THE IZBET SARTAH OSTRAKON
 
MUSINGS OF A STUDENT SCRIBE

Brian Colless

    (1) ’ l m d ’ t  [  ] | ’ ‘
    (2) k t t n ‘ r h. ’ t b ’ z |[n b ‘ ] t.  ` l t. t.   (3) s. m q m r q

    (4) ‘ w p b n h. g ’ t l h |d z q n?  ‘ t ‘ ’ ‘ l ’ h. l d ‘ l m
    (5) ’ B G D H M W H. Z T. Y K L  N | S P ‘ S. Q R Sh T 

http://storiadigitale.zanichellipro.it/media/images/070.jpg

That link provides a good photograph of the document.

The ostrakon to be studied here is a piece of a large jar; it was discovered in 1976 at ‘Izbet S.art.ah (near Aphek), possibly Eben-ezer (1 Samuel 4:1); it was published by Moshe Kochavi (1977). The sherd contains five lines of alphabetic script, the fifth being a copy of the alphabet (Kochavi 1977:4-6); the technical term for this is "abecedary", though my word is "abagadary" (or "abgadary"), because the first four letters are 'ABGD (see Abagadary). 

On archaeological and palaeographical grounds the object can be dated to the 12th or 11th century B.C.E. (Kochavi 1977:12). It is deemed to be an exercise of a student practising his alphabetic writing skills with a stylus on hardened clay (e.g., Kochavi 1977; Demsky 1977; Naveh 1978; Cross 1980; Dotan 1981; Sass 1988). I will argue that the words on the ostrakon affirm this, and that lines 1-4 constitute a coherent continuous text, not simply random letters or jottings. Dotan (1981:162-169) has also attempted to extract a meaning from the inscription, interpreting it as Hebrew. 

A sketch of the text is offered (based on the original drawing of Kochavi), together with my proposed reading of the letters on the ostrakon (the vertical bar indicates the break-line, where the two pieces of the sherd have been joined). Photographs are provided by Kochavi, Dotan, Cross, and Sass. I have seen the sherd in the Israel Museum, where it is on display, and Alexandros Zaharopoulos made a colour photograph for my use.


PALAEOGRAPHICAL ANALYSIS
    Previous analyses have been made by:
    Kochavi (1977:6-12), with table of letters, fig. 4;
    Cross (1980:8-12), with a comparative table of signs, fig. 13;
    Puech (1986:171-173), with a comparison table, fig. 8;
    Sass (1988:67-69), with a table of letters, p. 185.
These all need to be consulted, and I will add my own comments, as I deal with such matters as the order of the letters in line 5, the relative frequency of the consonants in the inscription, and the pairs of signs which are difficult to distinguish from each other. 

LETTER ORDER
    Some letters are not in the usual order of the ’abgad system, notably Z and H., ‘ayin and P, and M. It seems to me that the student-scribe (or a teacher) has corrected some errors in his alphabet, with a small vertical mark at each point.

M (apparently) is in the position where W is expected, through confusion of bilabial consonants (Cross 1980:10). The W has possibly been inserted here subsequently, either diagonally between the lower lines of E (h), or below the line on which the letters stand, where an enigmatic vertical stroke is found. There seems to be a correction mark above the M, which is not part of the character.

The small vertical mark appears between H. and Z, presumably indicating that their order should be reversed. Or is it in fact part of the left side of Z? Is Z a short letter consisting of two vertical curved strokes, or does it have the same height as the letters on either side of it, and the shape of a copper ingot (Cross 1980:10-11)? Such a double triangle figure possibly appears in lines 2 and 4.

A mark appears at the space where M should stand, between L and N. Is a Waw inscribed here below the line, or beneath L (other possibilities for the missing W are mentioned above)?

There does not seem to be a correction mark between P andayin (this order is known to be acceptable in Hebrew practice, Demsky 1977:17-18). However, there is a diagonal line above P, emanating from the tail of the fish Samek, and a vertical line above ‘ayin, which seems to be forming a support Samek, with three cross bars (Puech shows a number of strokes here). Neither the fish nor the support are clearly discernible, but if they are really there, then the scribe may be showing that there are two forms of Samek (the Egyptian Djed pillar, a spinal column, a support for the body; and the fish, cp. Arabic samk 'fish').

