The Multics development team produced many different series of documents describing Multics and our plans and progress in building it, beginning in the mid 1960s and ending when the system was cancelled in mid 1980. Some documents are available on this site:
- Multics Design Notebook, section I: Introduction by Professor Corbató (13K, 11/30/64, posted 10/10/01)
- AW17: Multics Commands and Active Functions pocket guide (81K, 04/01/80, posted 05/07/99)
- AG91: Multics Programmers' Manual: Reference Guide Table of Contents (170K, 1984, posted 11/28/98)
- Reserved Segment Name Suffixes AG91 Appendix E (8K, 11/24/98)
- Multics System Programmer's Manual Table of Contents
(17K, posted 10/07/05, 839 sections, 54 online)
- BA.1 MSPM- General Information (5K, 07/26/68)
- BA.2 Summary of Multics Technical Policy (5K, 07/26/65)
- BA.3.00 Documentation Conventions (8K, 05/19/66)
- BB.2 System Module Interfaces (PL/I Subset for System Programming) (19K, 9 figures, 6/24/66)
- BB.2.01 EPL Subset for System Programming (6K, 01/31/69)
- BB.3.01 Multics Standard Tape Format (9K, 06/02/67)
- BX.1.00 Multics Command Language (18K, 10/29/68)
- Multics Technical Bulletins Index (234K, 11/05/87, 719 documents)
- MAB-048 Rules for the Multics Change Review Board (20K, 08/12/85)
Sections describing the development document series follow.
Multics Design Notebook
The very early plans for Multics were described in a set of Project MAC memos called the Multics Design Notebook. These memos were written in 1964 and circulated to the system builders and to other OS researchers to explain what MIT wanted from its next machine. In Section I: Introduction Prof. Corbató mentions that the choice of GE as the system vendor had already been made as of November 64. These memos have been scanned by The Multics History Project (MHP) but are not available online.
The Multics System Programmers' Manual (MSPM) was intended to be the primary design document for Multics. Most of the MSPM was written between 1965 and 1969. The MSPM has been scanned by MHP but is not available online. The Table of Contents is online, and a few of the 838 sections have been placed on this site:
- BA.1 MSPM- General Information
- General statement of what the MSPM was for. (5K, 07/26/68)
- BA.2 Summary of Multics Technical Policy
- Key technical directions, e.g. segments don't overlap, need permission to write master mode. (5K, 07/26/65)
- BA.3.00 Documentation Conventions
- Labeling and content of MSPM sections. (8K, 05/19/66)
- BB.2 System Module Interfaces (PL/I Subset for System Programming)
- Data formats and programming conventions. Note that many of the formats described in this section are specific to EPL and were changed radically when later versions of the PL/I compiler were introduced. (19K, 9 figures, 6/24/66)
- BB.2.01 EPL Subset for System Programming
- Definition of the REPL subset of EPL. (6K, 01/31/69)
- BB.3.01 Multics Standard Tape Format
- The format of Multics "native" tapes. (9K, 06/02/67)
- BX.1.00 Multics Command Language
- The definition of the Multics command language. Interpretation of commands, Shell, Listener, special characters. (18K, 10/29/68)
Multics Programmer's Manual
The Multics Programmers' Manual (MPM) was begun at Project MAC in 1965 as the follow-on manual to the CTSS User's Guide. As described below, these manuals were taken over by Honeywell, and assigned Honeywell document numbers.
During the early development phase of the project, 1965 to 1968, there were a number of other documents that didn't fit into the MSPM. Technical white papers, tool documentation, and so forth were assigned a "repository number" and published to the project. There were three series: M for MIT Project MAC, B for Bell Labs, and G for General Electric. There were about 100 documents in each series, but the online table does not list those that were later obsoleted.
Multics Planning Notebook
The Multics Planning Notebook (MPN) was a multi-section management document which described Multics production milestones in the 1967-68 time frame. These memos have been scanned by MHP but are not available online.
Checkout Bulletins and Related Documents
Multics Checkout Bulletins (MCBs), and their successors Multics Technical Bulletins (MTBs), were design documents created by the development staff proposing, explaining, and expanding ideas for changes to Multics. As described in "The Multics System Programming Process," one or more MCBs would be written as the first step in introducing a significant change to the system. MCB numbers started at 1 and went up to over 1000, before the series was replaced by MTBs (because Multics "checkout" was finished). The MTB series began in 1973 and continued on into the 1980s. (An index of 719 MTBs from 1973 to 1987 (100K, 08/17/96) is available online.) At first, MCBs also covered task assignments and schedules, but later these kinds of documents were put into a different series, Multics Task Reports (MTRs). Some of these memos have been scanned by MHP but are not available online.
In 1971-73 Multics Staff Bulletins (MSBs) were produced: they began as a set of memos listing tasks to be done, and later added technical memos on various system improvements. Multics Task Reports (MTRs) contained group progress reports from 1973-1983. There were also Multics Administrative Bulletins (MABs) describing rules and practices to be followed. Multics Operating Staff Notes (MOSNs) informed the system operators at MIT of procedural changes; many of these were combined into a Honeywell manual, the Multics Operator's Handbook.
A later series of documents called Multics Alternative Documentation (MAD) was produced by and for site analysts.
A series of memos were written about 1970 describing the proposed system design of the "follow-on" machine to the 645. This machine eventually became the Honeywell 6180. The series was called Multics Hardware Design Memos (MHDM). MHDM-12 by Corby was the one cited in the MIT/GE agreements.
MIT Project MAC produced many Technical Reports and Technical Memos. MIT theses that were done using MAC facilities were often published as TRs or TMs. The MIT Library 2000 project scanned a large number of these documents and the images and abstracts are available online on a server provided by the MIT Libraries. See the Multics Bibliography for pointers to these documents.
Project MAC's Computer Systems Research Group produced a series of memos, including 55 Multics Performance Logs (MPLs) from 1969-1971. CSR also produced an internal set of CSR_RFC documents from 1973 to 1985: up to about 1977, some of these were about Multics.
The MIT site was the primary system development machine for Multics during the 1970s, used by CISL programmers. MIT Information Processing Center produced several sets of internal memos documenting its management of the Multics service, including Multics Installation Bulletins (MIBs) from 1973-1978. and 211 Multics System Status (MSS) memos from 1977-1984, mostly logging system crash analyses.
Multics Change Documents
In order to make any change to Multics, even a one-line fix, a programmer had to fill out a Multics Change Request (MCR) and have it approved by the MCR Board. This practice was instituted in 1973, and is documented in "The Multics System Programming Process". The online list of MCRs has not survived, but lists of approved MCRs from 1973 through 1980 have been saved by MHP, as well as MCR packets and minutes from the Phoenix MCRB from 1980-1987.
Once a change request was approved, the programmer filled out a "yellow form," which told the program librarians which source modules to pick up, how to recompile them, and how to integrate them into the operating system. The yellow form required the signatures of the programmer, programmer's manager, auditor, and documentation representative. The MCR and yellow form were replaced by electronic versions in the 1980s, near the end of the system's development. MAB-048 describes the rules for the MCR board as of 1985 and contains a sample MCR form.
When Honeywell turned Multics into a commercial product, one of its steps was to take over the MPM and produce it as a four-volume set of manuals, numbered AG90, AG91, AG92, and AG93. Honeywell eventually produced almost 100 manuals describing Multics. Many of these manuals have been scanned and put online by Al Kossow. Manual AW17 was a brief summary of commands.
Papers, Books, and Articles
Over 300 conference papers, books, and articles about Multics were written during the system's lifetime, many by members of the development team.