There are several kinds of fragmentations of signs into parts which are not functionally independent in cuneiform writing. Here is a list of some of these. These are from incomplete notes, so some may have to be corrected, and additional citations added for some. This is an older page, and other pages probably cover much of this in a more focused manner.

These are to be distinguished from actual historical changes, in which what once was a single sign comes in another era to be written as a sequence of two signs (allowing normal line break between them, for example, where that would earlier not have been normal in the middle of the single sign). For this second type, please see the page on Historical Change (not yet on line).

***

For signs having four parts arranged in a 2x2 square can be encoded and even named as single signs (no components specified), or can be treated as a left-right sequence of components treated as independent signs (SIGN OVER SIGN).(SIGN OVER SIGN). But a fuller study of the vocabulary indicates that greater productivity and closer correspondence of encoded characters to signs used arises from treating them in most instances as indivisible single signs, and naming them either simply as SIGN, or else as (SIGN.SIGN) OVER (SIGN.SIGN). For that fuller study, please click here.

***

Signs which bear no relation to the supposed parts

"ASH.ASH" (Labat #1) is simply HAL (Labat #2), no ASH involved, one sign, not a sequence of two.

BULUG3 (Labat #60b) is not a sequence PAP.PAP. It is the combination of the whole together which is the gestalt, see ZATU #62.

***

Signs which might etymologically have been compounds at one time, but whose parts fused rather early (the first sets of examples), or whose parts were replaced by different parts *only in the one sign*, so that encoding them as a sequence of parts would break historical continuity between earlier and later forms (the second set of examples). In all of these there is little or no reason to want the encodings to be different for different eras, violating our usual principle to encode what is equivalent across stylistic and historical change using the same character code, so far as this is not unreasonable.

DIR Labat #123 'excess' perhaps once from MA2+A (boat + water)? Whether or not, now an indivisible sign.

MASH2 (goat) perhaps from MASH ('1/2' used as phonetic?)+"HI.gunu" (testicles, used to signal male animal) ?

SHE.SUM was originally SUM x SHE ? But fused early so all three vegetal stems are parallel in some Uruk forms.

ZI2 Labat #147 was originally AB x PA, later fused

AM a high frequency syllabary sign, fused from GUD x KUR in Old Bab., Old Assyr.

A$$UR was in its origin a phonetic spelling ASH+SHUR. Should it be encoded as a single atom? I would bet this sign never separates across line break, was regarded by users as a single sign, not as a special reading of a sequence of signs.

***

Ellermeier #333b2 SHESH&NA later has a shape more of SHESH&KI. (I am starting to use the ampersand "&" to signal fused compounds, with apology to Steve Tinney's usage, because the contrast of the living ligatures of the type "fi, fl" with the fused ones like "&" is such a direct pointer to using "&" for this meaning. I'll try to make sure it is always clear in context.

Apparent SAG is replaced by apparent KA in Labat #19 and #18, specific to these two signs, which are therefore best regarded as indivisible?

U5 = HU x SI originally had a lower part as in HU x DUR2 'seat', later looks like HU x SI 'horn'.

U x UM or DUB became U x URUDU,
but UM or DUB alone did not become URUDU. So not U.UM or U.DUB, but a single atom later U x URUDU.

IL2 Labat #320 'raise', 'be high'
Sumerian and Old Babylonian it looks like 'raise' GA (milk product) above the head SAG.
By Neo-Assyrian, the second part reformed and resembles TUN3 'axe'.
Thus at no time should it be fragmented into parts *as signs*, never mind what it was when a *picture*

***

Some examples where alternate namings seem to reflect different understandings by different assyriologists, but where treatment as a single sign may avoid the difficulty?

GIR2.gunu as a late form of KISHIK ?