 CONSONANT FREQUENCY
If the first four lines constitute a Canaanite text, then there are some oddities in the relative distribution of consonants. In another connection I have compiled a table of consonant frequency for West Semitic writings, including Sinai proto-alphabetic inscriptions, early Phoenician inscriptions from Byblos, Biblical Hebrew prose (Colless 1992:94-95, 100). There are variations for the positions of several signs, which can be explained by particular circumstances, but this is a useful tool for decipherment. The occurrences of each consonant are ranked over the whole five lines of the ostrakon, so that each letter should have at least one instance. Anomalies in the results will be addressed in the discussion.

TABLE OF CONSONANT FREQUENCY
        (BB=BYBLOS    SN=SINAI    HB=HEBREW    IS=IZBET SARTAH)

         BB SN HB IS
    ’    07  07  02  02 (9)
    B   04  02  08  08 (4)
    G   17  15  20  16 (2)
    D    16 17  14  08 (4)
    H    15 13  13  16 (2)
    W    13 19  05  16 (2)
    Z    19  11  21  13 (3)
    H.   08  12  15  08 (4)
    T.   21  22   22  08 (4)
    Y    09  20   07  06 (6)
    K    05  09   09  16 (2)
    L    01   05   04  04 (7)
    M   02  04   01  07 (5)
    N    06  03  12  02 (9)
    S    18  14  18   21 (1)
    ‘     11  08  10  01 (10)
    P    14  18  16  13 (3)
    S.   22  16  19  16 (2)
    Q   20   21 17  08 (4)
    R    12   10  03 13 (3)
    Sh   10  06  06  21 (1)
    T    03   01 11  04  (7)
There are over eighty letters in this sample, and the most frequent are ‘ayin and ’aleph. The letter ’aleph certainly scores highly in Hebrew, but ‘ayin is normally around the middle of the range in West Semitic texts. My explanation for this irregularity will emerge later in the discussion, but anticipating this it may be said that five of the ten occurrences of the sign ‘ayin are apparently used as word signs for ‘yn, "eye" (noun and verb). This adds five more instances of Y and of N; so Y moves from the lower end of the table to the high position it normally merits, while N moves to its expected level near the top.
    The consonants T. and Q are usually at the bottom of the scale, yet they both stand in eighth place here. If the four lines are merely made up of arbitrary signs and not sequences of words, then it may well be that the scribe felt the need to practise forming circle-letters (‘ayin, T., Q). However, if this is a coherent text, the presence of the word t.t. ("clay") in line 3, preceded by a word ending in T., will seriously distort the results in this limited sample. Similarly, by chance there are three words containing Q in my reading of the inscription.
    In Hebrew prose Waw occurs frequently as a conjunction, and it also functions as a marker for the vowels U and O; similarly Yod appears often as a vowel-marker, and as a verbal prefix. In the text of this ostrakon, there are no vowel-markers (matres lectionis), no verbs with y-, and perhaps a single case of w as "and". Other apparent abnormalities will be discussed in the next section.

PROBLEMATIC PAIRS
From the start it has been acknowledged that there are sorting problems with regard to several pairs of similar signs on the ostrakon, and here we need to consider G/P, B/L, Q/W, Q/R, D/R, M/Sh. 

G/P. Both letters have a vertical stem with an acute angular curve at the top. The only occurrences are in the first half of Line 4. The deciding factor is the length of the stem, apparently: P is shorter than G in the ’abgad line, and this distinguishes them in the text, even though the oblique line of the G goes to the right in one case and to the left in the other. Both signs appear in the second half of the frequency scale, as would be expected.     

B/L. Again we might invoke the relative height of each letter as a criterion: B is much smaller than L in line 5, but they are almost identical in shape, though the B is squarer and the L is rounder. Accordingly the likeliest B in the text would be the fourth sign in line 4, being as short as the adjacent P, and smaller than the following N, H., G, while the tenth letter would be L, standing as high as the H. In line 2, the ninth letter is fairly square and comparatively small, and could be B. In the second half of line 2, the presumed L is an inverted form of the L in the ’abgad line. This suggests that more than one scribe has been at work on this tablet, as assumed by Kochavi (1977:4) and Demsky (1977:18-19), though it may simply be a variant from the same hand. The interpretation I am proposing for the text allows four instances of B (one hypothetical, presumed to be obliterated, in line 2), and seven cases of L; this puts each of them in an appropriate position on the scale of frequency.

