Nicholas Sims-Williams and Ursula Sims-Williams published “Two new folios of the Book of Zambasta” in Reference Sims-Williams and Sims-Williams2023. These folios came to light when their current owner noticed similar leaves in the online International Dunhuang Project database. These additions bring the total number of known folios of the main manuscript of the Book of Zambasta (Z) to 209. Although the new leaves contribute just a little less than 1 per cent of the current total, they contain noteworthy new features and are thus not only a surprising but also a welcome contribution to the study of the language. One folio pertains to Chapter 2, about the conversion of the magician Bhadra and was the only folio missing from Chapter 2 (Emmerick Reference Emmerick1968a: 10). The other folio contributes lines 6–17 of Chapter 24, which still has significant gaps.

In the new folio from Chapter 2 there is a word which has been seen before but has been puzzling. The first hemistich in line 36 reads:

Sims-Williams and Sims-Williams translate:

None of this is controversial except for the phrase ma ne dīñu “let it not be apparent”.

This phrase is similar to ma ne dīñi, which occurs later in the same chapter:

Emmerick translated:

Emmerick assumed dīñi to be 1. sing. subjv. act. from de-Footnote ⁴ “see” + -īñi 1. sing. subjv. act., but that inflected form, involving synchronic e+ī→ai, would be *daiñi as in de- + -īndä → daindä 3. pl. pres. act. “they see”, or de- + -īmä → daimä 1. sing. pres. act. “I see”. dīñi is formally more plausibly 1. sing. subjv. mid. from dä-Footnote ⁵ “appear” with ä+ī→ī as in pajä- + īndä → pajīndä Z 2.190b “they request” or nä- + -īndä → nīndä Z 3.60a “they sit”. Because of that formal possibility, a few years ago Hitch translated the same hemistich as:

With regard to dīñu in the new folio, Sims-Williams and Sims-Williams now propose to see:

This is a clever idea and formally correct. Both de- + -āña- and dä- + -āña- would regularly result in the ambiguous dyāña-. But it does not seem likely that this potential ambiguity is involved in this form. The same phonological conditions, e+ā and ä+ā, apply when the abstract noun suffix -āmatā- is attached to these stems. de- + -āmatā- and dä- + -āmatā- both result in the ambiguous form dyāmatā-, which is in fact attested in both meanings, “sight, beholding” from de- “see”, and “appearance” from dä- “appear”, e.g.:

Similarly, the potential ambiguity of dyāña- was likely not an obstacle to it being used in both meanings.

With regard to understanding dīñu and dīñi, Hitch now thinks that all of the previous efforts, including his own, involving de- “see” or dä- “appear” are incorrect. The shape dīñ- occurs twice more in Khotanese. One of these instances, SI M 14.1r6 burjate dīña is unclear and not translated by Emmerick and Vorob’ëva-Desjatovskaya Reference Emmerick and Margarita1995 (SDTV III: 212; identified as Z 21.112 in Maggi Reference Maggi1998: 287–8). There appears to be no shape similar to or related to burjate in Khotanese and thus it remains unclear. This instance of dīñ- is not helpful to the current discussion and may be set aside.

The remaining instance of dīñ- is in the Old Late Khotanese Avalokiteśvaradhāraṇī (SDTV III: 239–250). The supplicant appeals to Avalokiteśvara for deliverance and then gives a list of six negative instructions or prohibitions relating to the manner of the deliverance:

Emmerick and Vorob’ëva-Desjatovskaja translate just the first three prohibitions:

These three occur in a parallel set of negative commands earlier in the text:

Mauro Maggi points out that in the phrase ma aña yana the meaning of yan- “make” is preferably given as “deem, consider, regard” in the same way that Latin facio “I do” can also mean “I consider”. He offers the translation, “do not consider otherwise (my promise and great resolve)”.Footnote ⁶

The last three prohibitions in A 20v5 have been tentatively translated three times by H.W. Bailey as “do not overthrow (?) me, do not seize(?) me, do not treat me violently(?)” (DKS: 274 s.v. bahauji), as “do not make me fall, do not seize me, do not rend me” (DKS: 158a s.v. dīñä), and as “do not cast me down, do not seize me, do not rend me(?)” (DKS: 157a s.v. dārañā “tearing(?)”. The meanings of all three commands were in doubt for him. Similarly, Emmerick earlier translated these three as, “Do not cause me to be beaten(?), do not weaken me, do not …” and noted, “The meaning of these words is obscure” (Reference Emmerick, Gnoli and Panaino1990: 107–08 and fn. 25).

