Canonization in the Public Arena: an Analysis of the Events

The events described here are an example of the social location of non-institutionalized aspects of literary canonization.Footnote ¹¹ Institutionalized canonization implies an official and normative definition of a literary corpus as the canon. It establishes the boundaries of the canon, defining which works are included and which are excluded. In many cases it also secures the stabilization of the text, as it defines various mechanisms for that purpose. Institutional canonization is frequently accompanied by a foundational narrative about the origin of the canon, either by presenting its legendary, sacred, or revelatory source or by documenting its (imaginary) public affirmation by recognized bodies or key historical figures.

However, the process of canonization is not only of an official nature. It also involves various means through which the canonical status of a work is manifested, performed, and mediated to the public. Canonical works are subject to a continuous process of interpretation that ensures their relevance for society; their importance is marked in public by various means, such as the physical appearance of the books. These works are generally discussed in a manner that informs people about their importance. Their content is frequently mediated through public performance or educational means, as these works are taught, translated, dramatized, and interpreted.

Similar forces and social dynamics also act to establish the cultural value of literary corpora or works that are not subject to full and institutionalized canonization, yet enjoy certain cultural importance. Aggadic midrash is a good example, being subject to some aspects of canonization, but not granted the same canonical status of the Hebrew Bible, or even the Mishnah or the Babylonian Talmud. The events presented in our fragment both reflect the role and status of aggadic midrash and take part in establishing its canonicity. First, we learn that the people involved in this process considered aggadic midrash to be an authoritative source. In the first event an argument from the midrash was accorded authority. It was used to justify the content of the piyyut, and was accepted by the individuals involved. It is important to notice the two stages in the dialogue between the cantor and his opponent. First, the opponent challenged the cantor for not identifying Elijah with Phineas. The cantor defended this view by referring to the biblical verse in 1 Chronicles. The answer did not satisfy the opponent—the biblical verse was not considered by the protagonist to provide sufficient support. It was at this moment that the cantor suggested the sources in Bereshit Rabbah. The rabbinic source was accepted as evidence.

This short dialogue is emblematic of the complicated relationship between the biblical and rabbinic canons. On the one hand the Bible is considered the primary canon and holy text; on the other hand the rabbinic tradition is the authoritative source for biblical interpretation. This creates an interesting situation. The biblical text is the primary canon of rabbinic Judaism. Its authority and importance are manifested in many ways, including by rabbinic interpretations. At the same time the biblical text is also dependent on its rabbinic interpretation. It is only through the rabbinic tradition that the meaning of the biblical text can be established. To that extent, the rabbinic interpretation holds greater authority than the biblical text per se. In Halbertal's discussion of the tension between the two corpora in the curriculum he quotes a late reflection on the topic:

This sort of dynamic might be manifested in the events documented in our Genizah fragment. The biblical verse can be established only on the basis of its rabbinic interpretation. Halbertal's discussion justly focuses on the Babylonian Talmud as the most canonical rabbinic work. In the events documented in our fragment the authoritative rabbinic text derives from the aggadic corpus—Bereshit Rabbah— and not from the Babylonian Talmud. The aggadic text is granted authority usually ascribed to the Talmud.

A slightly different rhetoric of authority is implied in the detailed discussion of sources that follows the narrative portion of the fragment. Here we find clues pointing to the common hierarchy between the Babylonian Talmud and other rabbinic texts, according to which their authority is dependent on the verification of their view in the Babylonian Talmud, as stated by Amram bar Sheshna Gaon: “Every baraita which the Amoraic rabbis do not contest in the Talmud—the halakhah follows them”.Footnote ¹³ In the first stage of the discussion the author announces that the text from Bereshit Rabbah persuades him that Elijah is not Phineas, and this opinion is further established for him when he finds a support for it in the Talmud (ENA 637.4b, lines 6–7). Just as Rav Amram bar Sheshna asserts that non-Talmudic halakhic texts may be approved if they can be supported by the Talmud, our text turns to the Talmud to approve the aggadic statement. The language used by our author here echoes geonic discussions that question the reliability of aggadah. Our text reads ותאכד ענדי דלק מן קולהום פי תלמוד (and this view became established for me on the basis of their statement in the Talmud …);Footnote ¹⁴ a similar expression is found in regard to the authority of aggadah in a statement by Sherira Gaon: והנכון מהם מה שמתחזק מן השכל ומן המקרא (and the true [statements] among them [the aggadic statements] are those that can be validated on the basis of logic or on the basis of the Bible).Footnote ¹⁵ These statements are examples of the cultural work of the language of canonization. The same mechanism is at work in both statements. The authority of aggadic literature as a body of knowledge and traditions is dependent on one's ability to validate it on the basis of another source of knowledge. Sherira Gaon dictated that the validating bodies of knowledge are the Bible and logic; the author of our text turns to the Talmud. We should also note that cognate words are used here to mark the relations between the different bodies of knowledge (the fragment: תאכד [became established]; Sherira: מתחזק [can be validated]).Footnote ¹⁶

