The Land That Flew: In Search of Oḍḍiyāna
Prologue: Three Travellers
The first traveller is a Chinese monk, ill and homesick. It is the spring of 520, and Song Yun has come over the iron-chain bridges of the Hindu Kush into the country his records call U-chang. The valley is at its loveliest: birdsong, butterflies over the flowers, evergreen trees in a mild climate. All of it makes him ache for China, and he falls seriously ill. After a month he finds relief by obtaining charms from the locals. He is a Buddhist and an imperial envoy, and he has been healed by a spell.
The second traveller is the Kashmiri pilgrim Jñānanetra, or so tradition says, in the middle of the ninth century. He walks into a cremation ground called Karavīra and stays there, worshipping and waiting, until the Goddess reveals herself. She comes as Maṅgalā, “the Auspicious One", surrounded by a retinue of yoginīs. Out of that encounter comes a whole Kashmiri Śaiva tantric theology of the Goddess: the Krama system. Its scripture places the place of revelation in the northern quarter, a great seat called Oḍḍiyāna, frequented by Siddhas and yoginīs.
The third traveller is a Tibetan, Orgyenpa Rinchen Pal (1230–1309). He arrives in the same valley after Buddhism there had all but vanished. He reaches Dhumathala, which he calls the heart of Uḍḍiyāna, and finds a self-born sandalwood image of the goddess Maṅgaladevī. He weeps, and he writes a guidebook that Tibetans use for centuries.
The three are separated by seven centuries and three civilizations, and they are looking for the same place. That place is the subject of this essay. Our Guru Padmasambhava is said to have been born there. It is called the Land of the Ḍākinīs and the source of the yoginī tantras. For Kashmiri Śaivas it is the Northern Seat, and for Śrīvidyā it is the central bindu at which the Goddess Mahātripurasundarī — the Great Beauty of All the Three Worlds — herself sits in the eternal embrace of Kāmeśvara, the Lord of Love. Yet where Oḍḍiyāna was has been contested for a century. Some Tibetan teachers now put it in Odisha, on the opposite side of the subcontinent.
My argument, in brief, is that Oḍḍiyāna was a real valley: the Swat valley of northern Pakistan and the hill country around it. It was then turned, by pilgrims, poets and tantric authors, into something that could float. Both halves of that sentence matter. The valley explains why the name attached to the north-west. The floating explains why it could later be found elsewhere.
I. A Name That Will Not Sit Still
The word appears as Uḍḍiyāna, Oḍḍiyāna, Oḍiyāna, Udyāna, Oḍyāna, Uḍḍayana and more. Tibetans write it O rgyan or U rgyan, and these are attempts to render the Sanskrit, not translations.
Three explanations of the name run through the sources, and they compete.
The garden. Udyāna means a park or orchard, and this is the sense 19th century Sanskritists record, as the name of a country in the north of India. It suits the Chinese transcriptions and a land of vines, forests and flowers.
The flight. Tantric authors preferred a more dramatic derivation from ud plus the root ḍī, “to fly". The Kubjikāmata tells it as a story: the goddess Kujeśvarī says that because she flew up (oḍḍitā) within Oḍḍīśa, the place will be known as Oḍḍiyānaka. A second text, the Ṣaṭsāhasrasaṃhitā, tells it slightly differently. In the incomparable Prof. Sanderson's rendering, the Siddha Uḍḍīśa ascends into the sky there. This root is the one tradition gives to ḍākinī, the “sky-goer". The Sarvabuddhasamāyoga, one of the earliest sources for the gloss, explains the ḍākinī as the power of moving everywhere through space. So the name seems to promise flying women.
The people. A third layer is older and much less romantic. A Kharoṣṭhī relic inscription records Senavarma as king of the Oḍi, and a study of Barikot places it in the third quarter of the first century. Two very early Swat inscriptions, of Senavarma and of his father Ajitasena, use the form Oḍi. On the ground, the Italian mission glosses Udegram as “Ora/Oḍi" in Greek and Indian sources. A natural reading is that Oḍi was the name of a people and a place, and that “-yāna", “flying" and “garden" are all later re-castings of it. That reading is a hypothesis, but it fits the evidence better than either learned etymology does.
