Culture
A Brief History of Email: Ray Tomlinson, the @ Sign and What Came Before
Who invented email? MIT's 1965 MAIL command, Ray Tomlinson's 1971 network mail and the @ sign, SMTP, MX records, Hotmail and Gmail, with every date sourced.

Ray Tomlinson sent the first email between two computers in late 1971, and he couldn’t tell you what it said. “The test messages were entirely forgettable and I have, therefore, forgotten them,” he wrote years later on his page at BBN. “Most likely the first message was QWERTYUIOP or something similar.”
That’s a fitting start for a technology whose history is full of firsts that turn out to have a predecessor. Tomlinson didn’t invent email, as his own account makes clear. What he did was send it across a network and give it the @ sign, which is enough to make him the name most people know. The rest was assembled over two decades by dozens of people, many of them arguing in memos about header fields and how to write the time.
Here’s the story, with dates checked against the original RFCs, the accounts of people who were there, and the most thorough technical history of the subject, Craig Partridge’s 2008 paper in the IEEE Annals of the History of Computing.
Email history at a glance
| Year | What happened |
|---|---|
| 1965 | MIT’s CTSS gets a MAIL command, written by Tom Van Vleck and Noel Morris |
| 1971 | Ray Tomlinson sends mail between two computers on ARPANET and picks the @ sign |
| 1972 | Larry Roberts writes RD, an early program for reading and managing a mailbox |
| 1973 | RFC 561 proposes the first standard headers: From, Subject and Date |
| 1974 | John Vittal starts MSG, which popularises Answer, the ancestor of the reply button |
| 1977 | RFC 733 becomes the official message format |
| 1979 | CompuServe opens to home computer owners, email included |
| 1982 | SMTP (RFC 821) and the RFC 822 message format are published |
| 1983 | MCI Mail launches at $1 a message, and DNS is specified |
| 1984 | POP, for fetching mail from a server, is proposed |
| 1986 | MX records let mail route by domain name |
| 1988 | IMAP2 is published as RFC 1064 |
| 1989 | MCI Mail connects to the internet, and “You’ve got mail” is recorded for the company that became AOL |
| 1992 | MIME adds attachments and text in other alphabets |
| 1996 | Hotmail launches free webmail |
| 2004 | Gmail launches with 1 GB of storage, on April Fools’ Day |
Before networks: MIT’s MAIL command, 1965
By 1965, the Compatible Time-Sharing System (CTSS) at MIT let hundreds of registered users share its IBM 7094s from dial-up terminals and keep files on disk. People being people, they started leaving each other notes. According to Tom Van Vleck’s history, users created files with names like TO TOM in shared directories, and the recipient would log in later, look for the file and print it out.
A staff planning note from around the turn of 1965, by Pat Crisman, Glenda Schroeder and Louis Pouzin, proposed a proper MAIL command. Nobody was free to write it, so two new members of MIT’s research staff, Van Vleck and Noel Morris, did it themselves that summer. MAIL was added to CTSS in August 1965.
You typed MAIL, the name of the file to send and the recipient’s problem and programmer numbers, and the message was appended to a file called MAIL BOX in their home directory. The login command learned to announce YOU HAVE MAIL BOX. Messages were limited to one disk record, 2,592 characters, because disk space was scarce. There was no subject line, though conventions, Van Vleck says, “rapidly sprang up”.
Management initially resisted “letters” as a waste of resources. There was also a quieter worry that calling it MAIL would provoke the US Post Office into demanding a first-class stamp for every message. An MIT professor eventually asked someone fairly senior at the Postal Service, who told them not to worry.
Van Vleck is careful with the word “first”. “I do not know who first invented these applications,” he writes, only that he hasn’t found any documented computer mail that predates it. Other systems had their own messaging: SDC’s Q32 let users send each other messages in 1963, though it had no mail facility, and Van Vleck notes that IBM’s CP/CMS had email as early as 1966. The filmmaker Errol Morris, Noel’s brother, investigated the story for a five-part New York Times series in 2011, and left the headline as a question: “Did My Brother Invent E-Mail With Tom Van Vleck?”
