On the Edge between Literacy and Orality

Manuscripts and Performance of the Zoroastrian Long Liturgy

Oral Tradition, 35/2 (2022):211-50 

1. Today’s Performance of the Long Liturgy

The Long Liturgy (later LL) is the main Zoroastrian ritual. The central part1 consists of the recitation of the Gāϑās and the Yasna Haptaŋhāiti, a series of texts in Old Avestan, an Iranian language older than the one of the rest of the liturgy. In antiquity, this recitation was simultaneous with the pressing of a plant (known metonymically as haōma, “juice”), and an animal sacrifice, a small portion of which was then offered to the fire. In modern times, the sacrifice and the meat offering have been abandoned. This central section is preceded by an introduction during which the priest acquires the necessary qualification for the sacrifice, among others by drinking a previously pressed haōma. The whole is closed by the demand for a reward for the performance consisting basically of offerings to the fire (firewood and the dry residue of the pressing of haōma) and the water (a libation containing the haōma prepared in the central part of the ceremony).

A change in the self-understanding of Zoroastrians in Iran has shifted the focus from the ritualistic approach to a religious identity defined more in ethical and subjective terms (Mazdapour 2004; Ringer 2011). Furthermore, the celebrations in pilgrimage centers have increased vastly to the detriment of the rituals performed in the fire-temples. Accordingly, the importance of this liturgy has drastically diminished in the last fifty years, especially in Iran. The neglect, simplification, or substitution of rituals has had such an impact on their performance in Iran that we can affirm that they have almost completely disappeared. The text of the Yasna is sometimes read, mostly on the first day of the seasonal festivals (the gāhāmbār). The priests sit in front of a table and read the Yasna from a book, mostly from the description of the liturgy by Anklesaria (1888). The text does not even need to be recited in full, but may be abbreviated in different ways. Ritual actions have been reduced to a minimum.

In India, ritual practice has followed the same trend, albeit not to the same extent. Only the Yasna and Vīdēvdād (the latter mainly within the context of the ceremonies for the preparation of the Nērangdīn—consecrated bull’s urine used for purification purposes) continue to be regularly performed. In contrast to what happens in Iran, the texts are not just read from a book, but recited from memory within the context of a ritual performance that includes similar

1 On the structure of the LL, see Cantera 2020c.


ritual actions to those described in manuscripts that are several centuries old. Kotwal and Boyd (1991) have provided a description of the standard performance, the Yasna. Furthermore, several videotapes record parts of the performance of the Yasna as presently performed in India.2

2. The Sasanian Avesta: Deconstructing the “Sasanian Archetypes”

The Avestan texts preserved on manuscripts (the oldest extant ones dating back to the thirteenth century) are basically the same that are still being used in the LL and other Zoroastrian rituals. The exceptions concern mainly (1) certain variants that are no longer performed in modern practice, such as the Vīštāsp Yašt, (2) parts of former rituals that were no longer performed at the time of the extant manuscripts (from the end of the thirteenth century on), such as the Hādōxt Nask, and (3) two meta-ritual treatises, Hērbadestān (on the priestly teaching) and Nērangestān (a set of instructions for the proper performance of the rituals). Nowadays, the relationship between extant manuscripts and the Avesta as described in Pahlavi literature and the Persian Revāyats is problematic.

The eighth and ninth books of the Dēnkard and Chapter 28 of the Anthology of Zādspram, two works from the second half of the ninth or even tenth century, describe an Avesta that is quite different from the Avestan texts we encounter in the manuscripts.3 They describe an Avesta of twenty-one books (nask) that were arranged in two different ways: according to the twenty-one words of the Ahuna Vairiia, or in three groups of seven books.4 Their compilation is the result of a Sasanian rescue program instigated in their time by a Parthian and several Sasanian kings.5 This rescue program was necessary, as Alexander’s conquest had supposedly destroyed the written book of the Avesta, and seriously compromised the tradition.

Western scholarship has found difficulties in explaining the exact relationship between this “Great Avesta” and extant Avestan texts. Firstly, it seems that at least three-quarters of the original texts have been lost (Geldner 1896a:20). Secondly, the extant texts do not appear in the same arrangement as described in the Dēnkard. Only one of the twenty-one nasks has been preserved in the same arrangement as the one described in Pahlavi literature, the Vīdēvdād, with a further three in a similar arrangement, the Bayān Nask,6 the Vīštāsp Sāst, and to some extent

2 J. W. Boyd and W. R. Darrow have recorded the film, A Zoroastrian Ritual: The Yasna (1982). It is a staged production that reproduces the parts of the Yasna during which ritual actions (and not simply recitations) take place. A further videotape recording of some sections of the Yasna was prepared under the direction of Khojeste Mistree. The ceremony was performed by Mobed Asphandiar Dadachanji and Mobed Adil Behsania at Vatcha Gandhi Agiary, Gamdevi, Mumbai, in 2013. A first complete video of another staged performance of the Yasna is being prepared within the context of the Multimedia Yasna Project.

3 For summaries of the description, see Darmesteter 1892:III, x-xvi.

4 Accordingly, if the Great Avesta has to be imagined as written, at least two different versions existed, in which the twenty-one books were arranged in two different ways.

5 For a collection of the texts describing this process, see Bailey 1943:151-64; Cantera 2004a:106-63; Tremblay 2012:107-08.

6 See König 2012.


the Stōd Yasn. Furthermore, we have some sections of the Huspārom: the Hērbadestān and Nērangestān. Other extant Avestan texts are either scattered among several nask or simply missing from the Great Avesta. Since the beginning of the study of the transmission of the Avestan texts, it has been noted that the key for the preservation is their ritual usage (Darmesteter 1892:I, xxxviii-xxxix; West 1892:xxxix), meaning that the texts were preserved because they were actually used in the rituals. Most non-ritual Avestan texts were therefore lost at an unspecified time, although probably after the end of the ninth century, as the Dēnkard contains almost all twenty-one nasks of the Great Avesta.

The idea has always been of two realities, although only Kellens (1998) has formulated it with clarity, drawing the corresponding conclusions: a Great Avesta, consisting of what the Sasanian kings were able to rescue from a supposedly older “Greatest Avesta,” and a series of rituals. The differences among scholars concern the ontological position of each reality and its respective chronological position. The traditional view is that the ritual Avesta is a subsidiary collection extracted from the Great Avesta, and therefore later than that text. See, for example, the statement by X. Tremblay (2012:131):7

L’Avesta est une constellation de compilations de textes le cas échéant étronçonnés ou transposés destinée à la liturgie, aussi dénuée d’unité stylistique qu’unΤυπικόνou un Ευχολόγιον byzantin, uni seulement par la langue et la fin: accomplir soit le sacrifice qui donnera l’immortalité et la définitive sur les démons, soit les prières mineures qui permettent entretemps de contenir le mal. . . .

This understanding of the nature of the extant texts is combined with a view in which the written transmission has played a dominant role since the development of the Avestan script. Accordingly, the Great Avesta’s reduction to the extant texts used in the liturgies is explained by the loss of manuscripts. Only one or very few copies of each text would be available, whereby the loss of a manuscript would mean the loss of a text (Hoffmann and Narten 1989:17). Although Bailey has clearly shown that the oral transmission continued after the beginning of the written one, and was even more prestigious than the latter, the history of the transmission of the Avestan texts has been explained as an almost exclusively written one.

Kellens (1998:477-83) has revealed the shortcomings of this view. Firstly, the surviving texts are not directly sections of the Great Avesta, so we must explain how the liturgical tradition arose. The loss of manuscripts of the Great Avesta cannot have led to the appearance of the extant manuscripts, as the texts are presented in a different arrangement than the one they

7 The following statement by H. Bailey is also well known: “I may say that it is likely that our present Avestan texts go back to an edition after the fall of the Sasanian empire of the fragments saved from the first edition of about the middle of the sixth century A.D.” (1943:193).


supposedly contained.8 Secondly, the texts are not randomly arranged, but follow a ritual cursus. This argument has subsequently been further developed. Some years later, Kellens (2012) showed that the Avestan texts themselves followed a liturgical cursus similar to the one of the LL as it appears in the manuscript. This means that the liturgical cursus according to which the text is arranged already existed when the Avestan texts were composed in antiquity. Moreover, Kellens has progressively arrived at the conclusion that the text of the LL is a coherent and carefully arranged text. His new commented translation of the LL is a systematic attempt (especially in the later volumes) to reflect this coherence (Kellens 2006, 2007, 2010, and 2011; Kellens and Redard 2013).9 Thus, Kellens substitutes the model of linearity (the linear transformation of the Great Avesta into the extant Avesta) with one of simultaneity. In Sasanian times, there were two parallel collections: the Great Avesta and the Ritual Avesta. The extant Avesta known from the manuscripts derives from the latter and has no relationship with the former.10

The flaws in the traditional model detected by Kellens are obvious, and the alternative he proposes will certainly be upheld. Nonetheless, the availability of new materials (the liturgical manuscripts through the Avestan Digital Archive and the Nērangestān through the new edition by Kotwal and Kreyenbroek (1995, 2003, and 2009)) have led to a better understanding of the liturgical tradition’s historical development. Accordingly, Kellens’ standpoint on the Ritual Avesta should be reappraised. Furthermore, a new understanding of the integration of the apparently non-ritual texts in the liturgies has prompted a new view of the relationship between the Ritual and the Great Avesta.

According to Kellens (2012), the ritual cursus of the LL is contemporary to the composition of the Avestan texts, but the arrangement of the texts according to this cursus as we know it probably took place in Sasanian times, before the invention of the Avestan script. However, it is highly likely that already in antiquity, at a time when Avestan was still a living language, not only the ritual cursus of the LL but also its texts were arranged in the form they still have in the manuscripts and in modern practice (of course, with the inevitable changes that a performance over centuries involves). The Avestan version of the Nērangestān11 already followed the same version of the LL that is known to us through the manuscripts and modern practice (Cantera 2014:210-16). Furthermore, the variants of the LL attested in the manuscripts constitute a complex ritual system that adapts to different performative contexts. The creation of such a

8 Geldner also supposes a Yasna-liturgy from the Sasanian period onward (Geldner 1896a:18 n. 4 and 1896b:xxxiiii). The difference with Geldner is that whereas Kellens attributes the origin of the Yasna-text to the Sasanian period, Geldner thinks that the Yasna-text was arranged later. The distinction between the Yasna-ritual (the “cursus liturgique”) and the Yasna-text also appears in Kellens, albeit dated some centuries earlier: the Yasna-ritual is already known to the composers of the Avestan texts and not only to the authors of the Middle Persian literature. The Yasna-text is Sasanian for Kellens, but post-Sasanian for Geldner.

9 See also the review by Cantera (2016). Ahmadi (2018) argues against this. For a reply to his criticism, see Cantera 2020c.

10 The idea that the extant Avestan texts represent a kind of “prayer book” has already been expressed several times; see Spiegel 1882:605; Nyberg 1958:23; Panaino 1999 and 2012:84-86.

11 TheNērangestān has recently been made accessible in a reliable edition and translation by Kotwal and Kreyenbroek (1995, 2003, and 2009).


system in Late Antiquity, a thousand years after the demise of Avestan as a living language, is extremely unlikely.

Kellens’ “Ritual Avesta,” from which our manuscripts derived, consisted of two “anthologies”: the recitative of the LL and the collection of short rituals (Kellens 1998:477, 479). He does not conceive it as a book, but as two “Stammhandschriften” with an independent transmission (Kellens 1998:488). Indeed, his “Ritual Avesta,” the two “Stammhandschriften,” suffers partly from the same problems as Hoffmann’s archetype: it is a theoretical starting point for the written transmission from which our manuscripts stem, but neither the exact content of these anthologies can be defined nor the exact relationship between our extant manuscripts and this pristine composition. The Great Avesta is, for Kellens, a “réalité presque aussi théorétique que l’Avesta d’Andreas” (1998:488). The same could be said concerning the Sasanian “Ritual Avesta.” How should we imagine the collection of the LL? Did it contain the description of one or all the variants of the LL? If only one, which one? How and when were the others copied for the first time? From which source?

In fact, the variants of the LL in Sasanian times were not limited to the Yasna, Visperad, and Vīdēvdād, as Kellens seems to assume, but many other variants existed, some of which have been preserved in the manuscript tradition: Vīštāsp Yašt and Dō-Hōmāst. Others existed, but they have been lost: the Bayān Yašt and the Hādōxt, among others. Moreover, did the “Stammhandschrift” contain a scheme of a neutral performance of a variant LL or the specific performance for a day and with a particular dedication? The idea of the “archetype” or “Stammhandschrift” of the LL does not sufficiently consider the LL’s enormous dynamism (see below). The problem is compounded further still for the short rituals.12 Does Kellens assume the existence of a sort of Tamām Xorde Avestā, even though there is no evidence of anything similar before the copy of Ms. E1? There is evidence that the rituals existed, and that at a certain point their descriptions were transcribed close to the oral transmission. But which rituals and what the manuscripts looked like remain purely speculative. The simplest hypothesis is that the manuscripts were not very different from the extant ones in their arrangement and disposition. The extant manuscripts do not form two collections, but a much more complex arrangement with at least three different categories (LL, Drōn Yašt, and minor rituals), with different types of manuscripts in each category. They reflect the three basic types of past and present rituals.

The liturgical manuscripts are not the scions of the two “Stammhandschriften”; instead, they continue a long oral tradition of describing the liturgies. Novices in the priestly schools learned the texts by heart together with the basic instructions for their proper performance. The ritual instructions could be learned separately from the texts as collections (as we find in the Nērangestān13) or together with the liturgies. No Avestan versions of these descriptions have survived, but they can be imagined because the Nērangestān contains the Avestan versions of certain instructions (Darmesteter 1892:I, xciii; Cantera 2014:191-92, 211-16). The Avestan descriptions were abandoned and substituted by the Middle Persian ones we know from the

12 On the historical constitution of the manuscripts of the Khordeh Avesta and of the Yašts, see König 2012, 2015, and 2016.

13 Chapters 28-33 contain summaries of these complete descriptions in the form they were transmitted orally.


manuscripts in Sasanian times, exactly in the same way as the ritual instructions in Middle Persian were later substituted in India by others in Gujarati. Thus, the manuscripts are simply written versions of descriptions of the liturgy that had been circulating orally since antiquity and continued being learned and recited in the priestly schools in Sasanian and early Islamic times. We cannot be sure whether the descriptions of the LL were all written down at the same time or not, and even less so whether or not they were all transcribed in Sasanian times. The answer depends on the assumed purposes for the transcription (see below). In any case, it cannot be ruled out (and seems indeed more likely) that different liturgies were penned at different times. The LL, as the longest one, is a prime candidate for being the first ritual to be written down.

