Ched-ai/voynich-manuscript-metadata

Voynich Manuscript Metadata

Dataset Summary

This dataset contains structured metadata about the Voynich Manuscript (Beinecke MS 408), a famous 15th-century codex held at Yale's Beinecke Rare Book & Manuscript Library. The dataset includes three tables: pages, folios, and quires, providing comprehensive codicological information.

Dataset Structure

Configurations

This dataset has three configurations:

• pages: Page-level metadata (226 records)
• folios: Folio-level metadata (102 records)
• quires: Quire-level metadata (18 records)

Pages Configuration

{
  "page_id": "f1r",
  "folio_id": "f1",
  "side": "recto",
  "quire_id": "qA",
  "currier_language": "A",
  "hand": "1",
  "illustration_type": "herbal",
  "section": "herbal_a",
  "has_text": true,
  "line_count": 31,
  "is_foldout_panel": false,
  "panel_number": null
}

Pages Fields

Field	Type	Description
page_id	string	Unique page identifier (e.g., "f1r", "f85v3")
folio_id	string	Parent folio identifier
side	string	"recto" or "verso"
quire_id	string	Quire identifier
currier_language	string	Currier's statistical language classification (A/B)
hand	string	Scribal hand per Davis 2020
illustration_type	string	Type of illustration
section	string	Manuscript section
has_text	bool	Page contains transcribed text
line_count	int	Number of text lines
is_foldout_panel	bool	Part of a foldout page
panel_number	int/null	Panel number if foldout

Folios Configuration

{
  "folio_id": "f1",
  "folio_number": 1,
  "quire_id": "qA",
  "is_foldout": false,
  "panel_count": null,
  "is_missing": false,
  "pages": ["f1r", "f1v"]
}

Folios Fields

Field	Type	Description
folio_id	string	Unique folio identifier
folio_number	int	Numeric folio number
quire_id	string	Parent quire identifier
is_foldout	bool	Folio is a foldout
panel_count	int/null	Number of panels if foldout
is_missing	bool	Folio is missing from manuscript
pages	list[string]	List of page IDs

Quires Configuration

{
  "quire_id": "qA",
  "quire_number": 1,
  "folio_count": 8,
  "is_complete": true,
  "folios": ["f1", "f2", "f3", "f4", "f5", "f6", "f7", "f8"],
  "section_primary": "herbal_a"
}

Quires Fields

Field	Type	Description
quire_id	string	Unique quire identifier
quire_number	int	Numeric quire number
folio_count	int	Number of folios in quire
is_complete	bool	No missing folios
folios	list[string]	Ordered list of folio IDs
section_primary	string	Dominant manuscript section

Manuscript Sections

Section	Description	Currier Language
herbal_a	Plant illustrations with text	A
herbal_b	Plant illustrations, different style	B
astronomical	Circular diagrams, zodiac	Mixed
biological	Human figures, tubes, pools	B
cosmological	Foldout pages with complex diagrams	B
pharmaceutical	Small plant parts, containers	A
recipes	Dense text, few illustrations	A

Dataset Creation

Source Data

Metadata extracted from:

• IVTFF transcription file headers (ZL3b-n.txt)
• voynich.nu section classifications
• Davis (2020) scribal hand identifications
• Beinecke Library catalog

Processing

Built by the Voynich Computational Analysis Toolkit (VCAT).

Considerations

Limitations

• Scholarly disputes: Some classifications are contested
• Section boundaries: Not all scholars agree on section divisions
• Hand identification: Based on Davis 2020; other scholars may differ
• Pre-1.0 schema: May evolve in future versions

Attribution

All metadata includes implicit attribution to the ZL transcription and voynich.nu. Disputed classifications are noted in the source documentation.

Related Datasets

This dataset is part of the Voynich Computational Analysis Toolkit (VCAT):

• voynich-eva: EVA transcription text
• voynich-transcription-mismatch: Cross-transcription comparison

Datasets can be joined on page_id, folio_id, or quire_id.

Quick Start

from datasets import load_dataset

# Load pages metadata
pages = load_dataset("Ched-ai/voynich-manuscript-metadata", "pages")

# Load folios metadata
folios = load_dataset("Ched-ai/voynich-manuscript-metadata", "folios")

# Load quires metadata
quires = load_dataset("Ched-ai/voynich-manuscript-metadata", "quires")

# Count pages by section
from collections import Counter
sections = Counter(p['section'] for p in pages['train'])
print(sections)

# Find foldout pages
foldouts = [p for p in pages['train'] if p['is_foldout_panel']]
print(f"Foldout panels: {len(foldouts)}")

Licensing

Processing code and dataset structure: MIT License

Underlying data: Compiled from IVTFF transcription headers (Zandbergen-Landini), voynich.nu section classifications, Davis (2020) scribal hand identifications, and the Beinecke Library catalog. Released as a research resource: cite the original scholars, and verify rights independently for commercial applications. See SOURCES_LICENSE.md for full provenance details.

Citation

@misc{voynich-manuscript-metadata,
  author = {VCAT Contributors},
  title = {Voynich Manuscript Metadata},
  year = {2026},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/Ched-ai/voynich-manuscript-metadata}
}

References

• Zandbergen, R. voynich.nu - The Voynich Manuscript. https://www.voynich.nu/
• Davis, L. F. (2020). How Many Hands Wrote the Voynich Manuscript? Manuscript Studies, 5(2), 271-305.
• Beinecke Rare Book & Manuscript Library. MS 408: Voynich Manuscript. Yale University.

Contact

• Repository: GitHub
• Issues: GitHub Issues