1971: Ray Tomlinson sends mail between two computers
By 1971 plenty of time-sharing systems had local mail, and ARPANET, funded by the US Defense Department’s Advanced Research Projects Agency, was starting to connect them. Ray Tomlinson was a programmer at Bolt Beranek and Newman (BBN) in Cambridge, Massachusetts, working on TENEX, an operating system for Digital’s PDP-10 computers. He had already written an experimental file transfer program called CPYNET.
In July 1971 Richard Watson of SRI published RFC 196, “A Mail Box Protocol”, a scheme for sending documents to numbered mailboxes attached to printers at other sites. Tomlinson told Partridge he found it “overly complicated because it tried to deal with printing ink on paper with a line printer and delivered the paper to numbered mailboxes.” Better to deliver a message to a person’s electronic mailbox and let them decide whether to print it.
The pieces were lying around. TENEX had a local mail program, SNDMSG, which appended messages to a mailbox file, and CPYNET could write files on another machine. “The idea occurred to me that CPYNET could append material to a mailbox file just as readily as SNDMSG could,” Tomlinson wrote. He added an append mode to CPYNET, a change he called “a no-brainer”, and folded its code into SNDMSG.
“The first message was sent between two machines that were literally side by side. The only physical connection they had (aside from the floor they sat on) was through the ARPANET.” After the forgettable tests came the first real message: “When I was satisfied that the program seemed to work, I sent a message to the rest of my group explaining how to send messages over the network. The first use of network email announced its own existence.”
The new SNDMSG shipped with the next release of TENEX in early 1972. TENEX was the most common operating system on ARPANET, so by Partridge’s estimate probably at least half its users now had network mail. It was crude. A message to several people on one host was sent once per recipient, and if the remote machine was down, SNDMSG reported a failure and gave up. Reading mail meant dumping your whole mailbox file to the terminal, and deleting a message meant cutting it out with a text editor.
Why the @ sign
SNDMSG needed a way to tell local mail from network mail, so Tomlinson added an at sign and the host name to the user’s login name. “I am frequently asked why I chose the at sign, but the at sign just makes sense,” he wrote. “The purpose of the at sign (in English) was to indicate a unit price (for example, 10 items @ $1.95). I used the at sign to indicate that the user was ‘at’ some other host rather than being local.”
Not everyone was delighted. On MIT’s Multics system, @ was the line-kill character: type it and everything you’d written on that line was erased. Multics users mailing other hosts had to type -at instead. A network mail meeting in February 1973 settled that “the standard form of the address is: user@host”, as RFC 469 records, yet the first header standard later that year spelled addresses out as TOMLINSON AT BBN-TENEX. For several years, Partridge notes, people wrote “at” in their headers and @ in the body of their messages. The symbol won in the end, and in 2010 the Museum of Modern Art added it to its design collection, captioned “Ray Tomlinson. @. 1971.”
From memos to conversations, 1972 to 1977
Email’s keenest early users were the people paying for the network. Partridge calls Larry Roberts, head of the ARPA office funding ARPANET, “probably email’s most important early adopter”. Early in 1972 Stephen Lukasik, ARPA’s director, took it up too, and soon tired of a mail reader that made him read his whole mailbox in order. One night that July, Roberts wrote RD, a set of text-editor macros that listed your messages and let you choose which to read. Colleagues rebuilt it as NRD, and then BANANARD.
In 1974 John Vittal, at USC’s Information Sciences Institute, began MSG, which became ARPANET’s most popular mail program. It had Move, Forward and, most importantly, Answer, which worked out who a reply should go to and copied the subject line across. Vittal had to decide what to offer, which Partridge sums up as “reply to everyone? reply only to the sender of the note?”, and it took three attempts to get right. Users at the time, Partridge writes, remember Answer as “converting email from a system of receiving memos into a system for conversation.” Every reply-all storm since traces back to that design question.
