The ^tug₂guz-za/^tug₂guz.za at Ḫattuša and Beyond

1. Introduction

The present philological contribution arises from the author’s recently concluded work on the Hittite economic texts. Among the problems encountered in working with these texts is the large vocabulary of international luxury goods that seldom appear elsewhere in the Hittite corpus. One of these items is the ^tug₂guz.za,^[1]which has been variously translated in Hittitological literature as “Zottendecke, [Zotten]stoff,”^[2] “robes,”^[3] “guz.za-Stoffe,”^[4] or not translated at all.^[5] Endeavoring to choose between these translations occasioned a recourse to the neighboring fields of Sumerology and Assyriology, where it was noted that an even wider range of interpretations prevails. Thus, since there is no recent comprehensive discussion of the ^tug₂guz-za/^tug₂guz.za, and indeed, since the explanations found in the main reference works are partially incompatible, the present contribution aims at providing a history of research, a critical review of secondary literature on the subject, a discussion of the problems of interpretation, and suggestions for future paths of investigation. It should be emphasized, however, that a full understanding of this important article of realia must await an interdisciplinary investigation of the cloth across the two-and-a-half millennia history of the lexeme in cuneiform corpora, and no doubt further contributions of research on ancient textiles in the associated fields of art history and archaeology.

The ^tug₂guz-za/^tug₂guz.za is a type of cloth or garment attested in various spellings from the Ur III to the Late Neo-Babylonian period. In the 3^rd millennium, the ^tug₂guz-za is attested as a thick, heavy cloth of medium to high quality that could apparently be worn on the body as a wrap in addition to being attested as a blanket. In the 2^nd to 1^st millennium, the ^tug₂guz.za appears as an internationally-attested luxury cloth used as a blanket, rug, or fine upholstery for seats and chariots. As with most ancient textiles, the bulk of attestations of the ^tug₂guz-za/^tug₂guz.za is found in lists and inventories, and thus little is known of the cloth’s outward appearance. Even a unique text providing a complete description of the amounts of wool and the labor necessary for crafting a ^tug₂guz-za (see the discussion of ITT 5, 9996 iii 4–rev. i 4 in section 4 below) has only afforded reconstruction of the cloth’s underlying weave. Interpretations of the visual characteristics have therefore hinged on the etymology of the Sumerian word guz-za and on glosses of the cloth in lexical lists. On the one hand, the lexical lists show that one meaning of Sumerian guz(-za) was ‘bristly, wire-haired’ (Akk. (ḫ)apparrû), leading to the original argument that the ^tug₂guz-za/^tug₂guz.za was a type of shaggy cloth traditionally designated as the “Zottenrock” in Mesopotamian iconography. On the other hand, neither direct glosses in the lexical lists nor descriptions in practical texts hint at any shagginess or bristliness for the ^tug₂guz-za/^tug₂guz.za. Instead, the lexical lists seem to implicitly connect the cloth with a different Sumerian guz-za meaning ‘bright red’.

2. History of Research

Research on the ^tug₂guz-za/^tug₂guz.za began with Landsberger (1934: 102), who, in a discussion of the pig variety šaḫ-bar-lum = ap-pa-ru-u₂, argued for the existence of a Sumerian lexeme lum= guz‘shaggy’ based on the lexical entries ^gu-uzguz= ap-par-ru-u ‘bristly, having wiry-hair’ (AO 7661 obv. i 36) and lu₂ siki-guz-za = ḫa-ap-pa-ar-ru-u₂ ‘a man with bristly/wiry (body)hair’ (UM 5.147 obv. 5), adducing as additional evidence the garment ^tug₂guz-za, which he translated Zottenrock – ‘shag-skirt’.^[6]Jacobsen (1939: 27) offered a different etymology: he transliterated ^tug₂guz-za as ^tug₂ḫus-sa₃ andconnected the garment with the color red based on the passage tug₂lum dub-še₃ (CTNMC 44 rev. 2), which he read as “for dyeing (dub) the cloth red (lum).”^[7] Comparing the interpretations of Landsberger and Jacobsen, Oppenheim (1948: 65) sided with Landsberger regarding the semantics of ^tug₂guz-za (“The decisive proof that this term [scil.^tug₂guz-za] denotes a special fabric of flocky and shaggy texture is offered by the equation in UM v 147:5 lú síg(!)-guz-za = ḫa-ap-pa-ar-ru-ú ... ‘man with a guz-fleece’ = ‘shaggy’ ... .”), while embracing the phonology suggested by Jacobsen (“guz = ḫunzu, ḫuzzu ‘shaggy’” and suggesting (loc. cit.) that guz is “probably to be read ḫuz, even ḫuz(z)a as Akkadian loanword in Sumerian ... .”). Although Oppenheim (1948: 66) conceded that “[i]t is not unlikely, as Jacobsen ... has already suggested, that the term guz(z)a denoted at the same time a certain treatment of the wool and a specific color,” he did not mention the color red or take a position on what the specific color was. Oppenheim concluded with the observation that the ^tug₂guz-za was frequently listed among ^tug₂ni₃-lam₂ ‘festival clothes’ and was in reality an expensive cloth.