Besides, Kellens does not clearly define the position and function of the Great Avesta. Its existence is beyond question, but the details of its contents escape us, and its existence as a written book is also uncertain (Kellens 1998:486, 488). The traditional view was, as we have said, that the Ritual Avesta derives from the Great Avesta. In my view, the contrary is true: the Great Avesta seems to be a rearrangement of the texts of the Ritual Avesta. There is a special performance of the LL in which dialogues between Ahura Mazdā and Zaraϑuštra are inserted into the recitation of the central part of the LL, the Old Avestan texts (Cantera 2013b). This is the most likely context for the oral performance of any Avestan text that is not purely ritualistic, and which is presented as a revelation from Ahura Mazdā. The manuscripts preserve only two of this most complex type of LL (Vīdēvdād and Vīštāsp Yašt), and only the Vīdēvdād is still performed, and solely in India. The traditional view is that this kind of ceremony is the result of a late expansion of the LL through the intercalation of books from the Great Avesta (Modi 1922:350-51). Malandra even contends that the ceremony was an innovation of the Islamic period (Malandra 2000). However, the intercalation of these texts is not the only specificity of this type of ceremony. There are also differences in the standard litanies of the LL. One of the most significant ones is the substitution of the mention of the part of the day when the ceremony is performed, the so-called ceremonial ratu, a formula specific to each ceremony (Cantera 2013b; Martínez Porro 2022). The ceremonial ratu in the dative is for the Vīdēvdād dātāi haδa.dātāi vīdaēuuāi zaraθuštrāi aṣ̌aōne aṣ̌ahe raθβe,and for the Vīštāsp Yašt haδa.mąθrāi zaini.parštāi upairi.gātubiiō gərəptāi mąθrāi spəṇtāi aṣ̌aōne aṣ̌ahe raθβe. Consequently, ceremonies of this nature are unlikely to be later creations. They originated at a time when Avestan was a living language.

Within this context, the nasks gathered in the Great Avesta seem to be the dialogues between Ahura Mazdā and Zaraθuštra removed from their ritual setting, as we can see clearly for the Vīdēvdād. Furthermore, the central texts of the LL that served as basis for the intercalations were also compiled in a nask, the Stōd Yasn.14 The Pahlavi literature classifies the twenty-one nasks of the Great Avesta into three groups:

14 The special position of this nask among the others is emphasized through its highly prominent position in both arrangements of the Great Avesta: as the first nask of the first group, the gāhānīg,in the arrangement of the nasks in three groups, and as the nask corresponding to the last word of the Ahuna Vairiia in the alternative arrangement.


gāhānīg: Stōd Yasn, Sūdgar, Warštmānsar, Baγ, Wašti, Hādōxt, Spand
hadāmānsrīg: Dāmdād, Waxtar, Pāzen, Ratuštāiti, Brih, Kaškaysraw, Vištāsp Sāst
dādīg: Nigādom, Duzd-sar-nizad, Huspārom, Sagādom, Vidēvdād, Cihrdād, Bayān Yasn

Only two nasks are preserved among extant Avestan texts: the Vīdēvdād and the Vištāsp Sāst in their corresponding intercalation ceremonies, namely, the Vīdēvdād and the Vīštāsp Yašt. It has hitherto gone unnoticed that the word serving as a designation for the group of the corresponding seven nasks is precisely the first word of the ceremonial ratu of the corresponding ceremony. Thus, the Vīdēvdād appears in the group dādīg, and the first word of its ceremonial ratu is data-. The Vīštāsp Sāst, in turn, belongs to the group called hadāmānsrīg, and the first word of its ceremonial ratu is haδa.mąθra-. Accordingly, we may contend that each nask is a dialogue between Zaraθuštra and Ahura Mazdā inserted into the performance of the LL, and that the classifications of the nasks into three groups has a ritual origin: it is based on the common ceremonial ratu used in the corresponding ceremonies. It seems that there were at least three different ceremonial ratu, but one of them, corresponding to the ceremonies into which the gāhānīg nasks were inserted, has not survived, perhaps because these ceremonies have ceased to be held. In fact, it is likely that many of them were no longer widely performed even in Sasanian times. Hence the need to salvage the texts that were dispersed (the insistently repeated term pargandag) in the different regions by the different kings, starting with Valaxš and later by Ardašīr and Šābuhr (see DkM411.17). Clearly, whereas the LL’s structure was familiar to the entire priesthood, individual priests did not know all the nasks, but probably just one or two. The best-known nasks were probably the most frequently performed, such as the Vīdēvdād. Others were known by only a few priests in certain areas. Hence, the story told in the Abdīh ud sahīgīh ī sīstān that after the destruction brought to Sīstān by Alexander the Great, only a group of women and children knew the Bagān Yašt (Bailey 1943:161).15

The rituals that originated during the Achaemenid Empire and spread over vast areas, probably through the institutional support of the Achaemenid kings, continued to be performed until the Sasanian period. Centuries of ritual performance, partly under foreign rule, led to some changes (despite the conservative nature inherent to oral traditions) and the neglect of certain rituals. The basic variants of the LL were not threatened, as their knowledge was part of priests’ basic instruction. By contrast, the nasks were known only by a certain number of priests. Thus,

15 The Pahlavi narrative of the different attempts to gather the scattered texts sometimes refers specifically to written texts. In the famous account of the fourth book of the Dēnkard, written texts are explicitly mentioned. Thus, Valaxš set out to collect the surviving fragments of the abastāg ud zand copied by Darius III, both the written texts and the orally transmitted ones (har čē . . . pargandagīhā abar nibištag tā če uzwān abēspārišnīg pad dastwar mād ēstād . . . nigāh dāštan ō šahrīhā ayyādgar kardan framūd, “he ordered to preserve and send as memoranda to the countries everything that was extant scattered in written texts including as well the ones orally transmitted by the dastwar,” Dk4.16 [DkM412.7]). According to Dk3.420 (DkM406.3), Ardašīr gathered the books that were scattered (ham nibēg az pargandīh ō ēk gyāg āwurd, “he gathered and brought the books from the dispersion to one place”). Nonetheless, this information should be considered with caution, exactly like the information about the written Avesta of Achaemenid times (the one of Vištāsp on golden tablets, Dk3.420 [DkM405.17], Dk4.14 [DkM411.17], ŠE 3; or on parchment in golden ink, Dk53.2 [DkM437.17]; or the copies made by Darius III, Dk4.15 [DkM412.3]). These notes are anachronistic attributions to the past of the actual transmission modes of the time of the Dēnkard and all the other Pahlavi ninth-century sources. Within the Islamic context, they could at least in part be motivated by the interest to show that Zoroastrians have a protracted tradition (as already assumed by Bailey 1943:151).


those nasks that were less frequently recited (because the ceremonies in which they were embedded were more rarely performed) were almost or completely lost. Accordingly, some salvage programs were launched, although they were of only limited success in the long term.

The origin of the liturgical manuscripts is clear. They are the result of the progressive transcription of the traditional descriptions of the liturgies that had been transmitted orally since antiquity. The exegetical manuscripts in which the text of the liturgies or of some parts thereof were translated, first into Pahlavi and then into Sanskrit, require a different explanation. This is not the place for a detailed discussion, so I will simply outline the main stages in this parallel tradition. The archaisms of the language of certain Pahlavi translations of Avestan texts and other signs indicate that some of them had already been composed in the first centuries of the Sasanian era.16 The translation was also orally transmitted. It is probable that not all Avestan texts were translated, but only the ones whose content was relevant. This excludes the litanies of the LL and many of the short liturgies. Translations were probably produced for the same texts that were integrated in the Great Avesta. They were, basically, the central texts of the LL and the nasks. Thus, the translation of the Vīdēvdād, the Vištāsp Yašt, and at least some Yašt, would be the heirs of the translations included in the Great Avesta for the corresponding nasks. The translation of the Stōd Yasn was extended to become a translation of the complete Yasna. This process probably occurred as late as the beginning of the eleventh century. The introduction of the so-called combined manuscripts (including ritual instructions and Pahlavi translation) describes the joining process of a liturgical manuscript with one containing the Pahlavi translation (Geldner 1886-96:I, xxiv-xxvii; Cantera and de Vaan 2005). The manuscript containing the Pahlavi translation was probably a manuscript with a Pahlavi translation of the Stōd Yasn, and the translation of the sections of the LL that were not part of it were produced at that time. The same process involved the translation of some of the sections of the Visperad that do not appear in the Yasna.17

In any case, it seems that no extant Yasna manuscript with Pahlavi translation continues a Sasanian tradition of a Yasna with translation, but without ritual instructions. The exegetical manuscripts of the Yasna belong to two different groups:

  1. The first group includes a series of manuscripts copied in India in the nineteenth century, while harking back to an Iranian manuscript produced in the region of Kāzerun, a copy of which was sent to India, probably in the sixteenth or seventeenth centuries (Cantera 2012c). These manuscripts include ritual instructions in Pahlavi that are similar to the liturgical manuscripts.
  2. The second group consists of the manuscripts copied by Mihrābān Kayxōsrō in India at the end of the thirteenth century and the beginning of the fourteenth, as well as their copies. They do not include ritual instructions.

However, we have reasons to affirm that Mihrābān’s manuscripts were also extracted

16 See Cantera 2004a:164 ff., especially 220, where older literature is discussed.

17 The latter clearly complements the non-Stōd Yasna-sections of the Yasna. The fact that it includes, for no apparent reason, only the sections until Y54 might indicate that the translation of the Visperad is not a traditional one, but created once and for all.


from a manuscript similar to the ones in group 1, including not only the Pahlavi translation, but also the ritual instructions. First, we find ritual instructions concerning the number of repetitions and the speaker. Second, whenever the manuscripts in group 1 include ritual instructions, Mihrābān has left blank spaces. What is more, some blanks are filled in ms. 510 (K5), with the expected ritual instructions written in red.18 They are mostly the titles of sections or short instructions, such as the speaker or the position of the auxiliary priest when speaking. Sometimes, however, they are longer ritual instructions.19

These ritual instructions could have been added by a second hand, but even this would show that the blank spaces in the Yasna manuscripts written by Mihrābān were understood as corresponding to the texts written in red ink in the liturgical manuscripts. The absence of the red texts in Mihrābān may be accidental (it was intended to be copied later, but this never happened) or intentional (as the manuscript was not created for liturgical use, the ritual instruction could be omitted). In any case, the blanks show that Mihrābān’s Yasna manuscripts derive from ones that also included the ritual instructions for the performance.

In sum, there has been a continuous ritual activity from antiquity through to modern times. The manuscripts are the scions of the description of the ceremonies that were learned by heart in the priestly schools as part of the necessary apprenticeship for the performance of the rituals, and not automatic copies from a Sasanian archetype either of the Great Avesta or of a Ritual Avesta. The content-oriented texts of the rituals were also learned together with their Pahlavi translation, especially the dialogues between Ahura Mazdā and Zaraϑuštra that were intercalated in some performances of the LL. These texts were also separately arranged as structured collections. At least two arrangements are known: one after the twenty-one books of the Ahuna Vairiia and another in three groups of seven nasks each. This collection did not simply include the texts that were regularly performed, but is the result of several attempts to preserve texts that were under threat because their ritual performance was no longer common. The use of script would at least be a useful tool for the preservation of these texts, and probably not only for that.

18 A very interesting feature of mss. 500 (J2) and 510 (K5) is the numbering of quires in red ink. They indicate the end of a quire through the word rāyēnīd in 510 and rāyēnīdag in 500 at the bottom of the page and the number of the following quire in the left margin through the corresponding ordinal + ǰudēdād <ywdtydʾt> <<abāg>> zand. The indications are in red, and probably by a second hand. The quires consist mostly of twelve folios, but we find some of ten or even of nine folios, for example, 500 (J2) folios 5v, 17v, 29v, 39v, 51v, 63v, and so forth; 510 (K5): 10v, 22v, 34v, 46v, 58v, 67v, and so forth.

19 An extraordinary position is adopted by a long ritual instruction that appears in Y8.6-7 in ms. 510 (K5):

wehēnīgīhā (?) tā gyāg srišāmrūtīg gōwišn srōšdrōn xwardan ud dahān pāk kardan ud dast pad pādyāb kardan ud abar barsom nihādan ud ašemwohū wehēnīgīhā (?) cahrušāmrūtīg guftan yatāhūweryō 4 guftan

Until the place where he has to recite three times (the Aṣ̌əm Vohū). He should eat the srōšdrōn, make his mouth clean, wash his hand with pādyāb, put it on the barsom, recite four times the Aṣ̌əm Vohū, and recite four times the Yatāhūweryō.

In ms. 500, we find instead a blank space. However, in the group 1 manuscripts, we find the same instruction with only a few textual variants, mainly a different spelling of the difficult word wehēnīgīhā and the end that says yatāhūweryō bišāmrūtīg gōwišn, “he has to say the Yatāhūweryō twice.” Similar instructions appear in the liturgical manuscripts, too, but the wording of ms. 510 (K5) corresponds to the exegetical manuscripts in group 1.


3. The Invention of the Avestan Script and the First Manuscripts

When writing appears for the first time in an oral tradition, the function of manuscripts can be multifarious, together with the consequences its appearance has for the oral transmission. The manuscripts may appear as a tradent, an alternative to memory for preserving and handling texts in fading oral traditions, or as an instrument in the process of memorizing texts that continue being transmitted mainly orally, or as tools for the aural performance of a text, etc. Even within a tradition, manuscripts might enter into the oral tradition with different functions depending on the nature and role of the texts. We have reasons to assume that the introduction of writing pursued different goals for the tradition of the Great Avesta than for the Ritual Avesta.

Manusčihr, the head of the Zoroastrian priesthood in Iran in the second half of the ninth century, informs us in one of his letters that Wehšābuhr presented the twenty-one nasks of the Avesta to the priestly assembly organized by Xōsrō I at the beginning of the sixth century. The assembly agreed on it and, consequently, they were transcribed and sealed (NM1.4.17):20

wehšābuhr pad hanǰaman ī anōšag-ruwān xusraw ī šāhān šāh ī kawādān wīst ud ēkān xbazišnīhā ōwōn nimūd kū ōwōn menišn padiš ēstād hēnd u-šān nibišt ud āwišt.

Wehšābuhr presented in the assembly of the king of kings, Xōsrō, son of Kawād, of immortal soul, the twenty-one parts so that all were unanimous, and he wrote and sealed them.

In fact, Xōsrō I’s reign is widely accepted as the most likely time for the invention of the Avestan script.21 As a consequence, the twenty-one nasks of the Great Avesta were purportedly written down. The nasks had been collected a few centuries before, a last time during the reign of Ardaxšīr I (224-42 CE). According to the Pahlavi account, the collection of the nasks was accompanied by the creation of written copies, but this information cannot be either verified or falsified. The only certainty is that the extant Avestan copies do not hark back to these alleged copies produced in the third century, but were copied directly from the oral transmission when the Avestan script was invented. After their recovery, some ceremonies might have been reactivated, as seems to have been the case for the Bayān Yašt.22 Other nasks were preserved for the importance of their contents as a source of authority in legal, ritual, and theological matters, but perhaps no longer used in ritual performances. These were probably taught and learned orally in the priestly schools, together with their Pahlavi translations, as we know was the case for the

20 See Bailey 1943:173; Kanga 1966:50, 56; Cantera 2004b:123. On this assembly and the transcribing of the Avesta under Xōsrō I, see Cantera 2004a:160-62; Huyse 2008; Rezania 2012.

