PRIMARY KEY on a *group* of columns imply that each column is NOT NULL?

If I define a primary key:

       name TEXT NOT NULL,
       address INET,
       PRIMARY KEY(name, address)

the definition (seen by \d) becomes:

 name          | text                        | not null
 address       | inet                        | not null

"address" is now not null, which I do not want. It seems unnecessary:
I just want the tuple (name, address) to be unique, which seems
possible even if some 'address' values are NULL.

It does not appear to be documented in
http://www.postgresql.org/docs/7.4/interactive/ddl-constraints.html#AEN1975.
Is there a workaround?