A Brief Report on the Nesar and Lang Kanjurs: Some Textual-Critical Observations Focusing on the Tibetan Gaganagañjaparipṛcchā

Abstract

The Tibetan Kanjur has long been recognized as both a symbolic embodiment of the Buddhist canonical literature and as a ritual object, resulting in the production of various versions that differ in content, arrangement, and specific textual formulation. Since the late 1970s, the provenance, lineage affiliations, and historical development of these Kanjurs have attracted significant scholarly attention. In this paper, I present the findings of textual-critical research on the Tibetan translation of the Gaganagañjaparipṛcchā (Ggn), focusing particularly on two manuscript collections preserved at Nesar Monastery in Dolpo, namely the Nesar and Lang Kanjurs. Both Kanjurs, possibly dated as early as the thirteenth century, lie outside the two main lineages, Tshal pa and Them spangs ma, and demonstrate strong connections with Local or Independent Kanjurs, notably those of Phug brag and Namgyal. By undertaking a close comparison of selected passages from the Ggn across twenty-one canonical witnesses, this study finds that, for the Ggn: (a) the Nesar and Lang Kanjurs possess a group of unique textual variants which distinguish them from all other known Kanjur and Proto-Kanjur editions; (b) the Lang Kanjur appears to have been based chiefly on the Nesar Kanjur or an exemplar closely related to it; and (c) the compilers of the Lang Kanjur also relied on at least one other manuscript, which seems to have preserved readings of greater accuracy. These findings highlight the importance of the Nesar and Lang Kanjurs for textual-critical investigation and for understanding the transmission history of the Tibetan Buddhist canon. Ongoing research into these Kanjurs will yield crucial evidence for constructing a more nuanced and historically informed account of the formation, adaptation, and regional diffusion of the Tibetan Buddhist canon.

1. Introduction: The Development of Textual Criticism in Tibetan Kanjur Studies

Textual critical studies and stemmatic analysis of the Tibetan Kanjur serve not only as initial attempts to reconstruct a hypothetical urtext or to produce critical editions, but also represent an independent field of scholarship that has cultivated its own methodological rigour, theoretical models, and research directions. The field of textual criticism, dedicated to determining the most accurate or trustworthy textual form through comparative evaluation of multiple witnesses, can trace its methodological lineage to the early 19th-century German philologist Karl Lachmann (1793–1851). Since then, the Lachmannian method, commonly known as stemmatics, and various approaches derived from it, have been extensively utilized in the study of textual traditions spanning ancient Greek texts to medieval vernacular works, as well as New Testament manuscripts and contemporary literary compilations. Within Buddhist Studies, and especially in relation to Tibetan canonical texts, textual criticism was integrated into academic practice only in the latter part of the 20th century. A significant advancement occurred in 1978, when Bodhipathapradīpa: Ein Lehrgedicht des Atiśa (Dīpaṃkaraśrījñāna) in der tibetischen Überlieferung, edited by Helmut Eimer, was published (Eimer 1978). In this publication, Eimer produced a critical edition of Atiśa’s Bodhipathapradīpa through detailed comparison of sixteen Tibetan versions drawn from different Kanjur recensions.1 This study provided not only a critically reconstructed Tibetan version, but also articulated a methodological model for textual critical research on the Tibetan Kanjurs, illuminating their intertextual and inter-versional dynamics. Eimer advanced this paradigm further in his 1983 monograph Rab tu ’byuṅ ba’i gźi, concentrating on the Pravrajyāvastu from the Mūlasarvāstivāda Vinaya and utilizing thirteen Tibetan textual witnesses. His contributions highlighted the methodological value of comparative textual criticism in Tibetan canonical research and established a foundation for scholarly critical editions of Kanjur texts.2

Alongside Eimer, another influential scholar in this field is Paul Harrison, who published The Tibetan Text of the Pratyutpanna-buddha-saṃmukhāvasthita-samādhi-sūtra in 1978 (Harrison 1978). His edition conducted a comparison among four different Kanjur editions, specifically the Derge, Peking, Narthang, and Lhasa editions. Nonetheless, from the standpoint of textual criticism—particularly regarding Tibetan Kanjurs—this work faced constraints due to two primary factors: the limited number of versions utilized and their exclusive association with the Tshal pa lineage. Consequently, Harrison’s edition did not represent the diversity of variant readings and transmission characteristics found in the Them spangs ma tradition. In response to this limitation, his subsequent work, The Samādhi of Direct Encounter with the Buddhas of the Present, published in 1990, included an appended apparatus critico based on the Stog Palace Kanjur, which serves as a representative of the Them spangs ma tradition (Harrison 1990). While brief, this supplement constituted a significant advancement toward a more systematic and inclusive methodology in the textual criticism of the Tibetan canon.3 Harrison’s other pivotal contribution to the study of the transmission and textual lineages of the Tibetan Kanjur was the publication of Druma-kinnara-rāja-paripṛcchā-sūtra: A Critical Edition of the Tibetan Text (Recension A) (Harrison 1992a). Within this research, he employed ten Tibetan canonical witnesses—Cone, Derge, Lhasa, Lithang, Dunhuang, Narthang, Phug brag, Peking, Stog Palace, and Tokyo—to create a critical edition. Through a thorough comparative evaluation of these versions, he produced a detailed stemma codicum that traced their relationships. A key methodological advancement introduced in this study was his division of textual variants into two categories: (1) transmissional variation, resulting from scribal errors or incidental modifications arising during transmission, and (2) recensional variation, which stems from intentional and systematic editorial revisions. This differentiation has since become fundamental in Tibetan textual criticism and continues to shape academic perspectives on the transmission of the Kanjur.4 Additionally, Harrison advocated for a practical guideline for future critical editions of Tibetan canonical texts: specifically, that rigorous textual criticism should compare at least two representatives from both the Tshal pa and Them spangs ma traditions. Supported by his empirical findings, this recommendation has gained broad acceptance and remains a standard editorial guideline in the discipline.5

