Map

86-DOS 1.00 and PC-DOS 1.00 Source Listings (Paterson Listings)

Wiki summarydosretro-computinghistorymicrosoftsoftware-preservation ↳ show in map Markdown
title
86-DOS 1.00 and PC-DOS 1.00 Source Listings (Paterson Listings)
type
summary
summary
Microsoft's April 2026 release of recovered, scanned, OCR'd 86-DOS 1.00 and PC-DOS 1.00 development listings from Tim Paterson, MIT-licensed
sources
early-dos-development-continued
tags
dos, retro-computing, history, microsoft, software-preservation
created
2026-04-30
updated
2026-04-30

On 86-DOS 1.00's 45th anniversary (April 2026), Microsoft's Open Source blog announced that the DOS-History/Paterson-Listings repository — a multi-year recovery effort led by historians Yufeng Gao and Rich Cini — has been licensed under MIT through a coordinated pull request. The listings come directly from Tim Paterson's personal archive and are now publicly available for study.

This sits alongside the earlier MS-DOS 1.25/2.0 (2018) and MS-DOS 4.0 (2024) source releases as part of Microsoft's incremental opening of early DOS history.

What's in the listings

  • The 86-DOS 1.00 kernel — Paterson's original 8086 OS, written at Seattle Computer Products before IBM picked it for the PC. This is the actual ancestor of MS-DOS, not a later reconstruction.
  • Several development snapshots of the PC-DOS 1.00 kernel — point-in-time working states preserved on paper as the IBM PC release was assembled.
  • Well-known utilities including CHKDSK.
  • Listings of the assembler itself — DOS-era development tooling, not just the OS.

The listings are assembler printouts with hand-written notes by Paterson preserved alongside the code. The Microsoft post frames them as "a printed commit history of a Git repository" — each listing captures the code at the moment it was printed, which features were implemented when, what errors were made, and how they were fixed. That kind of point-in-time view of development is rare for systems software of this era.

How the recovery happened

This wasn't a simple "scan a few documents" effort:

  • Locating the source — Tim Paterson kept his archive of printouts. Recovery required convincing him to share, then physically transporting the materials.
  • Scanning — multi-thousand-page assembler printouts in their original physical form.
  • OCR'ing — printer listings of dense assembly are notoriously hard for off-the-shelf OCR. Joshua Carsbrook (jscarsbrook.me/doshistory/) developed printer-listing-specific OCR for this work.
  • Transcription and validation — turning OCR output into compilable assembler.
  • Legal review — the MIT licensing required a pull request coordinated with Microsoft's open source program office.

The physical artifacts have since been donated by Tim Paterson to the Interim Computer Museum.

Why this matters

This is the earliest DOS source code anyone has yet recovered — predating the 1.25 release by about a year. For studying how operating systems were built in 1980-1981 (development practices, hand-edited code paths, the assembler ecosystem), the printed listings are a primary source the way committed git history would be for a modern codebase. They show what shipped and what was tried and rejected, in Paterson's own annotations.

The MIT license is the load-bearing detail for hobbyists, retro-computing projects, and emulator authors who want to use, port, or study the code without a research-only restriction.

Linked from