A tabulation of two-component signs in the ZATU list for archaic Uruk cuneiform
(EXCLUDING multi-component signs such as ENSI, ENKUM, NINKUM),
noting whether they continue into later cuneiform after Fara and Ur III, and what the arrangement of the components is both earlier and later, yields a pattern of high regularity, which can almost be called a rule.

This is important because it helps us to determine what are single signs of cuneiform writing, in LATER cuneiform as well as in EARLIER cuneiform. A small list of signs which should be added to the UTC encoding comes at the end of this message.
The total is 8 two-component signs from later cuneiform, and 13 two-component signs from archaic Uruk, which this approach argues function as single signs, but which are not yet in the UTC list.

The tables are summarized on this web page
http://www.CuneiformSigns.org/AdjacentAttached.htm
including complete lists of the signs referred to in each count, and their attestations in all lists so far included in the concordance.
(This page will later be linked to the ZATU triage, whose revision is not yet posted.)

The pattern is in summary this:
1. Signs which have components touching in archaic Uruk will be single signs (fused, infixed, parts touching) in later cuneiform.
2. Signs which have components not touching in archaic Uruk almost never survive into later cuneiform as single signs.
3. Signs which have components next to each other sideways (before the rotation) never continue into later cuneiform as single signs.
With the exception of course that paired components do occur (they are named in the formula SIGN over SIGN in the PSList, referring to the post-rotation positions of the components), and there are two signs listed with distinct components rather than a repretition of a single component: SIGN-one over SIGN-two).

There seem to be a number of sub-groups permitting other predictions.
Within group 1.,
(a) When a modifying component is nested above, or touching above the later form may be fused, or touching.
(b) When a modifying component is nested below, or touching below, the later form may be infixed.

The role of meaning transparency of components:
There are a number of Uruk combinations in the ZATU list having modifiers (?) like SHESH and MUSH3 above and to the left of a base (?) UNUG or AB or other signs, and these are not considered single signs in later cuneiform. Is that perhaps because the semantic relation remained transparent, in some cases as place names?
Consider by contrast IL2 = SAG with GA and a support above the head. This is similar to the preceding if we consider only the physical location of the components. But the components of this single sign do not remain transparent, it is later visually fused (and presumably it became functionally fused before it became visually fused). In general there seems to be little need to appeal to meaning transparency of components. The form of the signs alone is sufficient to determine whether we have a two-component sign or a two-sign compound. Already in archaic Uruk.

This formal distinction in the archaic Uruk period of course warrants further investigation. But it reinforces evidence from other features
of later sign form and behavior in text, such as the lack of division of components of single signs across ends of lines and the lack of
any significant spacing between components of single signs, even when there is lots of extra space within a single line.

The signs in the ZATU list which should then be added to any list of cuneiform single signs are then these:

A. Those persisting into later cuneiform:
UZ = HU w SHE
MUD = HU w HI
E3 = DU w UD
PAD3 = RU w IGI
ERIM2 = RU w NE
SHU w NIGIN2
SAL w NAM2
SAL w KUR

B. Those in archaic cuneiform which do not persist into later cuneiform:

(fused)
EN x NUN

(nested inside)
RU w NI
SI w GUD

(nested)
MURUB2 = SAL x ?LAGAR?

(attached above right)
PI w SHE
LAGAB (circle) w SHE
GISH w SHE

(attached above)
ARINA = Z379 w two BU
NAMESHDA (optionally not attached)
(SHUSH Krebernik =)   NAM2 w SHE
Z158 (EZINU) w SHE
"SHELU" = LU3 w SHE
KIN2(?) w SHE