As shown in N3916: http://std.dkuug.dk/jtc1/sc2/wg2/docs/n3916.pdf = L2/10-356, there exists a Latin letter which resembles the Cyrillic soft sign Ь/ь (U+042C/U+044C). This letter is part of the Jaꞑalif variant of the alphabet, which was used for several languages in the former Soviet Union (e.g. Tatar), and was developed in parallel to the alphabet nowadays in use for Turk and Azerbaijan, see: http://en.wikipedia.org/wiki/Janalif . In fact, it was proposed on this base, being the only Jaꞑalif letter missing so far, since the ꞑ (occurring in the alphabet name itself) was introduced with Unicode 6.0.
The letter is no soft sign; it is the exact Tatar equivalent of the Turkish dotless i, thus it has a similar use as the Cyrillic yeru Ы/ы (U+042B/U+044B). In this function, it is a part of the adaptation of the Latin alphabet for a lot of non-Russian languages in the Soviet Union in the 1920s, see e.g.: Юшманов, Н. В.: Определитель Языков. Москва/Ленинград 1941, http://fotki.yandex.ru/users/ievlampiev/view/155697?page=3 . (A proposal regarding this subject is expected for 2011.) Thus, it shares with the Cyrillic soft sign its form and partly the geographical area of its use, but in no case its meaning. Similar can be said e.g. for P/p (U+0050/U+0070, Latin letter P) and Р/р (U+0420/U+0440, Cyrillic letter ER). According to the pre-preliminary minutes of UTC #125 (L2/10-415), the UTC has not accepted the Latin Ь/ь. It is an established practice for the European alphabetic scripts to encode a new letter only if it has a different shape (in at least one of the capital and small forms) regarding to all already encoded letter of the same script. The Y/y is well known to denote completely different pronunciations, used as consonant as well as vocal, even within the same language. Thus, if somebody unearths a Latin letter E/e in some obscure minority language which has no E-like vocal, to denote a M-like sound and in fact to be collated after the M in the local alphabet, this will probably not lead to a new encoding. But, Latin and Cyrillic are different scripts (the question in the "Re" of this mail is rhetorical, of course). Admittedly, there also is a precedence for using Cyrillic letters in Latin text: the use of U+0417/U+0437 and U+0427/U+0447 for tone letters in Zhuang. However, the orthography using them was short-lived, being superseded by another Latin orthography which uses genuine Latin letters as tone marks (J/j and X/x, in this case). On the other hand, Jaꞑalif and the other Latin alphabets which use Ь/ь did not lose the Ь/ь by an improvement of the orthography, but were completely deprecated by an ukase of Stalin. Thus, they continue to be "the" Latin alphabets of the respective languages. Whether formally requesting a revival or not, they are regarded as valid by the members of the cultural group (even if only to access their cultural inheritance). Especially, it cannot be excluded that persons want to create Latin domain names or e-mail addresses without being accused for script mixing. Taking this into account, not mentioning the technical problems regarding collation etc. and the typographical issues when it comes to subtle differences between Latin and Cyrillic in high quality typography, it is really hard to understand why the UTC refuses to encode the Latin Ь/ь. A quick glance at the Юшманов table mentioned above proves that there is absolutely no request to "duplicate the whole Cyrillic alphabet in Latin", as someone may have feared. - Karl Pentzlin
