Summary

  • RFC 1020, S. Romano and M. Stahl’s November 1987 Internet Numbers, recorded assigned IP network numbers and gateway autonomous systems under the DDN NIC Hostmaster. Independent IP-family networks were marked with an asterisk and directed to apply separately for permission to interconnect.
  • Entries named a responsible person and mailbox, with NIC handles for fresh WHOIS/NICNAME lookup when contact details changed. Old network numbers could remain temporarily in the list with T and a -TEMP name suffix. The printed register tracked assignments, contacts and transitions as different kinds of work.

The asterisk was doing more work than some network operators might have liked. In November 1987, S. Romano and M. Stahl published RFC 1020, a status report on Internet numbers. It listed networks already connected to the research and operational Internet alongside independent networks using the IP family. The independent ones carried an asterisk before their number. Their administrators had to apply separately for permission to interconnect, and those networks were to stay out of the Internet hosts’ and gateways’ working tables. The registry could hand out a number; it had not handed out a boarding pass.

The report’s title, Internet Numbers, sounds like a dry afternoon with a ledger. It is a remarkably clear description of how a growing network organized responsibility. The DDN Network Information Center’s Hostmaster at SRI International had assumed assignment of IP network numbers and autonomous-system numbers. Romano and Stahl thanked Jon Postel and Joyce Reynolds for continuing help. The RFC was to be updated periodically, and anyone needing current information was directed to Hostmaster. Even at publication, the printed book was a snapshot with a person and a process behind it.

Four letters on the left side of network entries divided use into Research, Defense, Government outside Defense, and Commercial. They were marked R, D, G and C. An independent network using IP could have a registered number in this book; the star identified its separate status. Interconnection was handled by a separate application. A network appearing in a registry, a network placed in working routing tables, and a network reachable through an agreed connection occupied different stages of the job. The notation made those stages visible without pretending the ledger itself could plug in a cable.

The table also needed human beings. Each registered network or autonomous system had a named responsible person and a network mailbox. Bracketed references pointed to a document and a NIC handle. That handle was a unique identifier in the NIC WHOIS, or NICNAME, service. The authors explained the problem in plain operational terms: people change electronic mailboxes. A reader looking for today’s address or phone number should query WHOIS/NICNAME or ask Hostmaster. The stable identifier pointed toward a moving contact, rather than freezing a 1987 mailbox as if people never changed jobs.

This was not a nicety for future historians. Network operations needed someone who could answer when an assigned number caused a conflict, an interconnection required coordination, or a listed route had gone stale. The printed ledger gave a name and an initial mailbox. The handle offered a way to refresh that information. A phone book that says “check the latest number” is better than a beautifully printed wrong number, especially when the recipient is responsible for a working network.

Numbers themselves could change. RFC 1020 said an old network number would stay in the list for a transition period when an assigned number changed. A T followed the old number; -TEMP was added to the network name. That small label preserved a trail from the previous identifier to the current one. It told readers which line in the table was transitional. For a particular renumbering, the operational outcome would be found in dated routing and host records. The book supplied a visible marker so nobody had to mistake an old row for a fresh allocation.

The autonomous-system section gave another layer of the naming job. Exterior Gateway Protocol used a 16-bit field for groups of gateways, and RFC 1020 registered its values. AS 0 was Reserved. AS 1 was listed as the BBN Core Gateways. AS 18 appeared as UTEXAS. The numbers and names belong to the November 1987 table. They show the registry giving stable labels to gateway groups, while the traffic those gateways carried remained a matter of operating evidence. Once again, a book could identify a thing without making it run.

Romano and Stahl spent pages laying out the then-current classful structure of IP network numbers, plus special functions such as the 127 loopback network. Those details matter to the 1987 table’s syntax. But its sharper historical lesson lies in the notes beside the numbers. The R, D, G and C categories told a reader what kind of allocation was recorded. The asterisk said an independent network needed an additional interconnection step. The NIC handle gave a way to find a current human contact. The T and -TEMP labels kept an old number legible during change. Four tiny notations did not replace engineers or operators; they told the engineers and operators where the outstanding work was.

There is a pleasing bureaucratic joke here. An organization can fill out the form, receive a number, appear in the official book, and still discover an asterisk beside its name. The joke becomes serious at the border between records and service. Someone has to decide the interconnection, update the tables, maintain the contact, and coordinate a renumbering without losing the network. A register is valuable precisely because it can tell these jobs apart and keep a trail of who is answerable for each one.

RFC 1020 is a 1987 status report. Its table gives the assignments and responsible contacts that its compilers recorded then. The current state belongs to later registry and operating records. The document itself points readers toward an update path through Hostmaster and WHOIS. That instruction is part of the story: an address book for a live network has to admit that people move, numbers change, and connection requires more than a line of ink.

The asterisk is the article’s small hero. It meant that a network could be counted in the number system and still need a separate decision at the gate. The contact handle and temporary number labels show the rest of the discipline. The Internet’s identity system worked as a ledger, a telephone book and a handoff procedure at once. No one wanted to route packets by calling Hostmaster each morning; everyone needed a way to know whom to call when the ink and the wires disagreed.