Palaeography of the Ancient South Arabian script. New evidence for an absolute chronology. In: Arabian archaeology and epigraphy 24 (2013), pp. 186-195

Arab. arch. epig. 2013: 186–195 (2013) Printed in Singapore. All rights reserved Palaeography of the Ancient South Arabian script. New evidence for an absolute chronology The chronology of the Ancient South Arabian civilisation in the first millennium BC has always been based on palaeography. In the past, however, palaeographical comparison of the extant inscriptions was in fact the only reliable tool for historical reconstruction of that period; absolute dates were confined to the alleged synchronism of the Sabaean ruler Karibʾil Watar in Assyrian sources in the early seventh century BC. Only recently have some Ancient South Arabian inscriptions come to light, which are dated according to an absolute (foreign) era and thus allow us, for the first time, to determine the absolute date of a particular ductus of the script in that period. Furthermore, the radiocarbon analyses of a number of inscribed wooden sticks from ancient Yemen have helped to establish a reliable chronology for this particular type of documentation as well (see Drewes et al., this volume). Based on this new evidence, the paper gives a general overview of the development of the Ancient South Arabian script from its emergence in the late second millennium BC up to the latest instances in the mid-sixth century AD. The different periods of both monumental and minuscule variants of this script are characterised by representative examples of established chronology in order to provide some reliable cornerstones for dating epigraphic material from pre-Islamic Yemen. Keywords: chronology of Ancient South Arabia, Ancient South Arabian script, palaeography, monumental script (musnad), minuscule script (zabur) Irrespective of an altogether fairly long tradition of scientific research, the penetration of central aspects of Ancient South Arabian epigraphy has been rather long in coming. Although the first inscriptions from pre-Islamic Yemen were discovered 200 years ago, their script was entirely deciphered only in the 1870s.1 That the script of these stone and rock inscriptions was only one (and probably the smaller) part of South Arabian writing culture was detected again only 100 years later by the discovery of everyday correspondence on wooden sticks. Likewise, the chronological framework of the written evidence had been disputed by generations of scholars, resulting in the establishment of two main chronologies, a ‘long’ and a ‘short’ one, and consequently in rather diverging proposals for the starting point of Ancient South Arabian civilisation 1 For early research history in this field see Stein, forthcoming. 186 Peter Stein Friedrich-Schiller-Universität Jena, Theologische Fakultät, F€urstengraben 6, 07743, Jena, Germany e-mail: peter.stein@uni-jena.de either in the early or in the mid-first millennium BC. These two chronologies, which influenced historical research over decades during the past century, were mainly based on two palaeographical ‘systems’: that of J. Pirenne (1956) and that of H. von Wissmann (mainly 1976 and 1982). Although the so-called ‘long’ chronology has meanwhile prevailed, neither the starting point of Ancient South Arabian writing culture nor the exact chronology of its early development could so far be determined. The only anchor in absolute chronology for the entire first millennium BC had been the alleged synchronism of the Sabaean Mukarrib Karibʾil Watar in the early seventh century BC (see below, phase B). All periods prior or after this particular ruler (up to the turning point of the Christian era) had only loosely been structured by means of internal chronology, almost exclusively based on pal- PALAEOGRAPHY OF THE SOUTH ARABIAN SCRIPT aeographical comparison of the extant inscriptions. The historical framework resulting from this internal chronology, as established by von Wissmann (1982) and nowadays widely accepted by the scientific community, has served as the main tool not only for dating inscriptions and the historical events mentioned therein, but also to determine the emergence of Ancient South Arabian civilisation (at least as far as its identity in written sources is concerned) about the (early) eighth–ninth century BC. In the past few years, however, a number of entirely new data have come to light, which enable us not only to redefine the starting point of Ancient South Arabian writing, but also reliably to fix the internal chronology of the first millennium by clear absolute dates for some palaeographically representative inscriptions. One part of this evidence is the correspondence on wooden sticks, a material that is essentially ready for radiocarbon analysis and has revealed particular data for the very early phase of writing. The other is formed by some recently published monumental inscriptions, which contain foreign dating formulae. These absolute dates in the Seleucid and the Nabataean eras, respectively, allow us for the first time reliably to assign the palaeographical ductus of particular inscriptions from the first millennium BC to a certain year. These spectacular new findings give reason enough to reconsider the entire chronological framework of the Ancient South Arabian script. In order to give a representative picture of the development in its entirety, the generally much better documented post-Christian periods are described rather comprehensively, exhibiting the ductus of selected inscriptions of established date, separated from each other by about two centuries. We thus get a sequence of a particular ductus of the monumental as well as the minuscule scripts, which