Q/W. When I first studied this document (Colless 1988:49-50) I thought I had found additional confirmation for my hypothesis that proto-alphabetic Q was a cord wound on a stick (Hebrew qaw, "line"); the Egyptian hieroglyph shows the string around the middle of the stick, with one end of the string protruding at the top, and the examples of Q in the Sinai inscriptions have at least one protrusion at the top (for example, 363 and 376; Colless 1990:5, 8, 9); the Q of the Arabian scripts has this shape, but in the Phoenician alphabet Q has been reduced to a circle on a stem, which was the original shape of Waw ("hook"), and so Waw had to open its circle at the top to distinguish it from Q (Colless 1988:36-37). There is a diagonal line emanating from the top of the Q in line 5 of this ostrakon, but editors disregard it, presumably seeing it as another of the cracks and scratches which litter the surface of the tablet, and I have accepted this as a possible explanation (Colless 1992:30). When I came to examine the inscriptions in lines 1-4, I decided that there are perhaps four examples of Q, none of which has an obvious projection at the top, and so the stroke on the Q in line 5 would be accidental; it projects too far above the circle to be an intentional part of the sign. As for W, there is only a stroke below the alphabet line at the point where W should stand, but M has apparently taken this position (as noted above). The one example of W in my interpretation is the second sign of line 4, indicating the conjunction "and"; its shape is not clear, but it may have an open top, as drawn by Cross (1980:8). 

Q/R. Having removed the top stroke from Q, how do we distinguish it from R, which is also shown by this scribe as a ring on a stem? In its pictographic form R was a human head (Colless 1988:50; 1991:30), and perhaps the head of R in this text is not a circle but an oval? However, as we grope around the four lines of text this rule does not seem to help us; the two instances I would select are the fifth characters in line 2 and 3 (the latter standing next to Q, in the sequence mrq).

D/R. The fourth sign from the end of line 4 looks like a typical R, with a neck-stem, and it is generally accepted as such; but I would argue that this supposed R could be D: it has been written small and it has sprouted a stem in the process (as on the Tel-Dan inscription from the 9th century), but it is a possible D. Cross has taken the fourth letter in line 1 as possibly R, but it is surely D. Another D stands beside the break-line in line 4.   

M/Sh. As noted above, M is apparently in the place where W is expected, after H. There is no real confusion between M (a wavy line representing water) and W (a hook), but Sh is a problem. The form of Sh in the alphabet line is not matched by any other character in the text, and following Dotan (1981:162) I accept all the wavy lines as M. The third letter in the first line could well be Sh, but it comes to a point at the centre, not a curve. The second and fourth characters in line 3 are quite small (and the first is very faint), but they qualify as M, while the large sign at the end of line 4 is a final flourish for M, rather than Sh. The phoneme M is usually at the top of the scale of frequency, but its presumed five occurrences here give it a fairly high position. In stark contrast, Sh stands at the bottom, rather than around the middle; this must be an accident, in a text where sh is not found as a relative pronoun or in any of the words chosen by the writer to express his ideas.

LETTERS AS LOGOGRAMS
In my attempts at reading Bronze Age alphabetic inscriptions I have stumbled on a principle that has escaped the notice of others: the pictophonograms of the proto-alphabet could function  not only as consonant signs, but also as logograms (Colless 1988:65; 1990:5). Each alphabetic sign could acrophonically represent the initial consonant of the word it denoted, or else the whole word. The characters were not simply phonographic but also logographic. The same seems to be true of the signs in the Canaanite syllabary, otherwise known as the Byblos pseudo-hieroglyphic script (Colless 1992:99); each glyph could be a logogram as well as an acrophonic syllabogram (initial consonant plus vowel); therefore it is in effect a logo-syllabary. 

