Please choose  from the categories below
Cuneiform Signs

Analysis and reports to support an international standard for computer encoding of the Cuneiform writing system

Research on the development of Cuneiform signs

 

Spacing in the Gudea Statues and in the Codex Hammurabi

This note reports that the treatment as single sign vs. as "lexical item" (a sequence of signs) was *already* taken full account of decades ago in Labat and other sign lists of the scholarly tradition. There is no need to revisit that issue, except to refine it. Where evidence is available, in lines with enough space to see distinctions, both in the Gudea Statue texts and in the Codex Hammurabi, the evidence correlates very strongly with the choice by Labat and Borger to classify as single sign vs. as a sequence of signs (implied for those glyph combinations which are not classed as single signs). This is a part of the confirmation that the traditional sign catalogs knew what they were doing, and for the most part *correctly* differentiated between single signs and sequences of signs.

Best wishes,
Lloyd Anderson
Ecological Linguistics

***

Citations from the Gudea Statues are in the form A to N followed by column and register ("line") in "Arabic" numbers. Citations from the Codex Hammurabi are in the form of Roman numerals (column) followed by register ("line") in "Arabic" numbers.

Already, there is a very high correlation between spacing behavior and the treatment of the standard traditional catalogs as a single sign. The treatment of the traditional catalogs probably took this into account, consciously or unconsciously. Those items marked by Labat as lexical items rather than as single signs all have loose spacing, and in one case where the opportunity presented itself, do split across end of line.

Few exceptions: Of the 16 items first studied, there are only three real exceptions. Two place names GIRSU and ABZU are written with glyphs touching yet considered lexical items rather than single signs by Labat. In reverse, one item SA4 is catalogued as a sign, yet has close spacing only once out of eight occurrences, has loose spacing in six, and is split across end of line ("indent") once.

From the Lucien-Jean Bord and Remo Mugnaioni publication *Les Statues Épigraphiques d Gudea*, Musée du Louvre I have surveyed the spacing of some characters which might be taken to be single or to be sequences. The publication does appear to consciously represent the original typography as faithfully as the authors could. Then the *Codex Hammurabi* was surveyed, and the results confirm the earlier ones.

The way I scanned was to look for unusually close spacings of parts which do exist otherwise as independent characters. This is by no means complete (not in such a short time), and I will have missed some spaced forms precisely because they did not stick out as having closer spacing of components relative to the wider spacing around them. If a line was too crowded with signs to distinguish closer vs. looser spacing with some reliability, then I disregarded it. Despite the limitations, this small survey still gives some indication. It also shows a very small amount of fluctuation. Most items examined are consistently treated one way or the other, are not a mixture.

First, both justification and "stretching" of signs can occur. The best example of stretching I found in the Gudea statues is on Statue B at B.7.49 (column 7 line 49), the sign ALAN. [Heimpel adds: <<Note that two registers earlier the same sign is unstretched even though spatial conditions are identical.>>] A good example from the Codex Hammurabi is the sign EL at viii.57 (less stretched) vs. at viii.60 (both more stretched *and* with more white space around it). The etymological parts of this sign (SAL and SI or its lookalike) are not separated during the stretching.

Normally in the Gudea statues, justification is used if a box contains only one line, while if it contains two lines, the second one is normally aligned right. The signs within a single line are normally spaced out to fill it, as for example even SIPA at F.4.7 (see below for other writings of SIPA) [Heimpel emphasizes that the context is identical in one of these other instances.]

**********************************************

1.
Items for which there is no spacing evidence for unity, and which Labat treats as a *lexical item* (text unit but not a sign).