Building on these foundational works, several important critical editions of canonical sūtras were issued in the following years. In 1993, Jens Braarvig released a critical edition of the Tibetan translation of the Akṣayamatinirdeśa, which constitutes the twelfth chapter of the Mahāsaṃnipāta corpus (Braarvig 1993a, 1993b). The year 1994 marked the publication of Peter Skilling’s Mahāsūtra (Skilling 1994), Jonathan Silk’s edition of the Prajñāpāramitāhṛdaya (Silk 1994), and Braarvig’s Bodhisattvacaryānirdeśa (Braarvig 1994). This momentum was sustained in 1995 and 1996 with Jeffrey Schoening’s edition of the Śālistambasūtra (Schoening 1995) and Siglinde Dietz’s work on the Lokaprajñaptiśāstra (Dietz 1996), respectively. Research in textual criticism within this domain has remained vigorous into the twenty-first century. In 2002, Michael Zimmermann provided a critical edition of the Tathāgatagarbhasūtra (Zimmermann 2002). Subsequently, Jens Braarvig and Ulrich Pagel produced a joint edition of parts of the Bodhisattvapiṭaka (Pagel and Braarvig 2006) in 2006. In 2013, Hiromi Habata published a critical edition of the Mahāyāna Mahāparinirvāṇasūtra (Habata 2013), which was followed by Robert Kritzer’s work on the Garbhāvakrāntisūtra in 2014 and Ruxin Chen’s edition of the Nandimitrāvadāna in 2018 (Kritzer 2014; Chen 2018). Most recently, James Apple compiled a critical edition of the Avaivartikacakrasūtra in 2021 (Apple 2021), and Yixiu Jiang undertook research on the Svapnanirdeśa in 2024 (Jiang 2024).

2. Background: Transmission Lineages and Recent Research Focus

Based on relatively recent scholarly classifications, Tibetan Kanjur editions are typically divided into four main lineages: the Tshal pa (1347–1351) group, the Them spangs ma (1431) group, a mixed or hybrid group, and a group classified as independent or local (Tauscher 2015, pp. 108–9). Of these, the Tshal pa lineage—which originates from a Kanjur compiled at Tshal Gung thang Monastery between 1347 and 1351—and the Them spangs ma lineage—deriving from a manuscript Kanjur created at Rgyal rtse in Gtsang in 1431—are considered to represent the two canonical prototypes. These two lineages served as the foundational bases for replication and reproduction over subsequent centuries and constitute the core of most extant canonical editions. The Tshal pa lineage further divides into two principal transmission lines: the so-called “Peking line,” which encompasses Kanjurs printed in Beijing under Qing imperial patronage; and the ’Phying ba stag rtse line, descending from a manuscript Kanjur housed at ’Phying ba stag rtse in ’Phyongs rgyas. By contrast, the Them spangs ma group is comprised exclusively of manuscript Kanjurs, with notable examples including those from Ulaanbaatar, Stog Palace, Shey Palace, and Shel dkar (London), as well as newly digitized collections from Bhutan. The third category includes “mixed” or conflated Kanjurs, which incorporate content from both the Tshal pa and Them spangs ma traditions. Within this category, the Lhasa and Narthang Kanjurs are particularly prominent. A fourth group, frequently identified as “independent” or “local,” consists of Kanjurs produced in geographically remote regions, where textual transmission from the two dominant lineages was limited.6 Representative examples of this group include the Kanjurs of Phug brag, Bathang, and O rgyan gling.7

In recent years, scholarship has increasingly investigated Kanjur and sūtra collections beyond the two mainstream traditions, Tshal pa and Them spangs ma, thereby highlighting the diversity and dynamism of regional manuscript practices that have played a significant role in Tibetan Buddhist textual culture. This methodological and conceptual shift is grounded in a recognition that the canonical corpus is not static, but emerges from historically contingent, complex, and regionally situated processes of transmission, adaptation, and compilation (Skilling 1997b, pp. 100–1). The pioneering efforts of scholars such as Helmut Tauscher, Markus Viehbeck, and Bruno Lainé have been pivotal to this development, as their research provides systematic documentation and critical analysis of what are termed “regional” or “peripheral” Kanjur and sūtra collections located outside the established mainstream traditions. One particularly influential advancement in this scholarly field is the identification of the “Mustang/Ladakh group” as a distinct textual lineage (Tauscher 2015; Tauscher and Lainé 2015; Viehbeck 2020; Luczanits and Viehbeck 2021, pp. 341–61). First identified by (Tauscher and Lainé 2015, especially pp. 463–64), this assemblage is not simply a subset of the broader “independent” or “local” Kanjur category, but rather constitutes a unique regional transmission system. The Mustang/Ladakh group is defined by the circulation of textual fragments, the formation of regionally focused compilations, and the utilization of adaptive approaches to canon construction, all of which are shaped by the cultural, institutional, and geographic diversity of the Himalayan borderlands. This group—or more precisely, this network—includes not only the Early Mustang Kanjur, which survives today in the form of its catalogue (Eimer 1999), but also a range of sūtra collections from Dolpo, as well as the Kanjurs of Hemis and Basgo in Ladakh. The collections associated with this group exhibit pronounced textual and structural similarities, indicating the presence of an interconnected lineage distinct from both Tshal pa and Them spangs ma traditions. As Viehbeck observes, textual evidence from the Western and Central Himalayas “represents another important line of transmission next to the mainstream lineages (Viehbeck 2020, p. 256),” and it is now appropriate to regard the Mustang/Ladakh group as a third major stream in the evolution of Tibetan canonical literature.

