actually, i think i'm the leader with the new code i posted...or maybe i missed something.

And here is the code that 'found' Allen's flaw. Logically, it should work the same as his original, but it fails.

Code:

function f($)
  dim $h,$c,$o,$t
  $f=ubound($)
  for $c=0 to $f
      for $h=0 to $f
         $t = a+$[$c]+$[$h]
         if len($t)=2*len(join($))/($f+1) and $h<>$c
             $o=$o+$t
         endif
      next
   next
  for each $ in split($o,a)
    $h=ubound(split($o,$))
    if $h>$t
      $t=$h
      $f=$
endfunction