E-11 = "SU.DU" most common in sequence -TA-E11
     A.3.1 B.6.40 C.3.15 D.4.16 F.1.8 F.3.2 H.2.6
     Split across line end at B6.40.
     Not given a separate number by Labat; under L459

E3 = UD.DU
     B.8.7 (2x) C.4.6 (2x) D.2.6 E.3.10 G.1.15
     Kept on one line at G.1.15, even though the "UD" could go on the preceding line. Possibly because it respects word-boundary.
     Not given a separate number by Labat; under L381

**********************************************

Now two items with minor fluctuation, but mostly appear visually *as if* a single sign. These *are* listed as signs in the scholarly tradition.

2. SA4
The reading SA4 for the sequence visually like MUSHEN.NA2 -- This appears as two signs in Gudea Statue E column 9 line 4 (E.9.4). The MUSHEN is at the end of the first line within the block, the NA2 on the next line (justified right to show end of block ). In most cases these are on the same line, but spaced quite loosely: B.7.18, C4.2, D.5.8, H3.6, I.5.7. N.3.6 is medium spacing.
In one case, A.4.3, they are spaced quite tightly: tighter than the spacing in the preceding or following lines, presumably by the same scribe.

Because of the loose spacing of the majority of examples, I am ambivalent about what *this* set of evidence should lead us to conclude. The sign is treated as a single sign in the standard catalogs B138, L82. The close spacing of one example may be the anomaly. Or the line break in the other case may be the anomaly, a substitution of two different characters for the single character whose parts look like those two. That last is a resoning I do *not* like, it adds a layer of complexity between character sequence and surface representation which is precisely what should not ever be done unless there are good and necessary reasons.

I am just leaving that option open in case it turns out to be verifiable in some way. I would normally take the line break as determinative, and then argue in this case the catalogs made a mistake? What is the normal treatment elsewhere? Also fluctuating, and allowing line breaks? More common in some eras or styles, rare in others? We don't know until we tabulate actual data. Or we can take the word of the catalogs, if they otherwise correspond.

**********************************************

3. SIPA and SHAB

SIPA occurs with full loose justification on Statue F: F.4.7, which is precisely what one would expect if this is a sequence of characters. However, it occurs on F.4.3 with rather close spacing, identical to that of the components of MASH2 in line F.4.6, which is known as a single character, and presumably by the same scribe. In F.4.6, the components of SIPA are both at the left end, then a large amount (!!!) of white space, then the only other character at the right end. Is SIPA a text unit or a single sign?

It occurs with tight spacing and touching parts (looks like a single character) in the Gudea statues at B.2.8, D.1.11.

In the Codex Hammurabi, SIPA occurs with parts touching, or no significant spacing between parts (spacing less than what are clearly spaces between signs) at Prol.iv.45, xv.46,56,75 and in the main text at xxii.61,78,82,84 and xxiv.43. In several of these there is signficant space around it which separates it from neighboring signs. It is divided by space just barely greater than inter-sign space only once, in the main text at ix.60. This last is a crowded line of the kind not normally providing clear evidence.

One way of treating this is to say that the character SIPA is present in most occurrences, but the sign PA followed by the sign LU (or SI6) in that order are present in the one occurrence where the parts are greatly separated (and in the CH main text at ix.60, the components might be loosely written, or it might be another writing of separate signs. This could manifest precisely that ambiguity of interpretation by scribes mentioned by Tinney across time periods, in which they may reinterpret single signs as sequences of familiar components and write those signs instead.

Are the divided writings anomalies? What evidence do other texts bring, in lines where there is sufficient spacing to distinguish treatments?

Another way of interpreting this is that SIPA is a lexical item but not a single sign. That is to say, one might argue that the keeping together of the parts is keeping a "text unit" together, not keeping together the components of a single character. I do not want to bias the consideration in either direction.

Borger (B468) and Labat (L295m) treat this as a single sign, not a sequence of signs.


SHAB = "PA.IB"
     E.5.10 F.5.18
     Borger (B466) and Labat (L295k) treat this as a single sign, not a sequence of signs.

