About TCP/IP and AX.25 Addressing (For Reference)
An AX.25 path is called "source routing" in general networking terminology. This is because the path that the packet will travel through the network is determined by the source of the packet. E.g. If I know routers Bob and Sally get me to Jim, my destination address would be "Jim" with a path of "Bob,Sally". In more typical routing scenarios (e.g. Internet Protocol aka IP) the source node only knows how to get to the first router in the path (almost always your "default gateway"), and that node knows how to get the next step toward the destination.
IP includes a field called TTL (time to live) which is really a hop count, not a time. The source of the packet sets the maximum number of routers to be traversed ("hops") and each router decrements the TTL by 1. If the TTL drops to 0, the packet is discarded. AX.25 has no such provision - it isn't needed, because the route length is fixed by the path.
Another element of AX.25 addressing is the "SSID" secondary station identifier. An AX.25 address is a callsign plus a 4 bit number (0 to 15) so that a single ham can have 16 stations on the network at the same time. For a bit of added confusion, the SSID 0 is often left off, so you'll see "KG4WSV-3", but rarely "KG4WSV-0"; that'll be abbreviated to "KG4WSV". On a side note "CALL-N" is just a convention for representing what is actually a fixed width binary address; every AX.25 address takes up exactly the same number of bits when it's going over the air.
In an AX.25 packet, the source address, destination address, and 0 to 8 path addresses are all standard AX.25 addresses.
APRS Historical Paths
In order to accomplish a broadcast address (which doesn't exist in AX.25 AFAIK), the addresses "RELAY" and "WIDE" were used. AX.25 repeaters were configured to answer to an alias of "RELAY" and/or "WIDE" in addition to answering to their callsign. This allowed a packet to travel in multiple directions (network-wise) and allowed a generic PATH to work everywhere, critical to making the network useful in a mobile environment.
Do NOT use "RELAY" or "WIDE" on the air today! If you find web pages or docs suggesting it, they're OLD!
APRS Paths Today
The WIDEn-N concept used by APRS is a mashup. Instead of a fixed address in the source path (e.g. "Bob" or Sally" or "KG4WSV") it uses a generic address of WIDEn-N, where "n" is the requested number of hops (like the initial TTL in IP) and "N" is the number of hops remaining (analogous to the current TTL in IP). note the "n" is purely documentation; the functional part of the count is the "N", the part of the address that's in those 4 bits representing the SSID. In addition to WIDEn-N, there are special uses, e.g. STn-N where ST is the 2 letter state abbreviation. Effectively the n-N method uses the 4 bits of the SSID part of the address as a hop count.
An added twist is that there's a hop count for each of these generic path elements and AX.25 allows for multiple addresses in the path, so paths of "WIDE2-2,WIDE2-2" or "WIDE4-4" would yield a total of 4 hops.
There are additional historical artifacts, like the hack to allow ancient crude hardware incapable of the WIDEn-N SSID decrement trick to participate in the network as an edge repeater, where a packet could enter the network if the transmitting station couldn't be heard by a more capable n-N repeater. A path that looks like "WIDE1-1,WIDE2-1" is the result, the typical path for a mobile station in the U.S.
Many of the confusing aspects of APRS networking are due to hacks to support hardware that should be retired, e.g. the whole "WIDE1-1,WIDE2-1" thing.
More about paths around the world: http://info.aprs.net/index.php?title=Paths