The Origins of E-mail

Despite common belief, e-mail actually pre-dates the Internet; in fact, existing e-mail systems were a crucial tool in creating the Internet.

E-mail started in 1965 as a way for multiple users of a time-sharing mainframe computer to communicate; although the exact history is murky, among the first systems to have such a facility were SDC's Q32 and MIT's CTSS.

E-mail was quickly extended to become network e-mail, allowing users to pass messages between different computers. The early history of network e-mail is also murky; the AUTODIN system may have been the first allowing electronic text messages to be transferred between users on different computers, in 1966, but it is possible the SAGE system had something similar some time before.

The ARPANET computer network made a major contribution to the evolution of e-mail. There is one report [1] (http://www.multicians.org/thvv/mail-history.html) which indicates experimental inter-system e-mail transfers on it shortly after its creation, in 1969. Ray Tomlinson initiated the use of the @ sign to separate the names of the user and their machine in 1972. The common report that he "invented" e-mail is an exaggeration, although his early e-mail programs SNDMSG and READMAIL were very important. The ARPANet significantly increased the popularity of e-mail, and it became the "killer app" of the ARPANET.

Growing popularity
As the utility and advantages of e-mail on the ARPANET became more widely known, the popularity of e-mail increased, leading to demand from people who were not allowed access to the ARPANET. A number of protocols were developed to deliver e-mail among groups of time-sharing computers over alternative transmission systems, such as UUCP and IBM's VNET e-mail system.

Since not all computers or networks were directly inter-networked, e-mail addresses had to include the "route" of the message, that is, a path between the computer of the sender and the computer of the receivers. E-mail could be passed this way between a number of networks, including the ARPANET, BITNET and NSFNET, as well as to hosts connected directly to other sites via UUCP.

The route was specified using so-call "bang path" addresses, specifying hops to get from some assumed-reachable location to the addressee, so called because each hop is signified by a "bang sign", i.e. "!". Thus, for example, the path ...!bigsite!foovax!barbox!me directs people to route their mail to machine bigsite (presumably a well-known location accessible to everybody) and from there through the machine foovax to the account of user me on barbox.

Before auto-routing mailers became commonplace, people often published compound bang addresses using the { } convention (see glob) to give paths from several big machines, in the hopes that one's correspondent might be able to get mail to one of them reliably (example: ...!{seismo, ut-sally, ihnp4}!rice!beta!gamma!me). Bang paths of 8 to 10 hops were not uncommon in 1981. Late-night dial-up UUCP links would cause week-long transmission times. Bang paths were often selected by both transmission time and reliability, as messages would often get lost.

This article comes from Wikipedia.org
http://en.wikipedia.org/wiki/Email