**********************************************

4. Items which ALWAYS occur as a single graph in the Gudea statues, with no space between them, and which are highly frequent ("passim"), but which may nevertheless be sign sequences not single signs (so treated by Labat etc.). (I doubt that I would have missed spaced occurrences of the parts at least of these two, since they were so salient to me. But I might have, I of course grant. A search of an electronic text can help.)

GIRSU (with components GIR2 and SU) passim
     This is *not* listed in the sign catalogs. B.1.1 etc. etc.
This is a place name. Perhaps for place names in texts *at least from the polities connected with those places* we should treat the writing as a single graph as not compelling evidence for status as a single sign (exceptional, so ligatured?). But if texts from other polities also write the components always together, then perhaps it is good default evidence. In other words, if that is the normal writing, not a special case. There is a sign GA2 x (GIR2.SU) (N2664 proposal U+120C3, Borger#391), so perhaps they were felt as a unit. Not conclusive yet, because containers do sometimes have combinations of independent signs infixed. But in this case, we should consider the possibility that the standard catalogs made a mistake by *not* listing this as a single sign.

Labat treats this as a lexical item, listed under GIR2 L010, not as a single sign.

Another relevant consideration is that NIN occurs in lots of contexts. When written in the sequence NIN GIRSU, the NIN is kept together and the GIRSU is kept together, but there is much space between each of these two complexes. So it is not simply that NINGIRSU is a single lexical item, a text element, even though NINGIRSU will be listed in dictionaries, as will also NIN and GIRSU.
It is something more than that about the characters or text elements or readings.

ABZU = "SU.AB" = AB X SU (AB with SU above)
    Fused in B.9.2.
This is another place name, a mythical one. Only one example, so I tend not to draw conclusions yet. I believe fused only at the Gudea stage, if I remember right, though at Uruk stage it is "grouped", and Uruk "grouping" normally does correspond to later fusing, infixation, other indications of single-sign status. Labat treats this as a lexical item, listed under AB L006, not as a single sign. [Heimpel adds: <<We call that thing anagraphic and trace it to the Early Dynastic time when the sign order in registers was regulated by considerations of space, not of meaning. Some groupings of that time survived, among them abzu.>>

**********************************************

5. Other items which are always kept together, and which are almost certainly single signs. First ones which are treated as single signs in N2664R.

U3 ("IGI.DIB" components look like IGI followed by DIB / LU)
     B.3.13 B.5.56 B.6.3 B.7.25 B.7.50 B.7.51 (2x) B.8.22
     B.8.34 B.8.37 C.2.20 F.2.5 F.2.12 I.4.1 N.shoulder.5
     Codex Hammurabi 216+ examples passim, here a substantial list:
     Prologue i.5,15,17,23,35,46,56;   ii.15,47;   iii.26,40,42;   iv.18,31,62
     v.9,21;   vi.1,23,33,37,47,49,51,53,59;   vii.10,21,24,33,51;   viii.41;   ix.38,40,47,67
     x.3,8,14,19,26,31,37,47,52,53,60,66;   xi.14,36,40,43,49,52,63,67;
     xii.5,7,(11?),13,17,19,23,24,27,29,31,35,37,40,42,46,50,52,54,57,58,60;
     xiii.11,29,32,39,43,49,50,55;   xiv.5,7,15,22,25,31,33,38,41,46,49,54,62
     xv.3,24;   xvi.20,45,53;
     Main text i.35,61;   ii.7,20,40,54;   iii.4,19,28,41,51,57,63,68;   iv.10,13,34,45,55,70,71;
     v.27,33,52;   vi.76,84,85;   vii.69;   ix.13,44;   x.12,75;   xi.36;   xii.40,53,60,72,80;
     xiii.3,49,58,69,70,78,85,91,92;   xiv.12,43,48,62,79,(84?);   xv.3,10,46,62;
     xvi.46,52,93;   xvii.1,4,11,16,31,36,56,62;   xviii.13,60,79;   xix.3,19,20,29,87;
     xx.34,39,48,52,79;   xxi.9,17,32,78,91;   xxii.17,23,29,37,46,59,63,68,73,77,101
     xxiii.80,83,93;   xxiv.7,24,31,39,51,64,69,86;   xxv.34,37,92;   xxvi.43,57,76,88;
     xxvii.3,16,30,55,66,74;   xxviii.2,72,81
Borger (B731) and Labat L455 treat this as a single sign not a sequence of signs.
The revision N2664R now treats this as a single sign.