21 This is a widely accepted notion; see Cantera 2004:160-2; Cereti 2008; Huyse 2008; Panaino 2012:79-82. Tremblay (2012:117) proposes a slightly later date, between 550 and 630. Kellens (1998:488) used to postulate a post-Sasanian date for the invention of the Avestan script, Hoffmann and Narten (1989:34) a much earlier date because of the (incorrect) assumption that the inscription of a sarcophagus in Istanbul shows that the Pahlavi cursive (on which the Avestan script is based) had already adopted its form in the fourth century.

22 At some point, it was known only by a group of women and a child in Sīstan (see above). Nonetheless, the Nērangestān describes it as a regularly performed ritual (Kreyenbroek 2008).


Vīdēvdād.23 Several oral versions of the most popular ones were circulating simultaneously and competing with each other (Cantera 2004:220-29). The preservation in memory of so many nasks, some of them no longer in ritual use, and furthermore in different versions by different schools, posed a great risk of loss. Some of them were lost even after they had been transcribed, and before the end of the ninth century.24 Accordingly, preservation of the many nasks (especially the ones less frequently performed ritually) might have triggered the invention of the Avestan script.

Furthermore, a corpus scattered among different priestly schools was difficult to control. Since the Avesta and its translation was one of the main sources of authority, one does not wonder that at some point there arose the temptation of limiting the freedom of interpretation, and the idea of fixing a canonical exegesis. Accordingly, Xōsrō proclaimed that the zand (“traditional translation and interpretation of the Avestan text”) should not be taught out of the established agreement (ZVY2.4). His attempt at preserving and controlling the interpretation of the Avestan texts is no surprise considering the turbulent times caused by the Mazdakite movement that used the Avestan texts and their interpretation for political goals that threatened Xōsrō’s position (Rezania 2012). The manuscripts were thus created for a dual purpose: first, they are a repository containing texts that were threatened in the oral tradition and in pursuit of completeness; second, they are authoritative, while seeking to define the canon for the proper translations and commentaries of the Avestan texts.

However, to that double purpose, the invention of the Avestan script would have been unnecessary. It is likely that Avestan texts were copied before the invention of the Avestan script in Pahlavi script, and even in others. Actually, the Avestan script has an almost unique peculiarity, as has been repeatedly emphasized, that is very illustrative about the main purpose of its invention:25 its phonetic (and not phonologic) character. It conveys the impression that it was created for the faithful reproduction of orally transmitted texts, in contrast to the shortcomings of the Pahlavi alphabet, which works well for reproducing content, but whose reading aloud is nigh on impossible.26 The Avestan alphabet consists of fifty-four letters27 (but more might have been created28). It distinguishes phonetic nuances without phonological relevance. For example, it distinguishes fifteen vowels (a, ā, ā̊, ą, ə̨, ə, ə̄, e, ē, o, ō, i, ī, u, ū), three palatal š (š, š́, ṣ̌), and

23 A complete English translation of the Pahlavi text and commentary of the Vīdēvdād has been published by Moazami (2014). For a partial edition and translation see Andrés-Toledo 2016.

24 The Avestan text and the Pahlavi translation of the Wašti Nask had, indeed, already been lost when the description of the Dēnkard was prepared (Dk8.12), and the same is true for the Pahlavi translation of the Waxtar Nask (Dk8.6).

25 Beginning with Morgenstierne (1942), who has laid the foundations for the present understanding of the Avestan script, and further developed by Hoffmann (1971 and 1986; Hoffmann and Narten 1989).

26 There are three main reasons for this difficulty: (1) the use of one letter for several phonetic values; (2) the use of aramaeograms; and (3) its orthographical conservatism.

27 Most of the manuscripts do not use all of them, but there are some letters whose usage is limited to certain types of manuscripts.

28 Some more letters might have been created, but had disappeared before the time of the extant manuscripts (Ferrer Losilla 2016).


three velar nasals (ŋ, ŋ́, ŋˇ). For a content-oriented collection, the need for an accurate pronunciation would not have been a priority. The correct pronunciation of the Avestan texts is, however, extremely important for their recitation in the ritual performances (Morgenstierne 1942). We know that the differences in the recitation were enormous across the different regions of the Sasanian Empire. The magnitude of this problem can hardly be overestimated. Some years ago, a version of one of the most frequently used Avestan prayers, which is repeated several times in each performance of any ritual, was recognized by Gershevitch (cited in Sims-Williams 1976:75-82) in a Sogdian manuscript in Manichaean script dated around the tenth century. The pronunciation is quite different from the one reflected in the manuscripts:

Aṣ̌əm Vohū in the Avestan manuscripts Aṣ̌əm Vohū in the fragment from Dunhuang 
aṣ̌əm vohū vahištəm astī
uštā astī uštā ahmāi
hiiat̰ aṣ̌āi vahištāi aṣ̌əm
[wrt]mwγštmyštʾy wštʾwštʾy
wštʾγmʾytwrtʾʾy ʾγwštʾyrtm

uštāyuštī uštāhmāy
iturtāi əxuštāyirtəm/

Besides certain “Sogdianisms,” such as the palatalization ištī instead of asti, we find more significant differences. The most intriguing one is the preservation of the group rt, where the manuscripts show ṣ̌. This letter appears where we etymologically expect a post-tonic rt.29 The loanwords in Middle Persian show that at the beginning of the third century the pronunciation was hr, later hl, and by the time of creation of the Avestan alphabet a sound similar to /ʃ/. However, this evolution concerned only the pronunciation of the Avestan texts in the region of Fārs. In other regions, the pronunciation evolved in a quite different way, as the Sogdian version of the Aṣ̌əm Vohū shows. Hence, it is very likely that the invention of the Avestan script reflected an attempt to generalize a specific performance of the rituals, even with a particular way of reciting the texts in Western Iran. The first manuscripts containing descriptions of the ceremonies might have been prescriptive: they probably respond to an attempt to create a homogenous performance of the rituals in the realms of the Sasanian Empire. Later, in the interplay of oral and written transmission, they would assume other functions, but the initial one seems, however, to have been purely prescriptive.

The transcription of rituals, on the one hand, and of a selection of the texts with their Pahlavi translation, on the other, did not put an end to the oral transmission. The rituals continued being performed without the use of any manuscripts. At least some of the nasks also continued being learned by heart in the priestly schools. Different schools kept their alternative versions and transmitted them orally, as the existence of such schools even in the zand-related ninth-century literature shows (Cantera 2020a). Yet this affected, however, the works that were most frequently required: the Vīdēvdād for discussion about purity, and the Hērbadestān and Nērangestān for questions concerning the ritual. Other legal treatises have not been preserved, probably because a compilation such as the Mādayān ī Hazār Dādestān (“The Book of the

29 On this letter, see Hoffmann 1986.


Thousand Judgements”) assumed the authoritative role of the legal nasks, and because they lost their relevance after the first centuries of Islamization because of the change in the legal system. For some nasks, the Pahlavi translation without the Avestan texts was also learned by heart, and parallel versions with different arrangements and epitomes were also composed and learned.30 Besides the content-relevant nasks, the nasks that were still performed were also learned by heart together with their Pahlavi translation, such as the Stōd Yasn and the Bayān Nask. Accordingly, the rest of the Great Avesta that has come down to us consists of a few nasks with their Pahlavi translations that are precisely the ones for which a sustained oral transmission can be assumed. The written Great Avesta has disappeared without trace. The texts that were assumed to have survived in written copies have been completely lost. The only texts that survived were those that continued to be activated orally and learned by heart for that purpose. Far from putting an end to the oral transmission, the manuscripts entered into a complex interface for the preservation and updating of the Avestan texts in ritual performance, in the priestly schools, and in the assemblies where theological, legal, or ritual issues were debated. However, for a long time they played only a secondary role there.

4. Manuscripts and Performance: The Role of Manuscripts in the Preservation and Performance of a Dynamic Liturgy

After the success of M. Parry and A. Lord’s oral-formulaic theory (Lord 1960; Parry and Parry 1971), literacy and orality quickly developed into antithetic concepts. According to this Great Divide, literacy and orality became not only two different and mutually excluding ways of conceptualizing and transmitting texts, but even antithetic cultural forces. It did not take long for such a deep division to be challenged (see Finnegan 1973 and 1977). First, both concepts, “orality” and “literacy,” have to be calibrated “by taking account in at least three areas: the tradition (whether Native American, Turkish, medieval English, or whatever), the genre31 (as closely as one can track this aspect across traditions) and the nature of the documents” (Foley 1991). Furthermore, neither concept is mutually exclusive, as they often participate together in an interface for preserving and activating texts. These are the two ends of a “cultural diglossia” (Stock 1983), in which literacy does not simply supersede orality, but instead both constitute a continuum. As K. Reichl says, “in the orality-literacy continuum tensions and combinations in many shades and hues are conceivable” (2015:38). In Iran, for example, the use of writing was for a long time limited to administrative purposes and political propaganda. Its use was, however, widely excluded for literary and religious texts. This started changing in Sasanian times, but the progressive transformation to a chiefly written textual production and

30 One surviving example is the Zand ī Fragard ī J̌uddēwdād (König 2010; Elman and Moazami 2014). This tradition also survives in the eighth and ninth books of the Dēnkard, with the former epitomizing the twenty-one nasks of the Avesta and the latter only three nasks of the gāhānīg group: Sūdgar, Warštmānsar, and Bay Nask, albeit in great detail. An analysis of the text of these two long epitomes in search of traces of literacy or orality is still pending.

31 This aspect merits special attention in the studies about the oral production of the Avestan texts whose ritual performance involves a different performative frame than the epic texts, for example.


preservation was not completed before the advent of Islam, and even at that time the diglossia continued.32 Third, texts are rarely exclusively oral or written. Orality and literacy might be involved in the composition, preservation, and activation of texts in a productive combination (Finnegan 1977:17).

The transition from orality to literacy is rather a complex process that is modulated in different ways in different traditions. In Ancient Greek, the Iliad and the Odyssey were composed according to the rules of oral-formulaic theory by aoidoi. From the sixth century on, when script was introduced in Greece, rhapsōidoi simply recited the texts they had learned by heart in an almost identical way in each performance. They learned the texts either from other rhapsōidoi or from written sources (Reichl 2015:19). In Islamic tradition, once the muṣḥaf codex had been produced, at the time of the third caliph ʿUthmān (654-66), the text of the Qurʾān moved from the realm of orality to that of aurality: the written text was now read aloud in public. Moreover, the writing down of the text did not mean the end of its memorization. As T. Herzog says: “The Qurʾān itself is memorized to this day and comes fully alive only in oral recitation (as a text intended for aural reception), while the written book of the Qurʾān often only serves as a prop for memory and a guarantor of the correct recitation of God’s word (with the exception of blind recitators, who of course have no written props)” (2012:31).

The same (or a very similar) premise applies to the Avestan texts of the LL. As we have seen, they are still being memorized in India and “come fully alive” only in the ritual performance. The Avestan texts were “composed in performance” according to the rules of oral-formulaic theory by priests with abilities similar to the ones of the Greek aoidoi.33 After the end of the Achaemenid period, the priests developed into a kind of rhapsōidoi that performed in the ritual the text they had learned by heart and had to update according to very specific rules for each new performance. Like the rhapsōidoi, they mostly learned the text from other priests and reproduced them verbatim in the liturgy. To that purpose, they probably developed techniques similar to the ones of the close tradition of Vedic poets.34 They might also have used written texts for the memorization, but there is no evidence of this before the invention of the Avestan script. The invention, despite its importance for the transmission, did not put an end to the oral transmission. Already Bailey (1943:149-68) has pointed out that the oral transmission continued even after the invention of the Avestan script, and was even held in greater esteem than the written one still in the ninth century. For the ninth-century authors, memory is more prestigious than the manuscripts as tradent of the texts.35 This fits well with the external information that

32 For the refutation of the oft-assumed divide between the orality of the Middle Persian tradition and the Islamic literacy, see Vevaina 2015. A very illustrative collection of essays on the role of orality and the interplay with literacy in the literary history of Iran has been assembled by J. Rubanovich (2015).

33 On the application of this method to the oral composition of the Avestan texts, see Skjærvø 1994, 1997 [2000], 1998b, 1998a, 1999, and 2012 .

34 For an overview of Vedic techniques, see Kiparsky 1976.

35 The Graeco-Roman world, among many other cultures, also knows a similar preeminence of memory (Cribiore 1996:42; Park 2009:65).


stresses that Zoroastrians did not use books, like the observations in the Syriac books collected by F. Nau (1927).36

Indeed, Bailey’s oversight regarding the ritual nature of the Avestan texts meant that he failed to notice that the prestige of orality not only concerned the preservation in memory, but also the oral performance. He quotes the text Panǰ xēm ī asrōān, “The Five Virtues of the Priests,” to show that “the trained memory is one of the necessary qualities for a good dastaβar” (Bailey 1943:158). The text, nonetheless, places the emphasis on the performance in worship:

čahārom yazišn ī yazdān rāst wāzagīhā narm naskīhā pad nērang yaštan

The fourth (virtue) is the performance of the ceremonies for the gods according to the ritual instructions with the correct words and the memorized nasks.

Even in modern times, the usage of manuscripts during the performance is not allowed, except for the Vīdēvdād ceremony. Although in Late Antiquity the learning of the nasks was one of the duties of priests and even educated laity,37 their learning by heart by the priests was abandoned at some point. The question 65 of the Dādestān ī dēnīg informs us that a priest in formation did know by heart five nask. Nonetheless, this capacity seems to have vanished between the ninth and the sixteenth centuries. Already the Revāyat of Kāmdin Šāpur, written in 1559 (928 YE), informs us that the recitation of the ceremony until the intercalation of the first fragard proceeds without the use of a manuscript. When the main priest has to begin with the recitation of the words mraōt̰ ahurō mazdā̊, “Ahura Mazdā said” (beginning of the first fragard), he then touches the manuscript with his right hand and reads until the end of the first intercalation. He should then continue reciting without using the manuscript, but before he does so he must wash his hands with purified water. This procedure is repeated at the recitation of any intercalation of the Vīdēvdād. By the sixteenth century, a manuscript was being used for the recitation of the fragards of the Vīdēvdād, but its use is limited to the longest part, the nask, whereas the proper ritual is performed without use of the manuscript. The manuscript is even considered an impure object in the ritual area. Thus, the priest has to wash his hands, polluted through the manuscript, before continuing the recitation after each use.

Liturgical manuscripts are basically excluded from the activation of the text in performance. Their role is somewhat confined, as in the case of the Qurʾān, to the preservation of the text and a prescriptive function as guides for correct performance. The latter seems to have been the main reason for the creation of the first liturgical manuscripts in Sasanian times (see

36 The information gathered by Nau and his conclusions were sharply criticized by A. Christensen (1936:515-17), but have been critically reinstated by Tremblay (2012:114-16).

