HashSet and TreeSet are implemented on top of HashMap and TreeMap 
respectively, in the obvious way: an object is a member of the set if and 
only if it's a key in the association map.  There are (at least) five ways 
to construct the map:

1. Map each key to itself.
2. Map each key to the map.
3. Map each key to its class object
4 Map each key to the map's class object
5.Map each key to some other distinguished object.

Both choose #5, using a private static object used for that purpose.  I'm 
wondering if this was an arbitrary choice, or if there's some (perhaps 
GC-related) reason for it, and, if so, what general principle it 
illustrates.