This repository is a slow-burn academic case study of one specific event and one specific object: the 1516 discovery, by the peasant Jan Czeczek, of a small dark figurine of Mary in a chalice-hollowed field stone in Gidle, central Poland, and the 500-year cult that formed around it. The case is studied here on three planes:
- Archival — direct work on the surviving primary sources, in particular the three printed Dominican accounts of the cult (Trebnic 1636, Tomasz z Pilzna 1645, Zagajowski 1724) and the carved narrative relief now mounted in the chapel, datable to c. 1651–1670.
- Phenomenological — a close reading of what the texts and the relief actually preserve about the discovery event and the recurring phenomena reported at the site over the next two centuries: luminous, aerial, olfactory, contact, and bodily effects, multi-witnessed and sworn before notaries.
- Comparative — a clearly-marked comparative reading of the Gidle record against (a) the social-historical study of Marian apparitions (Christian, Warner, Zimdars-Swartz, Blackbourn, Swann) and (b) the contemporary academic study of anomalous / UAP / "phenomenon" encounters (Vallée, Hynek, Keel, Pasulka, the SOL Foundation).
The methodology is consistent across all three: cite the primary source by line number; mark the epistemic tier of each claim (observation / reasonable inference / speculative); keep every retraction visible in the document; resist all metaphysical mappings the surviving evidence does not earn.
In spring 1516, near the village of Gidle (Sieradz Voivodeship, then Kingdom of Poland), the peasant Jan Czeczek went out to plow. His oxen stopped and would not move; when he beat them harder they knelt down on the ground. Walking in front of them he found, on a large field-stone shaped like a chalice, a small dark figurine of the Virgin and Child — „a około tego kamienia i obrazu jasność barzo miła i przyjemna” ("around that stone and image, a very mild and pleasant brightness"). He took it home; the household was struck blind; the blindness was reversed when they washed the figurine in water and used the wash-water to bathe their eyes. The figurine became a recognised wonder-working object, attracted Dominican custody in 1615, was papally crowned in 1923, and the basilica around it remains an active pilgrimage site today (Bazylika Wniebowzięcia Najświętszej Maryi Panny w Gidlach). Over the next two centuries the Dominicans catalogued ~300 sworn miracle reports at the site — including a class of recurring nocturnal luminous phenomena formally indexed as such, multiple daylight aerial-figure encounters, a sworn deposition of an anomalous material property of the figurine itself (a powerful fragrance produced by its dust when burnt), depth-rescue miracles (falls into wells, burial under earth), and exorcisms with multi-witnessed material residues.
The single best starting point is docs/synthesis.md
(currently ~1350 lines, ten numbered sections plus appendices). If
you read in order, the document moves from:
| § | Section | What it does |
|---|---|---|
| 1 | The traditional story (1516) | Trebnic's Polish text, the source genealogy, the Czeczek surname etymology |
| 2 | The two physical objects | Figurine + stone, what we know about each |
| 3 | The narrative relief — now firmly mid-17th century | 17 numbered panel markers + annotated image, dating from Cholewiński 2015 |
| 4 | A phenomenological / Pasulka-style reading | Re-reading the discovery event with attention to its multi-modal character |
| 5 | Recurring anomalies in the post-1516 cult | 7 sub-sections built from a systematic OCR sweep of Zagajowski 1724, with line citations |
| 6 | The core hypothesis | 4 sub-sections, decomposed and falsifiable; contrasted with strong-skeptical and strong-traditional alternatives |
| 7 | Supporting context | Earth-cult substrate, comparative earth-disturbance miracles, the 1516-vs-Nuremberg correction |
| 8 | What is observation, what is interpretation | 4 sub-sections; §8.4 is the methodological covenant for §9 |
| 9 | A comparative anomalous-encounter reading | 6 sub-sections; the 9-row mapping table is the centrepiece |
| 10 | Open questions | 8 numbered items with concrete next steps |
If you only have time for one section, read §3.1 Annotated reading of the panel (it makes the case visible) and §5 Recurring anomalies (it shows that the case is not a one-off).
| Path | What it contains |
|---|---|
docs/synthesis.md |
The main document. |
docs/references/ |
Local copies of primary sources where digital editions exist. Trebnic 1636 is not digitised; Tomasz z Pilzna 1645 and Zagajowski 1724 are present as full DjVu + PDF sets. See docs/references/README.md for the catalogue with snapshot dates and canonical URLs. |
docs/working-notes/zagajowski-anomaly-sweep/ |
Two layers of evidence for §5: (1) the 15 grep hit-lists from the recall-tuned anomaly sweep of the Zagajowski 1724 OCR (raw material §5 was built from); (2) claims-cross-reference.md — the gold-standard per-claim entries with page-image crops, raw OCR, fresh vision transcription, English translation, and uncertainty notes for each of the 10 line-anchored citations in §5. The full OCR dump (228 pages → 8250 lines) and the line→page index are committed alongside. |
docs/working-notes/pilzna-1645-vision-pilot/ |
6-page vision-transcription pilot for Tomasz z Pilzna 1645 — the earliest surviving printed account of the Gidle cult, printed mostly in Schwabacher blackletter that defeats djvutxt almost completely. Demonstrates the vision-OCR approach on the most degraded primary source and surfaces 1518-dated miracles (the earliest dated Gidle miracles preserved in print). |
scripts/ |
Reusable pipeline for archival-source cross-referencing — OCR → line→page index → wide page-image crops with red-rectangle overlays. Source-agnostic; the same scripts ran unchanged on Zagajowski 1724 and on Pilzna 1645. See scripts/README.md for the per-stage description. |
IMG/ |
Field photographs of the altar and the relief panel, taken in-situ at the basilica. |
IMG/annotated/ |
The annotated panel (panel-annotated.jpg, full-res; panel-annotated-2k.jpg, downscaled) plus focused crops keyed to the markers in §3.1. |
Binary artefacts (DjVu, PDF, JPG, PNG, ZIP) are tracked through Git
LFS; see .gitattributes for the patterns.
