These datasets make the site’s largest research projects inspectable, searchable, and citable. Each landing page keeps the data beside its field definitions, source boundary, transformations, version, checksums, correction route, and review status. A downloadable file is not a claim that every row has received independent expert approval.
Choose the dataset that answers the question
| Dataset | Coverage | Best used for | Do not treat it as |
|---|---|---|---|
| Global Watch Brand History Index | 449 supplied brand and organization records | Finding names, commercial systems, continuity labels, ownership notes, and research leads in the global brand encyclopedia | A legal company register, trademark opinion, complete ownership database, or independently verified registry |
| Global Watchmaking Tool History Index | 326 normalized historical tool and process records | Moving between stable WT identifiers, tool families, functions, periods, workshop contexts, precision regimes, and the complete historical paper | A complete catalogue, invention-priority ruling, safety manual, or row-level claim-source map |
| Future Watchmaking Tools 2026-2051 Index | 365 outlook records: 326 historical carry-forwards and 39 emerging capabilities | Comparing the documented 2026 baseline with conditional 2031, 2041, and 2051 scenarios | Measured adoption data, 365 independent forecasts, probabilities, engineering specifications, or purchase advice |
What every release provides
The landing page is the durable citation target. Current download aliases make the newest public edition convenient, while versioned aliases preserve the exact release a reader used. Generated manifests record file paths, byte sizes, SHA-256 hashes, schemas, counts, and transformations. Where a source package has a defect or transfer exception, the public release records it rather than copying the problem silently.
The brand index retains supplied research classifications. The historical tool release applies a bounded transform to a defective generated significance field and exposes that transform. The future-tools release reconstructs a missing raw JSON file in memory to verify its expected hash, then publishes a distinct 23-field derived index and corrected statistics from the canonical CSV.
These are structural and reproducibility controls. They do not establish factual accuracy, current legal status, safe operation, product performance, authenticity, valuation, or independent review. Source-family or source-key labels are research trails; consequential use requires opening the direct source, checking the relevant passage and date, and applying an appropriate specialist or official standard.
Reuse and corrections
No dataset-specific open-data licence is asserted unless its landing page says otherwise. Public access permits inspection and personal research; it does not silently grant unrestricted mirroring, modification, bundling, commercial redistribution, or rights in third-party standards, pages, names, images, or source text.
To challenge a record, use the corrections process and include the dataset version, stable record identifier, field, current value, proposed value, and exact direct source. Material changes receive a new version; dated prior files remain stable so corrections and changed forecasts are visible rather than rewritten after the fact.
Read the methodology for source roles and uncertainty labels. The separate research protocols describe proposed market studies that currently have zero-data or release-gated status; they are not results and are not mixed into these editorial history and scenario datasets.