Typographic Evidence on Single-Sign Status

from a range of published texts (drawings of tablets)

In a large portion of published texts of various types there is sufficient spacing between single signs so we can tell the difference between a sequence of two independent signs and a single sign which has multiple components. Because Cuneiform, unlike Chinese, does not have a fixed size of each character block, the task is not as straightforward as it is for Chinese. However multiple lines of evidence do converge. Here are results for older texts, from Ur III backwards, showing that the typographic evidence is on the whole highly consistent. The texts so far covered include the following. More Ur III texts are being surveyed to add to this.

[Ur III Sale] *Sale Documents of the Ur-III-Period* by Piotr Steinkeller 1989 -- 22 tablets for which new drawings are included, together with their transliterations

[Gudea] *Gudea and His Dynasty* by Dietz Otto Edzard -- so far the Gudea Cylinder Fragments 8+3+5+4 and Fragment 12 -- this will be added to for the full text of the Cylinders using a publication of them combined with Edzard's transliterations

[Girsu] *Altsumerische Verwaltungstexte aus Gursu/Lagash* by Joachim Marzahn -- 109 tablets, no transliterations
[Lagash] "Inscriptions from Lagash, Season Four, 1975-76" by Vaughn Crawford, JCS 29 no.4 Oct. 1977 pp.189-222 --10 tablets with transliterations
[Al-Hiba] *Inscriptions from Al-Hiba - Lagash* by Robert Biggs, Undena Publications 1976 -- 41 tablet drawings

[Fara Lit.] *Die Inschriften von Fara. II. Schultexte aus Fara* by Anton Deimel 1923 -- 6 tablet drawings
[Fara] *Die Inschriften von Fara. III. Wirtschaftstexte aus Fara* by Anton Deimel 1924 -- 23 tablet drawings
[Abu Salabikh] *Inscriptions from Tell Abu Salabikh* by Robert Biggs 1974. OIP 99 -- the first 90-some tablet numbers out of 515, not all surveyed as systematically as the other sources, because I began here.

[Nippur] *Cuneiform Texts from Nippur. The Eighth and Ninth Seasons* by Giorgio Buccellati and Robert Biggs (1968+) -- 58 tablet drawings varying from Early Dynastic I to Middle Babylonian

[Gilgamesh Parpola] In addition, there are a couple of references to Parpola's 1997 *The Standard Babylonian Epic of Gilgamesh*, to sample an expert computer rendering of later usage.

I may of course have missed some examples with components separated, but the method of searching I used should not lead to this being at all common. I include this disclaimer only in order to not be making absolute claims that there is no exception whatsoever. Single exceptions cannot overpower the massive evidence of normal usage.

The numbering of lines I use in referring to reverses of tablets may occasionally be in the wrong order, and occasionally include spaces by error.

The Abu Salabikh texts in addition appear to contain quite a number of new signs. Since I was not yet looking for these, and have not yet checked their readings in context with competent authorities or publications, these remain tentative for the time being. TU paired at 69.169, SAR x MU at 42.4.10, (TUR3 paired) x AN at 53.vii.8, GA paired, (MUSH paired) x KAK at 1.121, and surely quite a number more.

Signs with "U" (SHU4) or SHU2 as leftmost part (originally uppermost, in archaic drawings normally referencing a cover or the sky) All such combinations are single signs, I have never seen even one example where the "U" (SHU4) or SHU2 is separated from what follows it, either by substantial white space as between known independent signs, or positioned on a separate line. Here only a couple of examples which I happened to notice in the texts surveyed. Was not looking for these. Please click here for a fuller list.

