In Search of a Use for ARPANET
In 1971, the ARPANET ("Advanced Research Projects Agency Network") had just begun to emerge as the first larger network of computers. It was sponsored and created by the U.S. Department of Defense and would later lead to the development of the internet. But in 1971, the ARPANET was little more than connected computers, and those who knew about it searched for possible uses of this invention.
Richard W. Watson, for example, thought of a way to deliver messages and files to printers at remote sites. He filed his "Mail Box Protocol" as a draft standard under RFC 196, but the protocol was never implemented. In hindsight and given today's problems with junk email (and junk faxes before that), that's probably not all that bad.
Another person interested in sending messages between computers was Ray Tomlinson. SNDMSG, a program that can deliver messages to another person on the same computer had been around for about ten years already. It delivered these messages by appending to a file owned by the user you wanted to reach. To read the message, they would simply read the file.
SENDMSG + CPYNET = EMAIL
"Why not deliver text to a SENDMSG mailbox with CPYNET?" Tomlinson probably thought, and made CPYNET append to files (instead of replacing them). He then merged its functionality with that of SENDMSG so that it could send messages to remote machines. The first email program was born.
The Very First Network Email Messages
After a few test messages (containing the timeless words "QUERTYIOP" and maybe "ASDFGHJK"), Ray Tomlinson was satisfied enough with his invention to show it off to the rest of the group.
Delivering a display of how form and content are inseparable, the first "real" email, sent in late 1971, announced its own existence (the exact words have been forgotten, unfortunately). It also included instructions how to use the '@' character in email addresses.
(Updated September 2011)