The Hurrian songs of Ugarit

The Hurrian songs are about 36 fragments of clay tablets found in the 1950s in the Royal Palace at Ugarit, in a level of the 14th century BC⁠[1]⁠[2]. Laroche's edition in Ugaritica V numbers the songs h.2 to h.17, h.19 to h.23, h.25, h.26, h.28 and h.30; the smaller pieces carry the field numbers RS 19.164 with letter suffixes⁠[1]. Only h.6 is substantially complete. It is a single tablet of three joined fragments, RS 15.30, RS 15.49 and RS 17.387, now National Museum of Damascus 3916, 17 cm wide, 6.3 cm high and 2.5 cm thick⁠[1]⁠[2]. The Hurrian text runs in four lines across the top, continuing from the front over the edge onto the back in a spiral. A double rule separates it from six lines of Akkadian terms, each followed by a numeral between 1 and 10, and a final colophon states that this is a song in the nid qabli tuning, of the zaluzi type, written by Ammurabi⁠[1]⁠[11]. Four other fragments name composers with Hurrian names; h.6 names none⁠[1]. Dietrich and Loretz collated the tablet in 1975 and corrected Laroche's copy in places⁠[23]. The Damascus museum's online record gives a photograph and Krispijn's translation⁠[2]. No open high-resolution image set, RTI capture or machine-readable transliteration of the Ugarit tablets exists; the theory tablets CBS 10996 and UET VII 74 have CDLI records with catalogue data, and CBS 10996 has photographs and RTI⁠[24]⁠[25].

Three Babylonian tablets make the notation readable. CBS 10996, a Middle Babylonian coefficient list from Nippur, pairs fourteen Akkadian names with pairs of string numbers: seven pairs a fourth or fifth apart, which Wulstan called primary, and seven a third or sixth apart, which he called secondary⁠[4]⁠[7]⁠[6]. Nabnitu XXXII, a lexical list from Ur, names the nine strings of the sammu instrument, most likely a lyre, counting from the front and from the back so that strings 1 to 7 are joined by 8 and 9 that repeat 1 and 2⁠[6]⁠[26]. UET VII 74, an Old Babylonian text from Ur, instructs the player to pass from one tuning to the next by altering a single string, producing a cycle of seven tunings each named after a primary pair; a second manuscript, UET VI/3 899, confirmed the cycle in 2009⁠[5]⁠[27]. The seven tunings are heptatonic and diatonic, that is, seven notes to the octave arranged as in a Western major or minor scale, though the exact sizes of the steps are not recorded⁠[28]. Every term in the notation line of h.6, with the Hurrian spellings qablite, irbute, sahri, isarte, titim isarte, zirte, sassate and others, matches one of the fourteen names, so each term is securely a pair of strings⁠[8]⁠[1]. The colophon's tuning, nid qabli, is one of the seven⁠[15]. Across all 35 notated fragments, Rahn found that individual strings occur at near-equal rates and that pairs two strings apart modulo seven are more frequent than pairs three apart⁠[22]. The Hurrian text is a hymn connected with offerings to Nikkal; the word pairs such as usta mari that recur after numerals in several fragments are not understood⁠[1]⁠[21].

Every realization has to decide four things the tablet does not state: whether a term means two strings sounded together or in succession, what the numeral counts, which way the scale runs, and how the notation aligns with the words. Wulstan read each term as one note and produced a melody on a rising scale⁠[9]. Kilmer read each term as a dyad, two strings struck together, and the numeral as a count of repetitions; the 1974 recording on a reconstructed lyre followed this reading⁠[8]⁠[10]. Duchesne-Guillemin read the terms as melodic steps and the numerals as note counts, and explained the syllables repeated at line ends by the spiral layout rather than as refrains⁠[11]⁠[29]⁠[30]. Vitale showed that nothing in UET VII 74 says whether a string is tightened or loosened, so the direction of the scale had been assumed; he argued for a falling series, and West built a melodic reading on it, while Gurney in 1994 and Crocker in 1997 accepted that the question was open⁠[12]⁠[13]⁠[14]⁠[31]. Hagel took the numerals as durations and tested the tunings against Greek modality; Baciu's 2025 preprint builds a rhythm on that reading and compares it with Vedic verse, but the premise is contested and the work is not peer reviewed⁠[15]⁠[32]. Dumbrill rejects the seven-note frame and reads the system as nine-note⁠[16]. Halperin applied frequency statistics to the terms⁠[33]. Rahn's 2011 analysis, and Gibson's 2021 thesis and 2025 article, argue that the data support string relations but not pitches, rhythm or direction⁠[22]⁠[28]⁠[17]. No test has yet eliminated a reading, because each is internally consistent and the tablet gives no check such as a sum or a second copy.

Five positions are live. The first, Kilmer's, is that the terms are dyads played together and the numerals say how many times, so the piece is a sequence of two-note chords with a sung line above it; its evidence is that the theory texts define the terms as pairs and that the Nippur music-instruction fragments use the same terms for performance⁠[8]⁠[34]. The second, held in different forms by Wulstan, Duchesne-Guillemin and West, is that each term stands for a melodic motion or a single note, so the line is a melody; its evidence is that a two-voice texture is otherwise unattested in the region and that a melody fits the syllable count better⁠[9]⁠[29]⁠[13]. The third, Hagel's, keeps a melodic or dyadic frame but reads the numerals as durations, which would give the piece a rhythm; its evidence is the distribution of the numerals⁠[15]. The fourth, Dumbrill's, keeps dyads but inside a nine-note system; it has few followers⁠[16]⁠[28]. The fifth, argued by Bayer in 1978, Rahn in 2011 and Gibson in 2025, is that the notation fixes only which strings are involved and in what order, and that pitch, rhythm and direction cannot be recovered from it, so the honest output is a tablature rather than a score⁠[18]⁠[22]⁠[17]. Cutting across these is the direction question: Vitale, West and Gurney lean to a falling scale, the Berkeley school to a rising one, and Gibson to the view that the texts allow either⁠[12]⁠[14]⁠[28]. The Hurrian words are a separate problem on which Thiel and Krispijn agree in outline and differ in detail⁠[20]⁠[21].

A Babylonian or Ugaritic text that explains the notation, rather than the tuning, would settle most of it: a tablet that says what a numeral after an interval name means, or that pairs a notation line with a syllable count. The Nippur music-instruction fragments are the nearest thing and are too broken to decide⁠[34]⁠[35]. A second copy of any Hurrian song, or a Babylonian song in the same notation, would allow the alignment of words and terms to be tested; Rahn suggests h.6 and h.7 may be variants of one piece, which could be checked on the fragments⁠[22]. Short of new texts, three things can be done now. The 35 fragments can be transliterated into a single machine-readable table of terms and numerals, which no one has published, so that Rahn's and Halperin's statistics can be repeated and extended⁠[22]⁠[33]. The syllable count of the Hurrian text can be compared with the sum of the numerals under each reading, which is the one internal check the tablet offers⁠[11]⁠[8]. And the Damascus tablet can be imaged with reflectance transformation imaging, as CBS 10996 has been, to resolve the signs Dietrich and Loretz read differently from Laroche⁠[23]⁠[24]. The direction of the scale depends on the tuning texts rather than on h.6 and will be decided by Assyriology, if at all⁠[31]⁠[27]. For the Hurrian words, progress depends on the Boğazköy bilinguals and on new Hurrian texts, not on the song itself⁠[19].