The messages themselves were a mess, because every system wrote headers its own way. RFC 561, published in September 1973 by Abhay Bhushan, Ken Pogran, Ray Tomlinson and Jim White, proposed that every message start with From, Subject and Date, followed by a blank line. There was no To field. RFC 680 in April 1975 added To, cc, bcc, In-Reply-To and Message-ID, though it was never officially issued as a standard.
RFC 733, dated 21 November 1977 and written by Dave Crocker, John Vittal, Ken Pogran and Austin Henderson, was the one that stuck. It settled the format still used for a name followed by an address in angle brackets. Getting there took what Partridge calls “a multiday argument (rich with ad hominem remarks)” over whether the Date field should use 12-hour or 24-hour times, and whether midnight was “12am”, “12pm” or “12m”. The 24-hour clock won.
Email was reaching beyond the labs, too. The Queen sent one over ARPANET on 26 March 1976, pressing a key during a visit to the Royal Signals and Radar Establishment in Malvern, according to the Queen Elizabeth Prize for Engineering. And in May 1978 a message from Gary Thuerk of Digital Equipment Corporation advertising its new computers went to hundreds of ARPANET addresses at once. The network’s managers called it “a flagrant violation”, and it’s generally counted as the first spam.
1982: SMTP and the format we still use
At first, ARPANET mail rode on the File Transfer Protocol, which gained MAIL and MLFL commands for the purpose in 1972. One convention from then has never left: in the MAIL command, a line holding a single dot marked the end of the message. As the network prepared to move to TCP/IP, Suzanne Sluizer and Jon Postel proposed a dedicated Mail Transfer Protocol in RFC 772 in 1980. Community lore remembers it as ugly; Partridge’s verdict is that it wasn’t ugly, just complex.
In November 1981 Postel described something much simpler, and published it as RFC 821 in August 1982. The Simple Mail Transfer Protocol needs three commands to send a message: MAIL FROM, RCPT TO and DATA. Each command takes at most one argument, and each reply opens with a three-digit code whose first digit tells you whether things went well. The message still ends with a line holding a single dot.
That same month Dave Crocker published RFC 822, the revised message format. Its most consequential addition was the Received header, which every mail server handling a message must add, leaving a trail of where it’s been. Both arrived just ahead of ARPANET’s planned switch to TCP/IP on 1 January 1983, set out in RFC 801.
SMTP was built for a network of researchers who mostly knew one another, and it has no way to check that a sender is who they say they are. That’s why, forty-odd years later, we bolt SPF, DKIM and DMARC on top.
Addresses that route themselves: DNS and MX, 1983 to 1986
Early ARPANET machines found each other through a single file, HOSTS.TXT, maintained centrally and copied to every host. The names were flat, so only one computer on the network could be called FRODO. Elsewhere it was worse. The UUCP network of dial-up Unix machines, which began inside AT&T in 1978, relayed mail hop by hop when the evening phone rates kicked in, and senders had to spell out the route. Someone at Berkeley mailing Partridge in the 1980s would address it to ucbvax!ihnp4!harvard!bbn!craig.
In November 1983 Paul Mockapetris specified the Domain Name System in RFC 882 and RFC 883. Its first design for mail, with separate Mail Destination and Mail Forwarder records, could make messages loop or go undelivered. Partridge, then at CSNET, found the problem, and Mockapetris proposed a single Mail Exchanger record instead. Partridge’s RFC 974 set out the routing rules and Mockapetris’s RFC 973 updated DNS, both in January 1986. The part after the @ stopped being a particular machine and became a name that says where its mail should go. MX records still route every email today.
In late January 1986, at a meeting at SRI, representatives of the Internet, UUCP, BITNET and CSNET agreed to work towards plain user@domain addresses across all four networks. Partridge was in the room: “The end of the era of ihnp4!ucbvax!bob%princeton.csnet@csnet-relay.arpa was visible and exciting.”
