A question about postgresql 8.1 and UTF strings

Good day all,
I have postgresql 8.1 (on WINXP service pack 2).
I have a table t_users with the columns id (serial), username (
varchar(256) ), and password ( varchar(256) )
The database was created with UT8 encoding.
I have 5 users at the database, and some of them have non-latin (especially
hebrew) usernames, i entered these non latin values simply by using the
pgadmin gui (was i correct to do so?) to enter the non-latin strings.
I opened a jdbc connection using a standard postgres JDBC URL and selected a
recordset with a hebrew username.
I used String user = ResultSet.GetString("username") to get the String
(which contains 4 hebrew letters)
and then I used

System.out.println(user.getBytes().length) - however, instead of seeing 8
bytes (2 bytes per each character, 4 characters ), i saw 4 bytes ....
Can you please tell me what is it that I'm doing wrong?

Thanks

Yair