Oppenheim’s explanation was not adopted by the main Assyriological dictionaries: both CAD and AHw. follow the older reading ^tug₂lum-za – with ^tug₂sig₄(= lum-šeššig).za as an accepted post-OB variant – rather than ^tug₂guz-za/^tug₂guz.za (see the respective iʾlu(m) entries: CAD I–J 90, AHw. 373b). While neither dictionary engaged with the arguments directly, by quoting the same lexical entries used by Landsberger and Jacobsen elsewhere,^[8] both dictionaries appear to have implicitly rejected the arguments of Landsberger, Jacobsen, and Oppenheim. This is all the more significant as CAD was co-edited by Oppenheim at that time. Similarly, both dictionaries declined to mention Jacobsen’s suggestion for the reading tug₂ guz-za = tug₂ ḫus-sa₃ under the ḫuššû(m) entries (CAD Ḫ 261, AHw. 361b). Finally, CAD Ḫ 266, s.v. ḫuzzû and AHw. 356a, s.v. ḫunzû, ḫunzuʾu rejected the earlier translation as “shaggy” and defined ḫunzu, ḫuzzu instead as “to be lame, to limp”/“etwa ‘lahm’?,” while retaining the equation with gu-uzguz in the lexical sections of their respective entries.^[9] With the dictionaries leaving the definition and reading of ^tug₂guz-za/^tug₂guz.za an open question as of the mid-nineteen-sixties, further study of the cloth devolved to philological commentaries of various texts in which it is encountered.

The post-dictionary commentaries and studies on the ^tug₂guz-za/^tug₂guz.za can be divided into two groups. The first group follows CAD and AHw. in that they do not commit to an interpretation of shagginess for the ^tug₂guz-za/^tug₂guz.za, though with the difference that all acknowledge a reading guz for lum. Although the ^tug₂guz-za was nowhere explicitly treated in the most obvious reference work by Waetzoldt (1972) on the textile industry of the Ur III period, in his subsequent RlA entry on “Kleidung. A. Philologisch” Waetzoldt (1980–1983: 21) was cautious about the Zottenstoff equation: “Sollte ^túgguz-za der Zottenstoff sein, so müßte nach der aB Zeit dafür noch eine andere Bezeichnung existiert haben, da Gottheiten nach jüngeren Darstellungen ihn häufig tragen.” In the realm of Hittitology, Güterbock (1973) cited the original work of Landsberger (1934) for the reading of lum = gu-uz, but referred readers for the meaning to CAD I–J 90, s.v. iʾlu A “(a garment),” which makes no mention of Zottenstoff. The two editions of Hittite administrative texts pertaining to the palace-temple economy at Ḫattuša, which – as might be expected – contain the bulk of attestations to ^tug₂guz.za in the Hittite corpus, also belong to the first group: Košak (1982: 54) interpreted ^tug₂guz.za as “robes,” citing Güterbock (1973) and equating the term with Akkadian iʾlu and illūku,^[10] whereas Siegelová (1986: 37) acknowledged only illūku, with the more cautious translation “guz.za-Stoffe.” Finally, there is the contribution of Quillien (2013) on the 1^st millennium ^tug₂guz.guz/gu-uz-gu-uz, who, despite the phonological and ideographic similarities (^tug₂guz.guz being apparently a reduplicated form of the 3^rd and 2^nd millennium ^tug₂guz-za/^tug₂guz.za) and despite the shared usage as a covering for chariots and expensive furniture,^[11] did not propose any connection between the ^tug₂guz-za/^tug₂guz.za and ^tug₂guz.guz cloths.^[12]

The second group of commentaries are those that accept, at a minimum, a shaggy, flocky, or bristly appearance for the ^tug₂guz-za/^tug₂guz.za cloth, and at a maximum, identification with the Zottenstoff of Mesopotamian iconography. All take Oppenheim as a starting point, beginning with Kümmel, who maintained the shagginess of the cloth when commenting on its appearance in a substitution ritual for the Hittite king. Based on attestations in texts from Ḫattuša, Alalaḫ, Ugarit, and El Amarna, where the ^tug₂guz.za is clearly used as a blanket and furniture covering, Kümmel (1967: 77) wrote, in contrast to Oppenheim’s translation “Zottenrock,” that the ^tug₂guz.za was“jedenfalls nach allen Belegen nur Stoffart oder Tuch, nicht ein bestimmtes Kleidungsstück, so daß ich hier nach Ideogramm und Gebrauch etwa ‘Zottendecke, -stoff’ ansetzen möchte.” Kümmel’s translation “Zottendecke” was followed by Archi (1983: iv), and Rüster/Neu (1989: 245, no. 310). In Assyriology, Dalley (1980: 73) came to the same conclusion, translating the term as a rug or blanket with wiry hair when commenting on the appearance of the cloth in an Old Babylonian dowry.^[13] Discussing an Old Babylonian text concerning textile finishes, Lackenbacher (1982: 143) took Oppenheim’s translation of “flocky and shaggy textile” as a given (“presque sûrement un tissu poilu”) and attempted to situate this attribute in a range of textile treatments described in the text.^[14] By 2010, Waetzoldt had accepted a translation “shaggy cloth,” though without further discussion or mention of Zottenstoff.^[15] Nearer to the iconography of the classic Zottenstoff, Limet (1971: 15), in discussing the appearance of the ^tug₂guz-za in Ur III economic documents found at Tellō/Ĝirsu, cited Oppenheim to support his own translation “vêtement tissé à bouclettes” – a woven garment with “curls” or “loops.” Limet (loc. cit.: fn. 4) also cited the equation in the lexical lists of the ^tug₂guz-za with the illūku garment (CAD I–J 86, s.v. illūku 2. ‘a sumptuous garment’), and considered that it was the characteristic curls/loops that made the garment compatible with the idea of sumptuousness (loc. cit. fn. 5). Finally, the most recent, novel contribution to the “shaggy, flocky, bristly” interpretation for the ^tug₂guz-za/^tug₂guz.za is found in Durand (2009: 35–37), who argued that the 3^rd millennium ^tug₂guz-za was the Zottenrock garment, supposedly continued at Mari by the gizzu-cloth,^[16] whereas the 2^nd millennium cloth “^tug₂sig₄.za,” taken by most scholars as an equivalent of ^tug₂guz.za,^[17] was an unrelated cloth marked by a thick weft and used for fine upholstery.^[18]

