Summary

  • BITNET’s first CUNY–Yale connection made an existing IBM store-and-forward facility useful across institutions; the link depended on a leased circuit and campus equipment, not a new universal backbone.
  • Retrospective accounts describe a division of costs and a campus model in which users were not charged separately for network messages. “Free” to an account holder therefore meant that universities carried the infrastructure bill.
  • The network’s reach grew as participating institutions contributed connections and processing, while later central support shifted from an IBM grant toward member dues. Those records describe cooperation, not a universal mandate.

The first message remembered from BITNET was almost comically small: “Hi, Hank.” “Hi, Josh.” In CUNY’s Winter 1995–96 magazine, the exchange is dated 5 May 1981 at 11:30 a.m. and described as the first electronic-mail delivery. That is a valuable institutional recollection, not proof that no earlier computer mail crossed another network. Its stronger historical use is narrower: two computing centres had made a connection work, and their operators immediately faced the question of what to do with it.

The connection joined the City University of New York and Yale. CUNY vice chancellor Ira Fuchs and Yale computing director Greydon Freeman wanted scholars at different universities to communicate, and both campuses already had IBM mainframes and software capable of remote communication. The protocol family was associated with IBM’s Remote Spooling Communications Subsystem and Network Job Entry. In practical terms, a sending computer could pass mail or files to another node, which stored and forwarded the material onward.

The first link did not require every campus to install the same computer; it required compatible software and a circuit between participating systems.

The bill was part of the architecture

A CUNY account published in 2006, drawing on interviews with Fuchs and Freeman, says Yale paid for the leased lines from New Haven to Manhattan—about $200 a month in Fuchs’s recollection—while CUNY paid about $10,000 for two modems. The same account says the founders split costs. An IBM-authored education report describes IBM as helping with the data-communication costs between CUNY and Yale. These statements may concern different parts or stages of the arrangement; the available accounts do not provide invoices that reconcile them. The safe conclusion is that the link required real investment and that several parties may have contributed.

The apparent bargain for users was different. Freeman recalled that a person with an account on a campus mainframe could send messages and files without a separate charge for each network exchange; ordinary timesharing still had its own economics. This was not costless communication. It moved costs from individual messages to campus equipment, leased lines, staff work and institutional budgets. That distinction helped make the service feel open inside participating universities while keeping the decision to connect with their administrators.

The operating model could travel. The 2006 CUNY account says Fuchs approached 34 universities in the Northeast. A Cornell computing history, based on a participant’s recollection, says Cornell joined through Penn State and became the sixth university connected. It recalls a 9,600-bit-per-second leased line and an unusual tariff in which an interstate circuit could cost less than an intrastate one. The episode shows that network geography was shaped by telephone pricing as well as protocol design: the affordable path did not always follow the shortest map distance.

A cooperative network still had a control surface

BITNET’s rule of growth was reciprocal in a concrete sense. A new institution connected to an existing node and, in turn, made its computer and communications capacity available to others. Messages moved through intermediate systems rather than requiring a dedicated line from every sender to every receiver. The design spread access by accumulating local contributions. It also meant that route availability, computer time and maintenance remained tied to the institutions that operated the nodes.

Later organization did not erase that dependency. BITNET’s own retrospective history says IBM support in 1984 helped establish centralized network services, including the BITNET Network Information Center; after that grant ended in 1987, member dues supported the centre. The distinction matters. Shared technical services could be centralized while the physical and computing work stayed distributed across universities. The record identifies contributors and funding arrangements, but it does not show that every scholar, university or network user authorized the same terms.

Heng Lu Note 73 is useful here only as an interpretive question: participation and contribution can show who took part, but they do not by themselves establish who had authority to speak for everyone affected. Applied carefully, the question keeps BITNET’s cooperative reputation from becoming a claim of universal representation. The evidence supports a practical network built by institutions that chose to connect and pay. It does not tell us that all scholars had access, that every institution could afford to join, or that the first link was the first email in the world.

What the connection changed—and what it did not

BITNET lowered the cost of routine exchange for people whose institutions joined. Email, file transfer and later mailing-list services made scholarly contact less dependent on travel, postal delay or a local colleague’s access to a remote computer. A later Agrawal and Goldfarb study, using university connection dates and coauthorship records, estimated that pairs of universities were more likely to collaborate after both joined. That study is evidence about a measured association in a particular dataset; it is not a direct measure of every user’s experience or proof that connectivity alone caused each collaboration.

Nor was BITNET simply the Internet before the Internet. It was a cooperating academic network with its own operating conventions, storage-and-forwarding paths and gateways to other networks. Its historical importance lies partly in the institutional design: existing machines, a leased circuit, shared costs and an obligation to contribute made inter-university communication repeatable. A link became a network when other institutions could see a workable way to join—and when the bill, labor and responsibility could be placed somewhere specific.

Sources and limits

CUNY’s 1995–96 account gives the 5 May date, time and first-message recollection; its 2006 profile records the founders’ later accounts of costs, initial outreach and campus charging. Cornell’s computing history supplies a participant’s recollection of the 9,600-bps line, tariff and sixth-campus sequence. An IBM education report says IBM assisted early data-communication costs, while BITNET’s retrospective history describes the later grant and member-dues support for BITNIC. These sources differ in proximity and purpose.

The first-message claim is therefore attributed to CUNY’s retrospective; cost figures are reported as recollections; the exact allocation of initial costs remains unresolved.

Sources: CUNY Matters, Winter 1995–96; CUNY Matters, September 2006; History of Computing at Cornell, Internet-First University Press; IBM Education and External Relationships, ERIC ED337592; A Brief History of BITNET.