Collecting your mail: POP, IMAP and MIME
SMTP moves mail between servers that are always on. Once people had personal computers that weren’t, they needed a way to fetch mail from a machine that was. Joyce Reynolds proposed the Post Office Protocol in RFC 918 in October 1984.
At Stanford, Mark Crispin, a systems programmer there from 1977 to 1988, designed something more ambitious for working with mail held on a server: the Interactive Mail Access Protocol, IMAP. His obituary dates it to 1985, and many accounts give 1986. The second version was published as RFC 1064 in July 1988. It’s now the Internet Message Access Protocol, and still maintained.
The last gap was anything other than plain English text. SMTP was defined for 7-bit ASCII, and binary files had to be squeezed through tools like uuencode. In June 1992 Nathaniel Borenstein and Ned Freed published MIME (RFC 1341), which divides a message body into labelled parts, so attachments, images and text in any character set could travel through existing mail servers unchanged.
Email goes commercial: CompuServe, MCI Mail and AOL
Outside universities, email arrived as a paid service. On 24 September 1979 CompuServe opened its network to consumers with a modem, through a service RadioShack sold as MicroNET, and its subscribers were among the first to get email. Addresses were strings of numbers, like 72241.443@compuserve.com.
MCI Mail, launched in September 1983, priced email like post. According to a 2014 business history paper by Benjamin Schwantes, subscribers paid $18 a year, then $1 for an electronic message of up to 7,500 characters or $2 for one printed and posted. A first-class stamp cost 20 cents. Vint Cerf, co-designer of TCP/IP, was one of its architects, and in July 1989 MCI connected it to the National Science Foundation’s network, NSFNET. As Nextgov put it, “MCI did something no other company had done before when it connected its e-mail to the Internet.”
AOL gave email a voice. In 1989 Quantum Computer Services, the company that became AOL, wanted spoken alerts for its new software, and an employee’s husband, Elwood Edwards, recorded “Welcome”, “You’ve got mail!”, “Files done” and “Goodbye” on a cassette recorder in his living room. He was paid $200, NPR reported when he died in 2024. The phrase became famous enough to title a 1998 Tom Hanks and Meg Ryan film.
Webmail: Hotmail and Gmail
Until the mid-1990s your address usually belonged to your employer, university or internet provider, and you read it in a program on one computer. On 4 July 1996 Sabeer Bhatia and Jack Smith launched Hotmail, free email you could read in any web browser. Their investor Tim Draper suggested adding a line to every message users sent, “P.S. Get your free email at Hotmail”, and his firm later held it up as the case study for what it called viral marketing. Microsoft bought Hotmail on 31 December 1997, when it had more than 9 million members, for what MIT Technology Review reported as $400 million in Microsoft stock. By the time Microsoft moved the last Hotmail users to Outlook.com in 2013, the service had 400 million active accounts.
Google launched Gmail on 1 April 2004. The press release promised 1 GB of storage, “more than 100 times what most other free webmail services offer”, and had Larry Page recalling a user who “kvetched about spending all her time filing messages or trying to find them” and had to “delete email like crazy to stay under the obligatory four megabyte limit.” It summed up the development work in seven words: “Millions of M&Ms later, Gmail was born.”
Given the date, the tone and a storage claim that Time later said “seemed downright implausible”, plenty of people suspected a prank. It wasn’t one. Paul Buchheit had started building it in August 2001, and Google let people in slowly by invitation, with invites selling on eBay for $150 and more. Fifteen years on, Google said 1.5 billion people relied on Gmail, and recalled that it had launched on April Fools’ Day “despite the timing, not a joke”.
So who invented email?
Nobody, singly. It’s an unsatisfying answer, and the right one.