Building on the established insights and methodologies from prior scholarship, this study examines the transmission and textual history of Tibetan canonical literature by analyzing several passages from the Gaganagañjaparipṛcchā (hereafter Ggn). The Ggn is, similar to the Akṣayamatinirdeśa researched by Jens Braarvig, part of the Mahāsaṃnipāta collection.8 No Sanskrit original of the Ggn has survived, and so far no Sanskrit manuscript has been identified. Nevertheless, select portions can be consulted through quotations preserved in surviving Sanskrit treatises, such as Asaṅga’s Ratnagotravibhāga, Śāntideva’s Śikṣāsamuccaya, and Kamalaśīla’s Bhāvanākrama.9 Therefore, to access the complete text of the Ggn, one must turn to the extant Chinese translations: the Da fangdeng daji jing Xukongzang pusa pin (大方等大集經虛空藏菩薩品, T.397.8), translated by Dharmakṣema in the early fifth century; the Daji da xukongzang pusa suowen jing (大集大虛空藏菩薩所問經, T.404), produced by Amoghavajra in the mid-eighth century; and the Tibetan translation ’Phags pa nam mkha’i mdzod kyis zhus pa zhes bya ba theg pa chen po’i mdo,10 completed in the ninth century by Vijayaśīla, Śīlendrabodhi, and Ye shes sde.11

As a representative sample, this study draws on a comprehensive set of twenty-one textual witnesses from diverse transmission lineages or traditions referenced above, namely, Basgo, Cone, Derge, Gondhla, Hemis I, Hemis II, Lang, Lhasa, Lithang, London, Namgyal, Narthang, Nesar, Phug brag I, Phug brag II, Qianlong, Ragya, Shey, Stog Palace, Tabo II, and Urga.12 There is no extant version of the Tibetan Ggn from Dunhuang. This research pays particular attention to two Kanjurs that have been recently accessed, Nesar and Lang, which are preserved within the extensive manuscript collections at Nesar Monastery (Gnas gsar dgon pa), situated north of Nepal’s Dolpo District. Despite their significance for investigating early Himalayan canonical transmission, scholarly understanding of these collections remains incomplete. Through the analysis of these rare and relatively understudied witnesses, I seek to enhance ongoing assessments of Mustang Kanjur traditions and clarify their role within the broader narrative of canonical transmission in western Himalayan Buddhism. In particular, this work aims to further the dialogue on textual transmission processes, regional adaptations, and the interactions between central and local manuscript traditions, as emphasized in recent studies of Himalayan and Mustang-area Kanjurs.13

3. The Mauscripts of Nesar Monastery

The existence of the Nesar Kanjur was initially introduced to Western scholarship through the observations of the British Tibetologist David Snellgrove (1921–2016), who travelled to western Nepal in 1956. In his travel account, Himalayan Pilgrimage: A Study of Tibetan Religion by a Traveller Through Western Nepal, published in 1957, Snellgrove noted that the small village of Phijor, located north of the Dolpo District and consisting of approximately fifty households, maintained a complete set of the Tibetan Kanjur at Nesar Monastery.14 This pioneering documentation drew scholarly attention to a significant, previously unfamiliar canonical collection in the Himalayan borderlands.15 Soon after, the French anthropologist Corneille Jest (1930–2019) broadened Western perspectives on Dolpo by engaging in fieldwork in the region from 1960 to 1961. In his monograph Dolpo: Communautés de Langue, published in 1961, Jest, referencing oral accounts from his local informant Kagar Rinpoche, reported that the Kanjur housed at Nesar Monastery was initially commissioned by King Koleal of the Kingdom of Jumla (Dzumla) and later relocated to Dolpo. Although this oral evidence awaits validation through codicological or historical inquiry, it highlights the intricate patterns of patronage and textual transmission that influenced the development of local canons in the western Himalayas.16

Following the visits by Snellgrove and Jest, the Nesar Kanjur remained largely neglected for almost four decades, gaining renewed attention only in the late 1990s through the efforts of the Pritzker and Roncoroni families, who are widely recognized for their philanthropic and academic initiatives in both the United States and Switzerland.17 Thomas Pritzker’s first journey to Dolpo in 1978 initiated a sustained engagement with the area, particularly through projects such as the Pijor Library Project and the Revival of Vijer-Dolpo Project, both aiming to enhance local infrastructure and promote educational advancement. A significant turning point occurred in the summer of 1999, when members of these families travelled to Phijor and played a key role in the restoration of Nesar Monastery. On this occasion, they managed to obtain digital copies of all 642 volumes of Tibetan Kanjur manuscripts, collectively encompassing approximately 160,000 folios, safeguarded in the monastery’s library.18 These precious manuscripts subsequently underwent an initial academic assessment by Amy Heller, who created a preliminary catalogue for the collection. Heller’s early analysis indicates that the manuscripts housed at Nesar Monastery can be sorted into three main groups: (a) 456 manuscripts originally from Nesar (Gnas gsar) Monastery, (b) 98 manuscripts relocated from Lang (Glang) Monastery, and (c) 71 manuscripts acquired from Serkhang (Gser khang) Monastery. This evidence demonstrates that what Snellgrove described as “a complete set of the Tibetan Buddhist Canon” at Nesar Monastery comprised not a single, unified Kanjur, but a composite of several Kanjur collections originating from three monasteries.19 Nonetheless, until very recently, textual criticism and cataloguing on the Tibetan canon have generally referred to this combined material simply as the “Dolpo Kanjur,” without acknowledging their distinct sources. This treatment is evident in numerous contemporary academic publications, such as Clarke (2018), Chen (2018, pp. 109–10), and Schmithausen (2020), each employing the term “Dolpo Kanjur” to designate the entire collection.20