Some examples of proto-alphabetic pictophonograms acting in this way will be offered here. The Gezer sherd is a fragment of a terra-cotta cult-stand, inscribed K (kap, "hand") N (nakhash, "snake") B (bayt, "house" or "temple"); thus the object is appropriately labelled as a "temple stand" (Hebrew ken, "stand"), with the B-sign standing for the whole word bayt (Colless 1991:31-32). In this instance the word bayt would presumably have a genitive case ending (-i), but vowels do not matter in this system of transcription. In this respect it resembles Egyptian hieroglyphic writing, where the consonants of the word depicted can be used, by the rebus principle, for any other word that has the same consonants, irrespective of the vowels. This idea may provide a clue for interpreting the Izbet Sartah text, where some problematic combinations of ’aleph and ‘ayin may be resolved by invoking this principle. 

In the inscriptions from the turquoise mines of Sinai, a water source bears the name ‘(yn) ’m, "Spring of the Mother (Goddess)" (02=377; Colless 1990:13); and another text (32=357; Colless 1990:37-39) speaks of "filling the jug with water of the spring of the Mother", ml kd m(ym) ‘(yn) ’m, where two signs are to be read as logograms. The pictograph for an eye is here employed in the extended sense of "water-well", and the wavy line for water represents the word for water (though this may simply be mu, rather than mayim). 

Presumably this practice would have ceased when the meaning of the signs was forgotten. I had previously thought that the Izbet Sartah text was too late in time for this principle to work, but apparently it was applied, even though the letters are no longer pictorial.

INTERPRETATION OF THE TEXT
An interesting attempt to find continuous meaning in the first four lines has been made by Dotan (1981:162-169); initially Kochavi (1977: 6) and Demsky (1977:19) tried this approach, but both eventually decided that it was simply an exercise in writing the letters of the alphabet.

Click on this photograph to see it in an enlarged form


Here is my interpretation of the text, with spaces between the proposed words, followed by a translation.

    (1) ’lmd  ’t[t]   ’‘(yn)
    (2) k ttn ‘(yn) rh. ’t b ’z[n b ‘]t. ‘l  t.t.
    (3) s.mq mrq
    (4) ‘(yn) w p b nh. g ’t l hd zqn  ‘t ‘(yn) ’‘(yn) l ’(lp) h.ld ‘lm

    (1) I am learning the signs. I am seeing
    (2) that the eye gives the breath of a sign into the ea[r by a styl]us on clay
    (3) (which is) dried (and) polished.
    (4) The eye and the mouth, with the resting of the voice, have come to the splendour of old age. See, now I shall be seen for a thousand lifetimes of the world.  

There are many variables opening up the floodgates of speculation and subjectivism: no vowels, no matres lectionis, no word separaters.

Thus, line 4 could well begin with the name of the writer (instead of the  subject  of the verb as "The eye and the mouth with the resting of the voice"): `wp bn h.g, something like `iwpai bin H.agai (both attested Hebrew names) . The -w- would be part of the root, not an indication of a vowel o or u.

