1BSD Released — The Berkeley Fork of UNIX Begins

On 9 March 1978, the First Berkeley Software Distribution (1BSD) went out on tape—an add-on to AT&T Bell Labs' Sixth Edition UNIX rather than an operating system in its own right. It had been put together by Bill Joy, then a Berkeley graduate student, later a co-founder of Sun Microsystems.

Joy began assembling it in 1977. Marshall Kirk McKusick's first-hand account ("Twenty Years of Berkeley Unix") says that "early in 1977, Joy put together the 'Berkeley Software Distribution'" and, over the following year, "sent out about thirty free copies of the system"; 9 March 1978 is the date conventionally cited for the release itself. Dates in this stretch of BSD history commonly differ by a year between sources, and it helps to keep "when it was assembled" separate from "when the tapes went out".

From that small tape grew a half-century of UNIX lineage: SunOS, BSD/OS, FreeBSD, NetBSD, OpenBSD, and the foundations of macOS. The system software of the PlayStation 4 and 5 and the streaming edge of Netflix all trace back to the BSD line.

Why UNIX Grew at Berkeley

It began at the Symposium on Operating Systems Principles at Purdue in November 1973, where Ken Thompson and Dennis Ritchie presented the first UNIX paper. Berkeley's Professor Bob Fabry was in the audience and immediately set out to obtain a copy. A PDP-11/45 was bought jointly with the Mathematics and Statistics departments, and in January 1974 a Version 4 tape arrived and was installed by the graduate student Keith Standiford. When the machine started crashing, Thompson debugged the dumps remotely from New Jersey over a 300-baud acoustic modem—the first instance of the Bell Labs/Berkeley cooperation that followed.

AT&T, kept out of the computer business by antitrust consent decree, was distributing UNIX to educational institutions at cost. That universities received the source, not a commercial binary, was decisive.

In the autumn of 1975 Thompson took a one-year sabbatical at his alma mater and brought up Version 6 on Berkeley's newly installed 11/70. Two unnoticed graduate students arrived the same term: Bill Joy and Chuck Haley. They started with the Pascal system Thompson had hacked together, improving it until its error recovery and compile speed made it the students' environment of choice. Then, chafing at the ed editor on their new screen terminals, they built the line-at-a-time editor ex from em, obtained from Professor George Coulouris at Queen Mary College, London.

What Was on the Tape

1BSD was not a complete operating system but a bundle laid on top of Sixth Edition UNIX. By McKusick's account the contents were spare: the Pascal system, and—"in an obscure subdirectory of the Pascal source"—the ex editor. That was it. Interest in the Pascal compiler's error recovery was what generated the requests for copies in the first place.

The copies were free, and Joy acted as distribution secretary, mailing tapes and answering the phone himself. Sites that received it patched it further and sent tapes back. A bidirectional flow of source—what would later be called collaboration—was already happening, informally.

csh and termcap are not on this tape. Joy wrote vi only after cursor-addressable ADM-3a terminals arrived, and termcap grew out of the need to describe the differences between terminals rather than carry code for each. Both belong to the next release, 2BSD.

4BSD, TCP/IP, and Commercialisation

Key milestones that followed:

• 2BSD (May 1979) — vi, the C shell, termcap; roughly seventy-five tapes
• 3BSD (December 1979) — VAX support, virtual memory; nearly a hundred copies
• 4BSD (October 1980) — the product of an eighteen-month DARPA contract: job control, auto reboot, a 1K-block file system; about 150 copies
• 4.1BSD (June 1981) — the tuned-up kernel that answered the VMS benchmark critics
• 4.2BSD (1983) — TCP/IP, the socket API, and the fast file system
• 4.3BSD (1986), 4.3BSD-Reno (early 1990)
• Networking Release 1 (1989) / Net/2 (June 1991) — freely redistributable releases with AT&T-derived code removed

The CSRG did not exist until 1980. When DARPA decided to standardise its research sites on UNIX, Bob Fabry won an eighteen-month contract beginning April 1980 and set up the Computer Systems Research Group around it. 1BSD, 2BSD, and 3BSD all predate it—they came out of a period when Joy was effectively running the distribution single-handed.

Nor did TCP/IP arrive in 4BSD. Berkeley's production-quality stack came together across 4.1a (April 1982, local distribution only) and 4.2BSD (1983): Rob Gurwitz of BBN released an early implementation, Joy integrated and tuned it, and Joy with Sam Leffler then restructured the internals so several network protocols could run at once. In the same window McKusick built the fast file system and Joy finished the interprocess-communication facilities that became sockets. The "r-commands"—rcp, rsh, rlogin, rwho—date from here too. NFS was Sun's later work, not a 4.2BSD feature; it entered the Berkeley tree with 4.3BSD-Reno.

USL v. BSDi — A Rite of Passage for Open Source

Net/2 shipped from June 1991 for a US$1,000 fee, taken up by several hundred individuals and organisations. Bill Jolitz wrote replacements for the six missing kernel files and put out 386/BSD free over FTP within six months. Meanwhile BSDi, founded by CSRG people, began selling a source-and-binary system in January 1992 for US$995, advertising a 99 per cent discount against System V source—and a phone number, 1-800-ITS-Unix.

AT&T's subsidiary USL (UNIX System Laboratories) first demanded the number and the "Unix" claims be dropped, then sued in New Jersey federal court in 1992, alleging copyright infringement, misappropriation of trade secrets, breach of licence, and trademark dilution. Judge Dickinson R. Debevoise heard the injunction request in December 1992 and, some six weeks later, issued a forty-page opinion denying it and throwing out all but two of the complaints. The University of California promptly counter-sued in California state court, arguing that USL had failed to credit Berkeley code carried into System V.

Novell bought USL in 1993, and CEO Ray Noorda—who said publicly he would rather compete in the market than in court—pushed for terms. The settlement came in early 1994, and on the substance Berkeley won:

• Of the 18,000 files in Net/2, exactly three were removed, with minor changes to a few others
• About seventy files gained a USL copyright notice while remaining freely redistributable
• USL agreed not to sue users or distributors of the forthcoming 4.4BSD-Lite

4.4BSD-Lite followed in June 1994, and BSDi, NetBSD, and FreeBSD all rebased on it. After 4.4BSD-Lite Release 2 in June 1995, the CSRG disbanded.

The lawsuit left the open-source community with two lessons. First, the necessity of a culture that documents code provenance carefully and secures the right to redistribute by licence. Second, that resilience against corporate legal risk comes from multiple community-run derivatives running in parallel—the FreeBSD / NetBSD / OpenBSD model. And it is frequently observed that the Linux kernel gained ground during the years the litigation ran.

Where 1BSD Lives Today

Trace the modern descendants:

• macOS, iOS, iPadOS, watchOS, tvOS, visionOS — Darwin's XNU kernel combines the Mach microkernel with BSD-derived code, and its userland and TCP/IP stack are BSD too
• PlayStation 4, 5 — system software built on FreeBSD
• Netflix's streaming edge — runs FreeBSD for high-throughput TCP
• Junos (Juniper Networks router OS) — FreeBSD-based
• WhatsApp's servers — famously carried hundreds of millions of users on Erlang over FreeBSD with a very small engineering team
• Linux distributions — as McKusick notes, roughly half the utilities packaged with them are drawn from the BSD distribution

A small tape sent out in about thirty copies still underwrites the world's communications, media, and gaming infrastructure. As the first practical demonstration of "sharing source code as a way of building software", 1BSD precedes GNU, Linux, Apache, and Git in the genealogy of open source.