Our policy for a standard encoding for Cuneiform signs is to encode signs themselves, not readings of signs, and to encode all of the distinctions needed, even if they arose late or only in some usages, and even if they existed early but disappeared later. All scribal traditions which we end up attempting to cover (and I hope that will be all of them) are considered. This section gives a list of splits and mergers which we can consider. Please email Lloyd Anderson with any additional examples (which will then be posted to this site), or with any considerations such as given shortly below which bear on implementation.

There are implications for IMPLEMENTATION GUIDELINES from splits and mergers. It is good if we think about these from the very beginning, so we do not create some structure which is hard to implement, or to put that more strongly and positively, so we create the structure which is the best we can devise to minimize implementation work and to permit the easiest implementations, while supporting Cuneiform usage in as natural a way as possible (all of these goals hopefully coincide, as I will argue they do).

In each case of splits and mergers, there are AT LEAST the following possibilities. The first two treat the two characters as atoms.

1. One of two codes is to be used all the time, except in texts for the time and area where a split occurred, or where a merger had not (yet) occurred. In those texts only, use the second code for the sign which is distinct only there. Implementation guidelines are needed to choose which of the signs (where there is more than one) is to be equated to the single sign in texts where there is only one.

2. Use both codes as appropriate, according to rules which at least cuneiform specialists and preferably also non-specialists can understand. (This sort of situation is generally avoided in encodings, but will sometimes be necessary given that we are dealing with a writing system which underwent changes through time and space, therefore changes in the set of distinctions made, and given that we do not want to fail to support it by being unable to code the distinctions, and that users want what is known to be the "same" sign to be encoded the same way in texts of different eras.)

3. Encode the two signs as explicit variants, using a [variant selector] code following one of them to instruct a capable font to render with the other glyph instead. This may have the advantage that the [variant selector] can be disregarded, causing the two to be searched and sorted as equal in some contexts, so that the connection of the two is not lost.

4. Encode at least one of the signs as the other with a combining diacritic mark. This is one conceivable solution for the archaic number signs which have overlaid "texture" signaling what kind of thing is being counted or measured.

To take one example from Gary Beckman's comments, a sign distinction which exists (only?) in Hittite or in the Anatolian area.

HZL #23 UZ6 vs. #38 MASH2, both matching to Labat and Borger (ABZ) #76. Gary writes:

<< HZL 38 is the better choice for a direct continuation of ABZ 76,
with HZL 23 as the variant. Note that most combinations employ HZL 38>>.

This is one kind of consideration.

Added information: Friedrich did not separate these two (his #58). The separation was introduced by H. G. Güterbock, JCS 16, 1962 p.23, and adhered to by Rüster & Neu (HZL).

***

Some early splits and mergers are explicitly signaled by Prof. Dr. Manfred Krebernik. Hist lis (from OBO 161/1) is included in the page on "Fara Signs not yet identified" (please click above if you wish to go there), and will be integrated into the rest of this web site probably on October 27th. That list highlights some different ways in which distinctions appear or disappear.