illustrates the palaeographical development of the Ancient South Arabian script overall in fairly regular stages of 200–300 years. Proceeding from a particular inscription, the date of which is more or less certainly established, the main characteristics of each stage in the palaeographical development will be outlined. This description accompanies facsimile illustrations of the particular letter forms from the inscriptions in question (Figs 1–2). The Ancient South Arabian script: monumental and minuscule The so-called Ancient South Arabian script, consisting of twenty-nine consonant letters, was in use in the southern part of the Arabian Peninsula from the late second millen- Fig. 1. The development of the Ancient South Arabian monumental script. nium BC up to the late sixth century AD, just before the arrival of Islam. Although a close connection with the Northwest Semitic scripts in Ugarit and Palestine is beyond doubt, the actual genesis of the South Arabian alphabet has not yet been revealed. For a long time it was commonly accepted that the use of script in South Arabia arose, quite suddenly and in a rather elaborate way, in the eighth, perhaps even the ninth century BC, without any visible traces of primitive development or derivation. The earliest known inscriptions, written on stone blocks or rock surfaces, present themselves in a perfectly developed shape, which had only gradually to be improved towards the ‘classic’ style of the great monumental inscriptions under the Sabaean ruler Karibʾil Watar in the early seventh century BC. Apart from some isolated words painted on pottery from the late second millennium — their actual relation to the later stone inscriptions could not be clarified 187 P. STEIN script was incised with a pointed style, a practice that is known from wax tablets in the ancient world. Applied to wood, however, this method of writing is unique in the ancient Near East and beyond. A first comprehensive investigation of the palaeographical development of this particular script was published by J. Ryckmans in this journal more than ten years ago (2001). The essential accuracy of his relative chronology would fortunately later be proved by the radiocarbon analysis of representative sticks. It is very fortunate that the publication of these analyses coincides with that of the present paper (Drewes et al., this volume). It is important to know that both types of script are almost completely restricted to a particular writing support: the cursive script to the wooden sticks and the monumental script to all other material. The cursive script occurs randomly on rock surfaces or stone objects, perhaps the result of a scribe writing when the mood took him.2 The wooden sticks, however, are exclusively inscribed with the cursive script, with no single instance of any use of monumental letters on them.3 Since the monumental script has been known for a much longer time, we will begin with this. Fig. 2. The development of the Ancient South Arabian minuscule script. — thus far there was no evidence of any primitive development in the South Arabian script. Only in the 1970s did the first evidence of everyday correspondence from Ancient South Arabia come to light. Unlike the inscriptions on metal or stone known so far, this correspondence was written on wood — in the form of cigar-like sticks, cut from palm-leaf stalks or branches of other kinds of wood. Thanks to radiocarbon analysis, the oldest pieces can be dated to the earliest phases of Ancient South Arabian history and thus fill in the gap that was left by other epigraphical remains. These inscribed sticks comprise all kinds of legal and business documents, correspondence, oracular requests and even school writing exercises. These handy sticks are the most easily prepared writing material imaginable. Just cut off the tree, the wood is ready to be inscribed immediately, without any further preparation of its surface (apart from, in some cases, stripping off the bark around it). The 188 Palaeographical development of the monumental script (Fig. 1) The specimens of letter forms presented in the table here (Fig. 1) are drawn from particular authentic inscriptions. Each column represents a characteristic example of a certain chronological period, separated from each other by a span of 2–4 centuries. Since the letter forms are taken from particular inscriptions, not all positions could be filled in, leaving some gaps for one or other letter in a certain period. The classification of the periods was made on a practical basis, by naming the main phases of Ancient South Arabian history (called Early, Middle and Late Sabaic) simply by the capital letters B, C and D. The letter A stands for the oldest phase of the script, called ‘archaic’. It should be emphasised that this classification system is not compatible with other systems in the older literature, such as those of H. von Wissmann and J. Pirenne, mentioned above; similarly, any confusion with the chronologically 2 For the few instances, see the references in Stein 2010: 18 n. 15; 47 n. 204 (for ThUM 34). 