If a trend emerges from the above research history, it is that the original identification of the ^tug₂guz-za as a “flocky and shaggy textile,” despite its implicit rejection in the dictionaries, including the volumes that Oppenheim co-edited, has either been repeated uncritically or has provoked a certain level of circular reasoning: shagginess is invoked to explain certain attributes of the cloth, such as the fact that it was a fulled or carded garment, a feature which many other Neo-Sumerian textiles shared (see section 4 below), and these attributes, in turn, are used to support the definition of shagginess. In the end, although it is not impossible that the shaggy or flocky interpretation is correct, or even that the ^tug₂guz-za was the Zottenstoff as depicted in visual imagery, neither interpretation has been proven.

3. Orthography: lum (guz) versus lum-šeššig (sig₄)

Before further assessing the evidence for the ^tug₂guz-za/^tug₂gu-za/^tug₂guz.za/^tug₂guz(“sig₄”).za in the text corpora and lexical lists, it is important to confirm what has been suspected, namely, that they are spellings of the same cloth. Returning to the problem of the disappearance of the writing ^tug₂guz-za/^tug₂guz.za in post-3^rd millennium texts, the accepted solution has been to interpret the ^tug₂sig₄.za, attested from the OB period onwards, as its successor. It has been proposed that the guz (i.e., lum)and sig₄ (i.e., lum-šeššig) signs are interchangeable in many post-OB cuneiform scripts, going back at least to Fossey (1926: 1101, under “Confusions de signes” nos. 33583, 33591 [sig₄]: no. 33484 [lum]). This observation was cited in discussions of ^tug₂guz-za/^tug₂guz.za by Landsberger (1934: 102 fn. 2), Ehelolf (apud von Brandstein 1934: v, no. 13), Kümmel (1967: 76), Güterbock (1973: 85), and Rüster/Neu (1989: 246, no. 311). Sign confusion, and perhaps peripheral/chronological variation, was also part of the approach taken by the CAD and AHw. (again for the old reading ^tug₂lum-za).^[19]

In the Hittite context, at least, the lum and lum-šeššig are not interchangeable. Rather, the distribution of the signs seems to be graphemically conditioned: phonetic writings and Akkadograms use lum, while Sumerograms use lum-šeššig. Searching the Hittite Palace-Temple Administrative, Festival, and Cult Inventory Corpora reveals that all twenty-two instances of Akkadograms and Akkadographic-complements that should orthographically use lum are written, without exception, with lum.^[20] Conversely, all Sumerograms that should use lum (guz, ḫum) actually use sig₄,^[21] as do the Sumerograms that should use sig₄.^[22] Significantly, of the twenty-five attestations of ^tug₂guz.za, twenty-four are written ^tug₂sig₄.za and only one – appearing in a broken context and possibly being a different word entirely – is written with the lum/guz sign.^[23] In fact, the only word in Hittite texts showing significant lum/lum-šeššig interchangeability is za.ḫum, Akk. šāḫu (also sāḫu, šīḫu, sīḫu), a kind of drinking vessel. It has been held at least since Ehelolf (apud von Brandenstein 1934: v, no. 13) that the forms za.ḫum(=.lum) and za.sig₄ interchange seemingly at random, occurring sometimes in the same texts: e.g. KUB 27.13 obv. i 4, 12, 14, 24, 27 (za.sig₄) versus obv. i 35 (za.ḫum); VBoT 108 obv. i 16 (za.sig₄) versus obv. i 17 (za.ḫum). In the Hittite Palace-Temple Administrative Corpus (PTAC) (edited by Košak 1982 and Siegelová 1986), one finds an even split: three attestations with za.lum and three with za.sig₄.^[24] Since the texts of the PTAC are from the same time period (late New Hittite), chronological variation can be ruled out, though it may be noted that the texts showing za.sig₄ all come from the so-called Bildbeschreibung genre describing cult images and figurines, so that scribal or genre considerations could be at play. The Hittite Festival and Cult Inventory Corpora, in contrast, favor the spelling za.sig₄ (18×) over za.lum (3×) by a considerable margin,^[25] though the exact sign ratios and the distribution by genre or chronology requires further research. The Palace-Temple Administrative, Festival, and Cult Inventory Corpora comprise perhaps just over 50 % of the extant Hittite corpus, and contain the majority of texts where za.ḫum ‘comb’ would be expected. If no other lexemes are found contradicting the lum = Akkadographic, lum-šeššig = Sumerographic distribution in Hittite, then the interchangeability of the signs in za.ḫum ought to be explained differently: the versions of za.ḫum written with lum are actually Akkadograms, i.e., sa₃-ḫum/ḫu₅, while those with lum-šeššig are Sumerograms, i.e., za.ḫum(“sig₄”), where the quotation marks indicate the non-standard (outside of Hittite) use of the sig₄-sign for ḫum.^[26]