The “flying" explanation looks even less secure once the ḍākinī herself is examined. The name ḍākinī can also be linked to words for summoning and drumming rather than flying. The Sanskritic flight may therefore be a gloss on a much older, more local word. In practice both stories worked. The name gave every tantric author a ready-made image of the goddess in flight over a mountain landscape.
A further early witness deserves care. A Kushan-period inscription in the Mathura corpus records the gift of a pillar to Huviṣka's vihāra by the mendicant Jīvaka of “Udeyana". Popular summaries sometimes place that vihāra in the north-west. The inscription, however, belongs to the Mathurā material. It is best taken as evidence that a monk from the Swat country was travelling in the Kushan heartland at the height of the empire.
II. Through Chinese Eyes
The Chinese pilgrims are our earliest and most detailed outside witnesses. They do not use the tantric vocabulary, since they wrote before it existed, but they show what the valley was like before it became a myth.
Faxian, c. 400: a pious frontier
Faxian crossed the river and found himself in “Woo-chang". He judged it truly a part of North India, where people spoke the language of the Middle Kingdom and ate and dressed as in the heartland. He found five hundred monasteries of the Little Vehicle, which supplied travelling monks for three days before asking them to move on. He was shown the Buddha's footprint, which appears long or short depending on the viewer, the rock where he dried his clothes, and the site where he tamed a wicked dragon. He stayed for the summer retreat.
Song Yun and Huisheng, 519–521: the land of spells
Song Yun's party arrived in the twelfth month of 519 and left for Gandhāra the following spring (Chavannes's reconstruction of the itinerary). Huisheng stayed in the country for two years. Their account is our first evidence that Udyāna was a place where one went to learn magic. It comes through a story of dragons.
Far to the east, in a mountain lake, a hostile dragon killed a merchant with a spell. The local king of Pan-to, on hearing of it, handed his throne to his son and went to U-chang to learn spells. After four years he returned, enchanted the dragon into human form and banished it. A king abdicates to become a student of charms, and the school is in Uḍḍiyāna.
Song Yun also gives a sober picture of the valley's court and religion. The king was a vegetarian who honoured the Buddha morning and evening with drums, conches and lutes, and then held court after noon. Murderers were exiled to the desert mountains and not executed, and doubtful cases were tried by ordeal with drugs. The valley he describes is neither wild nor decayed. It is a settled kingdom and full of monasteries, with a running current of magic.
Xuanzang, c. 630: the ledger of decline and the legends of the nāga
Xuanzang is the fullest witness, and his account is doubled-edged. The valley he describes is about five thousand li in circuit, with abundant grapes and thick forests, and a people he finds clever but unapplied, who practise the art of charms. He records some 1,400 old monasteries along the Subhavastu, once housing eighteen thousand monks, now mostly desolate. His report is unusually direct on the internal tension. The monks who keep the moral rules purposely forbid the use of charms. So the magic was widespread enough that the disciplined clergy felt obliged to legislate against it.
He lists five Vinaya schools: the Sarvāstivādins, Dharmaguptas, Mahīśāsakas, Kāśyapīyas and Mahāsāṃghikas. He names Muṅgali as the capital where the kings mostly reign, and he tells the region's founding legends, which are among the most vivid in his book:
The master of charms. A man named Gaṅgi, in the time of an earlier Buddha, used his charms to restrain wicked dragons from flooding the country, and the people paid him a yearly measure of grain. When some stopped paying he became a poisonous dragon himself, at the source of the Subhavastu. The Buddha subdued him, and the river floods every twelfth year as his allowance. This is a mantra-master turned water-serpent, and the story is about the valley's magic as much as its religion.