To a certain extent we can see the relationship between aggadah and the Talmud as a merger of the two geonic principles quoted above. The three statements follow the same pattern: the authority of an external, less canonical body of knowledge is subject to validation by a higher body of knowledge. Yet, there is an interesting shift.

As we shall see below, our text is later than the geonic texts. It reflects an interesting process in which a model that was used for establishing the authority of the Babylonian Talmud over external materials in the context of halakhah is now exercised for marking the superiority of aggadic traditions in the Talmud. As such it demonstrates the process by which the cultural stance of the Babylonian Talmud was expanded from its initial role as a normative canon to its later function also as a formative canon, as brilliantly explained by Moshe Halbertal.Footnote ¹⁷ This process is not only an expression of the greater importance granted to the Talmud, but also of the increased importance of aggadic materials in this Egyptian milieu. In that sense, when the author of our text applies the model of examining an external text against the Babylonian Talmud, previously invoked only for halakhic texts, he raises the degree of importance of the aggadah, bringing it closer to that of the halakhah.

So far we have noticed three different reflections of canonicity in our fragment: first, the establishment of rabbinic authority over that of the plain meaning of the text. This was shown in the presentation of the aggadic midrash as evidence, while the biblical text alone was not accepted as a sufficient proof text. Second, rabbinic authority was granted to a work of aggadic midrash in a manner that seems to equate its importance with that of the Babylonian Talmud. Last, we analyzed in detail the mechanism that preserves the hierarchy between talmudic and aggadic sources, reflected in the need to support the aggadah with a text from the Talmud.

The importance of the incidents documented here goes beyond the documentation of common perceptions about the canonical relations among the various works. What we see here is not only how people rated the different literary corpora involved, but also an example of the social context within which these hierarchies were communicated.Footnote ¹⁸ These events demonstrate how literary preferences are encoded in and disseminated through public behavior. The events took place in the synagogue, in a public arena. It is quite possible that other people listened to the conversation between the cantor and his opponent. They heard the cantor basing his argument on biblical verses; they heard that his opponent (who was possibly his rabbi; the cantor addressed him with the title “יא רבי,” [oh, rabbi]) was not satisfied with it; they heard the name of the work Bereshit Rabbah, so even if they were laymen who never studied the work, at least they became aware of the existence of the book; they heard the opponent accept the argument that was based on the midrash. Such incidents are part of a wider means through which agents of the canon spread their canonical preferences among the public.Footnote ¹⁹ After being persuaded by the aggadic text, the opponent, who is also the narrator of our text, brings it again into the public arena, in the second event. According to the narrator, he challenged a different cantor who performed a piyyut expressing the common view that Elijah is Phineas. He uses this event as an opportunity to recall the earlier incident. The second event became another moment of social interaction that enabled the public dissemination of the hierarchical relationship between the various texts.

In that sense, it is important to note that the quotations of the Bible and the midrash, and the acceptance of the midrash, are presented in the account of the event itself, while the further support from the Talmud is presented later on in the text, as part of the scholastic discussion that follows. Here the author addresses the audience with his written piece, and offers further arguments in favor of the view that he accepted in public. The social impact of a written text is not the same as that of an event in the synagogue. In these events we can see the cantor and his opponent as agents of the canon, and the events as social opportunities in which they communicate to the wider public the sets of literary preferences they hold and the different cultural values they ascribe to the various texts involved.