37 In fact, not only priests learned the ritual texts by heart, as stated in the text Xōsrō ud rēdag, “Xōsrō and the Page,” where the page affirms that he has learned at school (frahangestān) a series of ritual texts including the most complex ones: the Yašt, the Hādōxt, the Bayān, and the Vīdēvdād (Bailey 1943:160). Apparently, the performance of rituals (even complex ones) was not an exclusive competence of the priestly class in Sasanian society. Anyone fulfilling a series of requirements, with some of the main ones including knowledge of the texts by heart and having the necessary ritual purity, could perform rituals.


above, section 3) and was never completely lost. Thus, the dismantlement of the Ātaš Bahrām in Sanjān during the second half of the fifteenth century caused a major break in the ritual tradition. Consequently, it triggered, on the one hand, the sending of letters (known as the Revāyat) to the Zoroastrian communities in Iran searching for advice and requesting liturgical manuscripts (Cantera 2014:154-62) and, on the other, the production of liturgical manuscripts in two Indian centres, Bharuch and Navsāri. In the former, Ardašīr Zīvā has created the first liturgical Yasna based on an exegetical manuscript copied two centuries earlier by Mihrābān Kayxōsrō (see below, section 5). In the latter, Āsdin Kākā produced a copy of at least each one of the main types of liturgical manuscripts, based partly on the manuscripts copied in India at the end of the thirteenth century by Rustām Mihrābān (Cantera 2014:153-54.). The instructive and prescriptive nature of the Indian liturgical manuscripts is informed by the fact that from the very beginning the ritual instructions were written in Gujarati38 in order to instruct the Indian priests that did not know Pahlavi how to perform the ceremonies.

The prescriptive value of liturgical manuscripts depends on the prestige of their scribe. Famous scribes were commissioned to copy manuscripts that were especially prestigious and invested with exceptional authority. Accordingly, some priests were professionals that were invited to other regions to produce authoritative copies that might be used there. A well known example is the case of the brothers Frēdōn and Wahrom Marzbān, who lived in Kermān, but copied most of their manuscripts in Torkābād (Cantera 2014: 93-96). The sending of manuscripts from Iran to India during the period of the Revāyats must also be seen under the same light. The prestige of the copyists depends, of course, not only on their technical abilities in the production of manuscripts (calligraphy, binding, quality of the paper, and so forth), but also on their reputation as authoritative priests or the fact that they were in possession of prestigious old manuscripts that allowed them to know how the ceremonies should be performed. Accordingly, admired scribes not only copied authoritative manuscripts, but also had the authority to sanction other manuscripts copied by other less renowned copyists. The celebrated Frēdōn Marzbān, for example, sanctioned at least one manuscript of the Visperad and the Vīštāsp Yašt (mss. 2010 + 5010), copied by Mānušcihr Ardašīr, before sending it to India, as well as the Vīdēvdād manuscript 4025.

This explains two striking facts about the transmission of the liturgical Avestan manuscripts: first, most of the extant manuscripts were copied by specific families of scribes, and second, most of the extant copies of each class of manuscripts seem to share a common ancestor. Prestigious manuscripts were not only better preserved, but also more often copied. In fact, the vast majority of the extant Iranian manuscripts copied before the seventeenth century correspond to the family of Bundār Šāhmardān39 (most especially to Marzbān Frēdōn and his sons). Even in the nineteenth century, the manuscripts of this group were still the basis for the production of manuscripts by Isfandyār Anuširvān. A similar situation applies in the first half of the eighteenth

38 Notwithstanding, some manuscripts copied from Iranian originals kept the ritual instructions in Pahlavi. This is the case of the so called combined Yasna manuscripts (for example, mss. 400 [Pt4], 410 [Mf4]) and the copies of a Vīdēvdād manuscript sent to India at the beginning of the sixteenth century (from which five copies survive: mss. 4020 [Mf2], 4070 [K9], 4080, 4110, 4120), among others.

39 About this family and the Pahlavi manuscripts they have produced, see König 2014; for this family’s Avestan manuscripts, see Cantera 2014:93-96.


century, when most of the extant manuscripts were copied by a single scribe, Rustām Goštāsp. He had access to manuscripts by a branch of the Bundār Šāhmardān family, particularly the ones copied by, or in possession of, Gōpatšāh Rostom, some of which, like the lost manuscript Jp1 of the Vīdēvdād, had been copied by Frēdōn Marzbān (Cantera 2014:105). When Mullā Firuz spent some years in Yazd in the 1770s, Rustām Goštāsp’s manuscripts still enjoyed considerable prestige. Accordingly, Mullā Firuz took several manuscripts copied by this scribe to Mumbai (Cantera 2014:175-78). One of the most important ones is manuscript D83, which contains a Yasna, a Visperad, several šnūman, and Drōn Yašt (Dhabhar 1923:14 -5), as well as three Khorde Avestā (Mf28, Mf29, Mf45) and a commentary on the Aṣ̌əm Vohū (Dhabhar 1923:19-20.).

This tendency to preserve and copy especially important manuscripts is the most likely reason why all extant Iranian Vīdēvdād manuscripts hark back to one or several similar manuscripts copied by Šahryār Irdešīr40 (Cantera 2014:96-104, 109-113), and why it is highly likely that all Vištāsp Yašt manuscripts go back to a single copy from which the ones by Rustam Mihrābān (at the end of the thirteenth century CE) and Xōsrōšāh Anušagruwān (in 1344) were made. We should not therefore conclude that the sixteenth century was a bottleneck in the transmission of the Vīdēvdād, and that only one manuscript was available. The same applies for the forerunner of all Vīštāsp Yašt manuscripts in the thirteenth century. Tremblay has expressed it perfectly through a biological analogy: “le prestige et donc le succès reproductif d’un individu puis de sa lignée, répété au cours de centaines de générations, finit par évincer totalement la descendance d’autres ancêtres” (2012:130).

A frequent function of manuscripts in many cultures is to support memory (see Cribiore 2001:213 and passim) and this, besides the prescriptive one, seems to have been a function of the Avestan liturgical manuscripts. The production of these kinds of manuscripts might have been much more abundant than the extant manuscripts suggest. Nevertheless, manuscripts copied for personal or family use and lacking the authority of the prestigious manuscripts written by certain families were not preserved over generations and were not copied outside the family, so most of them have been lost. Only manuscripts copied in the last phases of the transmission in the nineteenth century have survived.

The interactions between manuscript and memory are complex. Manuscripts might assist in the learning process, although at the same time they are conceived and can only be used by persons who know the texts, or at least important parts of them, by heart. Hence, the high number of abbreviations they contain that render the use of manuscripts possible only for readers that already know the text. Individual learning with the assistance of manuscripts has never superseded traditional oral teaching. The numerous abbreviations used in the manuscripts show, on the one hand, that the manuscripts could indeed be used only by persons who knew the texts and, on the other hand, the different focus of the sundry descriptions of the liturgies. The number and extension of the abbreviations depend on the variant of the LL they are covering and the type of manuscript. The text of the Yasna is the basic one, and its knowledge is presumed by the manuscripts of other liturgies. Nonetheless, even the Yasna manuscripts contain abbreviations that go beyond the ones of standard prayers. Thus, ms. 3 has:

40 A famous sixteenth-century scribe who has also copied an exegetical Vīdēvdād and a manuscript of the Dēnkard from which the most complete extant manuscript derives.


imą. haōmąsca. miiazdąsca. tā ō ī (“up to”) huuarəštā̊. mąϑrā̊. pairəca. dadəmahe. āca. vaēδaiiemahe.

instead of the complete text:

imą. haōmąsca. miiazdąsca. zaōϑrā̊sca. barəsmaca. aṣ̌aiia. frastarətəm. gąmca. huδā̊ŋhəm. hauruuata. amərətāta. gąmca. huδā̊ŋhəm. haōməmca. para.haōməmca. aēsmąsca. baōiδimca. imąm. aŋhuiiąmca. aṣ̌aiiąmca. raϑβąmca. ratufritīmca. gāϑanąmca. sraōϑrəm. huuarštā̊. mąϑrā̊. pairica. dadəmahi. āca. vaēδaiiamahi.

It also abbreviates Y12.13-4, Y18.9-10, and so forth.41 The abbreviations are much more important in other types of manuscripts. Manuscripts of the Drōn Yašt focus only on the sections that are different from the corresponding section of the Yasna (Y3-8). The same applies for the Visperad manuscripts, where we often find direct references to the manuscripts of the Yasna; for example, ms. 2010 abbreviates VrS0.5-10 with the indication čīyōn pad yašt nibišt, “As it is written for the yašt (=Yasna).”

The dynamism of the liturgy is a fundamental aspect of the complex interplay between manuscript and memory, but it has mostly been wholly disregarded. The activation of the Avestan texts in the ritual performance always entails an actualization of the text. The analysis of the liturgical manuscripts in recent years has revealed the dynamic nature of the Avestan texts. On the one hand, the LL can be performed in many different variants. On the other, each variant of the LL changes according to certain parameters in each performance. Each enactment differs from the previous one according to different parameters, such as the time of the performance and the specific god for whom the ceremony is performed.42 This changing information is encoded in the liturgy’s prose sections, a series of litanies that introduce the metrical sections (Cantera 2020c). Manuscripts comparable to the Christian missals, introduced in the thirteenth century, would have been needed for a full description of the different variants of the LL and their performance throughout the liturgical year. However, such complex manuscripts were never created in the Zoroastrian tradition, perhaps because of the exclusion of manuscripts from the performance. Thus, the priest needs additional information for this purpose that is not contained in the manuscripts. Nonetheless, except for the manuscripts of the Yasna, the manuscripts acknowledge, and partly reflect, the liturgy’s dynamism. This aspect is essential for understanding the nature of the liturgical manuscripts.

The standard editions present the Avestan texts as static. The new edition of the Avestan

41 Not all abbreviations are of the same nature. Manuscript 40 is a modern and heavily abbreviated copy of ms. 10 (Mf1). It contains a Yasna and a Farroxši. It is one of the copies made at the end of the nineteenth century by Erachji Sorabji Kausji Mehrejirana from the manuscript in the collection of Mulla Firuz held at the Cama Oriental Institute with the goal of producing a copy of the original for the Meherjirana Library. The obvious interest of Erachji is the ritual instructions that are frequently more detailed in this manuscript than in other Yasna manuscripts. He therefore omitted large parts of the Avestan texts, simply copying the necessary portions of the text for correctly locating the instructions within the liturgy.

42 The minor rituals do not contain this information in their own text (basically all the rituals included in the so-called Khorde Avesta or “small Avesta” manuscripts), because they were performed in combination with rituals in which this information is included such as the Drōn ceremonies or the LL.


texts we are preparing in the project Corpus Avesticum Berolinense seeks for the first time to reflect the dynamic nature of these texts, so that each ceremony is generated anew each time according to the performance parameters indicated by the user (Cantera 2019a). The variable texts are numerous (Cantera 2018). The text specifies three time coordinates at several moments of the performance:43 the part of the day or asńiia- ratu-,the date of the celebration (day and month), and the seasonal festival or yāiriia- ratu-, if the liturgy is performed during one of the six yearly festivals, the gāhānbār.44 The performance time sometimes has textual consequences beyond the standard time indications. This is, for example, the case of Y68.31 consisting of a collection of three quotations from Old Avestan. In Geldner’s standard edition of the Yasna, we find the following text without further instructions:

vohū. uxšiiā. manaŋhā. xšaϑrā. aṣ̌ācā. uštā. tanūm. [Y33.10c] (si bār “three times”)
imā. raōcā̊. barəzištəm. barəzimanąm. [Y36.6b] (si bār “three times”)
yaɱī. spəṇtā. ϑβā. maińiiū. uruuaēsē. jasō. [Y43.7a] (si bār “three times”)

Most manuscripts have exactly this text. Nonetheless, ms. 4045 introduces the following instruction:

har gāh dārēd 3 bār guftan ud ka yašt nōgnāwar ayāb sīh rōzag har sē gāh ī nēmrōz ud aybārag guftan har ēk 3 bār

43 For a survey of the moments in the liturgy when this information is provided, see Cantera 2018:21-22.

44 I have recently described the variations during the last festival of the year according to the Revāyat of Kāma Bohra (Cantera 2018:30-41). However, in the meantime, I have discovered a further one. In the Visperad and related ceremonies, the formula for “taking the word” (wāz gīrišnīh) used at the end of the Frauuarāne is, after the installation of the auxiliary priests, the “double taking of the word” (zōt yaϑā. ahū. vairiiō. yō. ātrauuaxšō. frā.mē. mrūtē. rāspīg aϑā. ratuš. aṣ̌āt̰cit̰. hacā. aṣ̌auuā. viδuuā̊. mraōtū. rāspīg yaϑā. ahū. vairiiō. yō. zaōtā. frā.mē. mrūtē. zōt aϑā. ratuš. aṣ̌āt̰cit̰. hacā. aṣ̌auuā. viδuuā̊. mraōtū.) instead of the one used in the Yasna (zōt yaϑā. ahū. vairiiō. zaōtā. frā.mē. mrūtē. rāspīg yaϑā. ahū. vairiiō. yō. zaōtā. frā.mē. mrūtē. zōt aϑā. ratuš. aṣ̌āt̰cit̰. hacā. aṣ̌auuā. viδuuā̊. mraōtū.). However, some Visperad manuscripts indicate at the first Frauuarāne of the second Drōn Yašt that the standard “taking of the word” of the Yasna is used here, as well in the Visperad. The usual one in the Visperad is used here only when it is a ceremony for gāhānbār (ms. 2109, Cantera 2019b):

agar xšnūman gāhānbār bēd zōt yaϑā. ahū. vairiiō. yō. ātrauuaxšō. frā.mē. mrūtē. rāspīg aϑā. ratuš. kardan hamrāspīg yaϑā. ahū. vairiiō. yō. zaōtā. frā.mē. mrūtē. zōt aϑā. ratuš. kardan agar ēzišn ī gāhānbār nē bēd zōt yaϑā. ahū. vairiiō. zaōtā. frā.mē. mrūtē. rāspīg yō. zaōtā zōt aθā. ratuš

If it is the šnūman for the gāhānbār, then the zōt (should say) yaϑā. ahū. vairiiō. yō. ātrauuaxšō. frā.mē. mrūtē., the rāspīg answers: aϑā. ratuš. The same rāspīg (continues with) yaϑā. ahū. vairiiō. yō. zaōtā. frā.mē. mrūtē., the zōt answers aϑā. ratuš. If it is not a ēzišn for the gāhānbār, then the zōt (should say) yaϑā. ahū. vairiiō. zaōtā. frā.mē. mrūtē., the rāspīg then yō. zaōtā and the zōt answers aθā. ratuš.

It is not clear to me why this difference should apply only to the Frauuarāne of the second Drōn Yašt, but the manuscripts indicate this alternative only at this point.

45 Very likely a modern copy of ms. 10 (Mf1) by Rustam Goštāsp that I have not seen until now. This manuscript takes a very special position among the Yasna manuscripts because it includes more and longer ritual instructions.