3 That the minuscule script in its earliest phases clearly resembles the letter forms of the monumental script is self-evident, due to the common origin of both types of script. PALAEOGRAPHY OF THE SOUTH ARABIAN SCRIPT different phases of the minuscule script should be avoided.4 As for absolute chronology, spectacular new data is presently available for the phases named C1 and C2. The particular inscriptions are dated to a foreign era and thus allow, for the first time, an absolute dating of the Ancient South Arabian script in the second half of the first millennium BC. A1: the oldest instances of Ancient South Arabian script are painted pottery sherds from Raybun in Ḥaḍramawt, which are commonly dated to the late second millennium BC.5 These contain isolated words (proper names) or even single letters, so that it has been impossible to determine the language behind them. The connection of these isolated items with the much more elaborate inscriptions of the following periods had remained unclear for a long time. Only the consistent sequence of the minuscule script in the early period (see below) could substantiate the assumption of an unbroken continuity between these early letter forms and the writing ductus of the following stage A2. A2: the second column in Figure 1 contains samples of the letter shapes of several inscriptions that are certainly later than those under A1, but palaeographically older than the following period B.6 The actual age of these forms is not firmly established; traditionally they have been dated up to a few generations before Karibʾil Watar, hence into the eighth, perhaps even the (late) ninth, century BC (von Wissmann 1982: 145–147). They just fill in the gap between the early sherds from Raybun (phase A1) and the ‘classical’ period c.700 BC (B), and may thus be attributed 4 As long as there is no direct chronological correlation between the palaeographical phases of monumental and minuscule script, they should not be numbered by the same system. Since a convincing numbering system (with Roman numbers) was established for the minuscule script by J. Ryckmans, we prefer an alternative classification system by letters for the monumental script. These letters must not be confused with the classification of the early South Arabian script by J. Pirenne (1956), who distinguished her phases of the script of the first millennium BC by the letters A–E and respective subcategories. 5 For the evidence and its dating, see Sedov 1997: 43–47 and 94–95. —The ceramics found in Yala, a small site not far south of the Sabaean capital Marib, which are usually referred to as a chronological parallel to the early Raybun pottery (see e.g. Simpson 2002: 157), are in fact of much later date (in his publication Garbini [1992] speaks of a span between 850 and 580 BC, given by radiocarbon data). The particular script these sherds are incised with forms not part of the earliest traces of writing in South Arabia, but rather resembles an advanced stage of the minuscule script (phase II, see below). 6 The examples are collected from different inscriptions (Pirenne 1956: pl. II and table 2; von Wissmann 1982: 64–75). to the tenth–eighth centuries BC. Together with phase A1, this pre-classical stage may be called ‘archaic’. B: the first phase of script that can be dated with certainty is shown in the third column of Figure 1. These are letters from one of the largest texts from Ancient Arabia — the famous res gestae of the Sabaean ruler Karibʾil Watar, son of Ḏamarʿalī, written on two 7 m-long stone blocks in the temple of the main Sabaean deity Almaqah in the city of Ṣirwaḥ, 40 km west of Marib (RES 3945 + 3946, see photographs in Robin & Vogt 1997: 96–97). Thanks to a synchronism with Assyria, this inscription can be dated to the early seventh century BC.7 The shape of the letters in this text is extraordinarily well developed, keeping strictly to the following geometric rules: —The script is strictly rectangular and of constant and parallel lineation. Within one text the width of all lines is identical. The round letters and segments, in particular the w, are perfectly circular. —The letters’ proportion (von Wissmann: ‘H€ ohen-Breiten-Index’) is 3:1, i.e. the height of each letter is three times its width (the circle-shaped letters ʿ and w do not follow this rule, of course). —The vertical structure of the letters is also in three parts: except for clearly symmetrical letters (such as m, n, f, ḍ and the like and, again, the circle-shaped ʿ and w), the signs are divided into three parts, each filling a square of equal size. Thus the upper parts of h and ḥ (the ‘cup’) as well as of ʾ and s, for example, are exactly half the length of the lower part of the respective letters. This plain and somehow perfect ductus, based on a strict geometrical system, has consequently been considered a ‘standard’ of script in the scholarly world, and taken as a measure for evaluating all other periods of Ancient South Arabian writing. C1: the next palaeographical step with an absolute date is marked by two inscriptions put up in the Sabaean cities 7 A Sabaean ruler of this name is mentioned in an inscription by Sennacherib c.685 BC. This synchronism, established by von Wissmann (1982: 147–149, referring to an observation previously made by F. Hommel 1927: 75–86) and followed by most scholars (see e.g. Robin 1996: 1113–1114), is now supported by a second inscription of comparable size at the same site. This inscription, written in a similar ductus as that of Karibʾil Watar, was set up by Yiṯaʿʾamar Watar, son of Yakrubmalik, who could be identified with a Sabaean ruler of the same name mentioned in the annals of Sargon II in 715 BC (see Nebes 2007). The fact that the only two Sabaean rulers mentioned in Assyrian sources bear the names of the authors of by far the two largest monumental Sabaic inscriptions from that period makes the proposed synchronism highly probable. 