SHUDUL "SHUDUN" (= u$-ti-lu in Ebla syllabically written form) (Manfred Krebernik, in his *Beschwörungen aus Ebla und Fara* p.142): This sign. Fara LAK #274, was originally a yoke for draft animals, did not contain SHU2 at all. Later it was reanalyzed and changed its form to *look like* SHU2 x DUN4) Borger catalogs this as B876 -- not yet in the Unicode proposal of February 2004, should be added

SHAGAN [Al-Hiba 15.v.2]

GIR4 = SHU4 x AD ("U" x AD), B683 [ ]

UL (so-called "U.GUD") -- [Gudea 8+3+5+4.iii'.4', 8+3+5+4.iii'.7'] [Al-Hiba 29.iv.8] [Fara 9123.ii.4, 9123.iv.2] [Fara Lit. 12148.ii.3] -- already in the Unicode proposal of February 2004

UL paired (UL over UL, or "U.GUD over U.GUD") -- [Fara 12735.vii.3, 12436.viii.8, 9072.vii.3, 12618.i.3] -- not yet in the Unicode proposal of February 2004, should be added. Borger B698 does not assign the paired form any number separate from the single component = sign.

*

NIN (so-called "SAL.TUG2") There are literally hundreds of examples of the sign NIN throughout cuneiform texts of various periods. In nearly 100% of them the point of the component SAL actually touches the line of the TUG2. In not a single case did I notice any substantial separation between the components. In a tiny number of cases, there is a slight separation, but it is always less than the separation between independent signs. NIN is clearly and obviously a single sign. -- [Ur III Sale 16 examples; 94**.17 with small opening] [Girsu 26 examples plus on tablet 69 another 61 examples] [Lagash 4H-T1.v.1', 4H-T1.vi.5', 4H-T1.vi.7', 4H-T7.xi.3'] [Al-Hiba 23 examples] [Fara 21 examples, a few of these to be excluded not NIN] [Fara Lit. 12626 with 19 examples, 12644 with 24 examples] [Abu Salabikh 1.65, 86.ii.1, 86.iii.4-8, 86.iv.6, 86.v.6-10, 86.vi.7, 86.vi.9] [Nippur EDIII-Sargonic 10.16, 10.20, 10.14; Ur III 27.8; Lugalzagesi 27.3] -- not yet in the Unicode proposal of February 2004, should be added

GEME2 (a reading in later periods; so-called "SAL.KUR") -- [Ur III Sale 44.7. Nippur EDIII-Sargonic 8.11] -- not surveyed in general but there are quite a number of cases in which this is obviously a single sign.

SIPA (so-called "PA.UDU") Notice that the sign's reading does not use the parts in that order, and is almost certainly retained from archaic cuneiform, SI with PA above it. Occurrences of this sign surveyed never showed the components of this sign with any substantial separation, though they are not uniformly touching. Although the separations can be slightly greater than for NIN, they are not as great as for late versions of ZU.AB (see below). In [Gilgamesh Parpola] SIPA is encoded as a single unit, and in renderings is therefore never divided by additional white space, though an accompanying syllabogram may be separated from it even when part of the same word SIPA-ma, as in Tablet I lines 59 and 71. -- [Ur III Sale 60.15'] [Girsu 9.i.3, 11.iv.16, 11.vi.3, 11.vii.11, 11.viii.1, 14.i.4, 15.iii.4, 25.iv.3, 36.ii.9, 40.iii.641.vi.5, 41.vii.7, 44.i.3, 59.ii.3, 66.viii.5', 66.xi.9, 70.ix.9, 71.ix.7, 71.x.17, 71.xi.7, 74.v.5, 74.v.6, 74.v.8, 79.xi.12, 79.ii.4'? last, 80.ii.1, 84.iv.8, 87.iii.7, 89.iii.3, 89.v.7', 89.iv.10, 99.vi.4, 99.vi.5, 101.ii.2] [Fara 12436.iii.4, ?12478.ii.4] [Nippur Kadashman-Enlil inscription in Sumerian 52.5; Middle Bab. 53.3] -- not yet in the Unicode proposal of February 2004, should be added

U3 (so-called "IGI.DIB") Parts are always touching, and never separated -- [Ur III Sale 60.2] [Girsu 7.ii.2, 41.ii.5, 41.iii.9, 69.xii.4, 84.vi.2, 103.i.9, 103.ii.6] [Al-Hiba 9., 36.5, 38.10] [Nippur 9.10] -- already in the Unicode proposal of February 2004

AR = so-called "IGI.IR" Surveyed in -- already in the Unicode proposal of February 2004