*

UL = DU7 or "U.GUD" or SHU4 x GUD
     B.5.14 B.5.48 B.7.6 B.8.27 D.2.6 E.3.10 F.1.8 H.2.3
     Codex Hammurabi 83 examples:
     Prologue   ii.42;   iv.49;   vi.27,30;   vii.6,10?,24,25,34,40,54,56,59,62,64
     ix.55;   x.68;   xii.9,28,30;   xiii.65,72;   xiv.12,17;   xv.6
     Main text   i.54;   iii.37,73,77;   v.41;   vi.36,73.76;   vii.16,61;   viii.25,27,42,55,78
     ix.25,29,38,41,43,50,57;   x.42,77;   xi.4,21;   xii.2,24,43,44,62,70,77;
     xiii.26,68;   xiv.30,52;   xv.15,17,42;   xvi.14,38,53,59,86,92,94;   xvii.6,7;
     xxi.51;   xxiii.98;   xxiv.15,16,17,83;   xxv.102;   xxvi.91
Borger (B698) and Labat (L441) treat this as a single sign, not a sequence of signs.
The revision N2664R now treats this as a single sign.

*

AR = "IGI.RI"
     Codex Hammurabi 111 examples passim, here a substantial list:
     Prologue   ii.64;   iii.14,21;   v.12,60;   vi.12,27;   viii.1,22,64,65,67,72;  ix.34;
     x.18,27,58;   xi.24;   xii.21,28,59;  xiii.16,31,40,41;   xiv.12,55;   xv.68,76;   xvi.47,52
     Main text   i.23,41;   ii.27,30;   iii.3,66;   iv.18,19;   v.76;   vi.36,47,53,54,60,73,83;
     vii.63,65;   viii.1,42,48;  ix.17,20,76;   x.19,25,59,74;   xi.32,39,58,81;
     xii.16,18,25,32,48,64,71.76;   xiii.4,20,22,44,45,49,52,67,83;   xiv.35,42,60,70,76
     xv.29,35,38,49,63,66,74,84,94;   xvi.10,22,49,64,74,81;   xx.75;   xxii.66;
     xxiii.86;   xxv.1,6,36,59;   xxvii.32,47;   xxviii.26,90.
Borger (B726) and Labat (L451) treat this as a single sign, not a sequence of signs.
The revision N2664R now treats this as a single sign.
     Examples at xv.84 and xvi.10 have lots of space, so better evidence.

*

MUL = "AN OVER AN AN"
     C.2.23 E.3.4
Borger (B247) and Labat (L129a) treat this as a single sign, not a sequence.
The revision N2664R now treats this as a single sign.

**********************************************

6. Other items which are always kept together, and which are almost certainly single signs. In this group, ones which are *not* treated as single signs in N2664R.

PA3 = "IGI.RU" = IGI x RU
Almost always very closely together, what one might call highly kerned or ligatured if it were not that this does not normally alternate with a sequence of IGI and RU spaced apart, and that breaking over ends of lines is (? rare or more likely ?) nonexistent in normal scribal practice. Better to call this and many other signs "fused", as is Danish <ae> which was recognized after much struggle by Danes as being a letter not a ligature (even if it arose historically via ligturization).
     B.2.8 B.3.14 D.1.11 E.1.20 Cylinder B.xiii.6 end
Borger (B725) and Labat (L450) treat this as a single sign, not a sequence of signs.
The revision N2664R does *not* treat this as a single sign.