189 P. STEIN of Marib and probably Nasqum, but commissioned by foreign traders from East Arabia. Both inscriptions are dated, according to the Seleucid Era, to c.300 BC.8 The third century BC marks the transition from the so-called Early Sabaic towards the Middle Sabaic period, a transition that corresponds with far-reaching political as well as linguistic changes.9 The script of the texts in question already shows the characteristics of the palaeographical development during the Middle Sabaic period, such as —the use of acute angles instead of the strictly rectangular formation of the classical style B (particularly remarkable with the letters ʾ and n); —the complete abandonment of any fixed letter proportion; —a distinct thickening of the extremities of all lines, which will develop later on into regular serifs. In consequence, the borders of all straight lines of the letters are no longer parallel (as they had been in the older script); —the shift from the former vertical three-part structure towards a bipartite form: from now on, the upper elements of letters such as h and ḥ, ʾ and s have the same length as their lower parts; —the opening of the letters m and š towards curved structures (instead of the former triangular shape); —the twofold curving of the formerly crescent-like r towards a boomerang form; —the increasing flattening of the circled letter w towards an ellipse. Apart from this, elliptic forms are occasionally also found with other circle-shaped elements, namely letters such as y and ṯ. 8 The specimens presented here are taken from the Marib inscription. This text consists of two fragments, the first of which, RES 3605bis = Ry 547, has already been known for a long time. The date ‘year two of King Seleucos (I.)’ in line 1 of the text, however, has only recently been established with the help of a second, still unpublished, fragment (DAI Marib Oase 2007-1), which was discovered by the German Archaeological Institute. The joined texts and the chronological results were presented by Norbert Nebes at the Seminar for Arabian Studies in July 2007 in London and a publication is in preparation. I am very much indebted to my colleague for providing me with drafts of all the relevant material. —The second inscription of similar content and palaeography, but probably put up in Nasqum in the Wadi al-Ǧawf, is dated to the seventh year of the very same king (see Prioletta 2011). The exact absolute dates of the given years are, however, not quite certain because of the dispute in reckoning the first years of the king’s reign (Prioletta 2011: 288–289). 9 As for the latter, some marked differences in the grammar of Sabaic between the Early and Middle periods have been observed (see Stein 2011: 1046–1047 and passim; for more details see Stein 2005). 190 C2: three centuries later, the characteristics of the Middle Sabaic script, as outlined above, have fully developed. The table shows specimens from a Sabaic-Nabataean bilingual inscription from Ṣirwaḥ which was written, according to a date in the Nabataean part, in the year 7–6 BC.10 C3: from the mid-first century AD onwards, the palaeography of Sabaic monumental inscriptions is more or less reliably established due to a certain relative chronology of the Sabaean kings. Column 6 in our table presents characteristic letter forms from inscriptions under the rule of Ilsaraḥ Yaḥḍib and his brother Yaʾzil Bayyin, kings of Sabaʾ and Ḏu Raydan, who reigned in the mid-third century AD.11 D1–2: the latest phase of Sabaic, and of Ancient South Arabian writing culture in general, is called Late Sabaic. It is characterised by the political hegemony of the Ḥimyar, a confederation from the southern Yemeni highlands, and by the predomination of a monotheistic faith. Many inscriptions are dated according to a fixed era. In palaeography, a consequent development of the Middle Sabaic features can be observed, resulting in the following modifications: —serifs become more and more prominent; —the formerly elliptic letter w is stretched to such an extent that it finally appears as two separate circles; —there is a strong tendency to cut the inscriptions in relief. In relief inscriptions, the letters are generally of a rather compact shape with thick lines, often resulting in an almost square proportion of the width and height of each letter. The last column (D2) shows the characteristic forms of the Late Sabaic relief script. The samples in Figure 1 are taken from inscriptions from the late fourth (D1) and the mid-sixth century AD (D2).12 10 DAI Ṣirwaḥ 2004-12 + Fragm., dated in the year 3 of Arethas IV (see the preliminary treatment in Nebes 2006). Again, my thanks to Norbert Nebes for allowing me to use his unpublished materials. 11 This date inn confirmed by the inscription al-Miʿsal 2, dated according to an absolute era to the year AD 253, which refers to a military campaign under the rule of those two kings (see Robin 1981: esp. 323 and 334–335, and most recently M€ uller 2010: 25– 26). The inscriptions the samples were taken from are the dedicatory texts E 69 and J 576 + 577, the latter being the longest inscription from pre-Islamic times ever found on the Arabian Peninsula (see photographs of the inscriptions in al-ʾIryanī 1988: 16 and Bron 1992: 84–87, respectively). 12 D1: Gar B. Aswal 1, a building inscription from the Ḥimyarite capital Ẓafar, written under king Ḏaraʾʾamar Ayman (c. AD 400; see Robin 2008: 184, 200); D2: C 541, the large stela erected in Marib by king Abreha in the year 548 (dated 658 Ḥimyarite era; this is one of the latest dated texts from Ancient South Arabia). Good photographs of the latter are published in Daum et al. 1999: 270. PALAEOGRAPHY OF THE SOUTH ARABIAN SCRIPT In summarising the evidence presented here, we may conclude that there are two contrary tendencies in the development of the Ancient South Arabian monumental script: 1. The clear geometric system all letter forms are based on during the classical phase in Early Sabaic times is gradually abandoned in the later periods. 