He should say three times the text of the corresponding gāh.46 When it is a Yašt Nōgnāwar or Sīrōzag, then the three gāh, the one of Bāmyazd,47 the one of noon and the one of aybārag (?) have to be recited, each one three times.

This instruction is better understood in the light of the instruction in New Persian we find in another manuscript of the Yasna, ms. 19:

agar hāwan bid vohū. uxšiiā. manaŋhā. xšaϑrā. aṣ̌ācā. uštā. tanūm. agar raftwan bid imā. raōcā̊. barəzištəm. barəzimanąm. agar uziran bid yaɱī. spəṇtā. ϑβā. maińiiū. uruuaēsē. jasō.

When it is hāwan, (he should recite) vohū. uxšiiā. manaŋhā. xšaϑrā. aṣ̌ācā. uštā. tanūm. When it is rapihwin, (he should recite) imā. raōcā̊. barəzištəm. barəzimanąm. When it is uziran, (he should recite) yaɱī. spəṇtā. ϑβā. maińiiū. uruuaēsē. jasō.

Regularly, the priest does not recite all three quotations, but only one depending on the time of the performance: Y33.10c in the morning, Y36.6b at noon, and Y43.7a in the afternoon.48 The three quotations are recited only in special ceremonies with the dedication for Nōg-nāwar and Sīrōzag, but this is only rarely indicated in the manuscripts.

Moreover, each ceremony can be held for different gods or set of gods, and this involves more complex textual variations than the time of the performance. The selection of the god depends partly on the calendar date of the celebration, as well as on the purpose of the ceremony. The rules are complex, and an important part of priestly instruction. The dedication is always announced at the beginning of the liturgy and repeated at the end of each ceremony. Besides, the dedicationappears as part of the litanies towards its end (Cantera 2018:25-27). The dedication consists of different parts:

  1. Ahura Mazdā and the Aməṣ̌a appear in almost all dedications of the LL and the Drōn Yašt with only a few exceptions, such as the dedication for Sraōša, or sometimes for Miϑra;
  2. the patrons of the corresponding part of the day;
  3. the actual god(s) of the dedication;
  4. the section for “all the gods,” known as vīspaēšąm;
  5. the frauuaṣ̌i-section, known as aṣ̌aōnąm.

Only the actual god of the dedication is mandatory. All the other sections might appear or

46 That is, the part of the day during which the liturgy is performed.

47 The god of the dawn.

48 This practice goes back at least to Sasanian times, as it is already alluded to in the Nērangestān (N29.6):

nōg-nāwar yašt-ē bowandagtar be kunišn u-š bowandagīh vohū uxšiiā manaŋhā imā̊ raocā̊ barəzištəm barəzimanąm ud ẏahmī spəṇtā θβā mainiiū uruuaēsē jasō har ēk 3 bār be gōwišn

The Yašt Nōg Nāwar has to be performed more completely. Its completeness consists of reciting each one of these texts three times: vohū uxšiiā manaŋhā (Y33.10c), imā̊ raocā̊ barəzištəm barəzimanąm (Y36.6b), and ẏahmī spəṇtā θβā mainiiū uruuaēsē jasō (Y43.7a).


not following complex rules that are never made explicit in the manuscripts, but are generally known by the priests. Furthermore, the number of possible dedications is high. They are mostly quite short, but some dedications are very long. Yasna manuscripts mostly contain the dedication of Minu-Nāvar. The Visperad manuscripts have the specific dedication for the seasonal festivals. Nonetheless, the latter often provide alternatives in some places (see below the example of ms. 2007). The information is, nonetheless, never complete and systematic. The Iranian manuscripts of the Vīdēvdād reproduce the ceremony for the god Sraōša, but sometimes an alternative dedication for Ahura Mazdā is indicated, although almost every dedication is possible, as the New Persian Revāyāt inform us. The manuscripts of the Vīštāsp Yašt are the most open ones. Most of them indicate in the first Frauuarāne that the dedication for the Daēna is required for the first performance of the day, but later on, every dedication is allowed (Cantera 2018:29 and 2020d; Martinez Porro 2022). At the rest of the positions where the dedication is expected, they just mention that the corresponding dedication has to be recited.

Some manuscripts try to complete, at least partly, the missing information regarding the dedications. The Yasna manuscripts by Rustam Goštāsp, for example, tend to include such information. His ms. 8, preserved today in the Fire Temple of Yazd and copied in 1706, includes, beside the Yasna, the Sīrōza and šnūman ī Minu-Nāwar. His ms. D83 at the Cama Oriental Institute is even more interesting (Dhabhar 1923:14-15). Beside a Yasna and a Visperad, it contains the Sīrōza and a series of Drōn Yašt from which almost only the dedications are copied (and the rest is abbreviated). I have recently seen a series of manuscripts of the Yasna at the National Library in Teheran (ms. 5-39196, 20570, 114-1689) that include a series of texts at the end similar to ms. D83. In fact, the longest dedication, the Sīrōzag,49 is copied in a few Yasna (mss. 8, 19, 87, 252) and Visperad (mss. 2109, 2698) manuscripts, and in some Khorde Avestā manuscripts.50 Furthermore, the manuscripts of the Drōn Yašt consist almost exclusively of long lists of dedications. However, in general, the impression is that the dedications, like the rest of the performative variations, were learned without the assistance of manuscripts for at least a thousand years after the creation of the first ones and that the partial inclusion of a list of dedications in the manuscripts is a late phenomenon, with the oldest example dating from the eighteenth century.

The presence of certain dedications entails further textual changes than just their bare text. The most relevant ones are two textual extensions when the dedication ends with the section for the frauuaṣ̌is: one is a text recited as well independently, called Stōm (Modi 1922:427-29). The other one is not known elsewhere (Y22.26-28 = Geldner’s Y23.1-3). The latter is recited in

49 The Sīrōzag is a single dedication (and not an independent text that might be recited separately) that includes the patron of the thirty days of the month extended by three further dedications: (1) Haōma, (2) Apąm Napat and the waters, and (3) Dahmā Āfriti, a closing one for all the material and immaterial gods. Its relative success in modern manuscripts is due not to the frequency of its use in the performance of Drōn Yašt or Yasna, but to the fact that it is a kind of catalogue of the most frequently used dedications, the ones for each patron of the day.

It has traditionally been regarded as a text that has an autonomous existence. Even E. Raffaelli in his book on the Sīrōzag recognizes that it is a dedication, but refers to it as “its ritual recitation,” assuming that there is also a non-ritual recitation (2014:4).

50 For the manuscripts of the Khorde Avesta, see Raffaelli 2014:66, but the list will be much longer in the analysis of this class of manuscripts being carried out by G. König within the framework of the Corpus Avesticum Berolinense.


the litanies āiiese yešti and aṣ̌aiia daδąmi51 in the Hōmāst (and also in the Drōn Yašt in the Visperad and related ceremonies).52 It appears with āiiese yešti after the dedication of Y22.25 (= VrS25.31 in the Visperad and in the second Drōn Yašt, VrS78.32), and with aṣ̌aiia daδąmi (in the second Drōn Yašt, VrS82.19 and after Y66.19 = VrS90.27).53

The longer extension in the yazamaide litany (after Y25.4 and VrS78.32), the Stōm, appears as well in the yazamaide litany of Y59 (Y59.19-28). When the frauuaṣ̌is are mentioned in the yazamaide in the LL, the standard verb yazamaide is substituted by staōmi zbaiiemi ufiiemi and extended with a series of adjectives for the frauuaṣ̌is (Y59.19):54

aṣ̌āunąm vaŋᵛhīš sūrā̊ spə̄ntā̊ frauuaṣ̌aiiō yazamaide staōmi zbaiiemi ufiiemi
yazamaide nmāńiiā̊ vīsiiā̊ zaṇtumā̊ dax́iiumā̊ zaraϑuštrō.təmā̊.

I praise, call, chant, and we made a yasna for the good, strong, and beneficent frauuaṣ̌is of the orderly ones, (the frauuaṣ̌is) that belong to house, the clan, the tribe, the country, and district of the zaraθuštrō.təma

(The text corresponding to Y59.20-28 follows.)

The Yasna manuscripts include both texts in extenso because the dedication of Minu-Nāwar ends with the frauuaṣ̌is. However, all the other manuscripts of the other liturgies omit these extensions. Nonetheless, some manuscripts mention the possibility of their appearance. Thus, ms. 2007 describes in first instance a Visperad to be performed during one of the six yearly festivals (gāhānbār), but almost systematically details the differences when it is performed with a dedication that ends with the mention of the frauuaṣ̌is. When the use of such a dedication does not involve major differences, it indicates that any šnūman can be generally used, and then describes the use of the dedication for the gāhānbār (for example, VrS27.43 in the litany āuuaēδaiiamahi of the Hōmāst):

ā[at̰]. d[īš]. āuu[aδaiiamahi]. xšnūman ān ī bēd wizārdan agar xšnūman ī gāhānbār bēd āat̰. dīš. āuuaēδaiiamahi. raϑβō. bərəzatō. yō. aṣ̌ahe. raϑβąm. aiiaranąmca.

ā[at̰]. d[īš]. āuu[aδaiiamahi]. He should perform the corresponding dedication. If it is the dedication for the gāhānbār, (he should say:) āat̰. dīš. āuuaēδaiiamahi. raϑβō. bərəzatō. yō. aṣ̌ahe. raϑβąm. aiiaranąmca.

51 This litany appears displaced at the end of the ceremony (Y66) when the actual offering of the libation prepared during the Hōmāst takes place (Cantera 2020c).

52 It corresponds to the text edited by Geldner as Y23.1-3.

53 The text corresponds to Geldner’s Y23.1-3 with āiiese yešti. The variant with aṣ̌aiia daδąmi is not edited in Geldner’s edition, but can be found in the edition of the Corpus Avesticum Berolinense (that is, in Y66.19).

54 The text appears in Geldner as Y26.


When the dedication for the frauuaṣ̌is entails extensions, then a different formulation is used (for example, VrS25.31-32, dedication in the litany āiiese yešti of the Hōmāst):

agar xšnūman dudigar bēd ka xšnūman ahlawān padiš bēd auuaŋ́hā̊. frauuaṣ̌aiiō. yā̊. paōiriia. ā̊ŋharə . . .

If it is a different dedication (from the one for the gāhānbār), when the dedication contains ahlawān (he should say): auuaŋ́hā̊. frauuaṣ̌aiiō. yā̊. paōiriia. ā̊ŋharə . . . (Y23.1-3).

Accordingly, the actualization of the Avestan texts in performance requires two kinds of knowledge, namely, the long parts of the core text and the variable sections, and their combination. The manuscripts basically contain the former and, sometimes, a few short and unsystematic instructions regarding the latter. Their understanding is restricted, however, to experts who know the texts alluded to and their combinations. This lore of ritual and textual knowledge was only transmitted orally until the eighteenth century, and even then it appeared in the manuscripts in only limited form. Thus, Avestan manuscripts play a very specific role in a mainly oral tradition: they function as a prescriptive tool for preserving the performance from changes and, at least from the seventeenth century on, serve as a tool for memorizing the texts. These circumstances have important consequences both for their production and for the methods of analysis that might be used for the Avestan manuscripts. They are simply a link in a chain in which oral methods and processes dominate. Their position and relevance have changed throughout history, but even in modern times, where manuscripts have been replaced by printed books, their place remains subordinate to that of memory and oral performance, except in Iran, where the oral tradition has completely dried up in the twentieth century.

5. The Interplay of Memory, Performance, and Written Sources in the Production of Manuscripts: The Priest as Scribe

The view of transforming an oral transmission into a written one as a one-time process has conditioned our understanding of the process of producing manuscripts. Once the first manuscripts had been produced, the creation of further manuscripts would be limited to copying one manuscript from another, and the oral transmission would have been abandoned or only residually maintained, but without influencing the written one. In this model, the scribe is exclusively an agent of the written transmission, who does not participate in the oral transmission. This model is, however, valid only for texts for which there was no oral transmission before the production of the manuscript, or whose oral transmission had been terminated through the production of the first written copies. In other scenarios, written and oral transmissions coexist and participate in a common interface for the preservation and review of texts, including the production of the manuscripts. In many traditions, scribes are not only agents of the written transmission, but also of the oral one, as we have seen for the Qurʾān in the previous section. Thus, the manuscript is not only the tradent, but shares this role with the scribe who knows the text he is copying by heart. In Kirk’s words, “the oral and the written vectors


intersect in the scribe” (2016:114). During their coexistence, both tradents, manuscript and memory, interact for the preservation and activation of the texts, albeit in different settings in the different traditions.

This is clearly the case of the Zoroastrian tradition. The scribes are the same priests that perform the rituals in which they recite the texts from memory, without the assistance of the manuscripts they produce (except, as we have seen, in the case of the Vīdēvdād ceremony in modern times). Nonetheless, the traditional view dominated the analysis of the Avestan manuscripts. Once the Avestan texts had been written down, the manuscripts would have behaved similarly to a purely written transmission. A manuscript is a copy of another one, and if there are any differences, these are due to error or contamination, that is, comparison of one source to another one (Geldner 1886-96:I, xlviib). For N. L. Westergaard (1852-54:19-20), the manuscripts were drafted in Sasanian times and distributed to the regions, but only the ones copied again in the region of Yazd-Kerman have survived. For Geldner, our manuscripts go back to the original manuscripts of a liturgical post-Sasanian redaction of the Avestan texts that occurred long after the edition of the Great Avesta (Geldner 1896b:xlvi). For Hoffmann and Narten, the existing manuscripts are again descended from the first Sasanian ones via a series of hyparchetypes for each class of manuscripts that must have existed around the tenth century: the theory of the hyparchetypes.55 Kellens’ modification of Hoffmann and Narten’s view replaces the Sasanian archetype through two “Stammhandschriften” that contained two ritual collections, and is therefore closer to Geldner’s approach (except for the simultaneity instead of linearity between Great Avesta and Ritual Avesta).

Despite the differences in detail, all the models operate within the same theoretical framework according to which once the first manuscripts had been produced, the process was limited to the simple copying of extant manuscripts (except for the second editorial process assumed by Geldner).56 The only relevant tradents of the text are the manuscripts, and the role of memory and performance is limited to the distortion of the original spelling of words through the so-called vulgatepronunciation. Hoffmann and Narten describe it in the following terms (1989:18):57

55 First, Hoffmann (1969) postulated a single archetype for all the variants of the LL. A few years later, H. Humbach (1973) did the same for the Vīdēvdād. He assumed two successive hyparchetypes: (1) a liturgical one whose relationship with the hyparchetype for all the variants of the LL is unclear; (2) an exegetical archetype. Kellens (1998:447 n. 449) further assumes a dependence on the liturgical manuscripts from the exegetical hyparchetype because of the glosses of the Pahlavi translation that sometimes appear in the liturgical manuscripts. However, as Ferrer (2012) has shown, this process affects only the Indian liturgical manuscripts, and even then, progressively. For a history of this theory, see Kellens 1998:466-73, and for general criticism, see Cantera 2012a:290-97; Tremblay 2012:118-25.