The following analysis will use this theoretical framework to assess the canonical relationship of midrash and piyyut demonstrated in the events discussed above. Havdalah liturgical poetry is not perceived in our text as an authoritative source of knowledge. Rather, it is seen as promoting a view expressed by the individual cantors. The cantors are asked to defend and justify the claims hinted at in the texts they performed. The midrash, on the contrary, is understood as the source of correct knowledge.

It is important to recall that a similar relationship between midrash and piyyut is also found in the literary genre of piyyut commentary. Here too, aggadic texts are perceived as the sources of knowledge performed in liturgical poetry.Footnote ²⁰ In that sense these commentaries propound the same hierarchy implied in the events presented in our fragment. However, there are two significant differences between the appearance of this kind of argument in our fragment and in the commentaries, apart from the different geo-cultural and historical contexts in which they appear.

The first difference refers to the social setting. Commentaries on piyyutim were written works, part of a wider intellectual project of the scholarly elite. The role of the midrash as the source for the piyyut was presented here only to scholars. The Genizah fragment presents events, and as such it documents the communication of this model to the wider public. Second, when this argument is presented as part of a commentary on the piyyut, it also functions as another canonical mechanism, one that marks the importance of piyyut. Commentaries are written on canonical works; even the physical appearance of the texts on manuscripts might mark the importance of the text commented upon, in its large, square font and punctuation, in comparison to the smaller, cursive font of the commentary. The existence of the genre of commentaries to piyyut groups this literary form with the Bible and the Talmud; while the aggadic midrash is quoted here as the source of the knowledge, the context marks it as secondary, as a supportive means for the interpreted piyyut. The use of the same model of argumentation in the public event in the synagogue in order to either defend or challenge the text of the piyyut is different from the use of this model in written commentaries.

It is also important to note that the piyyut discussed in our fragment was composed in the period of the Genizah, and would have had only a relatively marginal place in the liturgy. Such recent compositions were constantly being composed and brought to synagogues by ḥazzanim; the congregants likely felt free to critique these new compositions in a way they wouldn't have done with more ancient texts. In contrast, the scholarly commentaries discussed previously explicated piyyutim whose creation was more or less coeval with that of the midrash, such as Yannai's piyyutim from the Byzantine era. It is possible that, unlike more contemporary compositions, Yannai's works were perceived as being as authoritative as the classical midrashic works.

The analysis of the events quoted above demonstrates how social interactions play a role in the establishment of the canonicity of the various literary works and genres involved in them. The events recorded in the Genizah fragment reflect a growth in the cultural value ascribed to aggadic midrash, here exercised in a manner that marks it as more authoritative than piyyut. We shall now turn to two additional discussions that will help us to further establish the nature of these events as emblematic of the growing importance granted to aggadic midrash and to the lack of authority of Havdalah poetry. First, identifying the literary and historical context of the events recounted in our fragment will help us to locate them at the moment of a change in the perception of aggadic midrash in the East. Second, a survey of Havdalah poetry will demonstrate the lack of stability of these texts at the time, and will suggest that this textual instability was also motivated by the very same issue that was raised in the events: the identification of Elijah with Phineas and his presentation as the zealot who killed the adulterous couple. Textual stability is another aspect of the canonical status of a literary piece. The instability of Havdalah poetry and its susceptibility to changes inspired by aggadic traditions is another field in which the canonical relations between the two genres are exercised, manifested, and established.

The Literary and Historical Background of the Events

Let us turn to the historical and literary setting of these interesting incidents. At the core of the events is the debate about the identification of Elijah with Phineas. Where did the different views on this question come from, and where and when were they held? What is the historical context in which a cantor may arrive at a synagogue and perform a piyyut that included the view that Elijah is not Phineas, against the view that was commonly held by the audience? The identification of Elijah with Phineas might have its roots in early rabbinic traditions,Footnote ²¹ as hinted at in patristic literature and in pseudo-Philo, and also reflected in the fact that 1 Kings 18–19 is the haftarah for parashat Pinḥas (or for the Palestinian sidra for Numbers 25:10–26:51).Footnote ²² However, this identification is explicitly found only in later midrashim, including a text of Yelamdenu only preserved in later midrashic collections (and very partially in our Genizah fragment),Footnote ²³ Pirkei de-Rabbi Eliezer, and a text of Midrash Tehillim only preserved in another Genizah fragment.Footnote ²⁴ It is also supported by the identification of Elijah as a priest in the Babylonian Talmud and in Midrash Mishlei.Footnote ²⁵ The view that Elijah is not Phineas, including the identifications of Elijah with the tribes of Gad or Benjamin, first appears in Bereshit Rabbah, quoted in our Genizah fragment. It is repeated in Midrash Tanḥuma, and in Seder’ Eliyahu, and is also quoted by the Tosafists from an unknown geonic responsa.Footnote ²⁶ In addition, these views are frequently repeated in the borderline genre of midrashic commentaries/anthologies such as Lekaḥ Tov, Sekhel Tov, Midrash Aggadah and Aggadat Bereshit.Footnote ²⁷