The origins of the Hittite pattern of graphemically conditioned lum/lum-šeššig confusion are unclear. It had long been assumed that the confusion of these signs was strictly a phenomenon of the cuneiform periphery, attested, most notably, in late second-millennium texts from El Amarna, Alalaḫ, Ugarit, and Ḫattuša. However, Dalley (1980: 73) showed that the ^tug₂sig₄.za was also present at Aššur and Babylon. Dalley put forth the hypothesis that the original confusion lay with murgu₂ = lum and murgu = sig₄, leading her to adopt the readings ^tug₂murgu₂-za for the 3^rd millennium cloth and ^tug₂murgu.za for its later designation. Borger (2010: 229–230) expanded on Dalley’s hypothesis, arguing that the sig₄/lum-šeššig had always had a reading murgu(also mur₇) as one of its original readings,^[27] whereas the signs lum and murgu₂ were graphically distinct in the 3^rd millennium,^[28] before they coalesced into one sign in the 2^nd millennium, explaining why the sig₄/lum confusion did not occur before this date. To state the reasoning of Borger more explicitly, it seems that the second-millennium confusion of lum and sig₄ originated in a phonemic confusion about words written with murgu and murgu₂, so that at some point all words written with sig₄/murgu could be written with lum/murgu₂ instead. However, the comparative rarity of a phonemic and Akkadographic usage for sig₄/murgu (Borger 2010: 230, no. 906 rejects a value mur₇) meant that the reverse was not true: very few words written with lum could also be written with sig₄, since lum very rarely, and then only in Sumerograms, had the value murgu₂. Thus, the confusion sig₄ = murgu ≈ murgu₂ = (post-3^rd millennium) lum did not spread among Akkadian and Akkadographic writings.

Whether the graphemic distribution of lum for phonemic and Akkadographic writing and lum-šeššig in Sumerograms discussed above holds elsewhere in the Hittite corpus, and possibly in other peripheral corpora of the 2^nd millennium, remains to be seen. Confirmation will require inspecting the cuneiform of each instance, since modern transliteration conventions have obscured the distinction between the signs.^[29] One avenue for future research would be to test the hypothesis of Dalley and Borger by examining the lum/sig₄ confusion by lexeme. If the confusion indeed originated from the homophony of the murgu and murgu₂-signs, then it stands to reason that the lexemes with murgu and murgu₂ should show the earliest and greatest incidence of confusion.

Establishing where and when the lum/sig₄ confusion originated and whether it is graphemically conditioned also outside of the Hittite corpus would require a comprehensive review of the other 2^nd millennium cuneiform corpora, which is beyond the scope of the present study. However, a preliminary observation can be made: in addition to the anecdotal evidence of the spelling ^tug₂sig₄.za in the peripheral and later on also the Middle Babylonian cuneiform traditions, the only lexical lists still using the spelling lum-za in the 2^nd millennium, namely Late OB Ḫḫ. XIX (MSL 10: 143–144, no. 23) and Nippur Ḫḫ. XIX (MSL 10: 147, nos. 60–63), are both early and southern (see Table 1).

A brief excursus must be made here regarding the phonetic realization of ^tug₂guz-za, since the proof establishing that lum and sig₄ are read “guz” in ^tug₂guz-za technically remains indirect: So far, no entry *^tug₂guz^gu-uz-za has been discovered in a lexical list. Instead, virtual confirmation comes from the series ur₅-ra = ḫubullu, where the entry ^tug₂guz-za in one manuscript is duplicated by a “^tug₂gu-za” in another manuscript, which could only occur if the original pronunciation was something like /guz(z)a/. It had been thought that the equation ^tug₂guz-za = ^tug₂gu-za was proven earlier.As discussed by Limet (1971: 15 fn. 4):

Le terme túg-guz-za figure dans les ‘précurseurs’ de la série H.-ḫ. (SLT 217: II, 2–4; cf. maintenant MSL 10, p. 147, 60–63). En revanche, dans la série elle-même, il est orthographié gu-za (restitué à sa place, ibid., p. 135, 268–273) et expliqué dans la série H.-g. par il-lu-ku, lu-bar sa-a-mu ... .