B.2.8 and D1.11 both contain the kind of situation Wolfgang Heimpel was discussing, in which there would have been plenty of room for IGI at the end of the first line of the box, but the entire PA3 is on the second line. Since there would also have been room for the entire PA3 on the first line, perhaps there is a syntactic reason for the line division, namely keeping several signs together which spell the same *word* (and not leaving the second line with too few signs in it? several other boxes on these statues do not keep words together even when they could). The following two occurrences have space between the parts. Despite this, PA3 is *not* split across line boundary, even though that would produce a more balanced line density, and even though it would make no difference to word boundary. It therefore seems that a variant glyph is the best way to understand these, a glyph which because it is unitary cannot be split across the line break. Certainly better than forcing all of the ordinary cases to be more complex to handle.
     I.4.5 I.9.7

*

AGRIG = "IGI.UM" varying by one wedge with "IGI.DUB"
     D.1.13
Borger (B727) and Labat (452) treat this as a single sign, not a sequence of signs.
The revision N2664R does *not* treat this as a single sign.

*

NIN (with components SAL and NAM2) passim
     B.1.1 etc. etc. CH prologue at iv.37; main text xv.20, xxvi.81,85, xxviii.41,50 (parts touching in all cases).
Borger (B887) and Labat (L556) treat this as a single sign, not as a sequence.
The revision N2664R does *not* treat this as a single sign.

*

NA4
     B.6.7 B.6.60 G.3.2 H.2.6
Borger (B385-6) and Labat (L229) treat this as a single sign, not as a sequence.
The revision N2664R does *not* treat this as a single sign.

*

"SAL.KUR" (used for male and female together, thus people etc.)
     Codex Hammurabi examples (probably many more not collected here).
     Main text   xii.40,53,68,72,75;   xviii.45,51;   xxii.82;   xxiii.59,68,75,80,81,83,96
Borger (B698) and Labat (L441) treat this as a single sign, not a sequence of signs.
The revision N2664R does *not* treat this as a single sign.

**********************************************

7. Other items which are always kept together, and which are almost certainly single signs. In this group, ones which are *not* treated as single signs in N2664R.

There is every reason to assume that all signs with "U" = SHU4 or SHU2 as first component are single indivisible signs as reflected in the correlated spacing behavior and standard sign catalogs. These act as "container" components just as much as do components which completely surround an infixed component. Just as does the closely corresponding "roof" radical among Han CJK characters.

GUL = "U.URUDU" = SHU4 x URUDU or SHU4 x (AB x U)
     B.6.19 B.7.57 & Cylinder A.viii.26
     Codex Hammurabi example ?? Main text   xxvii.19
In B.7.57, there would have been plenty of room for SHU4 on the first line, but not for the entire GUL. On Cylinder A.viii.26
one of the two GUL could either have been put on the next line, or just the second component of it could have been, leaving SHU4
at the end of the preceding line. But these were not divided. Again examples of the kind Heimpel notes. The word boundary is not respected no matter which way these lines divide, so that is probably not a factor.
Borger (B682) and Labat (L429) treat this as a single sign, not as a sequence.
The revision N2664R does *not* treat this as a single sign.

*

SHAGAN = "U.GAN" or = "SHU4.GAN" (etymologically a "covered GAN jar-stand?).
     Codex Hammurabi examples (are these correctly identified? check published transliteration)
     Main text   i.13,16,29,33,38,46,55,63,64,69,70;   ii.3,6,9,13 .
Borger (B684) and Labat (L428) treat this as a single sign, not a sequence of signs.
The revision N2664R does *not* treat this as a single sign.

*

SHUDUN = "SHU2.DUL4" = SHU2 x UR $e$$ig
     F.3.12
Borger (B876) and Labat (L549) treats this as a single sign, not as a sequence.
The revision N2664R does *not* treat this as a single sign.