However, this dynamic has shifted with the advent of more thorough local fieldwork and systematic cataloguing initiatives. In 2018 and 2019, the rKTs (Resources for Kanjur and Tanjur Studies, University of Vienna) project team undertook comprehensive, onsite research on the Nesar manuscripts. Their investigation produced a historically informed catalogue that considers the provenance and historical context of each manuscript grouping. Accordingly, in the subsequent analysis of the Ggn, I adopt this classification scheme: volumes from group (a) are referred to as the “Nesar Kanjur,” those from group (b) as the “Lang Kanjur,” and those from group (c) as the “Serkhang Kanjur.” Of these three, only the Nesar and Lang Kanjurs preserve the complete Tibetan translation of the Ggn; the Serkhang Kanjur lacks this particular text.

The exact details concerning the initial creation of these manuscripts, as well as the motives behind the transfer of certain volumes from Lang to Nesar Monastery, continue to be unresolved. Still, evidence from bibliographic data found within the colophons and introductory notes of several manuscripts implies that the earliest surviving exemplars likely originate in the 13th century. This interpretation has gained support from recent radiocarbon dating, which places their production between the 12th and 15th centuries.21 Assuming this timeline is correct, the Nesar Kanjur—apart from the Proto-Kanjur compilations—constitutes one of the oldest extant Kanjur collections currently available to scholars. Hence, the scholarly importance and cultural value of the Nesar Kanjur, together with its related collections, should be regarded as greater than previously assumed, influencing not just the historical study of the Dolpo region but also scholarship on canonical development, modes of transmission, and manuscript practices across the Himalayas.

4. Analysis of Collation Results: Nesar and Lang Kanjurs22

4.1. Nesar Kanjur (mdo sde, ca. 212b1–303b5)

The Nesar Kanjur manuscript of the Ggn is written in the dbu can script and exhibits a consistently uniform appearance, with careful calligraphy maintained throughout. Each folio examined in the collation process contains precisely nine lines of text. At the centre of each folio are two circular marks indicating the string hole positions, a characteristic feature of Tibetan pothi-style manuscripts. Significantly, the small hole at the centre of these circular markers appears to result not from actual string usage, but rather from a compass or similar instrument employed in creating the marks, which points toward a solely visual or schematic function. The manuscript incorporates several features characteristic of old orthography (dag yig), such as ma ya btags (the superscribed ya on ma), da drag (“strong” or emphasized da), ’a rten (supporting ’a chung), and gi gu rlog (a reversed gi gu, see below).23

’di ni CDF1F2GoHHiHeJLNNgQRSTa2UXZ :: de ni Lg :: de nï Ns (Ggn 185, 1)

Various abbreviated forms (bskur yig) also occur sporadically, even in cases where there is no obvious spatial limitation. Although these abbreviations are attested, the overall rate of their use, at least in the portion of text analyzed here, appears to be lower than in other Kanjur manuscripts recovered from Western Tibet.

lam CDF1F2GoHeHiHeJLLgNQRSTa2UXZ :: laṃ Ns (Ggn 188, 3)24

par CDF1F2GoHHiJLLgNNgQRSTa2UXZ :: pa+r la He :: pa+r Ns (Ggn 198, 1)25

legs par byas pa CDGoHeHiHeJLLgNNgQRSTa2UXZ :: legs par byas pas F1 :: legs sa par byas pa F2 :: legs pa+r byas pa Ns (Ggn 199, 3)26

Another paleographic detail worthy of mention is the repeated occurrence of the letter ta written in a form that is almost indistinguishable from ha, particularly in examples such as yon tan . This distinctive shape of ta was absent from other textual witnesses consulted in this, leaving its origins unclear. It remains uncertain whether this reflects an individual scribe’s habit, a localized scribal practice, or a transitional calligraphic style within a particular lineage. These instances highlight the necessity of careful paleographic analysis when assessing Western Himalayan manuscript evidence.

The Nesar Kanjur provides twelve instances of distinct readings, or variants, that set it apart from all other witnesses represented in the collation. The majority of these variants seem to have emerged inadvertently during the transmission or copying process (see below). There is no indication of intentional, systematic alteration—specifically, recessional variation—intended to revise terminology or doctrinal content. One significant instance occurs at Ggn p. 166, line nr. 2, where the Nesar Kanjur uniquely preserves what is likely the correct reading within a recurring phrase:

śes rab rnam par dpyod do CDF1F2GoHHiHeJLLgNQRSTa2UXZ: śes rab rnam par spyod do Ns (Ggn 166, 2)

In this context, only Ns has spyod do meaning “to practice, to perform”, whereas the alternative dpyod do signifies “to investigate”. The former is not just more appropriate contextually, but it also receives support from the equivalent Chinese translation in T404 (xing bore 行般若; cf. T404, 634c01, with no direct parallels in Dharmarakṣa’s version). The following examples illustrate the range of variants identified:

Dittography:

phyogs kyi (CDF1GoHHiHeJLLgNNgQRSUXZ): phyogs kyis (F2): phyogs kyi phyogs kyi (Ns) (Ggn 151, 1)

Minor variation:

bar du rig (CDF1F2GoHHiHeJLNNgQRSTa2UXZ) :: rab tu rig (Ns) (Ggn 170, 3. Here bar du renders Skt. yāvat, but in Ns it is replaced by rab tu, which subtly changes the nuance. This variant recurs at several points in the work.)