By the end of the first millennium, the identification of Elijah as Phineas had become the dominant view, especially in the East and in the Islamic domain. This view was spread further by its place in the liturgy, as it prevailed in piyyutim for Havdalah; it also appears in Targum Pseudo-Jonathan, which was probably a conduit for wider distribution of midrashic traditions.Footnote ²⁸

This literary mapping is an important clue for the historical setting of the events recounted in our Genizah fragment. It explains the surprise provoked by the identification of Elijah with the tribe of Benjamin in the Judeo-Arabic-speaking community. Despite the presence of this view in known texts it was not common knowledge nor part of the collective memory in the East. The cantor who provoked the first debate was a stranger (“came to our city”) and he performed an imported piyyut. The piyyut presented a view that was not current in the cultural milieu of the audience: Elijah is not Phineas. We can trace the journey of this view: we can identify its Palestinian origins in Bereshit Rabbah; its continuous vitality is demonstrated in quotations in the late genre of midrashic anthologies/commentaries. This implies that it was known in Byzantium and Italy. These works are an important link in the chain of tradition from the Palestinian rabbinic milieu, through the Byzantine and Italian bridges to Ashkenaz, France, and Provence. Therefore it is likely that they played a role in establishing familiarity with the view that Elijah is not Phineas in these Jewish communities under the Christian domain.Footnote ²⁹

The next stage in the diffusion of this opinion was its move from the limited academic context of written commentaries to the public sphere of weekly performance of the piyyut. The evidence shows that this happened in Christian Europe. Other than the citation in our Genizah fragment, there is no documentation of the piyyut יה חזקי (God, my strength), which includes the identification of Elijah with the tribe of Benjamin in the Genizah source. Neither have I found any reference to it in other Eastern manuscripts and texts. The piyyut's style and content represent strong grounds for arguing that the piyyut is of Provençal or French origin. Furthermore, the piyyut is found in the later additions to Maḥzor Vitry.Footnote ³⁰ The chain of events presented in our fragment likely began when a cantor recited a French or Provençal piyyut that was based on a view that was popular in Europe, but much less accepted in the East.

Another clue to the literary and geo-cultural background of the events is found in the works of the French Tosafists. The strophe quoted in our Genizah fragment was also quoted by Rabbi Samson of Sens, in the name of R. Jacob (Rabbenu Tam). As in our Genizah fragment, Rabbenu Tam identifies Elijah with the tribe of Benjamin on the basis of the text in Bereshit Rabbah.Footnote ³¹ Not only did the cantor present an imported piyyut, of French or Provençal origins, he also defended the content of the piyyut with an argument that was previously known in French contexts.

As already stated, after the presentation of events, the Genizah fragment includes a long discussion about the identity of Elijah. An explicit Provençal context of the identification of Elijah with Phineas is also evident in the fragment: the text quotes a certain “R. Isaac of Narbonne of blessed memory” who also refuted the identification of Elijah with Phineas.Footnote ³² His view is preferred over that of one “Shmariah Ha-Kohen of blessed memory”, who was probably a local rabbi.Footnote ³³ The fragment also presents an argument against the identification of Elijah and Phineas that was common in Ashkenazic and French circles.Footnote ³⁴ If this is indeed the literary and historical context of the events presented in our Genizah fragment,Footnote ³⁵ it sheds light on the growing importance attributed to aggadic midrash. It was not only specific traditions that were imported from the Jewish communities of the Christian West to a synagogue in the Arabic-speaking East. The fragment also reflects the importation of cultural values, namely the greater importance that was granted to aggadic tradition by Western scholars.

