A Think Again Media Group publication
On This Day Quiz

ARPANET: what arrived in its first transmission?

ARPANET connection: the two letters and the work behind them

By Phil Springer, Editor · Published · On This Day Quiz

ARPANET connection: the two letters and the work behind them

The first ARPANET connection between host computers delivered two letters: lo. On October 29, 1969, Charley Kline at UCLA tried to log in to a computer at the Stanford Research Institute, where Bill Duvall was waiting. Kline later recalled typing an L, confirming its arrival by telephone, and then sending an O. The receiving system failed when he tried the next character. The intended login stopped short, but the two computers had exchanged data across the new network.

SRI's institutional chronology places the attempt at 10:30 p.m. and reports successful communication within an hour. The Computer History Museum preserves Kline's recollection of the individual letters and the programmers' phone conversation. Those retrospective descriptions identify what happened that night. Documents circulated before the ARPANET connection preserve a different kind of evidence: the proposed message formats, software responsibilities, and testing procedures that made such an attempt possible. They also show that failures and recovery were already part of the engineering work.

Elmer Shapiro's Network Timetable, preserved as RFC 4 and dated March 24, 1969, lays out an unglamorous sequence. Communication equipment needed locations, cabling specifications, power, and telephone coordination circuits. A host-to-IMP interface had to be designed, reviewed with system programmers, built, and tested. The timetable set target dates for installation and debugging before messages could pass between UCLA and SRI. These were scheduled tasks, rather than a retrospective list of completed milestones. The distinction keeps the planned ARPANET connection separate from the later date of its successful use.

The same timetable calls for checking message integrity, delivery order, and delay. It proposes testing abnormal conditions and deliberately losing and restoring communication links, IMPs, and host computers. Later diagrams add Santa Barbara and Utah, expanding the line between two sites into a small network. Telephone conference arrangements expand with it. The ARPANET connection depended on coordinated troubleshooting by people as well as communication between machines. Shapiro's plan treats an outage as something to isolate, report, and recover from, rather than an exceptional event outside the design.

Steve Crocker's RFC 1, dated April 7, divides software between the host computers and the Interface Message Processors, or IMPs. BBN specified the IMP software; the host groups had to agree on their own part. Crocker describes the proposals as tentative and invites reactions. He also names earlier meetings involving researchers from Utah, SRI, and UCLA. The ARPANET connection was being assembled through agreements among separate groups, with unresolved details still visible in a document that later acquired the first number in a long technical series.

RFC 1 distinguishes messages sent by hosts from the smaller packets carried between IMPs. Its proposed message header includes a destination, a link identifier, and a tracing bit. The destination IMP reassembles packets into messages. Error checking uses a checksum added during transmission and checked at reception. These mechanisms concern moving data through the communication system. Crocker separately describes the goal of using a remote host as though a teletype terminal had been dialed into it. Carrying a message and providing a useful remote session were related but distinct responsibilities.

Duvall's RFC 2 gives the remote session more structure. A control link between hosts would establish and break user links and carry status information. A user's primary link would behave like an ordinary teletype connection at the remote computer and use that machine's standard login procedure. Additional links could carry binary or character data between programs. In this proposal, an ARPANET connection could support different forms of interaction without requiring all traffic to behave like a person typing at a terminal. Login was an initial use, rather than the complete purpose of networking.

The proposed procedures also anticipate simultaneous requests and unanswered messages. If both hosts try to establish the same link, their network identification numbers determine which side yields. If confirmation fails to arrive within the prescribed time, the requesting host treats the other as being in error and sends a disconnect message. RFC 2 also calls for a standard eight-bit ASCII character set on primary and control links. Such provisions address coordination and representation: both ends need compatible rules for what a symbol means and what to do when the expected response never comes.

The two letters belong against that background of provisional software and planned tests. The first ARPANET connection was neither a finished public service nor a transmission chosen for a grand ceremonial phrase. It was an attempt to use another institution's computer through the network. SRI reports that Santa Barbara and Utah joined later in 1969, creating four nodes. The brief failed login and the successful retry mark a practical transition: remote access had moved from proposed procedures and installation schedules into an exchange the programmers could actually conduct.

The Question

Which two letters reached their destination during ARPANET's first message on October 29, 1969?

Show the answer

The Answer

"lo". Charley Kline was attempting to send "login" from UCLA to the Stanford Research Institute, but the receiving computer crashed after the first two characters.

Send this question to a friend

Read next

By clicking “Send me a stumper” you agree to On This Day Quiz’s Terms of Use and Privacy Policy.