Summary

  • Tampere University of Technology’s December 1986 .FI application framed a country-code domain as an administrative name space whose first intended use was mail; it forecast host counts and planned nameservers, rather than documenting a finished national Internet.
  • FUNET’s archive places the first international routing test in December 1988, through NORDUnet to US networks. The two records show naming and packet reachability advancing on different clocks, while UUCP, EARN, X.400 and local TCP/IP already connected parts of Finland’s research community.

A request for a name, not a national network

The application dated 4 December 1986 reads less like a declaration that Finland had joined the Internet than a careful case for who could keep a country’s names legible. Tampere University of Technology wrote to the Internet hostmaster seeking the .FI top-level domain. Its administration director, Seppo Loimio, signed the cover letter; Juha Heinanen was named as the technical contact. A separate endorsement came from Matti Ihamuotila, chairman of the Finnish University and Research Network project, FUNET.

The application cited the country-code provision in RFC 920. The memo, published in 1984, treated domains as administrative units for dividing name management. A country-code domain could identify a namespace; the definition did not promise that every host under that label shared a network, a protocol, or an international route. That distinction is visible in the Finnish application itself. It proposed .FI for Finnish hosts and said the domain and its subdomains would initially be used only for mail, with other Internet services a possible later step.

The form also asked for the machinery that would make the name usable. Tampere planned to use existing Unix nameservers running BIND, with the necessary .FI zone information in place by 1 January 1987. The authors said that two US Internet sites could help make Finnish domain information accessible to Internet-connected sites abroad. Their proposal therefore joined a local administrative responsibility to a distribution problem that crossed the Atlantic. The name needed an authority and a way to reach the wider naming system; neither fact, on its own, carried a packet to a Finnish machine.

A small set of networks behind one applicant

Tampere University of Technology could make the case because it already worked across several systems. The letter described it as a major node in SFNET, a Finnish UUCP network with X.25 and telephone links to academic sites. It was also a gateway between SFNET and FUDEC, which covered DECnet sites at Finnish universities, and participated in EARN, the European extension of BITNET, and an experimental X.400 network.

The campus had an Ethernet with more than ten TCP/IP machines, alongside a similar number of DECnet machines. It planned to link local networks at Finnish universities, and some in Sweden, using public X.25 service and Ethernet bridges over leased lines. The wording matters: some pieces existed, other links were planned, and those pieces did not all become one nationwide IP system merely because the country acquired a top-level label. The application’s estimate of 100 hosts initially, 150 after a year, 300 after two years and 1,000 after five was a forecast, not a census of connected users.

FUNET’s own chronology helps place the proposal. Its X.25 research network began in 1984. EARN supplied a fixed international connection in 1985. TCP/IP and Ethernet spread across the network in 1986, and fixed-line TCP/IP routers followed in 1987. In 1988 FUNET joined the Internet through the Nordic NORDUnet research network at 56 kbit/s. CSC, which operates FUNET, dates the first routing test to 1 December 1988: IP packets passed from Finland through NORDUnet to the United States, reaching NSFNET and ARPANET.

The sequence does not mean that Finns had no international communication before 1988. UUCP, EARN, X.400 and mail gateways already crossed institutional and national boundaries. Nor does the later route prove when every .fi name first resolved from every outside network. The evidence instead records distinct work: an application to establish a domain, a planned mail-oriented namespace, and a later IP routing test across the Nordic research backbone.

Two clocks, two kinds of evidence

IANA records 17 December 1986 as the .fi registration date. That is the root-zone record date. It is not the date of the 4 December request, the planned 1 January 1987 zone update, or the first global IP route. FUNET’s and CSC’s institutional histories place that routing milestone nearly two years later. Keeping those dates separate prevents a country-code registration from standing in for a national infrastructure rollout.

That distinction also clarifies what the application was trying to protect. A domain administrator needed to assign names and make name-related information available inside and outside the domain. The applicant argued that Tampere had the people, network experience and US contacts to do this. Its proposed first service—mail—was compatible with the mixed and partly store-and-forward systems already in use. Later packet routing expanded what the network could do, but it did not retroactively turn the earlier application into a route announcement.

The evidence leaves several details open. The original application projected host counts but does not tell us how many machines actually used .fi at each stage. The current IANA record gives the delegation date but not 1986 traffic. The 2023 account of the routing test is retrospective, while the application is contemporary and prospective. Read together, they support a narrower conclusion: Finland’s administrative name space and its international packet path were built through related institutions, but they were not the same service and did not begin at the same moment.

Sources