In his translation of A 20v4–5 ma ma dīñä “do not overthrow (?) me”, Bailey treated the second ma as the 1. sing. pronoun (DKS: 274a sv. bahauji). However, it seems best to regard the ma at the end of line 4 and the ma at the beginning of line 5 as dittography induced by the change of line and thus to read only one instance. Elsewhere in the text where “not … me” is intended in a list of prohibitions, ma ma is consistently used before the verb with every prohibition. In A 13r1–3 ma ma or encliticized maṃ “not … me” occurs eight times, and in A 20r1 they occur three times. If the phrase meaning “not … me” were intended in A 20v4–5 one might also expect all six prohibitions to be preceded by ma ma.

As suggested by Emmerick, the hapax legomenon bahauji may contain a denominal verb bahoj- derived through umlauting from the adjective bahojsa- “weak” attested in bahojsa Suv[Or]12.45 “(they will be) powerless” (Emmerick Reference Emmerick, Gnoli and Panaino1990: 107–08 and fn. 25). It is fairly common to find present stems derived from nominals through umlaut; e.g. āstana- “beginning” → āstañ- “begin”, gūrāsa- “quarrel” → gūrāś- “to quarrel”, parāha- “restraint” → pareh- “restrain oneself”, parysa- “servant” → parś- “serve” (more examples in Hitch Reference Hitch1990). Emmerick suggested a gloss “to weaken” and in this context a suitable translation for ma bahauji may be “do not disparage (my promise and resolve)”.

Emmerick apparently did not wish to connect A 20v5 dārañā with the OKh i-stem d(h)āraṇi- “a spell” (cf. Skt dhāraṇī), which is well attested in Old KhotaneseFootnote ⁷ but does not show an -ā ending on any other related stem shape (*dhārañā, *dhāraṇā, *dāraṇā). If the manuscript were damaged and only the akṣaras dārañ[were attested in this line, it is likely that all scholars would assume either dārañä nom.-acc. pl. fem. “spells” or dārañu acc. sing. fem. “spell”. The -ā is unmistakable in the photograph (SDTV III: plate 197(d)) but it also seems unlikely that we are dealing with an unrelated word. As Mauro Maggi suggests, A 20v5 dārañā is most economically explained as dārañä acc. sg. fem. (OKh. dārañu) + 3 pl. encl. pron. -ā(ṃ) (OKh. nä), and consequently he translates ma dārañā yana as “do not regard them (my promise and resolve) as (meaningless) dhāraṇis”.Footnote ⁸

Five of the six commands in A 20r4–5 may now be translated: “Do not forget (my promise and resolve), do not consider (them) otherwise, do not neglect (them), do not dīñä, do not disparage (them), do not regard them as (meaningless) dhāraṇis”. These five are all formally second person singular imperatives: yana 3× “make” shows 2. sing. imp. act. -a (SGS: 211), while anaṃdīśä “neglect” and bahauji “be weak” show 2. sing. imp. mid -ä or -i (< Old Khotanese -u, SGS: 213). The sixth word, dīñä, plausibly also shows the Late Khotanese 2. sing. imp. mid. in -ä. This accords with Emmerick’s insight: “Prohibition, both inhibitive and preventive, is expressed by ma with the imperative in both Old and Late Khotanese” (Reference Emmerick, Gnoli and Panaino1990: 97). It now seems certain that the form dīñu from Z 2.36 is simply a regular Old Khotanese 2. sing. imp. mid. in -u. The context of this word in Chapter 2 of Z supports these possibilities and offers a clue to the semantics.

The heretics do not believe that the Buddha is all-knowing and propose various schemes to test him. If he would succumb to poisoned food, or arson, or black magic, or fall into a trapping pit, he would not be all-knowing. Bhadra, a learned sorcerer, offers to deceive the Buddha through magic. The heretics approve of the offer and encourage Bhadra to begin. We then read:

The heretics offer to help Bhadra. All go to a disgusting cemetery. There, Bhadra conjures up a marvellous palace, complete with carpets and servants serving wine. He declares that he will invite the Buddha to the illusory palace and when the Buddha does not realize he is really in a cemetery, Bhadra will end the illusion and confront the Buddha with his failure to know he was being tricked, thus proving he is not all-knowing. The next day the Buddha and his followers calmly enter Bhadra’s illusory palace and all sit on illusory divine seats. Bhadra declares that the Buddha has been deceived and therefore is not all-knowing. He then tries to undo his magic to reveal the cemetery and prove his point but is unable to do so. He realizes his error and experiences remorse and shame. As Bhadra tries to leave, Vajrapāṇi tells Bhadra that it is not possible to leave since he has invited the Buddha. Vajrapāṇi tells Bhadra, “It is time. Strike the gong!” Then we read in Z 2.100:

Bhadra is further encouraged by Ānanda. Bhadra strikes the gong and declares to the world that the Buddha is indeed all-knowing.

