Summary

  • In September 1992, USP's working email path used a 2,400-bps dial-up link to Waikato in New Zealand; the full-IP routes discussed in the same correspondence were options, not an installed service.
  • A contemporaneous USP reply in November 1994 said the university then had a full Internet connection over a 2,400-bps leased line to Melbourne. ITU's 2004 retrospective dates a direct AARNet connection to 1993, and the sources do not resolve the difference.

Analysis

The useful distinction in Fiji's early network history is not between being connected and being disconnected. It is between a service that worked for one purpose and a broader route that was still being designed.

The University of the South Pacific had exchanged email with the University of Waikato in New Zealand before September 1992. The service used a 2,400-bps modem. That was enough to move messages through a store-and-forward arrangement: users could send and receive email without holding a continuous Internet path open. It was not equivalent to an always-on IP connection, and the surviving report does not establish that the service was open to the public beyond USP.

John Houlker's contemporaneous NSRC report preserves an email discussion from 9–10 September 1992 about the next step. Attempts to use V.32 modems at 9,600 bps had not worked reliably; the participants suggested line quality as one possible reason, not a measured diagnosis. A Telebit modem had been ordered and was in transit. Full Internet access was still a set of choices under examination: a leased line, sharing a DEC circuit, a connection through Australia, a satellite route through Hawaii and FIX-WEST, or SLIP. The list records engineering alternatives. It does not show that USP installed any one of them in 1992.

The discussion also exposes the economics beneath a seemingly small technical link. Houlker reported that a telephone call from New Zealand to Fiji cost about US$1 per minute, with a lower overnight rate. He separately described analogue Fiji–New Zealand circuits at roughly US$5,000 a month and noted that digital service was available between Fiji and Australia. These are statements in one 1992 technical exchange, not a tariff schedule or a complete account of USP's purchases. In particular, the existence of a digital Fiji–Australia circuit does not prove that USP used it for Internet traffic at the time.

The evidence changes by November 1994. In an update preserved by the Network Startup Resource Center, USP replied: “We now have a full Internet Connection.” The report identifies a 2,400-bps leased line from Suva to the AARNet hub at the University of Melbourne. It says an increase to 4,800 bps was planned for January 1995; that is a plan, not confirmation that the upgrade took place.

A later ITU case study says USP obtained a direct connection through AARNet in 1993. The November 1994 contemporary update says “now,” while the ITU account looks back to the prior year. One date may refer to an earlier commissioning step and the other to a later operating configuration, but no source located here explains that. The discrepancy should remain visible rather than be collapsed into a single launch date.

Nor was 1992 the beginning of digital communication at USP: a university history dates the first USP email to Waikato to 2 July 1990. The 1992 thread is a status snapshot, not a first-email claim. It records email relay already in use, a higher-speed modem experiment that failed, and full-IP access still being planned. By 1994 the surviving record describes a leased Internet line, though sources differ on when that state began.

Those distinctions matter because a university's message relay, its direct IP path, Fiji's public ISP service and the regional USPNet network were not interchangeable systems. The historical record is strongest when each milestone is tied to its actual service, endpoint and date—and when a proposal is not mistaken for a running circuit.

Sources