Research Unix Editions: First Edition to Version 7

“Research Unix” names the sequence of Bell Laboratories editions through which Unix changed from a small PDP-7 experiment into a source-distributed system that universities and manufacturers could study and port. An edition was not merely a numbered marketing release. Its programmer's manual, source tape, supported hardware, commands, libraries, and local users describe a moment in a moving research project. Reading the editions prevents two common errors: treating the 1971 system as if it already had later Unix features, and treating Version 7 as a finished universal standard.

What the numbers identify

The familiar First through Seventh Edition labels primarily follow editions of the Unix Programmer's Manual. The manuals are unusually useful evidence because they list system calls, commands, file formats, and sometimes installation material that a site could actually use. They are not exhaustive histories: a manual records an interface better than a hallway discussion, a locally patched machine, or the date a feature was first tried. Edition dates also do not mean every licensed site installed the same tape on the same day.

Bell Labs had used the PDP-7 for the 1969 beginning, but the PDP-11 made a durable multiuser system more practical. The PDP-11 family was a 16-bit minicomputer family, far smaller than later workstations. Memory, disks, terminals, and peripherals were scarce shared resources. Its constraints help explain compact programs and careful interfaces, but do not make every later practice an inevitable consequence of small memory. The system was repeatedly rebuilt as users needed editing, document preparation, program development, and administration.

Edition	Approximate date and principal setting	Historical significance
First	1971, PDP-11	First programmer's manual for a working multiuser Unix system.
Second–Third	1972–1973, PDP-11	Rapidly evolving commands, languages, and system facilities.
Fourth	1973, PDP-11/45 era	Documents the system around the rewrite of much of Unix in C.
Fifth	1974	A mature research system and a base for early external distribution.
Sixth (V6)	1975	Widely licensed with source; especially important in teaching and ports.
Seventh (V7)	1979	Last broadly distributed Bell Labs research edition; a key ancestor for several later lines.

First Edition: a documented small system

The First Edition manual is dated November 1971. It followed the move to a PDP-11/20 and records a system with a hierarchical file system, user programs, an assembler, compiler tools, editors, and a command interpreter. Familiar names can mislead. Early filesystems, process handling, and commands were simpler and differed in detail from later releases. The manual’s table of contents is better evidence for an interface than a retrospective claim that an early Unix machine “had everything Unix has now.”

Text processing was central to the institutional case for the PDP-11. Bell Labs’ patent department needed document preparation, and tools such as roff made the machine useful beyond operating-systems research. That workplace use supplied users, problems, and justification for maintenance. It is more accurate than saying an elegant design spread by elegance alone. Research staff, support staff, document users, and programmers all shaped which facilities survived.

Second, Third, and Fourth Editions: movement rather than a frozen origin

The Second and Third Editions appeared while the system was changing quickly. The short intervals matter: Unix was not designed once and then merely copied. Commands were added or revised, the filesystem and kernel developed, and language work progressed. The early shell was a command interpreter, not the full scripting environment users now associate with sh, bash, or zsh. Pipes likewise arrived after the initial operating system. Historical continuity should not erase sequence.

By 1973, Ken Thompson and Dennis Ritchie could report that Unix had been rewritten in C except for a small amount of assembly language. Ritchie developed C from B and related Bell Labs work; Thompson had used B in the early period. A rewrite was collaborative systems work, involving compiler and library work, kernel changes, testing, and machine-specific code—not a magic conversion performed by a language. The Fourth Edition manual is close to this transition. Its importance is that much more source could travel as C while bootstraps, device handling, context switching, and drivers remained tied to a machine.

The 1974 CACM paper by Ritchie and Thompson introduced Unix to a wider technical audience. It describes a time-sharing system and provides a contemporary architectural account. It should not be read as a promise that every later Unix product would retain exactly the same interfaces. In particular, memory sizes, terminal types, hardware support, networking, security facilities, and licensing changed substantially after the paper.

Fifth Edition and distribution beyond one laboratory

Fifth Edition is often associated with early university use. AT&T’s regulated status and the 1956 consent decree constrained ordinary entry into the computer business, so Unix was licensed to universities and research organizations, often with source. “Often with source” is important but should not be relabelled open source: recipients obtained copyrighted code under contracts and redistribution conditions. Still, source access gave licensees an unusually direct opportunity to learn from, repair, and adapt a real operating system.

Distribution was physical and social. A tape and manual were not a self-installing product. A recipient needed compatible equipment, a C compiler and bootstrap path, local knowledge of disks and terminals, people able to read the code, and permission under the licence. Users exchanged fixes and ideas through professional networks. This made adaptation possible, but also guaranteed variation. A local port or command was evidence of one site’s work, not necessarily a feature of the edition shipped from Murray Hill.