The nāga bride. A Śākya refugee married a nāga maiden at a mountain lake above Muṅgali. Her father gave him a sword, and with it he killed the king of Udyāna and took the kingdom. When she grew a nine-fold serpent crest in her sleep, he cut it off, and she cursed the royal line with head pains. The ruling dynasty is thus descended from a serpent-woman of a mountain lake.
The relics. Uttarasena, king of Udyāna and son of that Śākya youth, arrived too late at Kuśinagara and was treated with scorn, until the gods announced that the Buddha had promised him a share of the relics.
He adds the geography of sacrifice: the mountain where the Buddha gave up his body for half a verse, and the place where a former Buddha as King Śibi cut his flesh to save a dove. Song Yun records the same Jātaka landscape, including Vessantara's giving of his children and the offering of a body to a tigress. The valley is a holy land of past-life self-sacrifice, and in this it resembles Gandhāra generally.
And he makes one remark that should give any confident local-boundary theorist pause. He says the old capital of Udyāna stood at Daril, far up the Indus, which is the Darel of the upper Indus. “Udyāna" for Xuanzang was already a larger, looser thing than the valley of Muṅgali.
The later travellers
The Korean Hyecho, in the eighth century, reports that after Gandhāra he went north into the mountains, arrived in Udyana after three days, and noted that it was locally called Oddiyana. The Chinese monk Wukong travelled about 753 from Gilgit into Uḍḍiyāna in the Swat valley (Wuzhangna), and on to Gandhāra. By then the name was in local use and matched the Sanskrit tantric form.
What the pilgrims establish is a spell-culture: Brahmanical charms in Song Yun, “charms" in Xuanzang, and a mantra-learning Buddhism. Davidson calls this the first stage of Oḍiyāna's aura, an early collection of spells evident from the sixth century onward. Women appear later and elsewhere.
III. The Valley on the Ground
What can archaeology add to the texts? Quite a lot, though not everything one might hope.
The Italian mission began in earnest in the mid-1950s. Tucci came to Swat in 1955 after Tibet closed to foreign visitors, drawn by a region important in Tibetan history as Padmasambhava's birthplace. Domenico Faccenna excavated the great stūpa at Butkara from 1956, establishing that in the second century BCE it received Hellenistic architectural decoration, suggesting direct Indo-Greek involvement.
The sites map onto the texts remarkably well:
Udegram evokes Arrian's Ora, conquered by Alexander. Its hill of Gogdara is probably the oldest settlement, with proto-historic rock engravings.
Barikot is Bazira/Vajīrasthāna, and is now the urban centre that has given the most insight into ancient Gandhāran urban society.
Butkara I is the main Buddhist settlement of the region, identified by Tucci with Song Yun's Talo and Orgyenpa's Dhumatʻala.
Shingardar is traditionally the stūpa King Uttarasena built for his share of the Buddha's relics.
For the late period, Anna Filigenzi's survey of the rock sculptures is the main resource. Her monograph on Buddhist rock sculpture of late antique Swat/Uḍḍiyāna appeared in 2015. The Jahanabad Buddha near Manglawar is one of the region's largest rock-carved Buddhas, of the seventh century.
Here the honest student must record a puzzle. Archaeologists have been struck by how few Vajrayāna artefacts Swat has yielded, given its reputation as the Vajrayāna country. The usual explanations are decline, insecurity and the perishability of ritual material. Filigenzi's own conclusion is that Swat saw the emergence of Vajrayāna, the last offshoot of a long tradition that flourished elsewhere. The absence is real, and the sceptics are right to stress it. But early tantric practice, spell-craft and cremation-ground ritual leave little in stone.
Inscriptions and dynasties
The Turki Śāhi period supplies the strongest epigraphic anchor. The Gardez Gaṇeśa carries a dedication connected with a king Khiṅgāla of Oḍiyāna. Davidson, citing Kuwayama's re-reading dated to 753 or 765, treated it as having settled the identification with Swat. I would put it more cautiously. The reading depends on emendations of an inscription full of engraver's mistakes, and Sircar read the crucial word differently. It is strongly supportive but not decisive. Sanderson adds that the Jain Kumārapālacaritasaṃgraha also locates Uḍḍiyāna in this area. And the so-called Hindu Śāhis are argued to have been Uḍi/Oḍi rulers, the same people whose kings were known by the Kushan period. If that is right, the Oḍi name survived in the rulers of the region for a thousand years.