The Adaptation of Piyyutim

The series of debates about the identity of Elijah provides us with a context in which to understand the development of several Havdalah piyyutim known to us from the Genizah. The first piyyut quoted in our fragment belongs to a wider group of Havdalah piyyutim. A survey shows the instability of these texts, and especially their relation to the question of the identity of Elijah. The same issue that was at the core of the debates in our fragment was also subject to changes and variations. Surprisingly, there is another Havdalah piyyut that refers to the same biblical event in exactly the same place in the alphabetical acrostic. It too suffered from textual instability at a specific point in the poem. The textual development of this piyyut, אותותינו התמהמהו (The signs of our [redemption] are delayed) by Rabbi Yehudah Ha-Levi, is presented below.Footnote ³⁶

A—The genre of Havdalah piyyutim which begin with איש (a man)

The other piyyut mentioned in our story, איש אָזוּר אֵזוֹר עור (a man, girt with a girdle of leather), is well attested both in the European and the Eastern milieus. The piyyut exists in numerous versions. Before presenting the conclusions of the survey of these piyyutim, we should consider the liturgical context in which they were commonly found. Havdalah piyyutim such as these were one component in a wider liturgical category of compositions for the Havdalah service in the synagogue,Footnote ³⁷ a genre that has not been sufficiently studied.Footnote ³⁸ We do not have any early documentation of these compositions. Fleischer already noted that there is no reference to Havdalah in the early Palestinian siddur found in the Genizah, which he published.Footnote ³⁹ It is probable that the Havdalah compositions developed only around the turn of the millennium, as a replacement for earlier models of public poetical marking of the Havdalah in the synagogue, including poetic Shema blessings known from Sa‘adiah's siddur, rare Genizah fragments, and shiv‘atot havdalah for the Amidah, performed when moẓa'ei shabbat falls on the eve of festivals.Footnote ⁴⁰

The wider liturgical unit of Havdalah developed into a complex structure. In addition to the main benedictions defined in classical rabbinic works, Havdalah liturgy includes verses of comfort and salvation, blessings and petitions, magical formulas, and piyyutim, most of which relate to Elijah.Footnote ⁴¹ These Havdalah liturgies vary significantly both in terms of structure and content. This testifies to a long period in which these liturgies were not fixed. It is in this grouping that we find the piyyut discussed in our Genizah fragment, and as we shall see, it shares the textual instability common to Havdalah piyyutim in general.

Our piyyut, איש אזור איזור עור (a man girt with a girdle of leather), also belongs to a sub-genre of Havdalah piyyutim that share a common form and present aspects of the figure of Elijah. Each stich begins with the word איש (a man) followed by a word that contributes to an alphabetical acrostic:

The repeated word איש (a man) probably alludes to the biblical description of Elijah as the man of God איש האלהים)) (1 Kings 17:18).Footnote ⁴²

There are eight different piyyutim of this sub-genre in the Genizah, and an additional piyyut known to us from other sources. These piyyutim are anonymous, given the alphabetical acrostic. Indeed, it is doubtful whether the very concept of “author” is relevant here, as the distinction between an adaptation and an independent work is hardly possible in this case.Footnote ⁴³ Each line in the piyyut refers to an event or phrase related to the figure of Elijah. The reference is usually to the biblical account of Elijah, though it occasionally draws on rabbinic literature as well. Among these we find allusions to the figure of Phineas as well as references to Malakhi 3. Though the order and exact phrasings of these allusions vary, in most cases the piyyutim seem to be based on a pre-existent pool of allusions and phrases. For example, piyyutim 1 and 3 in the appendix share at least eleven motifs. In some cases the shared motifs are merely allusions to the same biblical reference but in many cases they are phrased in an approximate manner, implying direct borrowing from one variation to another, or the use of a common source, or at least an echo of the same tradition.