rtag tu byaṅ chub kyi (CDF1F2GoHHeHiJLNNgQRSTa2UZ) :: byaṅ chub kyi (Ns) (Ggn 202, 4. Notably, the corresponding Chinese is simply puti 菩提, T404, 635c21)

Accidental omission:

chos thams cad kyi de bzhin nyid (CDF1F2GoHHiHeJLLgNQRSTa2UXZ) :: thams cad kyi de bzhin nyid (Ns) (Ggn 157, 7. Compare with T404 parallels: yiqiefa zhenru 一切法眞如)

sems can gyi de bzhin nyid (CDF1F2GoHHiHeJLLgNQRSTa2UXZ) :: de bzhin nyid (Ns) (Ggn 163, 2. Compare with T404 parallels: youqing zhenru 有情眞如)

gzag pa sgom pa nas (CDF1HHiJLNNgQRSTa2UXZ) :: gzag pa nas (Ns) (Ggn 173, 4)

śin tu bsdus śiṅ bzuṅ ba (CDF2HHiHeJLNNgQRSTa2UXZ) :: śin tu (Ns) (Ggn 202, 4)

In summary, the Nesar Kanjur exhibits only a limited number of unique or distinctive readings. However, most of its textual variations are most plausibly attributed to standard transmission errors, such as unintentional omissions, dittography, or orthographic peculiarities, rather than deliberate or systematic editorial alterations. This observation indicates that the scribe(s) responsible for the Tibetan Ggn in the Nesar Kanjur mostly preserved the established tradition and did not intentionally revise or modify the text.

4.2. Lang Kanjur (mdo sde, ca. 256b5–263a5)

The Lang Kanjur constitutes a carefully assembled manuscript collection, produced in a consistent dbu can script across all folios. Each folio exhibits uniform size and meticulously rendered, regular handwriting. In contrast to the Nesar Kanjur, whose folios contain nine lines, the Lang Kanjur features a format of eight lines per folio. Each folio bears two round notches at the centre, designating the location for string holes, consistent with the traditional Tibetan pothi book style. As discussed in previous scholarship (Zimmermann 2003, pp. 104–5), the Lang Kanjur retains several archaic orthographic conventions, such as ma ya btags, da drag, and ’a rten spellings.27 Additionally, the manuscript frequently reveals evidence of correction, a matter that will be analyzed further below. Where spatial limitations arise from line breaks or the proximity of string hole notches, the scribe(s) often resorted to abbreviated forms, including the anusvāra-like abbreviation for m (ṃ), or vertically stacked suffixes that depart from standard formations, here indicated by a plus sign (+), as exemplified below:

byaṅ chub sems dpa’ rnams (CDF2GoHeHiHeJLNQRSUXZ) :: byaṅ chub seṃs dpa+’ rnam+s (F1) :: byaṅ chub seṃs dpa’ rnams (Lg) (Ggn 151, 4. sems appears as seṃs at the line’s end.)

gzugs (CDF1F2GoHHeHiJLNQRSTa2UX) :: gzug+s (Z, Lg) (Ggn 177, 2. A vertical linkage of the -s suffix at the line ending.)

byaṅ chub sems dpa’ (CDF1F2GoHHiHeJLNNgNsQRSTa2UXZ) :: byaṅ chub seṃs dpa’ (Lg) (Ggn 192, 1. sems is spelled seṃs just before the string hole.)

thams cad (CDF1F2GoHHiHeJLNNgNsQRSTa2UXZ) :: thaṃs cad (Lg) (Ggn 192, 3. thams appears as thaṃs at the end of the line.)

dbugs (CDF1F2GoHHiHeJLNNgNsQRSTa2UXZ) :: dbug+s (Lg) (Ggn 192, 3. Vertical suffix -s before string hole.)

These patterns indicate that the scribe(s) of the Lang Kanjur, involved in work on the Tibetan Ggn at least, likely utilized a degree of flexibility when confronted with spatial limitations, presumably to preserve both the clarity and legibility of the manuscript. The Lang Kanjur exhibits fifteen distinct readings within the analyzed sample, differentiating it from all other witnesses considered in the collation. Of these, three instances highlighted below are especially noteworthy, as they are not just ordinary variants or errors but may have potential implications for interpreting the text.

mi rtag par CDF1F2HHiHeJLNNgNsQRSTa2UXZ :: myi ces bya bar Lg (Ggn 151, 2)

rjes su śes pas CDGoHHiHeJLNNgNsQRSTa2UXZ :: rjesu śes pas F1 :: rjes su śes pas | F2 :: rjes su dran pas Lg (Ggn 161, 1)

ñi tshe ba’i lam thams cad las ’das pa’i phyir CDF1F2GoHHiHeJLNNsQRSTa2UXZ :: ñi tshe ba’i lam thams cad ’das pa’i phyir Ng :: ñi tshe ba’i phyir thams cad las ’das pa’i phyir Lg (Ggn 187, 2. In this case, phyir appears in place of lam. Since phyir recurs within the same sentence, and because the relevant Chinese translation conveys the notion of the way, dao 道 (cf. T404, 635a25–26, 能超一切偏異道故), this usage most probably stems from an error in the scribal process.