However, Limet’s claim that the ^tug₂guz-za of the Hh precursors is written ^tug₂gu-za (and thus = illūku) in the main series is not supported by the lexical lists known at the time. In MSL 10, both the ^tug₂guz-za entries of Nippur Ḫḫ. XIX (MSL 10, 147, nos. 60–63) and the ^tug₂guz_x-za (= ^tug₂sig₄-za) entries of RS Ḫḫ. XIX (MSL 10, 150–151, nos. 65, 179–187), fail to correspond in either structure or location in the series to the ^tug₂gu-za entries of Standard Ḫḫ. XIX (MSL 10, 135, no. 268–272), thus leading the editors of MSL 10 to decline an equivalence. Fortunately, Limet has since been proven correct by a Late Neo-Babylonian duplicate to Ḫg.^XIX (SpTU 3, no. 116), which was not yet available to the editors of MSL 10. As recognized by von Weiher (1988: 235) and Weiershäuser/Hrůša (2018: 222) – both without a comment on different spellings –, the [^tug₂]gu-za = illūku in Ḫg.^{D XIX} no. 414 (see VAT 10261 rev. vi 14 in Weiershäuser/Hrůša 2018: 217) and the ^tug₂gu-za in Standard Ḫḫ. XIX no. 268 are now duplicated by ^tug₂guz-⸢za⸣ in SpTU 3, no. 116, obv. i 31. This, with the new data regarding the 1^st millennium ^tug₂guz.guz garment, written ^tug₂lum.lum, and spelled phonetically ^tug₂gu-uz-gu-uz, confirms that the ^tug₂guz-za/^tug₂(guz.)guz(.za) garment was pronounced from the 3^rd to 1^st millennium something like /gu(z)za/.

4. Manufacture and Use of the ^tug₂guz-za/^tug₂guz.za/^tug₂guz.guz Between the 3^rd and 1^st Millennia

A search in the online Database of Neo-Sumerian Texts (BDTNS) reveals 2,356 instances of “guz-za.”^[30] Of these, the vast majority are the garment ^tug₂guz-za followed by a gradation of quality, e.g., sig₅, us₂, gen, 3-kam, 4-kam, etc. (see the discussion of Waetzoldt 1972: 46–49). Almost nothing on the color of the ^tug₂guz-za is preserved.^[31] The fabric was heavy and large, but not remarkably so (Waetzoldt 2010: 204–205; cf. the detailed breakdown of the data in Firth/Nosch 2012: 73–74). Rather, the outstanding feature of the ^tug₂guz-za seems to be the relatively high amount of wool allotted for weft-threads compared to warp-threads, averaging at a ratio of 4.5 parts wool for weft to 1 part for warp, when compared to the ^tug₂ni₃-lam₂ (1.5:1) and ^tug₂bar-dul₅ (1:1) (Waetzoldt 1980–83: 2; 1972: 124). Despite the greater amount of wool for weft-threads, the ^tug₂guz-za took on average less time to produce than the ^tug₂ni₃-lam₂, probably due to the use of thicker weft in the ^tug₂guz-za (Waetzoldt 1972: 139, later confirmed by Firth/Nosch 2012, Andersson Strand/Cybulska 2013). Weavers were correspondingly paid less on average for a ^tug₂guz-za than for a ^tug₂ni₃-lam₂ of similar quality (Waetzoldt 1972: 82). However, despite the lower average cost of the ^tug₂guz-za, the most expensive varieties could rival the ^tug₂ni₃-lam₂ in time to produce: a ^tug₂guz-za of the 1^st/2^nd/3^rd quality required 300/120/90 work days, respectively, versus 335/300/150 work days for a ^tug₂ni₃-lam₂ of the same ranks (Waetzoldt 1972: 140). In fact, the costliest garment in terms of work days in the Ur III period found by Waetzoldt was a ^tug₂guz-za (or at least a garment produced using the guz-za technique in some way): [1 ^tug₂... ^t]^ug₂guz-za-šar₃ (ITT 3, 6606 obv. i 3ʹ–5ʹ), which cost 1,200 work days to manufacture. In summary, the ^tug₂guz-za was classed among the most expensive garments of the Ur III period, but with a higher than normal range of qualities and cost compared to, e.g., the ^tug₂ni₃-lam₂.^[32]

The text ITT 5, 9996 iii 4–rev. i 4 (edited by Waetzoldt 1972 as T.32; new edition Waetzoldt 2010: 205) details the wool and labor necessary to weave a ^tug₂guz-za of the 4^th class. The unparalleled level of detail and complete preservation of the passage stimulated two articles reconstructing the fabric based on experimental data produced by the Danish National Research Foundation’s Centre for Textile Research in Copenhagen (see Firth/Nosch 2012: 68 fn. 8; Andersson Strand/Cybulska 2013: 116 fn. 10 for sources). The resulting reconstructions suggest that the greater weight of wool devoted to weft compared to warp in the garment was due to the thickness or coarseness of the weft, not its length. Firth/Nosch (2012: 70–71, with illustration) favored a balanced tabby weave, whereas Andersson Strand/Cybulska (2013: 118–120, with illustrations) suggested a gauze weave instead, though the authors of both articles acknowledged the possibility of the other’s conclusions. Both reconstructions result in a fabric with a high degree of openness. As acknowledged by Firth and Nosch, there is nothing in the text to suggest that the ^tug₂guz-za was shaggy, knotted, or looped. After a review of the various versions of the Zottenstoff interpretation by Dalley (1980: 73), Durand (2009: 35), and Waetzoldt (2010: 205), the authors cautiously suggested that: “if the description is correct, it might possibly imply that the level of fulling, which was required to compensate for the openness of the fabric, created a ‘fuzziness’ of the fabric” (Firth/Nosch 2012: 72). Having achieved their goal of establishing the feasibility of the amounts of wool and labor prescribed in the text under discussion, Andersson Strand/Cybulska judiciously declined to comment on the finished appearance of the cloth.