IV. The Valley of the Peris
Even today, anyone who walks up into Swat's high country meets a landscape of lakes, and each lake has stories. Above Utror lies Pari Lake at about 14,400 feet. It is called Pari, or Khapiro, from the belief that fairies live there and bathe in its cool, clear water. Nearby, Kundol Lake carries a folktale of a golden bowl that appears at its centre every month, shimmering like the moon, which no one has been able to touch. Mahodand Lake, too, is said locally to be inhabited by fairies and mystical creatures.
The sources for Swat speak of fairies living and bathing in the lakes, without specifying night. Moonlit bathing is a well-known motif of neighbouring northern lake-lore. At Saif ul Malook in Kaghan, fairies are said to descend to bathe under the full moon. And parī is a Persianate word belonging to the Muslim folk world of today.
Even so, the resonance is hard to ignore. Xuanzang's valley had a nāga maiden at a mountain lake. Padmasambhava was said to be born on a lotus in Lake Dhanakosha. The Krama's yoginīs are airborne women of a mountain land. Ḍākinīs, in the ancient view of the word's history, were magical women of summoning and drumming. None of this proves continuity, but it does suggest that a landscape of glacial lakes and feminine spirits of place can persist while the religious vocabulary changes over it, from nāga to ḍākinī to parī.
V. The Northern Seat: What Kashmir Knew
The strongest single witness to the location of Oḍḍiyāna is not Chinese or Tibetan. It is Kashmirian, and it comes from people who wrote in Kashmir, next door.
The Krama and its cremation ground
Alexis Sanderson's 2007 study, The Śaiva Exegesis of Kashmir, is the decisive treatment. Robert Mayer summarises its result: Sanderson surveyed the far-flung places identified with Uḍḍiyāna, concluded that it lay near Kashmir, and thereby reconfirmed Tucci's identification with Swat.
The Krama scriptures locate Oḍḍiyāna carefully. The Kālīkulakramasadbhāva opens with the northern quarter and a great pīṭha called Oḍḍiyāna. Inside it lies the Karavīra cremation ground, crowded with Bhairavas and Mothers. There Kālī stands on Bhairava, devouring the power of Mahākāla, void in form, infinite, adorned by fifty Rudras and attended by sixty-four yoginīs. Bhairava then asks the goddess why all these mighty yoginīs have assembled, and what secret they are hiding. The answer is the Krama, presented as instruction given by the Goddess to Bhairava, an inversion of the usual gender of teacher and pupil, formulating a secret oral tradition concealed in the innermost awareness of the yoginīs of Uḍḍiyāna, the Pīṭha of the North.
The text does not stand alone. The chapter colophons of two Krama scriptures say they emerged in the Northern Pīṭha and were promulgated by Śrīnātha. The Yonigahvara has a colophon speaking of the Uttarāmnāya and the Oṃkārapīṭha, and the Manthānabhairava itself treats Oṃkārapīṭha as another name for Oḍḍiyāna. So the seat had several names: Oḍḍiyāna, Oṃkārapīṭha, the Northern Seat, and in the Krama's own idiom the Praṇava-pīṭha.
The founder and the women
Jñānanetra, also called Śivānanda, was believed to have received the Krama revelation in Oḍḍiyāna directly from its yoginīs, the Pīṭheśvarīs, or from their leader Maṅgalā, also called Vīrasiṃhā Svāminī, and all Krama authors trace their lineage to him. Abhinavagupta's Kramakeli says Śivānanda received his instruction from the Pīṭheśvarīs in the Uttarapīṭha. Mayer dates Jñānanetra to about 850–900, and notes that he received his revelation at a tangible place, the Karavīra cremation ground in Uḍḍiyāna.