Zimmermann, in his 2003 article, “A Preliminary Analysis of the Tathāgatagarbhasūtra as Found in the Lang Kanjur,” proposed a close connection between the Lang Kanjur and Phug brag Kanjur, suggesting that both may derive from a common source (Zimmermann 2003, p. 104).28 However, Chen’s 2018 analysis of the Nandimitrāvadāna observes that this relationship may not hold for every individual text preserved in the Lang Kanjur.29 For the Ggn specifically, the Lang Kanjur shares at least three textual variants with Phug brag (see below). These shared features go beyond minor scribal irregularities, such as inconsistencies in punctuation (shad) or grammatical particles, which could independently arise or be incidental. Instead, these involve the omission of entire phrases or the use of alternative words, thereby offering stronger support for Zimmermann’s observations.

rmi lam lta bu daṅ | CDGoHHiHeJLNQRSTa2UXZ :: om. F1F2LgNgNs (Ggn 180, 1)

smig rgyu lta bu daṅ | CDGoHJLNRSTa2UXZ :: mig rgyu lta bu daṅ | Q :: smig sgyu lta bu daṅ | HeHi :: om. F1F2LgNgNs (Ggn 180, 1)

yoṅs su bskaṅ ba’i CDGoHHiJLNQRSUX :: yoṅs su skoṅ ba’i HTa2 :: yoṅs su skaṅ ba’i Z (scribal error of skoṅ ba) :: yoṅsu skor ba’i F1 :: yoṅs su skor ba’i F2LgNgNs (Ggn 195, 2. F1F2 and LgNgNs have skor ba for skoṅ ba)

Nevertheless, the evidence further highlights a broader context concerning manuscript interrelationships. Several distinctive readings found in the Lang and Phug brag Kanjurs are also documented in the Nesar and Namgyal Kanjurs. This observation indicates that the Western Himalayan Kanjurs—Lang, Phug brag, Nesar, and Namgyal—constituted a regionally interconnected manuscript tradition, defined by patterns of parallel copying, textual borrowing, and instances of cross-institutional exchange. Furthermore, both Phug brag and Lang Kanjurs, alongside Tabo, Basgo, and Hemis Kanjurs, often preserve multiple versions of the same text, which demonstrates complex internal textual diversity and the integration of different lines of transmission within a single monastic library. Considering these factors, future research should approach these Kanjurs as active participants in a wider network of canonical production and adaptation, rather than as isolated or narrowly “local” traditions. In this context, texts were regularly circulated and revised to accommodate changing ritual, scholastic, and institutional priorities. Therefore, a comprehensive understanding of the Lang Kanjur requires not only rigorous philological collation but also codicological, historical, and network-based analyses, drawing on updated fieldwork, cataloguing projects, and comparative manuscript study.

4.3. Relationship Between the Nesar and Lang Kanjurs

There are twenty-two variants that are shared exclusively by the Nesar and Lang Kanjurs, setting these apart from all other editions. This demonstrates a significant degree of textual affinity between the Nesar and Lang Kanjurs, which are both preserved within the same monastery library. Notably, these two Kanjurs frequently share readings with Local or Independent Kanjurs, especially those seen in Phug brag and Namgyal. The instances listed below serve as significant examples, while minor variants—such as straightforward omissions, insertions, or grammatical particle differences—are excluded. An illustrative example occurs at Ggn 168, 2 (rnam par dpyod pa nas CDF1F2GoHHiHeJLNNgQRSTa2UXZ :: rnam par spyod pa rig nas LgNs), where the correct reading, or one accurately reflecting the Chinese translation, seems to be preserved solely in the Nesar and Lang Kanjurs. Such variants underscore the value of the Nesar and Lang Kanjurs in textual-critical research concerning Tibetan canonical texts.

Lexical or structural variation:

bdag med par rig | rnam par śes pa źi bar rig | CDF1F2GoHHiHeJLNNgQRSTa2UXZ :: bdag med pa daṅ | rnam par śes pa źi bar rig | LgNs (Ggn 156, 2–3. This variant concerns the question of whether anātma (bdag med pa, “non-self”) and śānta (rnam par śes pa, “tranquility”) are represented as separate or as conjoined concepts. The corresponding Chinese translation unambiguously treats them as distinct: 知識無我知識寂靜 “knowing non-self, knowing tranquility”, cf. T.404, 634b01–2.)

rjes su śes pas CDGoHJNQRTa2U: rjes su śes pas | F2HeHiLNgSXZ :: rjesu śes pas F1 :: rjes su śes so || LgNs (Ggn 167, 3. Lg and Ns present an alternate sentence structure.)

rjes su śes pas CDF2GoHHeHiJLNNgQRTa2UXZ :: rjes su śes pas | S :: rjesu śes pas F1 :: rjes su rtog pas LgNs (Ggn 175, 2. In this case, Lg and Ns agree in reading rjes su rtog pas “by thinking/repeatedly considering” in a recurrent phrase, though elsewhere both Lg and Ns, like the other editions, read rjes su śes pas “by knowing/realizing”.)

smon pa med par rig CDF1F2GoHHiHeJLNNgQRSTa2UXZ :: smond pa med pa daṅ LgNs (Ggn 177, 3. The sentence structure differs.)

śes rab rnam par dpyod pa nas CDF1F2GoHHiHeJLNNgQRSTa2UXZ :: śes rab rnam par spyod pa rig nas LgNs (Ggn 168, 2. The corresponding Chinese translation is 而行般若 (T.404, 634c09), in which xing 行 appears to correspond more closely with rnam par spyod pa as preserved in LgNs, than with rnam par dpyod pa in other editions. This indicates LgNs provides a more accurate reading in this context.)

thabs la mkhas pa’i CDF2HJNNgQRSUZ :: thabs mkhas pa’i F1GoHiTa2X :: chos la mkhas pa’i LgNs (Ggn 197, 4. thabs la mkhas pa’i or thabs mkhas pa’i are likely to represent the more authentic reading here, especially when considering the Chinese parallel shanqiao fangbian善巧方便 (T404, 635c2, Skt. upāyakauśala)).

