That seemed like a simple, reasonable explanation, until I turned to the source and read Uriel's article for fact-checking purposes. It turns out that his article described sending an initial FIN packet followed by an ACK packet and then looking for discrepancies between their TTL values.
After hearing these conflicting facts, I used Nmap to run a Maimon scan, monitoring the session with the Ethereal packet sniffer. It turns out that the Nmap documentation correctly describes Nmap's behavior: it sends packets with both the FIN and ACK flags set. This mimics the second stage (FIN/ACK) of the three-way handshake used to tear down a TCP/IP connection. The setting also provides an alternative to FIN probes, which mimic the first step of the TCP breakdown handshake, and SYN probes, which mimic the first step of the connection setup handshake.
Why would you use Maimon's FIN/ACK probe? It's simply another way of eliciting responses from systems that are configured to cloak their presence on the network. Consider it one more weapon in your probe arsenal.
This was first published in April 2007