---
title: "Medieval Roman-canon law rated a single eyewitness and a private document as equal 'half-proofs,' each insufficient alone"
type: "claim"
status: "seedling"
source_url: "https://en.wikipedia.org/wiki/Half-proof"
source_title: "Half-proof (Wikipedia)"
source_author: "Wikipedia, 'Half-proof'"
source_date: "2026-07-09T00:00:00.000Z"
source_quote: "the Glossators of the 1190s such as Azo, who 'gives such examples as a single witness or private documents' [as half-proof]; a defendant faced with half-proof 'might be sent for torture to extract further evidence that could complete the burden of proof' [fragments of the Wikipedia 'Half-proof' entry; the entry does NOT define full proof]"
source_tier: 3
audit_status: "flagged (Tier-3 Wikipedia definitional entry; the torture-completion mechanism carries the capture's [unverified-mechanism] flag and is not cross-checked against a primary legal-history monograph) | 2026-07-10 cross-model audit (fable) CORRECTED: the Wikipedia 'Half-proof' entry was re-read in full — it supports the 1190s Glossators/Azo dating, single-witness-or-private-document as half-proof examples, and torture 'to extract further evidence that could complete the burden of proof', but it contains NO definition of full proof (two eyewitnesses or confession) and NO mention of intime conviction; the prior source_quote paraphrase attributed both to the entry. source_quote rewritten to what the entry actually carries; the full-proof definition and the intime-conviction lineage are now marked in-body as capture background needing their own source. The torture-completion mechanism gains Tier-3 support from the entry itself; [unverified-mechanism] (needs primary) STANDS"
provenance: "Promotion from 10-inbox/raw/2026-07-09-hop-half-proof-moral-certainty.md, 2026-07-09"
origin: "hop-batch"
derived_from: ["10-inbox/raw/2026-07-09-hop-half-proof-moral-certainty.md"]
date_created: "2026-07-09T00:00:00.000Z"
tags: ["legal-history","law-of-evidence","canon-law","testimony","medieval-law"]
audits: ["2026-07-09 claude-fable-5"]
drafted_in: ["2026-07-13-enlightenment-backwards","enlightenment-backwards"]
---


From the 1190s Glossators (e.g. Azo of Bologna) onward, the medieval Roman-canon law of proofs quantified evidence into fixed fractions. A single trustworthy eyewitness *or* a private document each counted as a "half-proof" (*semiplena probatio*) — the cited entry lists both as examples of the same category, neither enough on its own. "Full proof," which alone could ground a conviction, required either two eyewitnesses to the same fact or the defendant's confession `[unverified-definitional — a standard legal-history characterization (e.g. Langbein's *Torture and the Law of Proof*), carried by the capture; it does not appear in the cited Wikipedia entry]`.

This directly resolves the recurring question of whether testimony or documents deserve more trust: the canon-law answer was to rate both as equally weak in isolation, by definition, rather than to debate which was more reliable. The system is a striking counterpoint to two other traditions that solved *how much evidence is enough* differently — Mabillon's comparative document authentication ([[claim-mabillon-1681-founded-diplomatics-to-refute-forgery-charge]]) and the Anglo-American jury standard ([[claim-whitman-reasonable-doubt-began-as-juror-soul-protection]]). The graded-proof architecture here (half-proof as an intermediate short of full proof) is mirrored, unexpectedly, in the vault's own source-verification ladder — a re-read of a single mirror is `capture-verified` but not `verified-verbatim` ([[claim-gersick-1991-deep-structure-quotes-confirmed-verbatim]]); the isomorphism is argued in [[2026-07-11-hop-vault-tiers-revive-legal-proof]].

The rigidity had a dark corollary. Because a half-proof fell short of conviction but established strong suspicion, judicial torture reportedly existed as the licensed instrument for converting a half-proof into the "full proof" of a confession. `[unverified-mechanism -- needs primary]` This torture-completion mechanism is the load-bearing dark side of the system and is not yet checked against a primary legal-history source; see [[question-verify-half-proof-torture-completion-mechanism]]. The quantified-proof regime was later the explicit target of the French revolutionary standard of *intime conviction* ("inner conviction"), which replaced fixed fractional rules with the judge's untethered conscience (capture background — this lineage is not in the cited entry and rides on the same routed verification).

> [!note] Seek's commentary:
> The half-proof rule is, to my eye, the most honest answer the night's research turned up: it declines to rank witness against document and simply legislates both as insufficient alone. There is a cross-domain echo worth holding loosely — the modern intelligence-analysis move in [[claim-ach-step-5-instructs-analysts-to-disprove-not-prove]] is the inverse instinct, refusing to let any single confirming datum settle a question. Both are procedures for distrusting a lone piece of evidence; I note the resemblance without asserting a lineage.
> — Seek