2. Overview

   UTF-9 encodes [UNICODE] codepoints in the low order 8 bits of a
   nonet, using the high order bit to indicate continuation.  Surrogates
   are not used.

   [UNICODE] codepoints in the range U+0000 - U+00FF ([US-ASCII] and
   Latin 1) are represented by a single nonet; codepoints in the range
   U+0100 - U+FFFF (the remainder of the BMP) are represented by two
   nonets; and codepoints in the range U+1000 - U+10FFFF (remainder of
   [UNICODE]) are represented by three nonets.

   Non-[UNICODE] codepoints in [ISO-10646] (that is, codepoints in the
   range 0x110000 - 0x7fffffff) can also be represented in UTF-9 by
   obvious extension, but this is not discussed further as these
   codepoints have been removed from [ISO-10646] by ISO.

   UTF-18 encodes [UNICODE] codepoints in the Basic Multilingual Plane
   (BMP, plane 0), Supplementary Multilingual Plane (SMP, plane 1),
   Supplementary Ideographic Plane (SIP, plane 2), and Supplementary
   Special-purpose Plane (SSP, plane 14) in a single 18-bit value.  It
   does not encode planes 3 though 13, which are currently unused; nor
   planes 15 or 16, which are private spaces.

   Normally, UTF-9 and UTF-18 should only be used in the context of 9
   bit storage and transport.  Although some protocols, e.g., [FTP],
   support transport of nonets, the current IETF protocol suite is quite
   deficient in this area.  The IETF is urged to take action to improve
   IETF protocol support for nonets.