I found a small error in Technical Report #16, "UTF-EBCDIC." In Section 3.5, "Signature," there is the following passage: The signature character U+FEFF (zero width no-break space) of Unicode transforms into the I8-byte sequence X'F1 BF B7 BF' which maps to X'DD 73 66 73' in UTF-EBCDIC. When this sequence is displayed (erroneously) using different a [sic] single-byte EBCDIC code pages, it can be visualized as different character strings. In Latin-1 EBCDIC code page 1047 (and coincidentally also in Latin-1 code pages 500 and 37), this byte sequence appears as "����" (small letter u with grave, capital letter E with diaeresis, capital letter A with tilde, capital letter E with circumflex). If the 4-character I8-sequence contains two 0xBF bytes, and they both map to 0x73 (as of course they must), then they will not be displayed as the two different characters '�' and '�'. The text should read: ... this byte sequence appears as "����" (small letter u with grave, capital letter E with diaeresis, capital letter A with tilde, capital letter E with diaeresis). The stray "a" in the passage which I marked with "[sic]" was left in for accuracy, but it is not the error I was referring to. The TR contains several such typos, so it would be unfair to single this one out. -Doug Ewell Fullerton, California
