The problem is with jdk 1.4 and the JIS0208 character set encoding.  Its
not encoding double byte character data correctly.  I attached a test
program for SUN to run.  The output of this test using the i18n.jar file
is:

26085
26412
38651
27671
26666
24335
20250
31038

The output of the same test using the charsets.jar file is:

70
124
75
92
69
69
53
36
51
116
60
48
50
113
60
82

Test code to demonstrate the problem:

public class CIITest {

  public CIITest() {

      byte [] ciiarray;

      //
      //Create a byte array containing the 16 bytes that will be encoded into the JIS0208 character set.
      ciiarray = new byte[16];
      ciiarray[0] = 0x46;
      ciiarray[1] = 0x7c;
      ciiarray[2] = 0x4b;
      ciiarray[3] = 0x5c;
      ciiarray[4] = 0x45;
      ciiarray[5] = 0x45;
      ciiarray[6] = 0x35;
      ciiarray[7] = 0x24;
      ciiarray[8] = 0x33;
      ciiarray[9] = 0x74;
      ciiarray[10] = 0x3c;
      ciiarray[11] = 0x30;
      ciiarray[12] = 0x32;
      ciiarray[13] = 0x71;
      ciiarray[14] = 0x3c;
      ciiarray[15] = 0x52;

      try {
         String temp = new String(ciiarray, "JIS0208");  //convert the byte array into a string with the JIS0208 character set.

          for (int i=0; i<temp.length(); i++) {
             System.out.println("Pos " + i + ": " + (int)temp.charAt(i));
          }

      }
      catch (Exception e) {System.out.println(e);}

  }
  public static void main(String[] args) {
    CIITest CIITest1 = new CIITest();
  }
}

The customer can fix the problem by removing the charsets.jar file and adding the i18n.jar file (from a JDK 1.3 install) in the jre/lib directory where they
reside.  However they won't have this option in a client's environment so
its imperative for SUN to understand this issue and not give them this
as a solution.

Looks to be a duplicate of 4766311 (committed to fix Tiger)
  xxxxx@xxxxx   2003-08-11