If you mean sending a message to another person’s mailbox on a shared computer, the earliest documented example anyone has found is CTSS MAIL at MIT in 1965, proposed by Crisman, Schroeder and Pouzin and written by Van Vleck and Morris. If you mean mail between computers, with a user@host address, that’s Ray Tomlinson at BBN in 1971. Tomlinson was clear about the difference: “Single-computer electronic mail had existed since at least the early 1960’s,” he wrote, and SNDMSG was an example of it.
If you mean the email you actually use, a crowd built it. Roberts and Vittal worked out how to read and reply, Crocker wrote the message format and Postel wrote SMTP. Mockapetris and Partridge made addresses route themselves, Crispin gave us IMAP, and Borenstein and Freed gave us attachments. Plenty more names sit at the top of the RFCs linked above.
You may also come across V.A. Shiva Ayyadurai, who says he invented email as a teenager in 1978, with a program he called EMAIL. When the Smithsonian’s National Museum of American History acquired some of his materials in 2012, it issued a clarification, quoted by the historian Thomas Haigh: “In accepting these objects, the museum did not claim that Ayyadurai was ’the inventor of email,’ as some press accounts have alleged. Exchanging messages through computer systems, what most people call ’email,’ predates the work of Ayyadurai.”
The 1980s are still in charge
The Radicati Group, a market research firm, estimated more than 4.4 billion email users in 2024, over half the world’s population, and forecast around 392 billion emails sent and received every day in 2026. The screens have changed beyond recognition since Lukasik read his mailbox top to bottom. The plumbing mostly hasn’t. SMTP has been revised, most recently as RFC 5321 in 2008, and a message still ends with a line holding a single dot, as it did on ARPANET in 1972. IMAP’s latest revision, IMAP4rev2, arrived in 2021. The @ is still Tomlinson’s, and MX records still decide where your mail goes.
That’s true of every mail client, ours included. Kanmail reads your mail over IMAP and sends it through your provider’s own SMTP server, on protocols designed in the 1980s. If you’re curious why some clients use Google’s proprietary API instead, and what that choice costs, we’ve covered IMAP vs the Gmail API. And if your inbox still feels like Lukasik’s in 1972, one long list in the order it arrived, there’s our guide to running it as a kanban board.
Frequently asked questions
Who invented email?
No single person. The earliest documented email system historians have found is MAIL on MIT's CTSS time-sharing computer, written by Tom Van Vleck and Noel Morris in 1965 for users of that one machine. Ray Tomlinson at BBN sent the first email between two computers over ARPANET in late 1971 and chose the @ sign for addresses. The standards still in use came later, from Jon Postel (SMTP), Dave Crocker (the message format) and many others.
What was the first email ever sent?
Nobody knows the words. Ray Tomlinson sent the first network emails in late 1971, as test messages between two computers sitting side by side at BBN. He later wrote that they were "entirely forgettable and I have, therefore, forgotten them", and that the first was most likely "QWERTYUIOP or something similar". The first one sent to anyone else was his note telling colleagues how to send mail over the network.
When was email invented?
Email between users of one computer dates to 1965, when MIT added a MAIL command to its CTSS time-sharing system, and other time-sharing systems gained similar features soon after. Email between computers arrived in late 1971 on ARPANET. SMTP, the protocol that still carries mail between servers, was published in August 1982, and MX records, which route mail by domain name, followed in January 1986.
Why do email addresses use the @ sign?
Ray Tomlinson picked it in 1971 because, as he put it, "the at sign just makes sense". It already meant "at" a unit price, as in 10 items @ $1.95, and he used it to show that a user was at another computer rather than on the local one. Not everyone agreed at first: on MIT's Multics system, typing @ erased the line you were writing.
Is email older than the internet?
Yes. Email on a single computer dates to 1965, four years before the first ARPANET machines were installed in 1969. Network email was running on ARPANET from late 1971, more than a decade before ARPANET switched to TCP/IP at the start of 1983, the protocols the internet still runs on. The internet grew up with email already on it.