56 Skjærvø adopts a more nuanced position, postulating that the texts, once copied, “were probably corrected and edited for some time after” (2012:19). This “conscious interference” would have continued for centuries.

57 Geldner’s statement is quite similar (Geldner 1886-96:I, xlvii): “There are, especially, two influences at work which tend to detract from the fidelity of the manuscripts. On the one hand, the copyists knew the majority of their texts by heart. The oral text, however, had become more corrupted than the written text and keeps constantly crossing the latter. Scribes who read and copied word for word from the text before them, ran less danger than those scribes who grasped the entire sentence and wrote it off before looking again at the copy before them.”


Jeder Schreiber hatte gelernt, Avesta-Texte zu rezitieren, und zwar nicht aufgrund von Handschriften, sondern durch mündliche Weitergabe. In den meisten Fällen wird er das, was er kopierte, auswendig gekonnt haben. Mit diesem Klang im Ohr schrieb er ab und „verbesserte“ bewußt oder unbewußt seine Vorlage. Diese Erscheinung, die man Vulgata-Aussprache nennen kann, lagert überall unseren Handschriften. Selbst die ältesten sind davon schon schwer betroffen. Da der Schreiber jeweils schon wußte, wie es „eigentlich“ heißt, hat sich der Sinn für buchstabengetreues Kopieren nie voll entwickelt.

Although Hoffmann and Narten acknowledge that the scribe knew the text by heart, the changes introduced in the manuscripts seem to be limited to the spelling of the words, hence their designation of this phenomenon as vulgatepronunciation (Aussprache) and the insistence on the sound (Klang). As linguists, they were more concerned about the infinite number of small variations in the spelling of single words than about other more significant textual variations. The latter were simply explained according to the traditional methods applied for purely written transmissions. Moreover, the dependences of the manuscripts continued being analyzed exactly in the same terms established by Westergaard and Geldner: common “errors” reveal a common source, and when the data do not fit, we have to assume contamination from several sources (Geldner 1886-96:I, xlviib).

Scribal competence is, though, “memory-based” in the traditions in which manuscripts are ancillary to memory, as is the case of the Zoroastrian tradition (Kirk 2016:115). The role of the scribes of the liturgical Avestan manuscripts is not limited to simply copying a previous original as accurately as possible. Their goal is always to create a manuscript that represents the liturgy as it should be performed.58 To that purpose, they usually have two sources available: on the one hand, one or several written manuscripts, and on the other, their own experience of the performance and the text they know by heart. The preference they attach to one or the other source depends on many factors: personality of the scribe, authority of the available written sources, historical context (splitting of a community into two different schools, etc.), geographical differences, type of manuscript,59 and so forth. Thus, they might reproduce a previous copy more or less faithfully, modify it in order to adapt it to the current performance (modernization of the pronunciation, introduction of ritual or textual changes, etc.), or even create new manuscripts with no written source at all, or use manuscripts of other liturgies that are transformed into completely new manuscripts. We must abandon the notion of the copyist of the manuscript as a faithful reproducer of a previous copy, and replace it with one of a priest that produces a guide for the performance of the liturgy that serves both himself and others as an instrument for learning how to perform it and as an aid for memorizing the texts.

Scribes are active agents of the transmission. Sometimes, they can even create new manuscripts without any written source. Yasna manuscripts are the best examples of manuscripts probably made from memory, as it is the basic form of the LL, and the first and best learned by the priests. Some Indian manuscripts, such as ms. 231, reveal infinite phonetic variants that

58 Historical interest for past performances does not seem to play any role whatsoever.

59 Exegetical manuscripts are obviously less influenced by the ritual performance, so they reproduce major deformations of the text (for example, changes in the order of some folios) without correcting them (Cantera 2010).


reproduce the Indian pronunciation, and are quite far removed from the traditional spellings we find even in other Indian manuscripts.60 They seem to be closer to a record of a live performance than to a written source. Accordingly, we may confidently posit that they were copied directly from the scribe’s memory. In general, Indian liturgical Yasna manuscripts tend not to be copied from older liturgical manuscripts, but to be composed at least partly on the basis of the scribe’s memory and ritual knowledge. A good example is provided by ms. 100 (B3). Ardašīr Zīvā created a liturgical manuscript in the sixteenth century (ms. 100 [B3]) using as source Mihrābān Kayxōsrō’s exegetical Yasna manuscript 510 (K5).61 Its dependence is clearly revealed by a number of shared bizarre spellings (Cantera 2014:152). Nonetheless, the manuscript produced by Ardašīr Zīvā is very different from its original. Firstly, the initial ten folios (until 11v) include the Paragnā, the preliminary ceremony to the performance of the LL. This ceremony does not appear in any exegetical manuscript, and has most probably been added by Ardašīr Zīvā on the basis of his own knowledge. In fact, the Paragnā is very weakly represented in the manuscripts,62 except in the Indian liturgical Yasna manuscripts that probably follow the model of Ardašīr Zīvā. Secondly, whenever the liturgical text differs from the one reproduced in the exegetical manuscripts, as is the case of the beginning of the Old Avestan texts,63 Ardašīr does not follow his written source, but instead his ritual knowledge, copying the text as it is recited in the actual performance! He has also created for the first time (as far as I know) the set of ritual instructions for the performance of the Yasna in Gujarati. Thus, ms. 100 (B3) is a wonderful example of how the oral and the scribal vectors intersect in the person of Ardašīr Zīvā. Similarly, when Isfandyār Anuširvān copied two Yasna manuscripts (mss. 15 and 82) in the nineteenth century, he seems to have used a Visperad manuscript as a source that he has transformed into a Yasna manuscript with all the textual changes64 that involves (Cantera 2014:118-19).

Sometimes, the scribe, indeed, produced the manuscript of a liturgy on the basis of a different one. The case of the manuscripts of the Vīštāsp Yašt is quite interesting, as this liturgy is the one less frequently performed. Ms. F13 in the Meherjirāna Library is a copy of a manuscript by Āsdin Kākā that contains a Visperad Dō-Hōmāst (ms. 2065) and an abbreviated Vīštāsp Yašt (ms. 5030) (Dhabhar 1925; Martínez Porro 2013:74 and 2014:79). The original belongs to the Mulla Firuze collection and is preserved in the Cama Oriental Institute with the siglum D73 (Dhabhar 1923:87-8; Martínez Porro 2014:77). Āsdin Kākā’s manuscript is, in turn, a copy of a previous manuscript by Pešōtan Rām Kāmdīn. Martínez Porro (2020) has noted that the Visperad Dō-Hōmāst is ultimately a copy of Rōstam Mihrābān’s ms. 2000 (K7b). Moreover, he has

60 In his Ph.D. thesis, J. Martínez Porro has discovered similar features in some Vīdēvdād manuscripts (for example, 4220, 4370, 4410, 4425), but only outside the Vīdēvdād nask.

61 This scribe had at his disposal Mihrābān Kayxōsrō’s manuscripts copied in Khambhat, mss. 510 (K5) and 4610 (K1) (Cantera 2014:152).

62 About this ceremony, see Cantera 2020b. I have also published a first (not yet critical) edition of the ceremony in Corpus Avesticum Berolinense (https://cab.geschkult.fu-berlin.de/exist/apps/cab/pages/tools/ceremony_generator.html).

63 For these differences, see Cantera 2013a:27-29.

64 He occasionally did not completely succeed in the adaptation, as he maintained, for example, the number of barsom-twigs of the Visperad.


concluded from certain orthographic and paleographic features that Āsdin Kākā’s Vīštāsp Yašt (and consequently ms. 5030) also goes back to a lost original by Rōstam Mihrābān.

Interestingly, the initial ritual instruction is identical in the Visperad Dō-Hōmāst and in the Vištāsp Yašt:

barsom 33 tāg ud frāgām 2 tāg cīyōn pad yašt barsom bastan ud parāhōm kardan sāzišn hamāg xūb frāz nihādan parāhōm kardan rāy pad har hōmāst pad kamistīh 3 tāg ud urwarām pārag-ē ǰām andak-ē frāz nihišn

The barsom (should consist) of thirty-three twigs and the frāgām are two. He should tie the barsom like in the yašt, prepare the parāhōm, and arrange all the implements carefully. For the preparation of the parāhōm he has to arrange for each hōmāst at least three branches of hōm, a twig of urwarām, and a drop of ǰām.

This initial instruction is slightly different from the one we find in another famous Vīštāsp Yašt manuscript, ms. 5020 (K4):

barsom 33 tāg ud frāgām 2 tāg barsom bastan ud parāhōm kardan sāzišn hamāg frāz nihādan hōm pad kamistīh 3 tāg ud urwarām pārag-ē ud ǰām andak-ē frāz nihišn

The barsom (should consist) of thirty-three twigs and the frāgām two. He should tie the barsom, prepare the parāhōm, and arrange all the implements. He has to arrange the hōm, with at least three twigs and a branch of urwarām, and a drop of ǰām.

The main difference is that while ms. 5020 requires three twigs of hōm and urwarām for preparing the hōm, ms. 5030 calls for the same quantity for “each hōmāst.” This indication only makes sense in the previously copied Visperad Dō-Hōmāst. In this ceremony, and besides the standard Hōmāst, there is a second one after the first Yasna Haptaŋhāiti (Cantera 2020e). Accordingly, six bundles of three twigs of hōm and urwarām are required (three for each hōmāst), and not three. Thus, we may postulate that Rōstam Mihrābān’s Vīštāsp Yašt was, in fact, an adaptation of a Visperad Dō-Hōmāst.65 This probably required from Rōstām Mihrābān or the scribe of his original considerable ritual and textual knowledge, even perhaps the use of an exegetical manuscript of the Vīštāsp Yašt.66

The Iranian manuscripts of the Vīštāsp Yašt (the oldest are mss. 5010 and 5020 [K4]) hark back, according to their colophons, to an original by Xōsrōšāh Anōšagruwān (Cantera 2014:113-14), who copied it in 1344. Although both manuscripts are copied from the same

65 An alternative hypothesis could be that the first ritual instruction was taken from the Visperad Dō-Hōmāst because, for example, the first page of the original Vištāsp Yašt manuscript was lost. We should consider that not the whole introduction to the manuscript is identical to the one of the Visperad, but only this first ritual instruction. The preceding section is different in both manuscripts.

66 A similar phenomenon lies at the origin of another liturgical manuscript of the Vīštāsp Yašt (ms. 5102). It contains the introduction of a Visperad, revealing that it was probably “reconstructed” on the basis of a Visperad manuscript.


original, one has the ceremony in extenso and the other in extremely abbreviated form. Thus, the scribes were able either to extend an abbreviated original or to shorten a complete one. Both procedures require a profound knowledge of the ceremony. More interesting for us is the fact that ms. 5010 has the same ritual instruction that we find in ms. 5030 (the copy dating back to Rōstam Mihrābān), whereas ms. 5020 records the modified one. Hence we should conclude, first, that the Iranian manuscripts also have the same origin as Rōstam Mihrābān’s Vīštāsp Yašt,67 and second, that either Wehmard Frēdōn, the scribe of ms. 5020, or Wahrom Marzbān, as his source, corrected the initial ritual instruction when noticing that it is inappropriate for a Vīštāsp Yašt.

The beginning of the Vīštāsp Yašt provides us with another good example of the changes that the scribes introduced in the manuscripts. Martínez Porro (2022) has revealed a difference between the Vīštāsp Yašt manuscript going back to Rōstam Mihrābān and the Iranian manuscripts. Whereas the Iranian ones (mss. 5010 and 5020) mention that the first performance of the day has to be held with the dedication for the Dēn, ms. 5030 has a similar instruction but with a variant of this dedication, the one for Dēn and Māraspand. As both branches seem to have a common source, this implies that either the Indian or the Iranian branch has adapted the dedication to their actual practice.

Manuscripts are a factor of conservatism in a ritual tradition that, like all ritual traditions, experiences the dilemma between change and continuity. However, conservatism is only possible to a certain extent. The limit is always the link to the real performance. If manuscripts stray too far from daily practice, they become useless. Accordingly, on each occasion the scribes have to define their position in the axis between the two functions of liturgical manuscripts: prescriptive and mnemonic. We can easily understand the dilemma in the case of the adaptation, or not, to the modern pronunciation. The confusion between ī and ū provides a good example. We already find at least one instance of it in Mihrābān Kayxōsrō at the beginning of the fourteenth century: ms. 500 [K5] ỵaēš.yạṇtūm instead of ẏaēšiiaṇtīm in Y9.11. The oldest extant Safavid Iranian liturgical manuscripts show that both sounds have merged together in the recitation at the beginning of the seventeenth century. However, the manuscripts copied before 1622 still use two

67 This could be confirmed by another striking fact. The regular wāz gīrišnīh of the Frauuarāne is the so-called double wāz gīrišnīh in the solemn ceremonies (Visperad, Vīdēvdād, and Vīštāsp Yašt) (Cantera 2016:53). By contrast, before the installation of the auxiliary priests and in the Yasna, we find the wāz gīrišnīh that starts with the zaōtar saying yaθā ahū vairiiō zaōtā . . . and continues with the auxiliary priest saying yaθā ahū vairiiō yō zaōtā . . . . The only exception to this rule is the Frauuarāne of the second Drōn Yašt (VrS79.0). Here the manuscripts of the Visperad and the Vīdēvdād are the same as in the first Drōn Yašt. Only the manuscripts of the Vīštāsp Yašt (with the exception of ms. 5102) have the double wāz gīrišnīh. Interestingly, the indication in ms. 5010 is the same used by ms. 5030 and the standard one in Rōstam Mihrābān’s manuscripts: t ud rāspīg wāz gīrišnīh dōgānag kardan, “The zōt and the rāspīg have to make the double wāz gīrišnīh.” Ms. 5020 copies the formula in extenso. There are two possible explanations: (1) the manuscripts of the Visperad and the Vīdēvdād have taken the Frauuarāne of the Srōš Drōn and used it incorrectly in the second case; (2) there is a real ritual difference between the Drōn Yašt and the rest of the ceremony, and the simple wāz gīrišnīh is correct in the all the manuscripts, except the ones of the Vīštāsp Yašt.

Although there is no certainty, I consider the first hypothesis more likely, as the scribes are usually well aware of the difference between the first and the second Drōn Yašt, and copy them faithfully. Accordingly, the dissimilar wāz gīrišnīh in all Vīštāsp Yašt might be a common error harking back to Rōstam Mihrābān’s manuscript or his source. If, as it seems, all liturgical Vīštāsp Yašt manuscripts go back to one produced by Rōstam Mihrābān or his source on the basis of a manuscript for the Visperad Dō-Hōmāst, its late creation could perhaps explain the bad state of preservation of the fragards of the Vīštāsp Yašt.


different letters for ī and ū, but frequently in the wrong positions. The manuscripts copied after 1622 all have ī, even when they copy from sources still distinguishing both sounds. They no longer claim that the distinction is maintained in the recitation and adapt to the actual performance.