Some of those piyyutim were extremely popular and as such they are preserved in many manuscripts and fragments. The variants of these piyyutim illustrate some trends and preferences in the portrayal of the figure of Elijah. In a few cases they reflect an effort to avoid the presentation of the harsh and zealous aspects of Elijah:

It is very plausible that different perceptions of Elijah were at work here. The identification of Elijah with Phineas, and especially the portrayal of Elijah-Phineas as killing the adulterous couple, amplify the representation of Elijah as a zealot. The effort to minimize the zealous aspects of Elijah's personality may have been a factor in the shaping of the two unique piyyutim that avoid the portrayal of Phineas's killing of Cozbi and Zimri (piyyutim 3 and 6 in the appendix). One piyyut uses an ambiguous expression that either implies the killing of the prophets or that of the couple, or intentionally implies both (5 in the appendix). Of special importance in that context are two rare versions of this text. T-S NS 107.28 (4 in the appendix) is a unique variation of the piyyut איש אזור (a man girt) that provoked the debate in our fragment. Not only does this version avoid the portrayal of the killing of the adulterous couple, it also lacks any reference to Phineas, except for the last line, where it mentions the covenant of peace, pointing out an opposite aspect of both figures. ENA 586.7 is another unique piyyut of this genre, and it avoids any reference to the figure of Phineas. It is possible that this unique version was inspired by the very same issue that provoked the debate documented in our Genizah fragment: the wish to avoid the identification of Elijah with Phineas, and especially the portrayal of the killing.

We have seen that Havdalah piyyutim of the איש (a man) genre, such as the ones we have just examined, were very popular. Many piyyutim were written in this form, differing from each other, yet drawing upon a shared reservoir of motifs. Different motivations may have contributed to this development. It may have been necessary to present new materials to the audience, a motivation that was viewed by Fleischer as one of the main influences in the development of Hebrew liturgical poetry.Footnote ⁴⁴ In other cases we may assume that stylistic constraints were at work: the need to simplify a piyyut that was used in such a common liturgical event as the Havdalah ceremony, the move from blank verse to rhyme, or from strophes of three lines to a fixed rhyme covering all lines of the poem, and so on. We should also take into account that conceptual motivations were at work here. Some liturgical changes might be related to the community's reluctance to portray Elijah as a zealot. It is possible that the same motivation influenced the identification of Elijah with Phineas, and especially the explicit portrayal of the act of killing the adulterous couple. This might be the reason for the choice of another event for the letter tet, especially in the case of the version documented in T-S NS 107.28, that hints at Phineas only by referring to the covenant of peace. More important is the case of ENA 586.7, איש אשר היה מלאך מעיקרא (a man who was an angel from the very beginning), which leaves no traces of the figure of Phineas.

If indeed this assessment is correct, and the identification of Elijah with Phineas did play a role in shaping the different versions of the piyyut, then the present survey provides another angle on the debates documented in the Genizah fragment. This is not to suggest an simplistic cause-effect relation, namely that the events documented in the fragment shaped the perceptions regarding Elijah, and that the poems were changed accordingly. The debate about the identity of Elijah was much wider than the events presented here, as we saw in the survey above. As indicated above, the debate dates back to early rabbinic works and contemporaneous non-rabbinic works. The polemical nature of this discussion is prominent in later midrashic collections in which rabbis' targets were probably Karaites or Samaritans.Footnote ⁴⁵ It is further developed in medieval biblical commentariesFootnote ⁴⁶ and the borderline genre of midrashic anthologies and biblical commentaries. What we have here are probably two different attestations of the debate: one is a documentation of vivid events in a synagogue; the other is the influence of the debate on the versions of piyyutim.

For the student of the reception of literary genres, the significance of the connection between these two distinct sets of evidence goes well beyond the specifics of the debate about Elijah. The narrative of the events presented in the Genizah fragments teaches us that midrashic works were considered to be primary biblical sources, while piyyutim had no authoritative value. A similar perception of the status of piyyutim is implied by their textual instability. The degree of a text's stability is also a measure of its canonical role: authoritative texts enjoy a higher degree of stability. The instability of Havdalah piyyutim demonstrates no sense of authorship, or of importance ascribed to a specific text. The text of the piyyut was open to improvisation and manipulation, possibly on the basis of different views in the midrash.