2. On the other hand, the ornamentation of letters by non-geometrical features such as serifs, curves, etc. is continuously increasing in the post-classical periods. The palaeographical development of the minuscule script (Fig. 2) The second mode of writing in Ancient South Arabia is the so-called minuscule script, which was almost exclusively used in daily life and scratched on wooden sticks. A particularly helpful feature of this kind of script is the fact that its writing support can be subjected to radiocarbon analysis. Indeed, such analyses have been made on a number of inscribed sticks from the collection of the Oosters Instituut at Leiden (see Drewes et al., this volume); the data that emerged from this, however, are still not precise enough to give a clear overall picture of the chronological development of their script.13 Evidence for absolute chronology, be it dating formulae or historical clues in the text, is found only from about AD 200 onwards. In any case, the palaeographical development of this particular script was already well established by Jacques Ryckmans in the 1990s.14 The numbering of the phases follows the system introduced by Ryckmans (2001). The specimens shown in 13 As can be seen from the data presented by Drewes et al. (this volume) and which is made use of in the following paragraphs, there is some striking chronological overlap between single pieces from one or other palaeographic stage of the minuscule script. This overlap may lead to the assumption that several palaeographical phases should not be taken as chronologically successive but rather as a contemporary variation. On the other hand, we have a number of historical dates confirming the turning point from phase IVa to IVb c. AD 300 (see below, with note 26), while the (comparatively few) 14 C dates for this period tend to be 1–2 centuries earlier. Although a contemporary existence of both ductus may not be excluded even in this case, it seems much less probable. Additional analyses of further sticks in future may help to clarify this apparent contradiction. For our purpose, however, the question is only relevant for the early periods since, particularly for phases IVa and IVb, we have taken the specimens for our palaeographic tables from text which can be dated by absolute chronology. 14 Ryckmans’s palaeographical sequence could mainly be confirmed by radiocarbon analysis of selected sticks, as is pointed out by Drewes et al. (this volume). the tables are taken from minuscule inscriptions in the Munich collection;15 for practical reasons, the letter forms of each period are mostly combined from more than one inscription. Again, there are some gaps left in the illustration as the inscriptions taken as models for certain periods accidentally lack specific letters. In contrast to this, the letter ẓ generally disappears from the script in the later periods from IIIa onwards, being replaced by the letter ḍ.16 I: the earliest phase clearly resembles the letter forms of the contemporary monumental inscriptions on rock and stone. Apart from the slightly bent shape, certainly due to the structure of the convex writing surface, the letters are more or less the same as in the archaic phase of the monumental script (phase A2). As for the age of this phase, we have 14C dates ranging between the eleventh/tenth and the (late) fifth centuries BC (the majority between 800 and 500).17 By far the oldest of these dates (the stick L 24, 14C dated to 1055–901 BC)18 corresponds well with the archaeologically fixed date of the earliest instances of monumental script from Raybun (phase A1, see above). II: the second phase of the minuscule script, divided into four subcategories (IIa–d) and called ‘transition’ by Ryckmans (2001), is characterised by a beginning inclination of the letters combined with a loss of their geometrical appearance. Marked features of this phase are: —The morphology of the letters is still based on their (archaic) monumental form (phase I), but increasingly affected by inclination and bending of lines. In the later periods (IIc–d), some letters start to open up, producing horizontal ends of formerly vertical lines. Characteristic indicators for the development within this phase are the letters l, m and q. 15 Bayerische Staatsbibliothek in Munich. The Ancient South Arabian minuscule inscriptions from this collection are quoted according to either their siglum X.BSB in Stein 2010 or their library signature Mon.script.sab. (unpublished). 16 This phenomenon of the minuscule script, which contrasts with the contemporary orthography of the monumental inscriptions, is probably based on sound shift in speech (see Stein 2010: 40–41). 17 The specimens in the table are taken from Mon.script.sab. 103 (a school alphabet), considering the two Minaic oracular requests Mon.script.sab. 74 and 76 (see also the alphabet in Ryckmans 1997). 18 This piece had been tentatively ranged in phase IIa by Ryckmans (2001: 225 fig. 1). A closer look at the inscription, however, reveals that its palaeography is definitely more archaic (cf. Stein 2010: 46 with n. 195). 