Here it should be pointed out that the question of the level of fulling and/or teaseling of the ^tug₂guz-za has attracted a certain amount of circular reasoning related to the cloth. That the ^tug₂guz-za underwent at least some level of fulling is confirmed, according to Soriga (2017: 31), by texts from Ĝirsu listing the garment as receiving the tug₂ sur-ra and tug₂ kin-di-a treatments involving oil and alkali. However, this is not a distinctive feature of the ^tug₂guz-za, since a host of other fabric types, including the ^tug₂ni₃-lam₂, underwent the same processes as a finishing step (cf. tables 9.1–4 in Firth 2013: 143–46). By contrast, the statement of Soriga (2017: 31) that: “the Old Babylonian tablet AO 7026 and a lexical text demonstrate unequivocally that the shagginess of the ^tug₂guz-za resulted from the raising of the nap of the cloth (Akk. mašāru) by the fullers with at least two different kinds of teasels” cannot be followed. Although the cited lexical lists indeed demonstrate that mašāru is a verb for carding/teaseling,^[33] the application of this treatment to the ^tug₂guz-za is again not distinctive, since it is shown in AO 7026 that the garment is only one of many types that undergo mašāru. The second type of teaseling proposed by Soriga for the ^tug₂guz-za is šartum leqûm, lit. “hair taking.” In AO 7026, the ^tug₂guz-za is one of two cloths — the other being the ^tug₂bar.dib siglaḫarītum (Lackenbacher 1982: 148) — that underwent šartum leqûm. As discussed by Lackenbacher (1982: 144):

šartum leqûm pose un problème car leqûm, ‘prendre’, a d’innombrables sens comme tous les verbes de ce type et c’est un verbe transitif. Je pense pas qu’il s’agisse d’enlever le poil, car les tissus guz.za ainsi traités sont «flocky and shaggy», mais de l’extraire, et šartum leqûm pourrait être l’équivalent du français «tirer à poils», qui signifie non pas tirer le poil mais tirer pour (obtenir) le poil ... .

Far from unequivocal, it can be argued that Lackenbacher’s interpretation of šartum leqûm (along with turrukum, laqātum pānum, and laqātum lā pānum) is based on the desire to end up with a “flocky and shaggy” cloth.^[34] Taken literally, šartum leqûm should mean to remove hairs (the kemp, or non-crimped hair, being undesirable in a finished wool product); this is the apparent interpretation of CAD Š₂ 129, s.v. šārtu 2.c, which is translated “to pick off hair.” At a minimum, neither the interpretation of Soriga (2017: 31 fn. 45) that “[s]ince šartum leqûm is one of the last operations before the seizing (Akk. puššuru) of the cloth, in this step the hair has to be further brushed and curled,” nor the further comparison of the process to the rattinatura finishing treatment applied to the Italian cloth panno casentino, in which the hairs of a cloth are rubbed and pressed until the appearance of animal fur is achieved, possess any independent evidence in AO 7026.

The sartorial usage of the ^tug₂guz-za in practice in the Ur III period remains unclear. Waetzoldt (1980–1983: 21) classified the ^tug₂guz-za among the “Stoffsorten, die meist zu Ganz-Körper-Gewändern gewickelt wurden,” while noting also that the term guz-za (without determinative) could further qualify other garments. The most common garments modified by guz-za among the attestations in the BDTNS are ^tug₂aktum guz-za (46×), ^tug₂gu₂-e₃ guz-za (27×), ^tug₂bar-dul₅ guz-za (25×), and ^tug₂ni₃-lam₂ guz-za (4×). It is interesting to note that the most frequent of these “modified guz-za”, the ^tug₂aktum, might not have been worn on the body at all, but possibly was sort of a heavy rug (Waetzoldt 1980–1983: 22: “Wegen des großen Gewichts wohl kein Kleidungsstück, vielleicht eine Art Teppich; auch für Schlafzimmer ...”). This anticipates the primary usage of the ^tug₂guz.za in the 2^nd millennium.

In the 2^nd millennium, the ^tug₂guz.za is attested almost exclusively as a piece of blanket or upholstery among the gift and palace inventories of Amarna, Ugarit, Alalaḫ, and Ḫattuša (Kümmel 1967: 77).^[35] The same usage appears in the Old Babylonian dowry text discussed by Dalley (1980: 73: “since sig₄.za is used ... for burial in this text, the meaning rug or blanket seems fairly certain”). This use as an upholstery is supported by the appearances of the ^tug₂guz(“sig₄”).za in the Hittite palace administrative texts, where the cloth is attested exclusively as a furniture covering, appearing e.g., among the objects of a bedroom suite (KBo. 18.170 obv. 7ʹ//KUB 42.43: 11ʹ and KUB 42.57: 7ʹ), as a covering for chairs (KUB 42.59 rev. 21ʹ–22ʹ; KUB 52.96 obv. 5); and on a chariot (KUB 52.96 obv. 1). There is no evidence for the use of the ^tug₂guz.za/^tug₂guz(“sig₄”).za as a personal garment in the 2^nd millennium.

In the 1^st millennium, the ^tug₂guz.guz/^tug₂gu-uz-gu-uz is attested primarily as a covering for chariots and fine furniture, though the fabric occasionally appears as a personal garment worn as a coat (Quillien 2013: 22).