The changes introduced are often more significant than simply modernizations of the pronunciation. We have already seen the changes introduced at the beginning of the Vīštāsp Yašt concerning the ritual instructions and also the dedication. Even more important changes are sometimes introduced. The case of the manuscript K11 (ms. 110 + 2220) is most illustrative. Geldner (1886-96:I, xxxvi-vii.) has rightly recognized that this manuscript must have been copied from an Iranian original. Its scribe, Dārāb Hirā Cāndā, was one of the addressees of the Revāyat of Bahman Isfandyār that was accompanied by a manuscript of the Visperad and one of the Vīštāsp Yašt (Cantera 2014:166-67). However, he has adapted his Iranian original to the Indian practice and introduced the necessary changes in the text. Some years ago, I have noted some textual divergences between the Iranian and Indian performance of the Visperad (Cantera 2014:266-67), namely, in VrS70 (after the Airiiaman Išiia) and VrS93 (following the recitation of the Spəṇtā.maińiiū Hāiti at the end of the Āb-zōhr). Both are probably Indian innovations. Although Dārāb Hirā Cāndā has used an Iranian original for ms. 2220, he follows Indian practice, and in both passages has the standard text of the Indian performance. He did not just copy his Iranian original, but adapted it to Indian practice. He introduced ritual instructions in Gujarati and changed the text wherever necessary to fit Indian practice. Despite the quest for ritual information among the Iranian priesthood, this Indian priest was not ready to fully adapt Indian practice to the Iranian one. Accordingly, the manuscript he produced did not just reproduce the Iranian original. He created a guide to the performance as he believed it should be performed.

This capacity for adapting the manuscripts to the actual performance is, of course, imperative for the tradition of liturgical manuscripts. The contrary would have meant its end, as manuscripts would become obsolete sooner or later. The idea of a Sasanian liturgical archetype and several hyparchetypes that were mechanically copied until modern times is untenable. The comparison of the liturgies as attested in the manuscripts with the Nērangestān shows, as well as obstinate ritual continuity (Cantera 2014:199-216), also the inevitable variation (Cantera 2014:248-58). There are numerous examples of modernization of the text, but the most impressive is probably the adaptation of the manuscripts to the disappearance of the animal sacrifice in the LL. At the time of the drafting of the Nērangestān, the animal sacrifice and the meat offering to the fire were a component of the LL in some of its most solemn performances. The main features of the rite are described in Chapter 47 of the Nērangestān (Kotwal and Kreyenbroek 2003:198-218; Cantera 2014:255-57 and 2022:73-89; Panaino 2017). When the animal was killed during the performance, the slaughter happened at some moment after the consumption of the drōn (a sacred bread). To that purpose, the priest bringing the animal is summoned with pasauuaŋhəm āstaiia,“I put in place the conductor of the animal.” Before Y34.20, the priest in front of the fire says: aētā̊sə.tē ātarə zaōϑrā̊, “these offerings are for you, o Fire.” The offering then took place during the Yeŋ́hē.Hātā of Y34.20 and Y35.1-2. The Sasanian manuscripts of the LL should at this point have contained ritual instructions similar to the ones found in N47.39. Nevertheless, none of the extant manuscripts shows any trace of these sections.


Why should they? What would be the purpose of including sections that are no longer performed?

The influence between manuscript and performance is reciprocal. The manuscripts intend to consciously shape the performance, and the performance, in turn, leads to conscious and unconscious changes in the manuscripts. Accordingly, the spread of variants should not be attributed exclusively to the copying process. The prevailing ritual practice in a community plays a role in this regard that I consider even more relevant than the copying process. We have detailed numerous generalizations of variants that go far beyond the boundaries of manuscript classes or groups of related manuscripts. They can best be explained as variants that have entered the performance, and from there have spread to manuscripts of different classes. There are numerous examples (Cantera 2012a:305 ff.). Remember, for example, the aforementioned generalization of the Indian variants in all Indian manuscripts of the Visperad and the Vīdēvdād.

The contrary is also true. Errors in the written transmission might enter ritual practice and thus jump to other classes of manuscripts. As I showed some years ago (Cantera 2012a:305-06), a copy-error in V3.14 produced two different variants in the recitation of the Vīdēvdād. The original text as it appears in the Iranian manuscripts and in the oldest Indian ones, such as ms. 4200, 4210, 4240, is:

spaiieti draōšəm
spaiieiti yātuγnīm
spaiieiti aṣ̌auuagnīm

The (daēnā) atones for thievery, it atones for killing through wizardry, it atones for the killing of the pious man.

Some Indian manuscript have one of these two variants:

spaiieti draōšəm
spaiieiti yātuγnīm
spaiieiti aṣ̌auuagnīm
spaiieti draōšəm
spaiieiti auuaγnīm
spaiieiti yātuγnīm
spaiieiti aṣ̌auuagnīm

The variant B is clearly an error of transmission: a scribe omitted two lines and copied the end of [aṣ̌]auuaγnīm after draōšəm.68 He noticed the omission and completed the missing text after using dots to delete the incorrect auuaγnīm. A later scribe, perhaps the scribe of ms. 4250 (O2), the oldest known witness of this error, overlooked deletion through dots and copied auuaγnīm as part of the text. The mistake is reproduced in manuscripts 4370 [L5] (1792) and 4400 (1802). They might have been just copies of ms. 4250 (O2), although this is very unlikely because ms. 4400 is, together with 4360 and 4320 (L2), one of the few Indian manuscripts that do not share the error V9.14 paoruuahe instead of the correct grauuahe (Cantera 2014:47) or the

68 For the details of this explanation see Cantera 2012a:305.


innovation V9.29 arəδβəm instead of the correct arəδəm (Cantera 2014:48). It seems instead that this variant entered the performance and from there spread to the manuscripts. Part of the Indian tradition introduced spaiieiti through the parallelism with the preceding and following sentences in an attempt to correct the text. The oldest attestation of this variant is already found in Surat in 1759 (ms. 4320 [L2]), and already has a Middle Persian translation in 1815 in Navsāri (ms. 4670). This variant was the most frequent reading in nineteenth-century manuscripts, independently of their genealogical origin. Moreover, this passage is repeated again in V8.29, and the manuscripts recorded exactly the same distribution. This copy-error obviously does not suffice to explain the variant in both passages. The process is much more complex, and the manuscripts’ influence on the performance and vice versa is a factor that cannot be disregarded.

We are thus in a better position for understanding the bizarre error Y12.3 ziiā̊iienīm discovered and correctly explained by Hoffmann (1969). It is clearly a mistake that took place in the course of the written transmission: while copying ziienīm,one copyist mistakenly wrote ziiā̊ instead ziie. He noticed his error, deleted iiā̊ with deletion dots, and completed the word correctly. Further copyists did not notice the deletion dots and copied it ziiā̊iienīm. Hence, he deduced the necessity of a common written hyparchetype for all Yasna manuscripts. Yet other explanations are also possible. This bizarre spelling appears in all kinds of manuscripts in which this passage is included: liturgical, exegetical, Yasna, Visperad, Vīdēvdād, and so forth.69 Therefore, we must suppose that the spreading did not take place solely through automatic copying. It appears even in the Yasna copies that seem to have no written source, but to have been copied from memory. Thus, this variant must have entered at some point in history the ritual practice, perhaps through the prestige of the scribe that made the original mistake or through other reasons about which we can only speculate. Thus, the performance can have triggered as well the spreading of the variant to all classes of manuscripts. Although we know today that it is the result of a mistake, it was felt by the priests as the right one.

The reciprocity of the influence between manuscripts and memory/performance is the missing link in the apparent discrepancy between philological argumentation and historical plausibility that Tremblay has perfectly described (2012:130):

En particulier il semble qu’il y ait contradiction entre la démonstration philologique d’un goulet d’étranglement de la tradition, une corruptèle d’un seul manuscrit ayant entraîné tous les autres dans sa chute comme s’ils étaient copiés sur lui, et la démonstration historique qu’il n’a jamais existé de manuscrit unique de tout l’Avesta, ni une lignée unique remontant à un seul manuscrit adamique.

The emphasis on the written transmission and the disregard of the influence of memory as an alternative tradent of the text (reduced to a source of multiple minimal but annoying mistakes and corruptions) is responsible for the discrepancy between the philological postulate of

69 It is unclear to me how the supporters of the hyparchetypes theory explain the spreading of this error from one archetype to manuscripts of different classes and different liturgies.


the existence of hyparchetypes,70 and the historical improbability of a single common source for manuscripts belonging to different classes and groups. The Avestan texts survive mainly through their activation in the performance and in the memory of the priests that have to perform them. The manuscripts are simply an auxiliary tool in this process, and not the focal point. Therefore, the analysis of the transmission of the rituals in Avestan languages should eventually consider that ritual practice has shaped the manuscripts at least as much as the manuscripts have shaped ritual practice.

Freie Universität Berlin


Ahmadi 2018
Amir Ahmadi. “On the Coherence of Yasna: A Critical Assessment of Recent Arguments.” Bulletin of the School of Oriental and African Studies, 81.1:57-82.

Andrés-Toledo 2016
Miguel ángel Andrés-Toledo. The Zoroastrian Law to Expel the Demons: Wīdēwdād 10-15: Critical Edition, Translation and Glossary of the Avestan and Pahlavi Texts. Wiesbaden: Harrassowitz.

Anklesaria 1888
Tehmuras Dinshaw Anklesaria. Avesta: The Sacred Books of the Parsis. Part I. Yasna bâ Nîrang. Bombay: Fort Printing Press.

Assmann 1983
Aleida Assmann. “Schriftliche Folklore: Zur Entstehung und Funktion eines Überlieferungstyp.” In Schrift und Gedächtnis: Beiträge zur Archäologie der literarischen Kommunikation. Ed. by Aleida Assmann, Jan Assmann, and Christof Hardmeier. Munich: Fink. pp. 175-93.

Bailey 1943
Harold W. Bailey. Zoroastrian Problems in the Ninth-Century Books. Oxford: Clarendon.

Boyd and Darrow 1982
James W. Boyd and William R. Darrow, dir. A Zoroastrian Ritual: The Yasna. Fort Collins: Colorado State University.

Cantera 2004
Alberto Cantera. Studien zur Pahlavi-Übersetzung des Avesta. Wiesbaden: Harrassowitz.

Cantera 2010
______. “Lost in Transmission: The Case of the Pahlavi-Vīdēvdād Manuscripts.” Bulletin of the School of Oriental and African Studies, 73.2:179-205.

70 In this light, the common errors adduced as arguments for the hyparchetypes should be better explained considering factors such as the prestige of single scribes and the reciprocal influence between manuscripts and performance.


Cantera 2012a
______. “Building Trees: Genealogical Relations Between the Manuscripts of Wīdēwdād.” In Cantera 2012b:279-346.

Cantera 2012b
______, ed. The Transmission of the Avesta. Wiesbaden: Harrassowitz.

Cantera 2012c
______. “Los manuscritos de Yasna con traducción pahlaví e instrucciones rituales (abāg zand ud nērang).” In EOS700. Séptimo Centenario de los Estudios Orientales en Salamanca. Salamanca: Universidad de Salamanca.

Cantera 2013a
______. “The Old Avestan Texts in the Videvdad and Visperad Ceremonies.” In Le sort des Gathas et autres études iraniennes: In memoriam Jacques Duchesne-Guillemin. Ed. by Éric Pirart. Leuven: Peeters. pp.25-48.

Cantera 2013b
______. “Talking with God: The Zoroastrian ham.paršti or Intercalation Ceremonies.” Journal Asiatique, 301.1:85-138.

Cantera 2014
______. Vers une édition de la liturgie longue zoroastrienne: Pensées et travaux préliminaires. Paris: Association pour l’Avancement des Études Iraniennes.

Cantera 2016
______. “The Taking of the wāz and the Priestly College in the Zoroastrian Long Liturgy.” Journal Asiatique, 304.1:47-63.

Cantera 2018
______. “Textual Performative Variation in the Long Liturgy: the Ceremonies of the Last Ten Days of the Year.” Dabir, 6:16-47.

Cantera 2019a
______. Corpus Avesticum Berolinense. Berlin: Corpus Avesticum Berolinense. https://cab.geschkult.fu-berlin.de

Cantera 2019b
______.  The Ritual Instructions in Pahlavi for the Performance of the Visperad. Berlin: Corpus Avesticum Berolinense.

Cantera 2020a
______. “Ast kē ēdōn gōwēd: Les opinions divergentes et la prise des décisions doctrinales dans le zoroastrisme sassanide et post-sassanide.” In Quand les dualistes polémiquaient: Zoroastriens et manichéens: Actes du colloque organisé au Collège de France, Paris, 12-13 June 2015. Ed. by Flavia Ruani and Mihaela Timuş. Paris: De Boccard. pp.31-64.

Cantera 2020b
______. “Avestan Texts in Contexts: Fragments Westergaard 6 and 7 and the Paragnā.” In Zaraθuštrōtəmō: Zoroastrian and Iranian Studies in Honour of Philip G. Kreyenbroek. Ed. by Shervin Farridnejad. Irvine: University of California. pp.69-105.

Cantera 2020c
______. “Litanies and Rituals. The Structure and Position of the Long Liturgy


within the Zoroastrian Ritual System.” In Aux sources des liturgies indo-iraniennes. Ed. by Céline Redard, Juanjo Ferrer-Losilla, Hamid Moein, and Philippe Swennen. Liège: Presses Universitaires de Liège. pp. 195-282, 377-400.

Cantera 2020d
______. “The Ritual Instructions in Pahlavi for the Performance of the Vīštāsp Yašt.” In The Yasna: A Dynamic Edition: Corpus Avesticum Berolinense. Ed. by Alberto Cantera. Berlin: Corpus Avesticum Berolinense.

Cantera 2020e
______. “A Brief Note on the Possibilities and Limitations Involved in Reconstructing the Historical Performances of the Avestan Liturgies: The Case of the Dō-Hōmāst.” Estudios Iraniaos y Turanios, 4:7-28.

Cantera 2022
______. “The Offering to Satisfy the ratu (miiazda ratufrī): the Dual System of the Animal Sacrifice in Zoroastrian Rituals.” In The Reward of the Righteous: Festschrift in Honour of Almut Hintze. Ed. by Alberto Cantera, Maria Macuch, and Nicholas Sims-Williams. Wiesbaden: Harrassowitz. pp. 39-96.

Cantera and de Vaan 2005
Alberto Cantera and Michiel de Vaan. “The Colophon of the Avestan Manuscripts Pt4 and Mf4.” Studia Iranica, 34.1:31-42.

Cereti 2008
Carlo Cereti. “On the Pahlavi Cursive Script and the Sasanian Avesta.” Studia Iranica, 37.2:175-95.

Christensen 1936
Arthur Christensen. L’Iran sous les Sassanides. Copenhagen: Levin and Munksgaard.

Cribiore 1996
Raffaella Cribiore. Writing, Teachers, and Students in Graeco-Roman Egypt. Atlanta: Scholars Press.

Cribiore 2001
______. Gymnastics of the Mind: Greek Education in Hellenistic and Roman Egypt. Princeton, NJ: Princeton University Press.