191 P. STEIN —The previously circled parts of letters lose their character in favour of straight and cornered lines, in particular y, ṯ, s: and ẓ (and again, q). —The letter ʾ (alif) has lost its double-angled top in favour of an inclined simple stroke. —In phase IId, the letter ḏ finally loses its horizontal bars in favour of a small stroke between the bottom ends of the two verticals. Overall in phase II we have 14C dates between c.800 and 400 BC. It thus corresponds more or less with the classical phase B of the monumental script. The examples labelled IIa and IIc–d here are characteristic of Ryckmans’s subgroups IIa and IId, respectively.19 The sub-division of phase II into at least three subcategories (a, c and d) is still to be reliably established.20 IIIa: the third phase presents the minuscule script in its developed shape. The development of phase II continues, leading to the following characteristics: —Most former verticals have inclined to curved lines, tapering off to horizontal ends. Unlike in earlier periods, these ends generally point to the left, thus resulting in some kind of base line in writing (the only exception is made by q and partly by m, ḍ and z).21 All box-, hook- or fork-shaped letters of earlier times have developed into a marked vertical (with the bottom bending to the left) and an almost closed, round corpus to the right of it. Irrespective of their different 19 The letter forms categorised as phase IIc here are found under IId in Ryckmans’s table (2001: 225 fig. 1). The marked difference from other characters of this phase (our IId, see Fig. 2), however, speaks for a clear distinction. On the other hand, Ryckmans’s phase IIc, represented by only two poor examples (ibid.), could not firmly be verified on the basis of the combined evidence of both the Leiden and the Munich collections. The letter forms of our samples are taken from Mon.script.sab. 102 (alphabet), 246 (account) and 419 (Minaic letter, all IIa); Mon.script.sab. 308 and 662 (alphabets), considering the Sabaic incantation Mon.script.sab. 7 (IIc); and Mon.script.sab. 160 (alphabet), considering the Sabaic and Minaic business documents Mon.script.sab. 78 and 588 (IId). 20 The subcategory IIb, as introduced by Ryckmans (2001: 225 fig. 1), is only based on one single — however well established — example, which is accompanied by a few mostly fragmentary inscriptions of similar ductus in both the Leiden and Munich collections. It remains an open question whether this particular ductus is really representative of a separate palaeographical ‘phase’ or just some variant of one of the other phases IIc and IId, respectively. 21 Letters of this and the following periods often seem to be connected with each other by such a line, although connecting letters are not a regular phenomenon of cursive writing in Ancient South Arabian. 192 origin, the basic shape of these two elements is more or less identical in all letters. —The originally circled letters ʿ and w have got a leftoriented vertical stroke as well. In consequence, the letters ʿ and b, for example, are now more or less identical in shape, distinguished only by their different size. —The letter d consists of a long vertical and a small closed element at some distance to the left, the latter looking like a small ʿ. 14 C dates of this phase range mainly between 500 and 100 BC,22 covering the Early Middle Sabaic period (cf. phase C1 of the monumental script). IIIb: This ductus is quite exceptional, as it does not fit into the straight palaeographical order. Even though we have (only two) 14C dates from the first–second century AD, a historical connection with phase IVa seems rather problematic since, from a palaeographical point of view, the ductus appears much older: apart from a marked accentuation of verticals, the basic shape of the letters is that of phase IId (with some tendency towards IIIa).23 Since many representatives of this particular ductus are written in the Minaic and Amiritic languages, this ‘phase’ could have been more a local than a chronological phenomenon.24 IVa: this phase, covering large parts of the Middle Sabaic period, presents the cursive character of the minuscule script very clearly. The letters are markedly curved and inclined, emphasising verticals (projecting lines of particular letters) as well as horizontals (mainly at the imaginary bottom line). Any similarity with the contemporary monumental script (phases C2–3) has disappeared. The basic shape of letter formation remains unchanged, only two marked differences from the preceding period IIIa can be observed: —The long, right stroke of the letters h and t projects far beyond the bottom line (in contrast to its orientation by the horizontal line, as most other letters, in the preceding phase). 22 The letter forms are taken from the Sabaic letters X.BSB 98 and 99, compared with the Minaic legal document Mon.script.sab. 554. 23 Another striking feature is the existence of the letter ẓ, which was generally omitted (i.e. replaced by the letter ḍ) as early as phase IIIa (see above, with note 16). Furthermore, a number of Minaic texts are written in this ductus, while this language is commonly thought to have died out in the late second century BC. 24 The examples are from the letters X.BSB 96 and Mon.script.sab. 142 and 640, written in Minaic and Amiritic respectively. PALAEOGRAPHY OF THE SOUTH ARABIAN SCRIPT —The letter d has a tripartite form: two (instead of one in IIIa) verticals accompany the small closed element. The latter gradually gets smaller and finally completely disappears in the following phase IVb. For this phase, we have 14C dates between 200 BC and AD 250, supplemented by additional data within the texts.25 IVb: the last phase of the minuscule script starts c. AD 300. This date is supported by chronological data from the inscriptions themselves, since texts of this late period are often dated according to the absolute era of the Ḥimyarite dynasty.26 14C dates, however, range from the (late) first to the early sixth century AD. The ductus of this late phase is very specific. Although resulting from a consequent development from phase IVa, the letter forms are characterised by the following peculiarities:27 —The horizontals are shortened in favour of a strong accentuation of all vertical lines. Thus the script has a generally rather crude appearance (the extent of which obviously depends on the material structure of the writing support, see note 28). —The letters ʾ and k, previously composed of two curved, continuously drawn lines, are ‘broken’: they are now formed by three separate, almost vertical lines which either keep clear of or cross each other (Ryckmans 2001: 233 ‘N’- or ‘‫’א‬-like form). —The letter d has finally lost its third (left) element, resulting in the formation of merely two verticals. —The letter m is made up two small bent strokes forming a kind of small hook; this shape has already emerged in the course of the previous phase IVa. —The distinct features as well as the size of certain letters become reduced to a large extent so that a number of signs appear to have a more or less identical shape (especially ʾ and k; ʿ, b, l and y; and sometimes ḥ and w). 