5. Attributes of the ^tug₂guz-za/^tug₂guz.za: Lexical Glosses and (Local) Allographic Equivalents

Further attributes of the ^tug₂guz-za/^tug₂guz.za can be garnered, with due caution, from the lexical lists and also from the local allographic equivalents of the term in the peripheral corpora of the 2^nd millennium. Borger (2010: 455, no. 900) lists three such glosses or equivalents next to ^tug₂guz-za: the lexical glosses iʾlu and illūku (citing Limet 1971: 15–16 and “die Wörterbücher”), and the Mariote equivalent ḫa/urūru, arrūru (Durand 1985: 161; AHw. 329a, 1559b, 1544b). A fourth term, not listed by Borger (2010), is the sub-gloss tunaniba found in the Practical Vocabulary Aššur (PVA) 251: ^tug₂sig₄^tu-na-ni-baza = iʾ-lu. A fifth term, g/kizzum, was proposed by Durand (2009: 35–36) as a revised local equivalent of the ^tug₂guz.za at Mari. All five terms have their complications, which are discussed below.

6. guz-za as Textile Descriptor ‘(bright) red’ in Lexical Lists

As discussed in section 2, the initial identification of the ^tug₂guz-za as Zottenrock by Landsberger (1934) and Oppenheim (1948) was based on the adjective guz-za ‘hairy, shaggy’ applied to animals and men in lexical lists. Since then, no independent evidence has come to light confirming that the ^tug₂guz-za or the ^tug₂guz.za was in any way hairy, shaggy, or otherwise similar to the cloth in visual depictions that modern scholars call “Zottenstoff.” Instead, the only adjective “guz-za” found applied to textiles and wool in the lexical lists means ‘(bright) red’, and there is evidence, which to my knowledge has not yet been pointed out, that this adjective was perceived as identical to the ^tug₂guz-za, at least in the logic of the 2^nd millennium lexical lists.

In the Standard Ḫḫ. XIX and the Ḫḫ. XIX Forerunners one encounters a repeated sequence of five adjectives describing textiles as, for example, in the Standard Ḫḫ. XIX nos. 106, 109–112 (MSL 10, 130):

Standard Ḫḫ. XIX

Entries elsewhere in the Standard Ḫḫ. XIX show that ḫuš-a had two Akkadian readings: ḫuššû and ruššû.^[45]Commentaries to Ḫḫ. XIX reveal that the distinction between ḫuššû and ruššû, was one of brightness, as can be seen in Ḫg.^{D XIX} 414–416 // Late NB Ḫg.^XIX (SpTU 3, no. 116) obv. i 31–33:

Ḫg.^{D XIX}

and in Late NB Ḫg.^XIX (SpTU 3, no. 116):

The same stereotyped, five-descriptor sequence also appears in the Ḫḫ. XIX Forerunners as, for example, the RS Ḫḫ. XIX nos. 44–48 (MSL 10, 149–150):

RS Ḫḫ. XIX (RS 20.32)

and the Emar Ḫḫ. XIX 6ʹ–10ʹ (handcopy Emar 6/1, 241; edition Rutz 2013: 197–198):

Emar Ḫḫ. XIX (Msk. 7498l)

The position held by ḫuš-a / ruššû ‘(bright) red(dish)’ in the Standard Ḫḫ. XIX is occupied by guz(“sig₄”)-za and gu-za in the Emar and Ras Shamra Forerunners, respectively.

The exact shade of red denoted by guz-za comes from a gloss of the five-descriptor sequence in another version of the Emar Ḫḫ. XIX, Msk. 7498j, 8ʹ–12ʹ (handcopy Emar 6/1, 465; edition Emar 6/3, 134):

Emar Ḫḫ. XIX (Msk. 74190j)

The equation gu-za = ša illurri in Msk. 74190j confirms the meaning guz(“sig₄”)-za/gu-za ‘(bright) red’ in the five-descriptor sequence, by specifying that the shade of red was comparable to that of illūru, a flowering plant generally accepted as the poppy anemone or windflower,^[46] known for its scarlet bloom, which also lent its name to the cosmetic rouge illūr pāni (CAD I–J 87, s.v. illuru).

The use of guz-za as a color term was not restricted to the peripheral Ḫḫ. XIX Forerunners. The Late OB Ḫḫ. XIX (MSL 10, 143–144), for example, shows guz-za, written with clear lum (guz) and not lum-šeššig (sig₄), functioning as a free-standing color adjective. Consider the following pair in sequence:

Late OB Ḫḫ. XIX (BM 92611)

Although the full, stereotyped, five-descriptor sequence is not attested in the Emar Ḫḫ. XIX, RS Ḫḫ. XIX, and Standard Ḫḫ. XIX, if Late OB Ḫḫ. XIX nos. 22–23 and 27–28 are taken to represent shortened versions, then the guz-za in no. 23 must be one of the shades of red. Since ḫuš ‘red’ appears elsewhere in the Late OB Ḫḫ. XIX (e.g., no. 28 and no. 30: ^tug₂ni₃-lam₂ ḫuš), guz-za must be the bright red.