B—The textual evolution of אותותינו התמהמהו (The signs of our [redemption] are delayed) by Rabbi Yehudah Ha-Levi

The findings of this survey of איש אזור (a man girded) piyyutim are not sufficient in and of themselves to demonstrate that the specific instability around the letter tet is related to the question of Elijah's identity. Since all components of the poem were subject to improvisation we cannot dismiss the possibility that the motivation for the variations in the letter tet was not the specific question of the identity of Elijah, but rather stylistic considerations or the wish to avoid depicting him as zealous. However, the textual evolution of another Havdalah piyyut, אותותינו התמהמהו (The signs of our [redemption] are delayed) by Rabbi Yehudah Ha-Levi, supplies us with a better case for claiming that the identity of Elijah, and especially the portrayal of Phineas's zealous killing, was a source of unease and anxiety, which played a role in the development of new versions.

Recently Binyamin Bar-Tikva has pointed out that this alphabetical piyyut is commonly preserved in a shortened version, up to the letters zayin–ḥet (see also Davidson Aleph 2077).Footnote ⁴⁷ He has also described three full versions that contain the rest of the alphabet:

1. (1) a version found in a Genizah fragment, in which the first eight letters of the acrostic were in one hand, and the letters following tet were completed with another hand;

2. (2) a version documented in Yemenite manuscripts;

3. (3) a late completion of the abridged version by Rabbi Shmu'el Ha-Rofe.

Bar-Tikva discusses the authenticity of the three full versions and points out the quality of the Yemenite version in terms of its style and language. However, the late date of the Yemenite manuscripts leads him to prefer the Genizah fragment as the authentic text.

Of particular significance for our purposes is the fact that the shortened version stops just short of the letter tet, the same letter that provoked the debate in our Genizah fragment. The reason for this may be connected to the contents of the tet strophe in the unabridged versions of the story. They appear as follows:

Here, again, the Yemenite version uses the letter tet to allude to the zealous act of Phineas, while this reference is missing in the two other versions. The Genizah version, as well as that of Shmu'el Ha-Rofe, does not include any allusion to Phineas and therefore avoids the identification of the two figures. It is possible that in the abridged versions the same mechanism was at work: the original version was that of the Yemenite manuscripts, which portrayed the zealous act of killing the adulterous couple. The poem was shortened before the letter tet, because of reluctance to present the identification of Elijah with Phineas, and especially to portray Elijah as performing the killing.

The argument for the antiquity of the Yemenite version is based on some textual observations that should be added to Bar-Tikva's helpful survey: The version of the Yemenite manuscript is also found in Maḥzor’ Aram Ẓova, as recently noted by Shulamit Elizur,Footnote ⁴⁸ as well as in one Genizah fragment, Mosseri V 138.1. Now, since this version has much earlier documentation than the seventeenth-century Yemenite manuscripts, the main argument raised against its authenticity is undermined. It is much more reasonable to assume that this version is indeed the original version of Ha-Levi, as reflected in its style.Footnote ⁴⁹

Now we can suggest that Ha-Levi wrote his poetry on the basis of the common and prevailing view in his time and place that identified Elijah with Phineas, using a motif already documented in earlier Havdalah poetry, such as the work of his predecessor Ibn Gabirol.Footnote ⁵⁰ The text was then shortened so that it ended right before the letter tet. The abbreviated version did not include this portrayal of Elijah as Phineas. Later poets completed the shortened version producing new versions that avoided the identification of Elijah with Phineas. If indeed the debate about the identity of Elijah was a factor here, it might testify to an earlier influence of the debate on Havdalah poetry, as the shortened version was documented in the early Diwan of Rabbi Yehudah Ha-Levi, redacted by Ḥiyya in the eleventh century (assuming that the text of the Diwan is an authentic reflection of the first redaction. One should also consider that the poem was shortened by a later scribe who copied the Diwan).Footnote ⁵¹

The contribution of these findings concerning Ha-Levi's poem is twofold. First, they offer a model for understanding similar evolutions of Havdalah poetry. The text of the piyyut was changed exactly at letter tet, the point at which Elijah was identified with Phineas and was portrayed as slaying the adulterous couple. As such, it strengthens the suggestion that the issue of the identification of Elijah and his zeal was at work in both cases. Second, it further demonstrates the instability of the texts of piyyut and their susceptibility to modification. Here we find a process of revision in a text of a highly celebrated poet. His text was not considered authoritative by those who were involved in the transmission of his poem, and was thus subject to modification.