Romanisation schemes are a reading aid, just like furigana.  They
     convert kanji and kana to something you can, supposedly, read and
     when pronounced actually resembles Japanese  pronunciation.   The
     reading  part  of this is fairly easy, but the pronunciation part
     is a lot harder.  It depends heavily on the target language  and,
     to  make  matters  worse,  opinions  on  what best represents the
     Japanese pronunciation in a given target language differ,  widely
     at times.

     [ed.: I'd like to include a striking example or two.  Any ideas?]