Field Notes Explorer

A full-text discovery engine for the written history of the natural world.

This is an AI experiment using modern vision models to transcribe historical field books, transforming both standard printed texts and previously untranscribed handwritten journals into a single, fully searchable archive.

Index currently building: The AI vision pipeline is still transcribing the archives. There are currently 936 works indexed (currently processing fieldnotesjulyde00rems). Check back frequently as new transcriptions are added periodically!
936
Indexed Works
143,203
Total Pages
102,713
Transcribed Pages
Taxonomic Index

Locate all document pages referencing a specific species.

About the Index

Bridging the Manuscript Gap

While printed historical records have long been searchable through traditional Optical Character Recognition (OCR), primary biological data has often remained opaque. The personal, handwritten field notes and journals of naturalists have historically been invisible to search engines, functioning essentially as unindexed images.

This project utilizes advanced, open-weight Vision Language Models (VLMs) to overcome this barrier. By automatically transcribing both standard typography and complex handwriting, the platform places manuscript field notes on equal footing with printed records.

Discovering Primary Data

By processing these images into machine-readable text, the Explorer surfaces dates, locations, and observational data that were previously hidden. Furthermore, the pipeline extracts and indexes scientific nomenclature, creating a cross-referenced network of species mentions across the archive.

Open Data Provenance

The primary sources indexed within this discovery layer are drawn from the expansive, open-access collections digitized for the Biodiversity Heritage Library and hosted on the Internet Archive, alongside the incredibly generous work of other institutions. This site operates as an independent utility designed to expand access to these vital public domain resources.