25 The specimens are from X.BSB 32 and 41, two Sabaic legal documents, which can probably be dated to about AD 200 (after Stein 2010: 48–49.); see also X.BSB 20 and 61 of similar date (ibid.). 26 See Stein 2010: 47 with nn. 204–205. The first inscription written in this ductus is around AD 300, while the so far last dated text was written in the year 522 (dated 5322 Ḥimyarite era). Thanks to a fairly broad documentation of inscriptions and external chronological data in this period, the transition between phases IVa and IVb is quite clearly established (see the palaeographic table in Stein 2010: 48–49). 27 The samples are taken from the Sabaic legal document X.BSB 46 and the letter X.BSB 152, dated to the years 522 and 554 Ḥimyarite era = AD 412 and 444, respectively. Conclusions (Fig. 3) It is clear that the script of the wooden documents, although of common origin with that of the monumental inscriptions, has quickly developed into a separate cursive. Due to the needs of fluent writing, this so-called ‘minuscule’ script tends more and more to markedly curved and inclined lines, as well as emphasising partly horizontals, partly verticals, dependent on the convex surface of the writing support.28 At the end of this development, the basic letter forms have hardly anything in common with their ‘ancestors’ more than 1000 years before.29 In contrast to this, the geometric ductus of the monumental inscriptions chiselled in stone remained basically unchanged, forming some kind of ‘capital’ letters which were altered only by ornamentation, but not by a marked deviation of their basic shape. The synchronic development of both monumental and minuscule script is illustrated in Figure 3. The individual periods of both types of script, as described above, are now arranged according to their chronological order. Proceeding from the ductus of those inscriptions that can be dated with certainty (with the approximate date given above the particular column), the remaining specimens of uncertain date (mainly the different phases of the minuscule script in the first millennium BC) are grouped in between. What does this survey tell us about the use of script in Ancient South Arabia? The result is simple and not unexpected: the script used for everyday purposes such as writ28 In contrast to stressing horizontals in the earlier phases of the minuscule script, a markedly vertical stretching can be observed in the last phase (IVb). This feature, which is often accompanied by a remarkably large letter size, is obviously dependent on the structure of the most common writing support in that period. The soft wood of the ʿUshar tree (Calotropis procera), consisting of thick horizontal fibres, seems to hamper fluent writing in a fine, level-oriented cursive. 29 Incidentally, the striking deviation of the minuscule script in the later periods was the reason for the difficulties in deciphering the wooden documents: from the discovery of the first pieces to the publication — i.e. transliteration and translation — of the first text, almost twenty years had passed. The person who dedicated himself most intensively to this subject from the beginning was the Jordanian scholar Mahmud Ali al-Ghul. Unfortunately it was not granted to him to harvest the fruits of his work due to his early death in 1983. The first publication of an inscribed wooden stick from ancient Yemen was made by ʿAbdallah (1986) (a receipt written in Sabaic). A first, albeit small, monograph on the topic was prepared by Ryckmans, M€ uller and ʿAbdallah (1994), including the publication of sixteen inscriptions. For further details on the research history, see the introduction in Stein 2010: 17–20. 193 P. STEIN Fig. 3. Monumental and minuscule scripts in synchronic comparison. ing letters, contracts, and accounts, is simplified in view of its purpose. The characteristic curving and inclination of the letters clearly emerge from the needs of cursive writing. Script in a public space, on the other hand, has been designed not for comfortable handling but rather for representative effects, for visual impression. While these effects were sought for in strict geometrical patterns in the early style (B), this was soon given up in favour of styling and ornamentation, by using additional elements such as curves, serifs, and others, in the later periods. Both types of script have a common origin in the late second millennium BC, but subsequently developed into two largely separate systems, which differ completely both in the mode of writing and (consequently) in the appearance of the letter forms. Both types always run parallel through the history of South Arabian writing. Although they have continuously been diverging 194 from each other, they remained representatives of a common writing culture — revealing texts in one and the same language, and ceasing to be in use only immediately before a new mode of writing spread over the Arabian Peninsula: the Arabic language