V6: a teaching system and a porting base

Sixth Edition, released in 1975, became particularly influential because many universities received it and because its source was compact enough to study. John Lions at the University of New South Wales wrote a commentary on the V6 kernel source for his course. The commentary helped generations of students connect system calls, filesystem code, process scheduling, and machine details. Its later circulation history is complicated by AT&T licensing; the important point is neither that it was secretly free nor that one book created Unix education. Licensed source, teachers, departmental machines, and a community of readers made V6 teachable.

V6 also made limitations visible. It targeted PDP-11-class machines, with their address-space and memory constraints. Its filesystem and process facilities were not designed for today’s disks, networks, threat models, or workloads. There was no built-in TCP/IP stack, container isolation, cryptographic package verification, or modern access-control framework. Studying it is valuable because it is a comprehensible historical system, not because it is a secure template for a current server.

Nonetheless, V6 carried recognizable Unix ideas: a small system-call interface, files and directories, processes, C programs, and independent utilities. The source could be adapted to other machines, but “portable” was relative. A port required an architecture-aware compiler, startup code, interrupt and device work, storage support, debugging, and local policy decisions. Portability lowered the cost of migration; it did not abolish hardware or make binary programs portable.

V7: a durable reference point, not a final destination

Seventh Edition, released in 1979, is frequently called the high-water mark of Research Unix distribution. It ran on PDP-11 systems and included a substantially richer collection of tools and languages, including the Bourne shell, awk, and make. Its manual and source became a practical reference for researchers and commercial licensees. V7 influenced both the 32V/early Berkeley path and, through later AT&T work, the commercial System III and System V lineage.

That description needs two qualifications. First, V7 is a common ancestor and source base, not a claim that BSD or System V simply copied an unchanged V7. Berkeley’s VAX work, DARPA-supported networking, local tools, and later replacement of AT&T code created a distinct trajectory. At AT&T, internal development moved through Programmer’s Workbench (PWB/UNIX), then System III, then System V. Second, V7 did not settle portability. It still reflected PDP-11 assumptions and a 1970s research environment, even as its C source and tools made future movement easier.

What V7 supplied	What later groups changed or added
C-oriented kernel and userland source	Ports for VAX, 32-bit workstations, x86, RISC systems, and later architectures.
Shell and tool-building culture	Different shells, package systems, graphical environments, and administration frameworks.
Processes, files, descriptors, and pipes	Networking, richer security, threads, virtualisation, and modern resource controls.
Licensed research distribution	Commercial licences, BSD-style licences, GPL-based systems, and other legal models.

Research Unix is a lineage, not a synonym

After V7, Bell Labs continued research releases, while commercial and university branches took their own paths. PWB/UNIX addressed internal programming-workbench needs and fed into System III; System III consolidated prior AT&T releases, and System V became the commercial line. Berkeley began with additions to licensed Sixth Edition Unix and then developed its VAX-oriented BSD releases. These paths exchanged ideas and code but should not be collapsed into a neat two-box family tree.

The modern relevance is practical. A manual page, a shell feature, or a C interface may descend from this era without having identical behavior on GNU/Linux, FreeBSD, macOS, AIX, or Solaris. POSIX later specifies selected common behavior; vendor manuals define extensions. Readers seeking the free-software context for GNU and Linux can continue in Open Source; the role of commodity Intel hardware belongs also in x86.

Reading the evidence carefully

Use edition manuals and preserved source trees for concrete questions: was a command present, what did a system call promise, and which files shipped? Ritchie’s retrospective history and the 1974 paper provide valuable interpretation from participants. Oral histories and later recollections add context, but may simplify dates or credit. A source archive can also contain a later reconstruction, patch, or site-specific tree. Naming the edition, machine, and document is better historical practice than saying simply “old Unix did this.”

For example, a command that appears in a V7 manual establishes that the command was documented in that release; it does not establish its behavior on every V7-derived vendor system. Conversely, an absence from a short manual index does not demonstrate that no one at Bell Labs had experimented with the idea. Histories become more useful when their claims stay proportionate to their evidence. The editions are primary artifacts for interfaces and distribution, while institutional records and contemporary technical papers explain why a particular system could acquire users.

They also make software change concrete. An edition contains omissions, awkward interfaces, and machine assumptions alongside enduring ideas. That is a reason to read it, not a reason to romanticize it. Modern portability comes from specifications, continuous testing, maintained toolchains, and explicit target support; the C-based movement of Research Unix was an important predecessor rather than a substitute for those practices.