The scientific transliteration system is roughly as
phonemic as is the orthography of the language transliterated. The deviations are with щ where the transliteration makes clear that two phonemes are involved, and џ, where it fails to represent the (monophonemic) affricate with a single letter. The transliteration system is based on the
Gaj's Latin alphabet used in
Serbo-Croatian, in which each letter corresponds directly to a Cyrillic letter in
Bosnian,
Montenegrin and
Serbian official standards, and was heavily based on the earlier
Czech alphabet. The Cyrillic letter х, representing the sound [x] as in
Bach, was romanized
h in Serbo-Croatian, but in German-speaking countries the native digraph
ch was used instead. It was codified in the 1898
Prussian Instructions for libraries, or
Preußische Instruktionen (PI), which were adopted in Central Europe and Scandinavia. Scientific transliteration can also be used to romanize the early
Glagolitic alphabet, which has a close correspondence to Cyrillic. Scientific transliteration is often adapted to serve as a phonetic alphabet. Scientific transliteration was the basis for the
ISO 9 transliteration standard. While linguistic transliteration tries to preserve the original language's
pronunciation to a certain degree, the latest version of the ISO standard (ISO 9:1995) has abandoned this concept, which was still found in
ISO/R 9:1968 and is now restricted to a one-to-one mapping of letters. It thus allows for unambiguous reverse transliteration into the original Cyrillic text and is language-independent. The previous official Soviet romanization system,
GOST 16876-71, is also based on scientific transliteration but used Latin
h for Cyrillic х instead of Latin
x or
ssh and
sth for Cyrillic Щ, and had a number of other differences. Most countries using Cyrillic script now have adopted
GOST 7.79 instead, which is not the same as ISO 9 but close to it. Representing all of the necessary diacritics on computers requires
Unicode,
Latin-2,
Latin-4, or
Latin-7 encoding. ==Table==