and script in the progression; train of Islam. Acknowledgements This article is based on a paper presented at the Symposium ‘Rationalizing Script: the simplification of characters and of writing systems’ at Ca’ Foscari University in Venice (Italy), 12–13 September 2011. The author is very much indebted to M.C.A. Macdonald (Oxford) for reading a draft of the article and for providing him with all the relevant material on the radiocarbon analyses long before its publication (on which see now Drewes et al., this volume). PALAEOGRAPHY OF THE SOUTH ARABIAN SCRIPT References ʿAbdallah, Y.M. 1986. Ḫat: t: al-musnad wa-nnuqus al-yamanīya al-qadīma. Dirasa likitaba yamanīya qadīma manq usa ʿala l-ḫasab. Al-Ḥalqa aṯ-ṯaniya. Al-Yaman al-Ǧadīd 15/6: 10–28. Bron, F. 1992. Mémorial Mahmud al-Ghul. Inscriptions sudarabiques. Paris: Geuthner. Daum, W., M€uller, W.W., Nebes, N. & Raunig, W. (eds.) 1999. Im Land der Königin von Saba. Kunstschätze aus dem antiken Jemen. (Exhibition catalogue). Munich: Staatliches Museum f€ ur V€ olkerkunde. Drewes, A.J., Higham, T.F.G., Macdonald, M.C.A. & Bronk Ramsey, C. this volume. Some absolute dates for the development of the South Arabian minuscule script. AAE 24. Garbini, G. 1992. Le iscrizioni su ceramica da ad-Durayb/Yala. Yemen 1: 79–91. Hommel, F. 1927. Geschichte S€ udarabiens im Umriss. Pages 57–108 in Nielsen, D. (ed.), Handbuch der altarabischen Altertumskunde. Vol. 1. Die altarabische Kultur (Copenhagen: Busck). al-ʾIryanī, M.ʿA. 1988. al-ʾIryanī 69. Rayd an 5: 9–16 (Arabic part). M€uller, W.W. 2010. Sabäische Inschriften nach Ären datiert. Bibliographie, Texte und Glossar. (Veröffentlichungen der Orientalischen Kommission der Akademie der Wissenschaften und der Literatur Mainz 53). Wiesbaden: Harrassowitz. Nebes, N. 2006. Eine datierte nabatäischsabäische Bilingue aus Ṣirwaḥ. JemenReport 37/1: 10. Nebes, N. 2007. Ita’amar der Sabäer: Zur Datierung der Monumentalinschrift des Yiṯa’’amarr Watar aus Ṣirwaḥ. AAE 18: 25–33. Nebes, N. in preparation. Gerrhaeans in Marib. A new fragment to a dated text. Archäologische Berichte aus dem Yemen. Pirenne, J. 1956. Paleographie des inscriptions sud-arabes. Contribution à la chronologie et à l’histoire de l’Arabie du Sud antique. Vol. 1. Des origines jusqu’à l’époque himyarite. Brussels: Koninklijke Vlaamse Academie voor Wetenschappen. Prioletta, A. 2011. The Sabaic inscription A20-216: a new Sabaean-Seleucid synchronism. PSAS 41: 283–294. Robin, C. 1981. Les inscriptions d’al-Miʿsal et la chronologie de l’Arabie méridionale au IIIe siecle de l’ére chretienne. Comptes Rendus des sceances de l’Academie des Inscriptions et BellesLettres: 315–339. Robin, C. 1996. Sheba. II. Dans les inscriptions d’Arabie du Sud. Cols. 1047– 1254 in Brend, J., Cothenet, E., Cazelles, H. & Feuillet, A. (eds.), Supplément au Dictionnaire de la Bible. Vol. 12, Fasc. 70 (Paris: Letouzey et Ane). Robin, C.J. 2008. Inventaire des documents epigraphiques provenant du royaume de Ḥimyar aux IVe–VIe siecles. Pages 165– 216 in Robin, C.J. & Schiettecatte, J. (eds.), L’Arabie à la veille de l’Islam. Bilan clinique. Table ronde tenue au Collège de France (Paris) les 28 et 29 août 2006 dans le cadre du projet de l’Agence nationale de la recherche ‘De l’Antiquite tardive à l’Islam’. (Orient et Méditerranée 3). (Paris, De Boccard). Robin, C.J. & Vogt, B. (eds.) 1997. Yémen. Au pays de la reine de Saba (Exhibition catalogue). Paris: Flammarion. Ryckmans, J. 1997. Un abecedaire sudarabe archaïque complet, grave sur un petiole de palme. Pages 11–36 in I primi sessanta anni di scuola. Studi dedicati dagli amici a Sergio Noja Noseda nel suo 65o compleanno 7 Luglio 1996 (Lesa: Fondazione Ferni Noja Noseda). Ryckmans, J. 2001. Origin and evolution of South Arabian minuscule writing on wood (1). AAE 12: 223–235. Ryckmans, J., M€ uller, W.W. & ʿAbdallah, Y.M. 1994. Textes du Yémen antique inscrits sur bois (with an English Summary). Louvain-la-Neuve: Peeters. Sedov, A. 1997. Die archäologischen Denkmäler von Rayb un im unteren Wadī Dauʿan (Ḥaḍramaut). Mare Erythræum 1: 31–106. Simpson, ST. J. (ed.) 2002. Queen of Sheba. Treasures from Ancient Yemen. (Exhibition catalogue). London: The British Museum Press. Stein, P. 2005. Linguistic contributions to Sabaean chronology. Archäologische Berichte aus dem Yemen 10: 179–190. Stein, P. 2010. Die altsüdarabischen Minuskelinschriften auf Holzstäbchen aus der Bayerischen Staatsbibliothek in München. Vol. 1. Die Inschriften der mittel- und spätsabäischen Periode. (Epigraphische Forschungen auf der Arabischen Halbinsel 5). T€ ubingen/Berlin: Wasmuth. Stein, P. 2011. Ancient South Arabian. Pages 1042–1073 in Weninger, S., Khan, G., Streck, M.P. & Watson, J.C.E (eds.), The Semitic Languages. An International Handbook. (Handbücher zur Sprach- und Kommunikationswissenschaft 36). (Berlin/ New York, De Gruyter). Stein, P. forthcoming. Wilhelm Gesenius, das Hebräische Handw€ orterbuch und die Erforschung des Alts€ udarabischen. In Schorch, S. (ed.), Biblische Exegese und hebräische Lexikographie. 200 Jahre ‘Hebräisch-deutsches Handwörterbuch’ von Wilhelm Gesenius. Conference at Martin Luther University HalleWittenberg, 14th–18th March 2010. von Wissmann, H. 1976. Die Geschichte des Sabäerreichs und der Feldzug des Aelius Gallus. Pages 308–544 in Temporini, H. & Haase, W. (eds.), Aufstieg und Niedergang der Römischen Welt II/9. (Berlin/New York: De Gruyter). von Wissmann, H. 1982. Die Geschichte von Saba’ II. Das Großreich der Sabäer bis zu seinem Ende im fr€ uhen 4. Jh. v. Chr. Herausgegeben von Walter W. M€ uller. € (Osterreichischen Akademie der Wissenschaften, Philosophisch-historische Klasse, Sitzungsberichte, 402). Vienna: € Osterreichische Akademie der Wissenschaften. 195