I have to use Japanese hankaku (single byte kana in (S)JIS) in mail subject and message,
so I tried the way that NO3L in Japan said, but I found there's a big problem.
In that way, mb_encode_mimeheader drop escape sequence from the head of encoded string from second line.
if you use long letter in subject, and if it includes multi byte string, It must crashed.
Therefore, if you want to use Japanese hankaku in mail subject, you might try like this.
Encoded subject letters in one line is not just 76 letters, so it is not based on RFC, but it work.
$intSubjectLength = mb_strlen($strSubject);
$intSeparateLength = 10;
for ($i=0; $i<ceil($intSubjectLength / $intSeparateLength); $i++) {
$arrSeparatedSubject[$i] = mb_substr($strSubject, $intIndex, $intSeparateLength);
$arrSeparatedSubject[$i] = mb_encode_mimeheader(mb_convert_encoding($arrSeparatedSubject[$i], "JIS", "EUC-JP"));
$intIndex = $intIndex + $intSeparateLength;
}
$strSubject = join("\n ", $arrSeparatedSubject);