*

Some instances on the Gudea Cylinders which a professional can look at,
but which without parallel analyzed transliteration I cannot handle.
Cylinder A.xi.24 (end of middle line) [Heimpel transliterates <<kalam-e zi-sha3-gal2 u3-ma-sum>>]
Cylinder A.xii.11 (end of middle line) [Heimpel transliterates <<giskim-mu ha-mu-u3-zu>>
Cylinder A.xiii.23 (MUL ? as if NAB.AN) [Heimpel transliterates <<uri3-she3 bi2-mul "stars", i.e. "twinkles.">>
Cylinder A.xxix.5 (is that GUL separated into "U" and the remainder?) [Heimpel: <<No. It is: SHIM na4 e2-a shu4-ga-bi>>]

*

Some items which are universally treated as single signs (and already in N2664)

EL = SIKIL = "SAL.SI"
     C.3.4 F.2.18
     Codex Hammurabi examples
     Prologue iv.22,(26?);   vii.(26?),40,54,59,62;    viii.47,57,60;   ix.58,61;   xii.4;
     xiii.9,15,30,55,64;   xiv.35,52;   xv.49,51,54,57,63,68;   xvi.2,4,19,23,42,47,51,67;
     Main text iii.7;   iv.2,10,17,19,22;   v.1,3,50;   ix.29,41,50;   x.68;   xiv.(5),17,55?
     xviii.71;   xix.6,14,24,34,37,71,73,77,80;   xx.25,69.72,77;   xxi.12,20,26;   xxii.14,81;
     xxiii.80,81,93,102;   xxv.44,57;   xxvi.8,30;   xxvii.41,64;   xxviii.67
Borger (B899) and Labat (L564) treat this as a single sign, despite apparent etymology.

ALAN (12034)
     A.3.2 B.7.12 B.7.22 B.7.24 B.7.47 B.7.49 stretched!
     B.7.58 B.7.60 C.3.6 D.4.17 E.8.19 E9.6 G.3.3
     H.2.7 I.5.1 N.3.2
     ?Cylinder A.xxiii.24,29, xxiv.3,
(B573, L358)

[Heimpel adds at the end, revising to a much more limited statement:

<<That some diri compounds do not get separated at line ends may be a matter of restricted convention. I am pretty sure I have seen the ru of the pa3 sign indented. Now, that I looked at what you observed, I must correct myself half-ways: I have never seen, and do not expect, separation of diri compounds across registers and unindented lines. >>]

***

[On the next item, which is a complex case and from which I do not draw any immediate conclusions, Heimpel adds: <<The writing is syllabic and ideographic at the same time: gish PI tug2 (not nam2) = PI which goes back to the drawing of an ear and gish tug2, its pronunciation>> He later adds <<the different spacing of gish pi tug2 may not be meaningful>>]

A possible distinction of reading depending on spacing. (This may even be a typo in the book, leaving off a "g" in the
transliteration of B.9.14, though there are both readings in the index of Labat, and the contexts before and after these signs are different. So they may actually have different meanings, and show that this sequence X Y is not equivalent to the fused form <XY> (a historical but not a synchronic digraph, just like Danish <ae>?). At any rate it shows the combination which would by some be noted "PI.NAM" in both cases, and typographically once totether and once separated, relevant to the topic of this note.)

I am very suspicious of this kind of evidence until experts confirm it, and as I have written several times, it is hard to find such contrasts even when a symbol system has them in measurable frequency, because we do not mentally register them... But if valid, it is quite conclusive. Our group has not been looking for it yet. To be valid, it would have to be confirmed in other texts too. Otherwise, this is merely an example of fluctuation.

F.2.9 GESHTUG2 =
    Appears as GISH PI NAM2 DAGAL KAM
B.9.14 GESHTU2 = "PI.NAM2" in close juncture
     Appears as GISH (PI.NAM2) HE2-
     (next line in same register) EM SHI GUB
Labat and Borger do not list this as a single sign, so there is an additional presumption against this one.