HUL (so-called "IGI.UR") -- [Ur III Sale 3.7, 5.8, 44.11, 45.20, 65.13/14/15, 125.13/14 ]

PA3 or PAD3 (so-called "IGI.RU") -- a single sign in early periods, meaning "to save", the IGI portion almost always nested inside the RU, sometimes the point of IGI even touching the innermost angle inside the RU -- [Ur III sale 65.18, S3.5] [Lagash 4H-T8.i.2', 4H-T1.iv.1'] [Al-Hiba 1.6, 3.i.10, 3.ii.5, 10.iii.3'] [Fara 12436.vii.6] -- not yet in the Unicode proposal of February 2004, should be added

ERIM2 = (so-called "NE.RU", RU with NE inside) -- [Abu Salabikh 65.56, 54.81, not surveyed systematically] If this originally represented the burning of a valley / fields (note value ERIM2 for war), then the combination was pictorial and not a syntactic sequence, which would be suggestive that it is a single sign. -- not yet in the Unicode proposal of February 2004, should be added

Shuruppak ("LAM.KUR.RU") -- [Fara 32 examples] [Fara Lit. 9128 Reverse.5] [Abu Salabikh 83.ii.15 is an example where the components are not overlapping, but still in a single stack, not more than one]

ABZU (so-called "ZU.AB") Notice that the sign's reading does not use the parts in that order, and is almost certainly retained from archaic cuneiform, AB with ZU as modifier above it. In Fara 12483.iv.2, lowest line of the reverse, the neck of AB is highly stretched, signaling the single-sign status, rather than permitting the parts to be separated. Would this conceivably relate to its meaning "the abyss(al sea)". ABZU is written as a single sign in Fara and Abu Salabikh. In archaic Uruk, the components appear to be separated as are components of virtually all signs. Is the positioning of those components already then fixed, the ZU above the AB? In later cuneiform, we appear to have a text unit (lexical item) but not a single sign. In [Gilgamesh Parpola] ZU.AB is encoded as the sequence of those two units in one lexical item, and in renderings is therefore divided by some white space, though not the maximum white spaces; so in Tablet VIII line 174 and Tablet XI lines 42 and 282.-- [Ur III Sale 94**.25] [Girsu 26.iii.3] [Lagash 4H-T1.v.6'] [Al-Hiba 3.i.2, 26.i.8'] [Fara 12436.iii.3, 9072.ii.8, 9122.iv.5, 12483.iii.7, 12483.iv.2.] [Fara Lit. 12148.ii.3] -- not yet in the Unicode proposal of February 2004, should be added to accomodate the Fara and Abu Salabikh usage, since there it is a single sign.

TUR3 (NUN x LAGAR) -- [Abu Salabikh 1.28, 46.vi.8?, 83.iv.8, examples where parts are superimposed, not sequential]

GIRSU and E3, examples where there do occur *some* instances with components separated, as separate signs. Given that there is fluctuation and historical change, it may be that in some cases signs are unitary single signs only in certain times and places, that in other times and places the single sign does not exist. By the principle of keeping all distinctions, whether later merged, or later split, a single sign rates status as a distinct encoded character in a computer standard if it occurs in any time or place as a single sign. The only exception would be cases where we might consider that a ligature was used only in a limited time and place. But as agreed, ligaturing is not to be used as a substitute for encoding single signs when the status as sequence of signsis is not normal..

GIRSU (GIR2 and SU as single sign in the majority of instances, though not in every one. The single-sign instances most often have the point of the diamond of GIR2 touching the point of the triangle of SU. In some cases, one diagonal side of the GIR2 runs along a diagonal side of the SU. In only a tiny number of cases are the parts of GIRSU written separately. (In quite a number of these examples, the context is NIN-GIRSU.)

