unable to understand query result

I have a database returing these result sets for two almost
identical queries:

#v+
$ select id, pid, length(pid), md5(pid) from product where pid  like '8000000';
 id   | pid     | length |               md5
------+---------+--------+----------------------------------
 3594 | 8000000 |      7 | 60b5792913f4acbccf45c281fa9e3c9f
(1 row)

$ select id, pid, length(pid), md5(pid) from product where pid  like '%8000000';
 id   | pid     | length |               md5
------+---------+--------+----------------------------------
 3594 | 8000000 |      7 | 60b5792913f4acbccf45c281fa9e3c9f
  722 | 8000000 |      7 | 60b5792913f4acbccf45c281fa9e3c9f
(2 rows)
#v-


There should be only one product with this id (no unique key,
because this is not a general constraint but just the current
situation), but there are two. Looking for it with query 1 I got
only one result, suspecting leading or trailing blanks I found the
other one using the wildcard.

But: What is wrong here? There must be some difference between the
two product ids, but with the same md5-hash there seems to be none.

This is by far too trivial to be a postgresql-bug, but what exactly
am I missing here?

Bye,
Stefan