as about fifty and his origin as Poitou, paroisse des Essarts Confirmed — a birthplace that matches independent sources. Isabelle, examined on 1 June, gives her age as thirteen; the clerk first wrote douze, struck it, and wrote treize Confirmed. That small correction, visible in the manuscript, is itself evidence: the court was weighing her age with care, and it is her age that would shortly spare her life.

A record this vivid tempts the researcher to quote it past what it proves. Isabelle’s testimony is the clearest case. She states, under examination, that her father forced her marriage and that her mother had not consented — a sworn statement, and citable as one.

What the conviction rested on is easy to misstate. Under the ideal of the period, a capital conviction was meant to stand on two unimpeachable eyewitnesses. This case had none to the killing itself. The neighbours across the river were earwitnesses — they heard La Touche cry out, “ha moy, hélas, au meurtre,” but saw nothing. The only near-eyewitness was Nicolas, a boy of ten. The conviction rested instead on the accuseds’ own confessions, drawn out across successive interrogations, joined to circumstantial “proximate indications” — the bloodied barn, the implements.

Isabelle’s reprieve is the same lesson from the other side. It reads as mercy, but the mechanism is procedural. Under the Criminal Ordinance of 1670, a corporal sentence required three judges and the most lenient opinion prevailed, and any sentence heavier than the amende honorable carried an automatic appeal to the Sovereign Council. Her youth gave that appellate court the ground it needed to spare her. Reading the arrêt without the Ordinance turns a rule of law into a sentiment — and misses why she, alone, survived.

A vivid 1985 narrative, widely reposted, supplied this case with dialogue, motive, interior thought — and a forename, “Marie,” that the primary record never gives. Each accretion had to be traced to a document or set aside. Three examples show the range:

The forename. Every primary heading — interrogations, sentence, appeal — writes Gillette Banne. “Marie” appears only in a derivative database and the reposts that followed it. Set aside. Rejected

The weapon. The retellings speak of a single implement. The record holds two: neighbours found a bloodied fléau (flail) at the scene, but the accused, in their own interrogations, describe the killing done with a houe (hoe). Two objects, two evidentiary roles — not interchangeable. Proven

The neighbour’s name. The earwitness is deposed as Baultier at Trois-Rivières (fol. 118) and appears as Gauthier at Québec. With both records in hand, this reads as one man written by two clerks — a scribal variant across two courts, cited as provisionally the same neighbour rather than resolved beyond doubt. Suggestive

A complete file is not an omniscient one. Two questions remain genuinely open, and are published as open rather than smoothed over. The will named in the execution notation — “Il y a un testament joint au dossier” — is referenced but not yet located among the imaged leaves; it is cited as referenced, not produced. And no baptismal record for Gillette at Argences, c. 1626, has been found; the registers for those years may not survive. Naming these gaps is not a weakness in the proof. It is the proof behaving honestly.