In working out his own interpretation Dotan presented a great variety of possible words, and three of his rejected items will be taken up here: ’lmd, ’t ("sign"), t.t. ("clay"). The word "clay" did not suit his context of named people transporting food and clothing to other people; and yet the ostrakon itself is made of clay. The word ’t  was taken to be a verb, root ’ty ("come" or "bring"), rather than the noun ’ot, "sign"; but the tablet is covered with "signs", and I take that to be the signification of the word in lines 1 and 2; but in one case, in line 4, I do see 't as a verb ("has come"). As for ’lmd, the root lmd ("learn") is highly apposite in this setting, where a student is learning to write. 
    Line 1
    ’lmd. 1. p. sg. imperfect, "I learn", or "I teach" (D-stem). Of course, by the principle of the letter as a logogram or a polysyllabogram (showing consonants but not vowels, as in the Egyptian system) only the first two letters needed to be written, namely ’aleph and Lamed, producing ’l(md); but that would be too ambiguous, suggesting the word El, God.
    ’t. Hebrew ’ot, "sign", used in later Hebrew for letters of the alphabet, with plural ’wtywt; interestingly, Puech takes the tall scratch after the T as Y; in any case an obliterated T has to be posited, for the plural form, "letters"; and in fact it is there below the T, though only its cross bar is clearly visible; compare the double T in the second position in line 2 . (Another possibility is that ’aleph taw means "the alphabet".) The space between the two words has faint marks, possibly representing H, and therefore the definite article, which is not a feature of old Canaanite; but there seems to be no other case of such ha- determination in the text, even though my translation uses "the" several times.
    ’‘. This strange combination ('aleph-`ayin) is known in Aramaic, as the word for "tree"; but this is not applicable in Hebrew; here the ‘ayin could be understood as a logogram, or better a "polyconsonantogram" (or even an ideogram for "sight"?), while the ’aleph provides the verbal prefix; hence "I am seeing"; the verb ‘yn is found in Ugaritic and in Hebrew, meaning "to eye" or "look at".
    Line 2
    k. Hbr. ki, subordinating conjunction, "that".
    ttn. 3. p. sg. imperfect, root ntn, "give". Note that one T stands above the other T, next to the K, but all previous interpreters have only recorded one T at this spot.
    ‘(ayin). Logogram, "eye"; cp. other possible instances in line 4.
    rh.. Hbr. ruah., "breath, spirit, sense, mind"; here it presumably means the sound that the sign represents (the phonetic value of the letter).
    ’t. As in line 1, "sign", "letter" (of the alphabet).
    b ’z[n]. Preposition b, "in(to)", or perhaps l, "to"; Hbr. ’ozen, "ear".
    [b ‘]t.. Preposition b, here "by (means of)"; Hbr. ‘et., "stylus, pen".
    ‘l t.t.. Preposition ‘al, "upon"; Hbr. t.it., "clay" or "mud".
    Line 3
    s.mq. Passive participle, qualifying t.t., "dried"; used in Hebrew of dry breasts and dried grapes (raisins).
    mrq. Passive participle, qualifying t.t., "polished", Hbr. mrq, "rub, polish". Could this refer to the "wet-smoothing" process applied to the surface of the sherd (Kochavi 1977:4) before firing (Sass 1988:66)? There is no conjunction between the participles, but see w in line 4; a faint possibility would be s.mq-m rq, "dried and empty", with enclitic -ma as the conjunction, and the Hebrew adjective req, "empty". Note that the English translation does not need a conjunction: "polished dried clay" (as against "clay that is dried and polished").
    Line 4 
As  noted above, the writer's name might be found here as `wp bn h.g, something like `iwpai bin H.agai. Or else it could be: "The eye and the mouth of Ben Haggai have come to the splendour of old age.
    ‘(ayin). Logogram, "eye"; cp. line 2.
    w. Conjunction, "and", as in Hebrew, Ugaritic, Aramaic, Arabic.
    p. Logogram, "mouth"; even though the pictophonetogram has lost most of the bottom lip by this time, apparently its meaning had not been lost (Semitic pu, "mouth", Hbr. peh, a  monosyllabic word, so the logographic principle need not necessarily be invoked here).
    b nh.. Preposition b, "in" or "with", and nh., an infinitive or a noun, root nwh., "rest, repose, settle".
    g. Noun, Ugaritic g, "voice"; found on the Beth Shemesh ostrakon (Colless 1991:47).
    ’t. In lines 1 and 2, I have taken this combination to be a noun, "sign", but here it seems to be the verb "come", 3. p. pl. 
    l hd. Preposition l, "to", with noun hd, Hbr. hod, "splendour, majesty".
    zqn. Noun, cp. Hbr. zoqen, "old age" (Genesis 48:10), or possibly "beard", Hbr. zaqan. Either the writer is learning to be a scribe at a mature age, or else he has now become a man with a beard.
    ‘t. Either a noun, "time, season", or an adverb, "now". The writing is not easy to read here, and the following ‘ayin may have a tail and be Q, hence ‘tq, "move on" or "grow old", and Hif. "copy out, transcribe"; all these ideas have interesting potential for the line of thought being pursued here.
    ‘ ’ ‘ . This sequence apparently supports the view that the signs are being practised at random; but taking the first ‘ayin as a logogram (as in line 2 and at the beginning of this line), vocative noun, perhaps with 1. p. sg. suffix, and the next two signs as saying "I see" (as in line 1) we have "Now, my eye, I am seeing". But another interpretation is offered below.
    l ’ h.ld ‘lm. The l is a preposition. The ’aleph could be a  rebus (polyconsonantogram): instead of "ox" it is "thousand" (or even "has learned"). Finally there are two more words with "time" connotations: h.ld, Hbr. h.eled, "duration, lifetime" or "world", and ‘lm, Hbr. ‘olam, "perpetuity" or "world". Thus, "for a thousand lifetimes of the world", or the like.

