The encoding name "UCS-2" is used several times in the specification. This is not an IANA encoding name. What is intended is ISO-10646-UCS-2. A note in the IANA character set list notes that this encoding doesn't specify byte order but that it is needed. However, there are no endian variants for this encoding as there are with UTF-16 (UTF-16BE and UTF-16LE). I suggest we remove all mention of UCS-2 from the spec, and use UTF-16 instead. The byte-order issues for UTF-16 are already discussed. This requires changing the description of utf16Width property as follows: Section 11: utf16Width Replace paragraph begining with "Specifies..." by this: Specifies whether the encoding 'UTF-16' should be treated as a fixed or variable-width encoding. 'UTF-16' is potentially a variable width encoding using either 2-byte codepoints, or matched surrogate-pair codepoints containing two adjacent 2-byte codepoints that are combined to compute the character code. However, it is historically common for users to specify 'UTF-16' when they mean the fixed-width subset which does not allow use of the surrogate pairs. When utf16Width='variable', then surrogate pairs are expected and assumed to encode a single character. When utf16Width='fixed', then surrogate pair codepoints are treated as individual character codes. Section 12.3.7.1.1 Character Width Remove "(e.g., UCS-2)" from table, replace with "(e.g., UTF-16 with dfdl:utf16Width='fixed')". Replace "variable (e.g., Shift_JIS, UTF-8, UTF-16)" with "variable (e.g., Shift_JIS, UTF-8, UTF-16 with dfdl:utf16Width='variable')" -- Mike Beckerle | OGF DFDL WG Co-Chair | Tresys Technologies Tel: 781-330-0412