Curated Measured Steel Transformation Datasets with Datum-Level Provenance (Material Simulator)

Curated, machine-readable steel transformation datasets (isothermal I-T/TTT, continuous cooling CCT, phase fraction, hardness) extracted from peer-reviewed publications and verified public-domain historical sources, with datum-level provenance: every value carries its exact source location, extraction method, extraction date, and the original publication's license and citation. Part of the Material Simulator project (https://materialsimulator.com), an interactive steel metallurgy education and reasoning-contract platform. This deposit is a versioned, citable snapshot of the project's "measured tier". See README.md for the full dataset index, per-source citations and extraction-fidelity classes. Version 1.2.0 contains 100 datasets, including the complete digitization of all 89 isothermal-transformation diagrams of the 1951 U.S. Steel Atlas of Isothermal Transformation Diagrams (2nd ed.) — every alloy-family section (carbon, manganese, nickel, chromium, molybdenum, silicon, copper, vanadium, multi-element, boron steels) plus the austenitizing-treatment comparison section. To the author's knowledge this is the first openly licensed, machine-readable, per-datum-provenanced release of that volume's complete diagram set. The Atlas was verified unrenewed (US Copyright Office renewal search plus an independent institutional rights determination) before extraction; data values were re-digitized, and no page image or figure is reproduced. The remaining 11 datasets cover modern CC-BY dilatometric CCT, phase-fraction, parametric-kinetics and hardness studies (grades S960MC, 42C, 42D, 23MnNiCrMo5-3, 54SiCr6, HRB500DW, a lean medium-Mn Q&P steel), and the digitized isothermal curve for near-eutectoid "Steel B" from Davenport & Bain's founding 1930 study. Extraction fidelity is declared per dataset and must be respected: table transcription (modern sources), calibrated figure digitization (Davenport & Bain 1930, uncertainty ±12 °C and a time factor ~1.6), and visual figure read-off against printed gridlines (the 89 Atlas datasets, wider declared uncertainty of roughly ±25 °C and a factor of 2-3 in time). The Atlas datasets are suitable for teaching, qualitative comparison, nose-location and alloying-trend work; they are not certified engineering data. Files: measured-steel-datasets.json (all 100 datasets in one file) and measured-steel-datasets.zip (the same datasets as individual flat files, one per dataset). Dataset identifiers are stable across versions and resolvable against the live API.