There is no one correct way to  write  anything  in  romaji  (see
     Q2.7).   As a logical consequence, romanisation flames are rather
     futile.  There are, at best, recommended ways.

     Perhaps the closest thing to a correct way  is  the  romanisation
     scheme  the  Japanese  government announced on December, 9, 1954.
     Based on the "gojuuonjun", it uses the following kana  to  romaji
     conversions:

     n  wa    ra   ya  ma        ha     na   ta       sa       ka   a
        -     ri       mi        hi     ni   ti /chi  si /shi  ki   i
              ru   yu  mu        hu/fu  nu   tu /tsu  su       ku   u
        -     re       me        he     ne   te       se       ke   e
        -/wo  ro   yo  mo        ho     no   to       so       ko   o

              rya      mya       hya    nya  tya/cha  sya/sha  kya
              ryu      myu       hyu    nyu  tyu/chu  syu/shu  kyu
              ryo      myo       hyo    nyo  tyo/cho  syo/sho  kyo

                            pa   ba          da       za       ga
                            pi   bi          -  /di   zi /ji   gi
                            pu   bu          -  /du   zu       gu
                            pe   be          de       ze       ge
                            po   bo          do       zo       go

                            pya  bya         -  /dya  zya/ja   gya
                            pyu  byu         -  /dyu  zyu/ju   gyu
                            pyo  byo         -  /dyo  zyo/jo   gyo

                                                               kwa
                                                               gwa

     The table does not stand by itself.  It comes with  a  couple  of
     rules as well, the first of which says that in principle you have
     to use the romanisation that's on  the  left  hand  side  of  the
     slash,  if  there is one of course.  The right hand side alterna-
     tive is only to be used for words with strong international  con-
     notations, those that are customarily romanised that way or if it
     strongly improves the  information  content.   Rather  vague  and
     prone to abuse, but that's what it says.

     The other rules are:
     *  To disambiguate the letter `n' in a word like `kinen', you use
        a  single  quote if it is the romaji `n'.  So `kinen' is to be
        interpreted as `memorial', whereas `No smoking' is `kin'en'.
     *  The "sokuon" (small tsu) causes doubling of the following con-
        sonant, as in `sippai'.
     *  The "chouon" (lengthened vowel) is represented by a caret,  ^,
        over  the  vowel.  However, in case the vowel is a capital you
        are at liberty to write the extra vowel, as in `Oosaka'.
     *  Finally, and curiously, you can capitalize all  Nouns  if  you
        please,  not  just  proper  Names and the initial Word of Sen-
        tences.  Sounds very German, if you ask me.

     By the way, the dashes in the table indicate the  fact  that  the
     use  of  these  kana  for  Japanese  words is discouraged. In the
     "gojuuonzu" these positions are either empty or filled with  kana
     from other columns.

     [ed.: I believe this  romanisation  scheme  is  known  as  "nihon
     shiki",  but  am  not  altogether sure about that.  Could someone
     verify this?]

     If you are familiar  with  romanisation  schemes,  you  may  have
     noticed  that  the above roughly encompasses the two perhaps most
     famous schemes: "kunrei shiki" and  "hebon  shiki".   The  former
     uses  the  alternatives on the left hand side of the slash, while
     the later opts for the right hand side, with  some  minor  excep-
     tions for both:

     *  Both systems miss the entries for `dya', `dyu' and `dyo'.
     *  "kunrei shiki" uses  `di'  and  `du',  whereas  "hebon  shiki"
        sticks to `ji' and `zu' for the same kana, thereby introducing
        ambiguity.
     *  Both systems have some extra romaji covering kana in  the  "wa
        gyou"  (wa  column) that have slipped into disuse, namely `wi'
        (both) and `we' ("kunrei  shiki")  or  `ye'  ("hebon  shiki").
        "kunrei shiki" also uses `wo' instead of `o'.
     *  Before the voiced versions of "ha  gyou"  romaji,  like  `ba',
        `pa',  etc.,  and  before "ma gyou" romaji, "hebon shiki" uses
        `m' instead of  `n',  leading  to  things  such  as  `shimbun'
        instead of `shinbun'.

     Both systems were invented in the 1880's.   A  strictly  Japanese
     invention,  "kunrei  shiki"  was the official romanisation scheme
     prescribed  by  the  government  in  1937.   "hebon  shiki"   was
     developed  by  an international group including James Hepburn and
     made its claim to fame when Hepburn adopted it for the third edi-
     tion  of  his Japanese-English dictionary in 1886.  Subsequently,
     his name got attached to the scheme, as in  Hepburn  system,  and
     went  through some minor revisions since then.  You may also find
     it referred to as "hyoujun shiki" (standard form!).

     Of the two, it is the least likely to be mangled in pronunciation
     by non-Japanese (but still a far cry from being idiot-proof).