The full commitments are in §8 of the synthesis. The short version:
- Primary sources first. Every substantive claim about the Gidle case is anchored on a sworn miracle record, a printed account, the relief panel, or a peer-reviewed secondary source — not on paraphrase, devotional retelling, or hearsay.
- Three epistemic tiers, marked. Observation (what the sources literally say), reasonable inference (what follows by ordinary historical-critical reasoning), and speculation (the phenomenological/comparative readings) are kept structurally distinct in the document, and the boundaries between them are argued explicitly rather than blurred.
- Retractions stay visible. Several earlier readings have been retracted as the evidence developed (the "wind-of-God" personification in the upper-left cartouche, now correctly identified as Sol; the literal "underground" reading of the panel's right-bottom corner, now correctly identified as a topographic vignette with an arched bridge). The retractions are kept in the document as visible audit trail, not silently removed.
- Comparative ≠ metaphysical. §9 reads the case against contemporary anomalous-encounter literature. That comparison is diagnostic and structural, not a claim about underlying ontology. §8.4 makes the boundary explicit: §§1–8 must remain readable and load-bearing for a reader who rejects §9 entirely.
A short note on provenance, written in the same epistemic register as the rest of the document.
The investigation began with a visit to the Gidle basilica made without prior knowledge of the site, the figurine, or the 1516 discovery story („I had no idea this place existed”). The friar's on-site description of the figurine and the surrounding altar art suggested that there was more here than ordinary devotional iconography; a follow-up conversation with Grok the same day — held privately as field material rather than published in the repo, in the convention researchers use for raw interview and conversation notes — pulled enough of the narrative thread that the project took shape over the hours that followed.
There are two honest ways to describe how the visit itself came about:
- By chance. The most economical reading. The visit was made incidentally; the basilica was encountered rather than sought; no prior interest in Marian apparitions, Polish Dominican history, or the comparative-anomalies literature was driving the choice.
- Guided. The reading the document itself does not foreclose. In the same comparative register that §9 applies to historical accounts of the same class, the chain of events that drew attention to this particular sanctuary may be of the kind that contemporary experiencer literature describes as prompted, directed, or otherwise non-coincidental.
The academic content of the project does not depend on which of these two readings is correct, and the document does not attempt to adjudicate between them — for exactly the reasons articulated in §8.4 about not importing causal/ontological language the surviving evidence does not earn. The structural parallel between this discoverer's experience and Czeczek's own — an ordinary person, going about an ordinary day, encountering something they were not looking for — is, however, worth naming, even if it is not worth arguing.
This document was assembled in a single working session on the afternoon and evening of 30 May 2026, following the basilica visit earlier that same day, in collaboration with Claude Opus 4.7 running in the Cursor IDE.
The division of labour is worth naming explicitly:
- The human author (a physicist by training) drove the investigation: the on-site visit, the original observations and photographs, every framing decision, all Polish-language verification, and — crucially — the corrective re-readings where the AI's first pass was wrong. Notable examples kept visible in the document as audit trail: the sun + moon iconography vs. the retracted wind-face reading (§3.1 marker #5); the bridge-and-pilgrim vignette vs. the retracted literal-underground reading (§3.1 marker #14, §5.3); the missed axe (§3.1 marker #17); the cylinder (§3.1 marker #16).
- Claude Opus 4.7 (Cursor IDE) did the bulk of the heavy-lifting in source discovery and download (Cholewiński 2015, Torenc 2010, the digital editions of Pilzna 1645 and Zagajowski 1724); the OCR pipeline (DjVu → text); the systematic 15-category anomaly sweep of the Zagajowski text; first-pass drafting of §§3, 5, 6, 8, and 9; and bibliography assembly. The model also caught several of its own factual errors before commit (wrong PhD institution for Pasulka, a hallucinated American Cosmic chapter title, an incorrect SOL Foundation founder list — all named in commit messages so they remain traceable).
The standard disclaimer applies: the human author is responsible
for every position taken in this document, and a careful reader
should treat AI-drafted material as starting points to be verified
rather than as final scholarship. The commits are designed so any
specific claim can be traced from final wording back to the
underlying primary source by following the git log.
A subsidiary observation, since it is honestly a relevant fact: that a synthesis of this depth (~1400 lines), an annotated relief panel with 17 cited markers, a systematic anomaly sweep of an 8250-line 17th-c. Polish OCR, and an end-to-end comparative-literature reading could be assembled inside a single working session is, in itself, a small empirical data point about how productively a trained researcher and a current-generation reasoning model can now collaborate on an archival case. It is not the project's purpose, but it is part of its situation, and worth recording alongside the rest.
Work in progress. Major artefacts (the synthesis, the panel
annotation, the primary-source archive, the anomaly sweep) are
complete to first-pass quality. Several open investigations remain
(§10 Open questions); the most important pending field work is a
visit to the Biblioteka Jagiellońska in Kraków to read the 1636
Trebnic original, which has never been digitised. See the Visiting
BJ section in synthesis.md.
This repository is best cited by commit hash:
Our Lady of Gidle (1516) — an archival and comparative investigation, repository at commit
<hash>, accessed<date>.
Substantive material in §9 (comparative reading) draws on published academic literature catalogued in the Comparative literature section of the synthesis bibliography. Those works should be cited directly where used.