Summary

  • The Chinese Academy of Sciences and IHEP credit an August 1986 email from Beijing to CERN as China’s first; their account describes satellite-assisted remote access, not a CSNET mail node.
  • A separate ICA service opened to Karlsruhe over X.25 in September 1986. In 1987 a CSNET/PMDF host at ICA sent a message into international scientific mail; it was drafted on 14 September and successfully transmitted on the 20th.
  • The 1987 route ran before NSF’s 8 November letter welcoming CSNET and BITNET email to China. It was a working email link, not full Internet access, which CERnet histories place in 1994.

“First” depends on what the network did

Two dates recur in accounts of China’s first email. The Chinese Academy of Sciences archive and the Institute of High Energy Physics (IHEP) place one milestone on 25 August 1986. The sender, Weimin Wu, was at a Beijing 710 Institute computer; the account says he remotely logged in over a satellite link to a CERN machine in Geneva and sent an email. CERN Courier identifies the recipient as Jack Steinberger. IHEP calls it China’s first email.

A second “first” belongs to a different system. In September 1987, a team at the Institute for Computer Application (ICA) in Beijing and colleagues at Karlsruhe built a CSNET/PMDF mail node. The message they composed was headed “First Electronic Mail from China to Germany.” It was meant to enter international scientific networks through a computer connection between Beijing and Karlsruhe, and its recipient list reached researchers in the United States and Ireland as well as Germany.

These accounts need not cancel each other. A message sent from China to a foreign research computer is not the same milestone as a mail host that can relay messages through an international network. The useful question is not which anniversary wins; it is which endpoint and service each claim describes.

The 1986 exchange depended on a scientific link

The CERN connection answered a practical need in high-energy physics collaboration. The CAS archive describes a satellite link and remote login from Beijing to CERN. It records an email, but does not document a CSNET host in China or a continuing international mail relay. The archive is therefore strong evidence for the 1986 email claim and limited evidence for broader network membership.

There was another step that year. Werner Zorn’s account says that on 1 September an email service over an X.25 remote dialog opened between ICA in Beijing and the computer-science department at Karlsruhe. This was more than one message: the service was available to multiple users. Yet a remote dialog with one foreign host still says less than a local CSNET node able to exchange messages with the wider research network.

The message had to cross several networks

The 1987 story is clearest when its six-day delay is retained. The China–Germany team composed the “Across the Great Wall” message on 14 September. The mail did not leave China that day. Jay Hauben’s reproduction of the original record and Zorn’s account say the first successful transmission was on 20 September.

At ICA, a sender used a 300-baud modem to reach an X.25 port at PKTELCOM Beijing. A satellite hop carried traffic to Italy’s ITAPAC packet network. A gateway passed it to Germany’s DATEX-P network, which delivered the message to a Siemens host at Karlsruhe. There, the local CSNET node relayed mail toward other destinations. The route was expensive; each international segment carried its own charges. This was store-and-forward email assembled from several separately operated links, not one continuous China-to-Internet circuit.

Permission followed a working path

On 8 November, Stephen Wolff, then a director in the US National Science Foundation, sent a letter welcoming the extension of CSNET and BITNET email to China. The letter came after the successful September transmission. That order matters: the approval gave the service a clearer institutional footing inside networks funded and coordinated in the United States, but it did not create the Beijing–Karlsruhe route or cause the test message to work.

CANET’s launch in March 1988 widened the institutional base. CERnet describes the China Academic Network as connecting dozens of research and education bodies and exchanging email through an X.25 gateway at Karlsruhe. Its history distinguishes that email service from full Internet connectivity, established later. A mail node could bring researchers into international exchange without making every campus, host or user part of a general IP network.

A later lens, not a claim about 1980s intent

Heng Lu’s Note 65, Running-Code Primacy, offers a useful later test: describe the implementation that ran, the participants who adopted it and the reach they actually gained before treating a policy letter or historical label as the whole system. That is an interpretive lens, not the vocabulary of the teams in Beijing and Karlsruhe. Applied here, it keeps three facts in view: the 1986 CERN email happened; the 1987 CSNET node enabled a broader mail service; and neither event by itself was full Internet access.

The record leaves one limit. Available accounts do not specify the complete protocol stack behind the 1986 CERN message or show that it supplied a continuing public mail service. For 1987, by contrast, the preserved mail headers, route description and subsequent CSNET approval identify both the service boundary and its relay. The distinction is technical, not ceremonial.

Sources