Darmesteter 1892
James Darmesteter. Le Zend-Avesta. Paris: Leroux.

Dhabhar 1923
Bomanji Nusserwanji Dhabhar. Descriptive Catalogue of Some Manuscripts Bearing on Zoroastrianism and Pertaining to the Different Collections in the Mulla Feroze Library. Bombay: The Trustees of the Parsi Punchayet Funds and Properties.

Dhabhar 1925
______. Descriptive Catalogue of All Manuscripts in the First Dastur Meherji Rana Library, Navsari. Bombay: Commercial Printing Press of the Tata Publicity Corporation.


Elman and Moazami 2014
Yaakov Elman and Mahnaz Moazami. “Zand ī Fragard ī Juddēwdād.” In Encyclopaedia Iranica. Ed. by Ehsan Yarshater. New York: Encyclopaedia Iranica Foundation. http://www.iranicaonline.org/articles/zand-fragard-jud-dew-dad

Ferrer-Losilla 2012
Juanjo J. Ferrer-Losilla. “Citations avestiques de la Traduction Pehlevie dans la tradition manuscrite Vīdēvdād-Sāde.” In Cantera 2012b:395-415.

Ferrer-Losilla 2016
______. “Preconsonantal Nasals in the Avestan Alphabet.” In Études de linguistique iranienne: In memoriam Xavier Tremblay. Ed. by Éric Pirart. Leuven: Peeters. pp. 157-84.

Finnegan 1973
Ruth Finnegan. “Literacy Versus Non-Literacy: The Great Divide.” In Modes of Thought: Essays on Thinking in Western and Non-Western Societies. Ed. by Robin Horton and Ruth Finnegan. London: Faber. pp.112-44.

Finnegan 1977
______. Oral Poetry: Its Nature, Significance, and Social Context. Cambridge: Cambridge University Press.

Foley 1991
John Miles Foley. “Orality, Textuality, and Interpretation.” In Vox Intexta: Orality and Textuality in the Middle Ages. Ed. by A. N. Doane and Carol Braun Pasternak. Madison: University of Wisconsin Press. pp.34-45.

Geldner 1886-96
Karl Friedrich Geldner. Avesta: The Sacred Books of the Parsis. Stuttgart: Kohlhammer.

Geldner 1896a
Karl Friedrich Geldner. “Avestische Literatur.” In Grundriss der iranischen Philologie. Vol. 2. Ed. by Wilhelm Geiger and Ernst Kuhn. Strassburg: K. J. Trübner. pp. 1-53.

Geldner 1896b
______. “Prolegomena.” In Geldner 1886-96:I, i-liv.

Herzog 2012
Thomas Herzog. “Orality and the Tradition of Arabic Epic Storytelling.” In Medieval Oral Literature. Ed. by Karl Reichl. Berlin: De Gruyter. pp. 629-52.

Hoffmann 1969
Karl Hoffmann. “Zur Yasna-Überlieferung.” Münchener Studien zur Sprachwissenschaft,26:35-38.

Hoffmann 1971
______. “Zum Zeicheninventar der Avesta-Schrift.” Festgabe Deutscher Iranisten zur 2500 Jahrfeier Irans. Ed. by Wilhelm Eilers. Stuttgart: Hochwacht Druck. pp. 64-73.

Hoffmann 1986
______. “Avestisch ṣ̌.” In Studia grammatica Iranica: Festschrift für Helmut


Humbach. Ed. by Rüdiger Schmidt and Prods Oktor Skjærvø. Munich: R. Kitzinger. pp. 163-83.

Hoffmann and Narten 1989
Karl Hoffmann and Johanna Narten. Der Sasanidische Archetypus: Untersuchungen zu Schreibung und Lautgestalt des Avestischen. Wiesbaden: L. Reichert.

Humbach 1973
Helmut Humbach. “Beobachtungen zur Überlieferungsgeschichte des Awesta.” Münchener Studien zur Sprachwissenschaft, 31:109-22.

Huyse 2008
Philip Huyse. “Late Sasanian Society between Orality and Literacy.” In The Sasanian Era: The Idea of Iran. Ed. by Vesta Sarkhosh Curtis and Sarah Stewart. London: I. B. Tauris. pp. 140-55.

Kanga 1966
Maneck Fardunji Kanga. “Epistle I, Chapter IV of Manuscihr Gošnjaman: A Critical Study.” Indian Linguistics, 27:46-57.

Kellens 1998
Jean Kellens. “Considérations sur l’histoire de l’Avesta.” Journal Asiatique, 286:451-519.

Kellens 2006
______. Études avestiques et mazdéennes vol. 1: Le Ratauuō vīspe mazišta (Yasna 1.1 à 7.23 avec Visprad 1 et 2). Paris: De Boccard.

Kellens 2007
______. Études avestiques et mazdéennes vol. 2: Le Hōm Stōm et la zone des déclarations (Y7.24-Y15.4 avec les intercalations de Vr3 à 6). Paris: De Boccard.

Kellens 2010
______. Études avestiques et mazdéennes vol. 3: Le long préambule du sacrifice (Yasna 16 à 27, avec les intercalations de Visprad 7 à 12). Paris: De Boccard.

Kellens 2011
______. Études avestiques et mazdéennes vol. 4: L’acmé du sacrifice: Les parties récentes des Staota Yesniia (Y27.13-Y59) avec les intercalations de Visprad 13 à 24 et la Dahmā Āfriti (Y60-61). Paris: De Boccard.

Kellens 2012
______. “Contre l’idée platonicienne d’Avesta ou les Considérations revisitées.” In Cantera 2012b:49-58.

Kellens and Redard 2013
Jean Kellens and Céline Redard. Études avestiques et mazdéennes vol. 5: La liquidation du sacrifice (Y62 à 72). Paris: De Boccard.

Kiparsky 1976
Paul Kiparsky. “Oral Poetry: Some Linguistic and Typological Considerations.” In Oral Literature and the Formula. Ed. by Benjamin A. Stolz and Richard S. Shannon. Ann Arbor: Center for the Coördination of Ancient and Modern Studies, University of Michigan. pp. 73-106.


Kirk 2016
Alan Kirk. “Manuscript and Memory.” In Q in Matthew: Ancient Media, Memory, and Early Scribal Transmission of the Jesus Tradition. London: Bloomsbury T & T Clark. pp.93-150.

König 2010
Götz König. “Der Pahlavi-Text Zand ī Fragard ī Juddēvdād.” In Ancient and Middle Iranian Studies: Proceedings of the 6th European Conference of Iranian Studies, Held in Vienna, 18-22 September 2007. Ed. by Maria Macuch, Dieter Weber, and Desmond Durkin-Meisterernst. Wiesbaden: Harrassowitz. pp. 115-32.

König 2012
______. “Das Nask Bayān und das Xorde Awesta.” In Cantera 2012b:355-94.

König 2014
______. “Die Pahlavi-Literatur des 9./10. Jh. und ihre frühe Kodex-Überlieferung (II): Die Pahlavi-Kodices der Bundār Šāhmardān-Familie.” Estudios Iranios y Turanios, 1:43-74.

König 2015
______. “Brief Comments on the So-Called Xorde Avesta (1).” Dabir, 1:17-21.

König 2016
______. “The Niyāyišn and the bagas (Brief Comments on the so-called Xorde Avesta, 2).” Dabir, 2:16-24.

Kotwal and Boyd 1991
Firoze M. P. Kotwal and James W. Boyd. A Persian Offering the Yasna: A Zoroastrian High Liturgy. Paris: Association pour l’avancement des études iraniennes.

Kotwal and Kreyenbroek 1995
Firoze M. Kotwal and Philip G. Kreyenbroek. The Hērbedestān and Nērangestān. Volume II: Nērangestān, Fragard 1. Paris: Association pour l’avancement des études iraniennes.

Kotwal and Kreyenbroek 2003
______. The Hērbedestān and Nērangestān. Volume III: Nērangestān, Fragard 2. Paris: Association pour l’avancement des études iraniennes.

Kotwal and Kreyenbroek 2009
______. The Hērbedestān and Nērangestān. Volume IV: Nērangestān, Fragard 3. Paris: Association pour l’avancement des études Iraniennes.

Kreyenbroek 2008
Philip G. Kreyenbroek. “The Term Bâgân Yasn and the Function of the Yashts in the Zoroastrian Ritual.” In One for the Earth: Prof. Dr. Mahyar Nawabi Memorial Volume. Ed. by Mahmoud Jaafari-Dehaghi. Tehran: Center for the Great Islamic Encyclopaedia. pp.81-94.

Lord 1960
Albert Bates Lord. The Singer of Tales. Cambridge, MA: Harvard University Press.


Malandra 2000
William W. Malandra. “VENDĪDĀD i. Survey of the History and Contents of the Text.” In Encyclopaedia Iranica. Ed. by Ehsan Yarshater. New York: Encyclopaedia Iranica Foundation. https://iranicaonline.org/articles/vendidad

Martínez Porro 2013
Jaime Martínez Porro. “La ceremonia zoroástrica de Vistasp Yast y sus manuscritos.” Boletín de la Sociedad Española de Iranología, 2:69-80.

Martínez Porro 2014
______. “Los manuscritos avésticos de la ceremonia Visperad.” Estudios iranios y turanios, 1:75-94.

Martínez Porro 2020
______.  “The Written Transmission of the Vištāsp Yašt Ceremony.” Studia Iranica, 49: 207-21.

Martínez Porro forthcoming
______. “The Timing of the Vidēvdād and Vištāsp Yašt Ceremonies.” In The Reward of the Righteous: Festschrift in Honour of Almut Hintze. Ed. by Alberto Cantera, Maria Macuch, and Nicholas Sims-Williams. Wiesbaden: Harrassowitz.

Mazdapour 2004
Katayoun Mazdapour. “Kontinuität und Wandel in den Ritualen der iranischen Zoroastrier.” In Zoroastrian Rituals in Context. Ed. by Michael Stausberg. Leiden: Brill. pp. 630-52.

Moazami 2014
Mahnaz Moazami. Wrestling with the Demons of the Pahlavi Widewdad: Transcription, Translation, and Commentary. Leiden: Brill.

Morgenstierne 1942
Georg Morgenstierne. “Orthography and Sound-System of the Avesta.” Norsk Tidsskrift for Sprogvidenskap, 12:30-82.

Nau 1927
François Nau. “Étude historique sur la transmission de l’Avesta et sur l’époque probable de sa dernière rédaction.” Revue de l’histoire des religions, 95:149-99.

Nyberg 1958
Henrik Samuel Nyberg. “Sassanid Mazdaism according to Moslem Sources.” Journal of the K. R. Cama Oriental Institute, 39:1-63.

Panaino 1999
Antonio Panaino. “Avesta. Religion in Geschichte und Gegenwart.” Handwörterbuch der Theologie und Religionswissenschaft:1024-26.

Panaino 2012
______. “The Age of the Avestan Canon and the Origins of the Ritual Written Texts.” In Cantera 2012b:70-97.

Panaino 2017
______. “Avestico rec. pasuuāzah-. Vecchie e nuove considerazioni a proposito dell’immolazione animale nella ritualistica indo-iranica.” In Anantaratnaprabhava. Studi in onore di Giuliano Boccali. Ed. by A. Crisanti, C. Pieruccini, C. Policardi and P. M. Rossi. Milano: Ledizioni. pp. 137-52.


Park 2009
Yoon-Man Park. Mark’s Memory Resources and the Controversy Stories (Mark 2:1-3:6): An Application of the Frame Theory of Cognitive Science to the Markan Oral-Aural Narrative. Leiden: Brill.

Parry and Parry 1971
Milman Parry and Adam Parry. The Making of Homeric Verse: the Collected Papers of Milman Parry. Oxford: Clarendon Press.

Raffaelli 2014
Enrico Raffaelli. The Sīh-rōzag in Zoroastrianism. A Textual and Historico-Religious Analysis. London: Routledge.

Reichl 2015
Karl Reichl. “Memory and Textuality in the Orality-Literacy Continuum.” InRubanovich 2015:19-42.

Rezania 2012
Kianoosh Rezania. “Mazdakism and the Canonisation of Pahlavi translations of the Avestan texts.” In Cantera 2012b:480-94.

Ringer 2011
Monica M. Ringer. Pious Citizens: Reforming Zoroastrianism in India and Iran. Syracuse, NY: Syracuse University Press.

Rubanovich 2015
Julia Rubanovich. Orality and Textuality in the Iranian World: Patterns of Interaction Across the Centuries. Leiden: Brill.

Sims-Williams 1976
Nicholas Sims-Williams. “The Sogdian Fragments of the British Library.” Indo-Iranian Journal, 18:43-82.

Skjærvø 1994
Prods Oktor Skjærvø. “Hymnic Composition in the Avesta.” Die Sprache, 36.2:199-243.

Skjærvø 1997 [2000]
______. “Eastern Iranian Epic Traditions III: Zarathustra and Diomedes—An Indo-European Epic Warrior Type.” Bulletin of the Asia Institute, 11:175-82.

Skjærvø 1998a
______. “Avestica II: Yokes and Spades and Remnants of the ‘Tripartite Ideology.’” Münchener Studien zur Sprachwissenschaft, 57:115-28.

Skjærvø 1998b
______. “Eastern Iranian Epic Traditions I: Siyāvaš and Kunāla.” In Mír Curad: Studies in Honor of Calvert Watkins. Ed. by Jay Jasanoff, H. Craig Melchert, and Lisi Oliver. Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft. pp.645-58.

Skjærvø 1999
______. “Eastern Iranian Epic Traditions II: Rostam and Bhīṣma.” Acta Orientalia Academiae Scientarum Hungaricae, 51.1-2:159-70.

Skjærvø 2012
______. “The Zoroastrian Oral Tradition as Reflected in the Texts.” In Cantera 2012b:4-48.


Spiegel 1882
Friedrich Spiegel. “Zur Textkritik des Awestâ.” Zeitschrift der Deutschen Morgenländischen Gesellschaft, 36:586-619.

Stock 1983
Brian Stock. The Implications of Literacy: Written Language and Models of Interpretation in the Eleventh and Twelfth Centuries. Princeton, NJ: Princeton University Press.

Tremblay 2012
Xavier Tremblay. “Ibant obscuri uaria sub nocte: Les textes avestiques et leurs recensions des Sassanides au XIIIe s. ad en particulier d’après l’alphabet avestique: Notes de lecture avestiques VIII.” In Cantera 2012b:98-135.

Vevaina 2015
Yuhan Sohrab-Dinshaw Vevaina. “‘The Ground Well Trodden But the Shah Not Found . . .’: Orality and Textuality in the ‘Book of the Kings’ and the Zoroastrian Mythoepic Tradition.” In Rubanovich 2015:169-90.

West 1892
Edward William West. Pahlavi Texts, Part IV: Contents of the Nasks. Oxford: Clarendon.

Westergaard 1852-54
Niels Ludvig Westergaard. Zendavesta, or the Religious Books of the Zoroastrians. Copenhagen: Berling Brothers.