Omissions:

bsam gtan bźi CDF2GoHHeHiJLNNgRSUXZ :: bsam gtan bźi pa Q :: bsaṃ gtan bźi F1 :: om. LgNs (Ggn 151, 2. Both Nesar and Lang Kanjurs omit the phrase bsam gtan bźi “the four dhyānas”, a crucial doctrinal term. The omission is validated through the Chinese parallel (T.404, 635a4–5 四攝四無量四禪四無色), which retains the expected passage. This supports the hypothesis of a textual loss during transmission, possibly due to a shared ancestor or the use of an incomplete or damaged exemplar.)

śes rab kyi ye śes kyis CDF1F2HHiHeJLNNgQRSTa2UXZ :: śes rab «kyi ye śes» kyis Go :: ye śes kyis LgNs (Ggn 195, 1. The lack of śes rab kyi “of insight” in both Lg and Ns becomes apparent when cross-referenced with the Chinese parallel zhihui 智慧 (T.404, 635b25, Skt. prajñājñāna), confirming the loss. This omission likely results from scribal error, which may compromise the doctrinal fidelity of the text.)

Insertion or duplication:

[bdag da$\stackrel{·}{n}$] sems can CDF1F2HHiHeJLNNgQRSTa2UXZ :: seṃs can Go :: sems can thams cad LgNs (Ggn 193, 1. Given the Chinese parallel woyouqing 我有情 (T.404, 635b17–18), it is plausible that thams cad in LgNs was an accidental interpolation, thereby altering the meaning. However, it remains uncertain whether this was a deliberate clarification or an inadvertent addition, particularly since sems can thams cad (Skt. sarvasattva) occurs with some frequency.)

mi ’jigs pa thob pa’i CDGoHHeHiJLNNgQRSTa2UZ :: mi ’jigs pa thoṅ pa’i X :: mi ’jigs pa’i F1F2 :: mi ’jigs pa myed pas thob pa’i LgNs (Ggn 195, 4. The inserted myed represents an instance of old orthography, ma ya btags, used for the negative med. Since the immediately preceding phrase is thams cad du sems god pa med pas, this can plausibly be interpreted as an accidental duplication of med. The corresponding Chinese phrase reads wusuowei 無所畏 “without fear”. See T.404, 635b26 for comparison.)

Reading shared with F1F2Ng:

rmi lam lta bu daṅ | CDGoHHiHeJLNQRSTa2UXZ :: om. F1F2LgNgNs (Ggn 180, 1)

smig rgyu lta bu daṅ | CDGoHJLNRSTa2UXZ :: mig rgyu lta bu daṅ | Q :: smig sgyu lta bu daṅ | HeHi :: om. F1F2LgNgNs (Ggn 180, 1)

yoṅs su bskaṅ ba’i CDGoHHiJLNQRSUX :: yoṅs su skoṅ ba’i HTa2 :: yoṅs su skaṅ ba’i Z (scribal error of skoṅ ba) :: yoṅsu skor ba’i F1 :: yoṅs su skor ba’i F2LgNgNs (F1F2 and LgNgNs use skor ba in place of skoṅ ba) (Ggn 195, 1)

Several noteworthy textual variants were ultimately identified, contributing valuable insights into the relationship between the Nesar and Lang Kanjurs (see below). In all such cases, the Lang Kanjur originally adopted readings that were identical to those of the Nesar Kanjur, which were subsequently considered erroneous. These readings were later revised or corrected to accord with the versions documented in other editions. This recurring phenomenon permits the formulation of the following hypotheses: (a) the compilation of the Nesar Kanjur predates the Lang Kanjur; (b) the Lang Kanjur was likely compiled with principal reliance on the Nesar Kanjur or a closely related source; (c) during the editorial process, at least one additional Kanjur manuscript was consulted for the Lang Kanjur, which offered readings distinct from those of the Nesar Kanjur; and (d) the additional manuscripts transmitted more reliable readings than the Nesar Kanjur. It should be emphasized, however, that these conclusions pertain specifically to the case of the Ggn and may not necessarily extend to other texts within these collections. Nevertheless, the textual data reveal evidence of systematic editorial intervention involving repeated corrections, signifying an intentional effort by the compilers of the Lang Kanjur to align their edition with more authoritative canonical models.

[mig yor lta bu’i] bar du rig CDF1F2GoHHiHeJLNNgQRSTa2UXZ :: «bar» rab tu rig Lg :: rab tu rig Ns (Ggn 170, 3. bar in Lg is a later addition inserted in the margin.)

de bźin du tshor ba daṅ CDF1F2GoHHiHeJLNNgQRSTa2UXZ :: «de bźin du tshor ba daṅ» Lg :: om. Ns (Ggn 172, 2–3. The omitted phrase de bźin du tshor ba daṅ (as in Ns) is inserted in the margin of Lg.)

sgom pa CDF1HHiJLNNgQRSTa2UXZ :: bsgom pa F2GoHe :: «sgom pa» Lg :: om. Ns (Ggn 173, 4. A later insertion in the margin adds sgom pa, which is otherwise absent in both Ns and Lg.)

rtag pa ma yin | CDF1F2GoHHiHeJLNQRSTa2UXZ :: rtag pa ma yin | rtan pa ma yin | Ng :: «rtag pa ma yin pa» | Lg :: om. Ns (Ggn 182, 1. The omitted phrase rtag pa ma yin pa (as in Ns) was later added in the margin of Lg.)

de mi chags pas CDGoHJLNRTa2S :: de «mi» chags pas U :: mi chags pa’i Z :: de chags pas HiNgX :: de chags par F1F2 :: de chags pa(’i)s Lg :: de chags pa’i Ns (Ggn 183, 5. The initial reading, as found in Ns, was the genitive form de chags pa’i; however, the gi gu was cancelled (marked by a line), and the text was revised to the instrumental form de chags pas. With the exception of F1F2 (which records -par) and Z (-pa’i), all other editions preserve the instrumental. Notably, every Kanjur from the Western Himalayan region, namely F1F2HiLgNgNsX, exhibits the same erroneous negative reading me.)30

kyis CDF1GoHJLNQRSTa2UZ :: kyi F2HiNgX :: «kyi» Lg :: om. Ns (Ggn 188, 1. kyi was not present in Lg originally, as in Ns, but was subsequently inserted in the margin.)