NINGIRSU as a single stack -- [Girsu 11.ii.12, 21.iii.1, 26.ii.2, 80.ii.2, 99.iv.5, 99.vi.6] [Lagash 4H-T3.i.1 4H-T3.i.8, 4H-T3.ii.5] [Gudea Fragm.12.i'.3', 12.i'.4', Fragm.8+3+5+4.iv'.2', 8+3+5+4.iv'.7'] [Abu Salabikh 82.i.8, 82Rev.iii.3]

NIN and GIRSU each as single unit, but not vertically above each other -- [Girsu 10.vi.3, 41.vi.8, 41.iv.3, 70.vii.3, 70.vii.14, 74.i.7, 79.v.4, 87.ii.5, 89.vi.6, 97.vi.ii]

GIRSU with components GIR2 and SU touching at points (NIN may be present but in this sub-list not in the same vertical stack) -- [Girsu 41.vi, 70.xii.2] [Lagash 4H-T7.i.1, 4H-T7.v.3, 4H-T3.i.9, 4H-T6.1, 4H-T2.1, 4H-T9.1 same vertical stack but NIN separated from GIRSU] [Al-Hiba 26.ii.7, 29.iii.7]

GIRSU with components GIR2 and SU touching on diagonal sides -- [Girsu 15.ii.4, 22.iv.4, 42.iv.2, 71.xi.5] [Lagash 4H-T1.iv.2'] [Al-Hiba 7.i.2', 15.vii.3']

GIR2 and SU separated -- [Lagash 4H-T3.ii.9, 4H-T3.iii.8] [Al-Hiba 1.9, and in reversed order! SU over GIR2 28.i.1]

E3 (DU with UD above or nested in the elbow space, or superimposed) -- [Girsu 14.xi.2', 17.v.1, 97.vii.3] [Al-Hiba 1.41] [Fara 9123.ii.4, 12745.iv.9, 9131.ii.2, 12596.vi.2] [Fara Lit. 12626.v.8, 12644.iv.5, 12626.vii.4, 9128.v.14] [Abu Salabikh 69.12, 69.253]

Kerning: Here are some examples of Kerning. An argument that we have an instance of "kerning" or "ligaturing" rather than a "single sign" etc. may rest on evidence that the items so combined are not in immediate constituency with each other, rather that their structural groupings are with other signs to right or left. Or that the combination is in free alternation with the sequence of two independent signs. These examples do *not* constitute evidence of status as a single sign.

DUMU kerned with following UR (and with some other signs ?) [Ur III Sale 7.5, 7.8, 44.10, 44.seal.iii, 60.6, 65.5]

LAL kerned? [Ur III Sale 3.3 LAL-IGI, 3.8 LAL-LU2, but not kerned LA2 45.8]
LAL x ?? = USAR3 [ur III Sale 5.6]

[d]EN Alternates with the sequence of the two signs. -- [Ur III Sale 3.8, 19.17, 37.22, 37.4, 94**.5, 94**.9, S3.4, S5.8] [Fara 9072.ix.9, 12608.vii.1, 12440.i.6, 12440.i.8] [Fara Lit. 12626.i.2, 12626.i.5, 12626.?, 12626.x.2, 12626.x.5, 12148.i.3]

EN has an elbow space which is used to nest a number of other signs. So for example EN with SIG7 nested is parallel to NIN with SIG7 in Abu Salabikh 1.64 vs. 1.65 (both lines further contain NUN with ME below and EZEN x KASKAL). 1.93 contains EN with NUN nested. And so on. -- EN x ME is in the Unicode proposal of February 2004, and probably is the most fossilized of the combinations. But even EN x ME, when infixed into LAGAB, is narrowed in later forms so it looks like ME followed by EN. These two forms should probably not be considered or encoded as distinct characters. Naming them so as not to imply that may be difficult.

NUN with ME probably remains two signs even if a syntactic or grammatical unit. AS1.52 has the two signs close together, while AS1.7 has much space between them.

IGI ligatured with "3" -- [Ur III Sale 19.4]

Mere Text Groupings which (?) are none of the above:

The typography of the suffix -am3 ("A.AN") at the end of the line in Ur III Sale 19.1 might reflect its grammatical unit status, but onctrast 37.2 and Piotr Steinkeller's long discussion of such forms in the volume.

PA.TE.SI seems in several cases to be grouped as PA.TE then SI, with greater space between the TE and SI than between the PA and TE. Chance? Or something for which there is an explanation? -- [Girsu 8.xi.4, 9.iii.4, 13.iv.3]