An electronic mail message can be manually traced to its origin if (1) all system administrators of intermediate machines are both cooperative and competent, (2) the sender did not break low-level TCP/IP security, and (3) all intermediate machines are secure.
Users of cryptography can automatically ensure the integrity and secrecy of their mail messages, as long as the sending and receiving machines are secure.
For example, if you inject a message through sendmail or qmail-inject or SMTP, you can simply type in a From field. In fact, qmail-inject lets you set up MAILUSER, MAILHOST, and MAILNAME environment variables to produce your desired From field on every message.
A modern Received line contains quite a bit of information. In conjunction with the machine's logs, it lets a competent system administrator determine where the machine received the message from, as long as the sender did not break low-level TCP/IP security or security on that machine.
Large multi-user machines often come with inadequate logging software. Fortunately, a system administrator can easily obtain a copy of a 931/1413/Ident/TAP server, such as pidentd. Unfortunately, some system administrators fail to do this, and are thus unable to figure out which local user was responsible for generating a message.
If all intermediate system administrators are competent, and the sender did not break machine security or low-level TCP/IP security, it is possible to trace a message backwards. Unfortunately, some traces are stymied by intermediate system administrators who are uncooperative or untrustworthy.
Unfortunately, the creator of the seal is often an insecure machine, or an untrustworthy central agency, but most of the time seals are kept secure.
One popular cryptographic program is pgp.