tshogs stsogs pas CDF2JLQRSTa2UX :: tshogs sogs pas HiZ :: tshogs bsags pas HN :: tshogs ltsogs pas F1 :: ’tshogs stsogs pas Go :: tshogs stsags pas Ng: s«ts»ogs pa’i stsogs pas Lg: stsogs pa’i stsogs pas Ns (Ggn 193, 1. The earlier version read sogs pa’i stsogs pas, which is consistent with Ns; nonetheless, this was subsequently changed to stsogs stsogs pas, in agreement with all remaining editions.)

śin tu bsdus śiṅ bzuṅ ba | rtag tu byaṅ chub kyi CDF2HHiHeJLNNgQRSTa2UZ :: śin «tu bzuṅ ba rtag» tu byaṅ chub kyi Lg :: śin tu byaṅ chub kyi Ns (Ggn 202, 3–4. In this case, Ns omits bsdus śiṅ bzuṅ ba | rtag tu and Lg appears initially to have followed Ns, but was later revised by adding back the omitted portion tu bzuṅ ba rtag in the margin.)

At present, it cannot be determined which edition(s) the editors of the Lang Kanjur consulted for review and correction. However, I provide below two instances—especially the first—in which the Lang Kanjur diverges from the Nesar Kanjur and instead reflects the textual variant found in the Gondhla manuscript collection:

snaṅ ba brñed nas CDF2HJLQRTa2UZ :: snaṅ ba rñed nas | HiNSX :: snaṅ ba rñed nas F1NgNs :: snaṅ ba chen po brñed nas GoLg (Ggn 186, 2–3. The reading in all other editions is snaṅ ba brñed nas, that is, “having attained the light;” in contrast, Go and Lg uniquely read snaṅ ba chen po brñed nas, which means “having attained the great light,” thereby conveying a different nuance.)

blta bas CDF1F2HHeHiJLNNgNsQRSTa2UX :: lta ba’i Z :: lta bas GoLg (Ggn 202, 1)

Even so, these two instances are inadequate for establishing any definitive textual relationship between the Lang Kanjur and the Gondhla proto-Kanjur. For a more persuasive evaluation, a wider-ranging comparative study incorporating lengthier segments of text is necessary to gather cogent evidence.

5. Conclusions

Since the pioneering research of Helmut Eimer in the late 1970s, followed by contributions from other scholars, the field of Tibetan Kanjur studies has undergone significant advancement. The two primary transmission lineages of the Tibetan Kanjur, specifically the Tshal pa and Them spangs ma traditions, are now well established, with relationships among different editions elucidated by stemmatic analysis, leading to several proposed stemmata codicum. In recent years, however, the expanded availability of additional manuscript collections has prompted more intricate and foundational questions. Since the mid-2010s, field investigations and subsequent studies led by scholars such as Helmut Tauscher, Markus Viehbeck, and Bruno Lainé have concentrated on newly found manuscripts in western Tibetan and Himalayan regions, especially Ladakh, Mustang, and Dolpo. This shift necessitates a reassessment of earlier scholarship, which largely prioritized the mainstream traditions and now requires an integrative approach that incorporates these recently surfaced textual sources. Notably, previous research into the Kanjur frequently relegated editions outside the Tshal pa and Them spangs ma lineages to the status of merely “local” or “independent.” As Peter Skilling already emphasized in 2012, continued research into these newly identified “local” manuscript traditions from Mustang, Ladakh, and Dolpo is crucial to addressing such issues and enhancing our comprehension of canonical transmission in Tibetan Buddhism (Tauscher and Lainé 2015, p. 463).

In this paper, I have examined the outcomes of textual-critical research on the Tibetan translation of the Gaganagañjaparipṛcchā, with an emphasis on manuscript traditions preserved at Nesar Monastery in the Dolpo district of western Nepal, specifically the Nesar Kanjur and Lang Kanjur. This study demonstrates, regarding the Ggn, that: (a) these two Kanjurs display several textual variants that set them apart from all other documented Kanjur and Proto-Kanjur editions, and at times provide more reliable readings that closely parallel the Chinese translation produced by Amoghavajra (T404) in the eighth century; (b) the Lang Kanjur seems to have been compiled chiefly using the Nesar Kanjur, or a closely connected textual tradition, as its base; (c) however, the editors of the Lang Kanjur subsequently referenced at least one additional manuscript during their editorial and revision activities. When considering the codicological data derived from the manuscripts themselves alongside recent radiocarbon dating findings, it appears that the Nesar and Lang Kanjurs date back to as early as the 13th century CE. Given that the Tshal pa and Them spangs ma mainstream traditions can be traced only to the mid-fourteenth century and the 1430s, respectively, and that the extant physical manuscripts from these lineages do not precede the 17th century CE, the Nesar and Lang Kanjurs are recognized as among the earliest surviving Kanjur collections. Moreover, these Kanjurs exemplify a unique tradition from the Western Himalaya, exhibiting pronounced differences from Tshal pa and Them spangs ma. Therefore, the Nesar and Lang Kanjurs are of great importance for future textual-critical research and the historical analysis of Tibetan Buddhist canonical literature. Ongoing research on these two Kanjurs will yield crucial evidence for reconstructing a more detailed and thorough narrative of Tibetan canonical transmission.