The contexts and structures of lines 36 and 100 are parallel. In 36 the heretics command Bhadra to not do something (ma ne dīñu) but to “begin” (ākṣuva) the illusion. In 100, Vajrapaṇi commands Bhadra to not do the same thing (ma ne dīñi) but then also to “begin” (ākṣva) declaring that the Buddha is all-knowing. It is likely that this parallelism is a deliberately used narrative tool. The parallelism also seems to confirm that the text originally had dīñu in both places and that dīñi in line 100 reflects Late Khotanese copyist influence (see also below).

The semantics of dīñ- should be something like “hesitate”, “delay”, or “wait”, an activity one should not do when one is exhorted to begin a task immediately. These semantics may be further refined by the etymology contributed by Derafshi (see below) which suggests “overthink” as a preferred meaning. Overthinking contains the concept of hesitation, and suits better the context of Z 2.36 which has matā- “idea, thought” earlier in the sentence, which now may read “What(ever) your idea may be, do not overthink (it)!” In both cases in Chapter 2 of Z it may be that Bhadra is told “Do not overthink (it)! Begin (now)!”. This meaning also fits with the six negative commands in A 20r4–5, which can now be completely translated as “Do not forget (my promise and resolve), do not consider (them) otherwise, do not neglect (them), do not overthink (them), do not disparage (them), do not regard them as (meaningless) dhāraṇis”.

As mentioned above, while dīñu in Z 2.36 is a regular Old Khotanese 2. sing. imp. mid. in -u, dīñi in Z 2.100 looks like a copyist emendation to a regular Late Khotanese 2. sing. imp. mid. in -ä ∼ -i. This idea is supported by the first word in the hemistich, cvä, which otherwise only appears in Late Khotanese.Footnote ⁹ Hitch earlier translated cvä as “if from this”, assuming underlying cu “if” + -ä 3.sg.encl.pron.Footnote ¹⁰ But cu + -ä otherwise has only one spelling in Z, cvī, which occurs 56 times and is also the most common one in Old Khotanese. A possible alternate spelling, cuī, does not occur in Z at all but is rarely found elsewhere in Old Khotanese, cf. Sgh[10]34.11, Suv[Or]12.12. Also against Hitch’s earlier interpretation is the fact that cvä, if cu + -ä, would be phonologically /čuə/ which is two moras, making the expected five-mora segment too long and thus unmetrical.

In his edition, Ernst Leumann emended cvä to *cu (E: 21, 100ˠ). The emended *cu tä tta ratä correctly fills a five-mora segment in the meter. Nicholas Sims-Williams, in his upcoming edition of Chapter 2 of Z, also emends to *cu but with other motivation. He notes, “cvä tä for cu tä. Cf. vasvätä for vasutä, v. 176. The scribe may have written cvä, intending it to be read as cᵘvä for cu tä. In restoring the regular orthography of the 2 sg. encl. tä he neglected to alter cvä to cu” (personal communication). As far as this author can tell, tä is never written *vä anywhere in Khotanese. The enclitic is written tä 25 times and te 8 times in Z. The phrase cu tä occurs 7 times in Z in that spelling. And Hitch does not think there is -v- for expected -t- elsewhere in Old Khotanese.Footnote ¹¹ However, both Sims-Williams and Leumann are, in the opinion of Hitch, correct to emend cvä tä to *cu tä. The simplest explanation for the spelling cvä is Late Khotanese influence. Consequently, because the hemistich already exhibits Late Khotanese influence, we should also with reasonable confidence emend the Late Khotanese dīñi to Old Khotanese *dīñu.

The three instances of dīñ- may now  be freshly translated:

The Khotanese  present stem dīñ- “overthink” can be etymologized as a denominal verb derived from an unattested Khotanese *dīnā- “thought, consideration” by means of the suffix *-i̯a- that caused the palatalization of *n > ñ. Khotanese *dīnā- is cognate with Avestan daēnā- “view, vision” and, like the Avestan word, goes ultimately back to Indo-Iranian *dʰai̯Hanā- Footnote ¹⁵ from the root *dʰai̯H- “to see; to think” (cf. Khotanese de-,Footnote ¹⁶ Avestan dī- “see”), which in Indo-Aryan is continued by Vedic dhī- “to perceive, think, reflect”.Footnote ¹⁷

