Summary

  • South Africa’s first full IP link followed an earlier email service. Rhodes University’s messages reached Portland through FidoNet store-and-forward exchange; the 1991 leased circuit to Randy Bush’s home office created a different kind of connectivity.
  • The line joined a local university network to an endpoint in a private home, then Rhodes carried traffic for other institutions and neighboring countries. It was a working bridge with a narrow operational boundary, not evidence of universal public access or a measured social transformation.
  • Rhodes University commemorates 12 November 1991, while its 2023 reconstruction places the first ping in late November and dates the first TCP/IP email to the morning of 12 November. The sources do not resolve that chronology into a verified event log.

At three in the morning, the far end was a house

The Rhodes University team had gathered to test a connection that crossed an ocean. John Stevens recalled waiting with Mike Lawrie, François Jacot-Guillarmod and Dave Wilson for a ping to travel from Grahamstown, now Makhanda, to Portland, Oregon. The connection did not answer. Sprint reported that the path stopped at Port Elizabeth, now Gqeberha. A Telkom standby technician was woken, found telephone links unplugged at an exchange and reconnected them. The first ping then passed.

The account is vivid because the failure was specific: the international path had reached South Africa, but a break in the domestic segment prevented the end-to-end test. Yet the far endpoint in Portland was not a university backbone or a carrier hotel. It was Randy Bush’s home office. Rhodes University’s 2023 account says Telkom and Sprint provided a low-speed leased line between Rhodes and Bush. Bush’s own retrospective describes the home connection as the first direct, full Internet connectivity to South Africa, distinguishing it from the email system that preceded it.

That endpoint changes the scale of the story. A university team, a domestic telephone path, an international carrier arrangement and a private office each supplied a different piece. Calling the result simply “South Africa joining the Internet” folds those boundaries into one event and hides the people and institutions that had to keep each segment working.

Email crossed the distance before IP did

In 1988, Rhodes computer scientist Pat Terry asked Bush how he and other colleagues seemed to exchange ideas quickly between meetings. Bush showed him low-cost computer networking. The first practical route was not a continuously connected IP path. It used FidoNet, a store-and-forward system: a computer could hold messages, pass them to another system during a scheduled exchange, and deliver replies later.

A 1989 notice in the Rhodes student newspaper, as recounted by the university, said that after three weeks more than 100 messages a week were being exchanged with Portland. The same account says much of the early traffic involved testing the service and establishing contacts. That number therefore describes reported message exchange at an early stage. It is not a count of people, a daily throughput measure or proof that all messages reached the wider Internet in real time.

The distinction matters technically and historically. Store-and-forward email made distance useful without keeping a live end-to-end path open. It could connect participants before their networks shared the same routing system. A successful email exchange showed that messages could cross the relay chain; it did not show that a Rhodes host had a direct IP route to a Portland host.

The local network had to exist first

The international line was only one layer. South African universities already had different computer environments and ways to exchange data. The North-West University history of UNINET says TCP/IP was selected as a common approach for linking unlike campus networks. It records an experimental Rhodes–Potchefstroom connection in 1990, before the direct international circuit described by Bush.

This sequence separates three things that are easy to collapse: a campus connection, an inter-university network and an international path. A stub or experimental line could test whether distant campuses could communicate without establishing that every university had adopted TCP/IP or had a commercial-grade route. Domain naming was another task. The 2023 Rhodes account recalls that the team chose .ac.za for academic institutions and .co.za for commercial organizations, and that Rhodes became the de facto .ZA administrator for a time.

A name in the domain system could identify a destination; it did not create the physical line or prove that packets could reach it.

In November 1989, Mike Lawrie asked Vint Cerf about .ZA. A later historical paper reproduces Cerf’s reply as saying that US policy appeared to restrict direct connection of a US government-sponsored network to South Africa, while indirect third-party connections seemed permissible. That is a later historian’s rendering of a participant’s correspondence. It is not a complete statement of US policy, a legal ruling or evidence that a government formally admitted the country to the Internet. The available account describes a practical route assembled across institutions, not a treaty ceremony.

A home office became a network boundary

The circuit to Portland gave the South African university network a route to an existing Internet-connected endpoint. Bush’s retrospective says the connection was established over a leased line to his home office. Rhodes’s later report names Telkom and Sprint as the carriers and describes the line as low-speed. None of the reviewed sources supplies a packet log, a detailed topology diagram or a continuous traffic series, so the exact routing and performance should not be reconstructed beyond what those accounts say.

The home office was both an enabling bridge and a dependency. It put a willing and technically capable participant at the far end, close enough to coordinate tests across time zones. Stevens recalled many calls taking place around 3 a.m. in South Africa, which was working time in the United States. A private endpoint also meant that the route depended on arrangements and equipment outside a national carrier backbone. The sources do not establish that it was the only international path in every sense, or that every communication used it.

They do show why this particular link depended on a chain of people and segments rather than a single “Internet switch.”

Bush later recalled that he would support universities and nongovernmental organizations that did not support or enforce apartheid. That was his stated condition for assistance, not a government policy or proof that every user met it. The same retrospective says Rhodes staff extended service to universities, small schools and NGOs in neighboring countries, while declining to claim what effect the service had on those societies. Charley Lewis’s study of South African Internet diffusion likewise treats access as contested and unequal; its interviews and archival evidence provide context, not a measure of the impact of this one circuit.

From Rhodes to Wits

Rhodes’s gateway did not remain the country’s principal international route indefinitely. The university’s 2023 reconstruction says international Internet and email traffic grew, Rhodes served most neighboring countries, and Eastern Cape telecommunications capacity could no longer handle the volume. The gateway moved from Rhodes to the University of the Witwatersrand. The reviewed public accounts do not provide a traffic series, an exact transfer date or a count of people affected by the move. They do establish that the operating center changed as demand and the constraints of the route changed.

That transition is the longer history hidden inside the memorable first ping. A link can be technically successful and institutionally fragile at the same time. The first packet answers whether a path worked at that moment. It does not establish capacity, redundancy, affordability, geographic reach or continuity under growth. Rhodes’s team made a route usable; the later move shows that the route’s institutional and telecommunications surroundings continued to matter.

Sources