The message obtained from the proposed interpretation for lines 1-3 seems fairly plausible: a student is learning to write the alphabet and muses over the art of writing on tablets. If line 4 is a continuation of this, we might suppose that the scribe's name is embedded in the long uninterrupted sequence of signs (bnh.g could be "son of Haggai", as Dotan has suggested), and the writer is saying that he will be remembered forever, because he has written something that will endure; but the translation "I am seeing" does not convey this idea. However, if we understand the first ‘ayin not as a vocative noun ("Eye!") but as an imperative verb ("See!"), and regard the ’alep-‘ayin combination as a Qal passive or Nifal ("I shall be seen"), then the expected meaning emerges before our very eyes: "See, now I shall be seen in perpetuity". 

REFERENCE BIBLIOGRAPHY

Colless, B. E. 1988, Recent Discoveries Illuminating the Origin of the Alphabet. Abr-Nahrain  26:30-67.
Colless, B. E. 1990. The Proto-alphabetic Inscriptions of Sinai. Abr-Nahrain 28:1-52.                                                           
Colless, B. E. 1991. The Proto-alphabetic Inscriptions of Canaan, Abr-Nahrain 29:18-66.
Colless, B. E. 1992. The Byblos Syllabary and the Proto-alphabet. Abr-Nahrain  30:15-62.
Colless, B. E. 1993. The Syllabic Inscriptions of Byblos: Text D. Abr-Nahrain 31:1-35.
Colless, B. E. 1994. The Syllabic Inscriptions of Byblos: Texts C and A. Abr-Nahrain, 32:59-79.
Colless, B. E. 1995. The Syllabic Inscriptions of Byblos: Texts B, E, F, I, K. Abr-Nahrain 33:17-29                                    
Colless, B.E., 1996-1997, The Syllabic Inscriptions of Byblos: Miscellaneous Texts,
Abr-Nahrain 34:42-57.                  
Colless, B.E., 1996, The Egyptian and Mesopotamian Contributions to the Origins of the Alphabet, in Cultural Interaction in the Ancient Near  East, ed. Guy Bunnens, Abr-Nahrain Supplement Series 
(Louvain)  67-76.                                                
Colless, Brian E.,
1998, The Canaanite Syllabary, Abr-Nahrain 35: 28-46.  

Cross, F.M. 1980. Newly Found Inscriptions in Old Canaanite and Early Phoenician Scripts. Bulletin of the American Schools of Oriental Research 238:1-20.
Demsky, A. 1977. A Proto-Canaanite Abecedary Dating from the Period of the Judges and its Implications for the History of the Alphabet. Tel Aviv 4:14-27.
Demsky, A. 1986. The ‘Izbet Sartah Ostracon Ten Years Later, in . I. Finkelstein, ed.,‘Izbet Sartah: An Early Iron Age Site. Oxford: 186-197.
Dotan, A. 1981. New Light on the ‘Izbet Sartah Inscription. Tel Aviv 18:160-172.
Garbini, G. 1978. Sull'alfabetario di ‘Izbet Sartah. Oriens Antiquus 17:287-295.
Kochavi, M. 1977. An Ostracon of the Period of the Judges from ‘Izbet Sartah. Tel Aviv
Lemaire, A. 1978. Abécédaires et exercices d'écolier en épigraphie nord-ouest sémitique. Journal Asiatique 266:221-235.
Naveh, J. 1978. Some Considerations on the Ostracon from ‘Izbet Sartah. Israel Exploration Journal 28:31-35.
Puech, E. 1986. Origine de l'alphabet. Revue Biblique 93:161-213 (esp. 170-172).
Sass, B. 1988, The Genesis of the Alphabet and its Development in the Second Millennium B.C. Wiesbaden. 33:17-29. 4:1-13.


Brian E. Colless, PhD ThD (b. 1936), Massey University, Palmerston North, New Zealand.