Breaking down the data from the lexical lists by time periods reveals the following pattern (Table 1): in the 2^nd millennium Ḫḫ. XIX Forerunners, guz-za in all its variant spellings (guz-za/guz(“sig₄”)-za/gu-za) was used for both the noun, ^tug₂guz-za, and the adjective, guz-za ‘(bright) red.’ In the 1^st millennium Standard Ḫḫ. XIX and the Late NB Ḫg.^XIX (SpTU 3, no. 116), guz-za/gu-za could only refer to the cloth, while ḫuš-a, in its reading ruššû, usurped guz-za/gu-za in the lexical lists as the adjective for bright red.

Since the adjective guz-za denoted a shade of red when applied to textiles in the Ḫḫ. XIX Forerunners, it ought to be asked whether the scribes of the lexical lists considered the noun ^tug₂guz-za to be derived from the root guz-za ‘red’, and not from guz = apparrû. Variation in the pattern of the five-descriptor sequence in the Ḫḫ. XIX Forerunners supports this. Among the eleven occurrences of the sequence “sal-la, ḫuš-a, guz-za, ni₃-mu₄, alam” for textiles and wool, there is only one significant deviation.^[51] In the ^tug₂guz-za sequence (RS Ḫḫ. XIX, nos. 184–187) the expected entry *^tug₂guz-za guz-za is deleted and not replaced. In fact, the sequence *^tug₂guz-za guz-za and its 1^st millennium reflex *^tug₂guz-za ḫuš-a (ruššû) are absent from all versions of Ḫḫ. XIX, Forerunners and Standard. In the Forerunners, the alleged ^tug₂guz-za ḫuš-[a] that is supposed to be attested in Nippur Ḫḫ. XIX (MSL 10, 147 no. 63) based on Source I (CBS 6580) does not exist: neither the handcopy (OIP 18 “Fragment”) nor photos (cdli.ucla.edu) for Source I show evidence for ḫuš-a. Instead, both confirm that the traces after the guz signs in rev. v 3ʹ and 4ʹ begin with a še-group, rather than the ḫi-group needed for a ḫuš sign.^[52] In the Standard Ḫḫ. XIX, the ^tug₂gu-za descriptor sequence (nos. 268–271) is a different set of adjectives based on social rank rather than color. Thus, the expected sequence *^tug₂gu-za ḫuš-a (ruššû) is also missing. This might be thought of as only a coincidence, except that the consistent absence of the same in the Forerunners suggests otherwise. The complete absence of *^tug₂guz-za guz-za and *^tug₂guz-za ḫuš-a (ruššû) across all versions of Ḫḫ. XIX suggests that the garment ^tug₂guz-za and the adjective ‘bright red’ (guz-za or ḫuš-a) were considered the same, since the most logical reason to avoid the combination would be that the sequence was perceived as duplicative.

There is no clear explanation for why the ^tug₂guz-zashould have an association with redness in the lexical lists of the 2^nd millennium. Although the 1^st millennium ^tug₂guz.guz could be made or decorated with red wool (Quillien 2013: 22; Gaspa 2018: 286), some texts in the 2^nd millennium certainly refer to the ^tug₂guz.za being made of other colors.^[53] This makes simple homophony reinforced by a scribal “folk” etymology the most likely possibility. Whereas the vast majority of cloths from the Ur III period were undyed, those intended for royal use were regularly dyed ḫuš-a (Waetzoldt 2010: 202). In other words, red would have be one of the only artificial colors of textile inherited from the Ur III period. Since red ^tug₂guz-zacloths would certainly have existed, and since an apparently homophonous adjective guz-za corresponding to ḫuš-a (ruššû, ša illurri) existed, the pattern of the ^tug₂guz-za in the lexical lists suggests that the scribes assumed that the cloth itself was, at least originally, a bright red garment. Whether or not this was true for the garments of the 3^rd millennium is unknown.

8. Conclusions

As stated in the introduction, the origins, original meaning, and evolution of the ^tug₂guz-za/^tug₂guz.za await further investigation in the various cuneiform corpora and in the visual imagery of the Ancient Near East. However, the above discussion allows for a few preliminary points.

1. Evidence from the Ur III period and from Mari suggests that the fabrication of the ^tug₂guz-za/^tug₂guz.za involved an abundant, thick weft that might have produced a striated or “ribbed” effect. Little else is known about the cloth’s appearance. The fabric was fulled, but there is no mention of the knots or the deep pile necessary to produce the layered ranks of loops visible in depictions of Zottenstoff.

2. Regarding function, the ^tug₂guz-za of the 3^rd millennium was first and foremost a heavy personal wrapping garment and secondarily a rug or blanket (cf. discussion of ^tug₂aktum guz-za). In the 2^nd millennium, the ^tug₂guz.za is thus far attested exclusively as a blanket or upholstery, whereas in the 1^st millennium it is again attested in the form of the ^tug₂guz.za/^tug₂guzguzu as a coat, but more often as a blanket or upholstery.

3. In the 2^nd millennium lexical lists, guz-za as an adjective clearly meant a shade of red when applied to textiles, and in the 2^nd and 1^st millennium lexical lists the noun ^tug₂guz-za had a non-trivial association with red (cf. the discussion of ^tug₂guz-za = illukū and the non-existence of the sequence *^tug₂guz-za guz-za/ḫuš-a above).

Thus, despite its association with Zottenrock/-stoff serving as the basis of discussion for over 90 years, confirmation that the ^tug₂guz-za/^tug₂guz.za was a shaggy, flocky, or bristly cloth has yet to emerge.