Formally, one may expect that Proto-Iranian *-ai̯(H)a-, like the diphthong *ai̯, results in Khotanese -ī- in all positions. ī < *ai̯ is common and is observed for instance in bīsa- “servant” < *u̯aisa-, Avestan vaēsa-,Footnote ¹⁸ and in the material suffix -īnaa- < *-ai̯naka- (cf. Middle Persian -ēnag). Other instances of ī < *ai̯(H)a are difficult to find.Footnote ¹⁹ It is found in *āyīna- “mirror” < *ā-dʰai̯Hanā- (cf. New Persian āyīna “mirror”) from the same root as *dīnā-. The word *āyīna- has complex attestations in Old Khotanese partly due to a variant *āyäna- with stem shortening. Beside the one locative singular spelling ātīña (Z 15.123), there is one locative singular case āyäña (Z 4.100) with -ä-.Footnote ²⁰ It is interesting to note that the more conservative variant ātīña (‒ ‒ ⏑) and the shortened variant āyäña (‒ ⏑⏑) co-existed, and that they were deliberately used in different metrical contexts.Footnote ²¹ The shortened *āyäna- form of the word occurs also in the instr.-abl. sg. āyärnna (intended *āyänna ← *āyänäna; Z 24.491, with intrusive r). A derivative of *āyīna- is better attested. āyīnaa- (< *-aka-) with the figurative meaning “example, simile” occurs seven times in the Book of Zambasta (five times āyīnaa- Z 19.86+, twice āyänaa- Z 19.89, 93). This derivative also features the ī∼ä variation in the stem.

Parallel to Proto-Iranian *ai̯(H)a > Khotanese ī, is Proto-Iranian au̯a > Khotanese ū which can be seen in the verb tsū- “go” (the long ū being conserved in the pres. ptcp. tsūmandaa- and the action noun tsūmatā- “going, career”) < PIr. *či̯au̯-a-, cf. OInd. cyavate etc.,Footnote ²² and in the verbal noun suffix -ūna- < *-au̯(a)na-, cf. Sogdian -wn (Degener Reference Degener1989: 168 § 28.A.4). Among the East Iranian languages, a similar sound development *-ai̯a- > -ī- is found in Bactrian, where it seems to occur only before nasals, for example in the denominal verb βιν- “to oblige, compel” < Old Iranian *upa-ayanā̆-.Footnote ²³

As for semantics, an important question is whether Khotanese is the only Iranian language to have kept the meaning “think” of the Indo-Iranian root *dʰai̯H-, like Vedic, or whether there are traces of this meaning in other Iranian languages as well. Although the root seems to be universally used with the meaning “see” in Iranian languages,Footnote ²⁴ a change in the understanding of the Avestan derivative daēnā- by several scholars in recent decades indicates another possibility. Almut Hintze (Reference Hintze, Williams, Stewart and Hintze2016: 77) summarizes this change thus: “rather than seeing with the eyes of the body, the noun daēnā- metaphorically denotes the activity of seeing with the inner eye of one’s mind and, especially, what is produced by such an activity, that is to say ‘thought’, ‘conviction’, ‘belief’, or ‘vision’”. The problem is whether these meanings that are in a close semantic relationship with “thought” are just the result of an Avestan metaphorical use (seeing with the inner eye) or whether they are the remnants, as in the case of Khotanese *dīnā-, of the meaning “to think” of the Indo-Iranian root from which daēnā- derives.

There is at least one occurrence of daēnā- in the Gāthās, which indicates that the meaning “to think” was not completely lost in Avestan:

As can be seen, Stanley Insler translated daēnā- as “conception”.Footnote ²⁶ The most important point to notice here is that the presence of the well-known Zoroastrian triad (thought, word, deed) in this stanza should not misguide one to understand manō “mind, thought” in relation to š́iiaoθanā˚ “action, deed”, and vacaŋhā˚ “word, speech”. Especially significant is the fact that š́iiaoθanā˚ and vacaŋhā˚ are both in the instrumental singular, whereas manō is accusative singular, the same case as daēnąm. So, manō should be understood in relation to daēnąm and vice versa. Insler cleverly translated manō as “mind” rather than “thought” in order to avoid giving the false impression of the Zoroastrian triad as a whole, as if the three elements were on one and the same level. Accordingly, daēnā- should be interpreted here with words semantically related to “thought” such as “conception, idea, belief”, which fit the context perfectly.

The occurrence of daēnā- in close connection with manah- “thought” in Y 48.4 may indicate that its meaning “conception/idea/belief” is not necessarily metaphorical but is a possible remnant also in Avestan of the old meaning “think” of the Indo-Iranian root *dʰai̯H-. This meaning is also apparently preserved in Kh. *dīna- “thought” that is at the basis of dīñ- “overthink”.

The semantic development “think” → “overthink” of Khotanese dīñ- is a fairly natural one. An example of a similar semantic shift is provided by the New Persian verb nigarīstan “look”, whose present participle nigarān nowadays means “anxious”.