What followed is the most remarkable feature of the lineage. Sanderson notes that the Krama lineage is distinguished from all others known to us by the prominence of female gurus. He lists them: Maṅgalā, Rūpānandā, Keyūravatī, Madanikā, Kalyāṇikā and Īśānī. One popular account adds that Jñānanetra initiated seventeen disciples, and that all three of his primary disciples were women, his successor being the siddhā yoginī Keyūravatī, known as the “Goddess K". This also raises an open question, whether Maṅgalā was a human woman so powerful that she was called a goddess, or an archetype in Jñānanetra's imagination, and if human, whether her sixty-four yoginīs were real women too. It cannot be settled, but the question is fair, and it is exactly the one the sources invite.
The lineage did not stay in the valley. Its radical non-dualism strongly influenced the schools of Recognition, Vibration and the Kashmiri Trika of Abhinavagupta, who was himself an initiate of the Krama. Some of the greatest works of Kashmir Śaivism therefore rest on a revelation received in a cremation ground in the mountains to the north-west.
A town called Maṅgalā
There is a striking convergence of names. One version of the story has the pilgrim Jñānanetra reach a town called Maṅgalāpura, probably modern Mingora, where nearly everyone was said to be a practising Tāntrika. Lamotte's older identification is different and arguably better anchored: Maṅgalapura is Mangalaor, the Meng-jie-li of Xuanzang, capital of the Uḍḍiyāna kings. Xuanzang's Muṅgali is the royal seat. And Orgyenpa, six centuries after the Krama, found a Maṅgaladevī image at the heart of Uḍḍiyāna.
So a Chinese pilgrim (seventh century), a Kashmirian scripture (ninth) and a Tibetan pilgrim (thirteenth) each turn up a “Maṅgalā" in the same valley. Maṅgalā is a common auspicious name, so I would not press the coincidence too far. But three independent witnesses do not usually converge by accident, and I know of no rival valley in which all three “Maṅgalās" can be found.
Kubjikā, the three Siddhas, and Śrīvidyā
The Krama is only one of several traditions that make Oḍḍiyāna a founding seat.
Kubjikā. In the Kubjikāmata the goddess, after passing the “Western Himagahvara", becomes incarnate in four Mahāpīṭhas: Oḍḍiyāna, Jālandhara, Pūrṇagiri and Kāmarūpa.
The Saṃvartā tradition. It teaches three original Siddhas and seats: Oḍḍīśanātha of Oḍḍiyāna, Ṣaṣṭhīśanātha of Pūrṇagiri and Mitreśanātha of Kāmarūpa, with Jālandhara as the “missing" fourth.
The Mañjuśrīmūlakalpa. This ancient Buddhist mantra-manual says excellent mantras can be mastered in the North: in Kāpiśa, Balkh, throughout Uḍiyāna, in Kashmir, Sindh and the Himalayan valleys.
Śrīvidyā. The Goddess Lalitā's Śrīcakra keeps the four-seat scheme, but the texts disagree on the assignments. The dominant account has her three śaktis symbolised as the mountains Kāmagiri, Pūrṇagiri and Jālandhara, with the apex, the very secret Oḍḍiyāna, at the centre of the Śrīyantra. The Lalitāsahasranāma commentary agrees that she abides in Oḍyāṇapīṭha and in the bindu. Others differ: one summary of Kashmirian Nityā worship puts Jālandhara, Pūrṇapīṭha and Uḍḍiyāna at the points of the triangle, with Kāmarūpa at its centre. The four seats are secure across the tradition. Which one sits at the centre is not.
One more parallel is worth noting. The Śrīvidyā navāvaraṇa worship in its eighth enclosure honours four Yuganātha gurus: Mitreśa, Uḍḍīśa, Ṣaṣṭhīśa and Caryā. Three of these match the Saṃvartā Siddhas of Kāmarūpa, Oḍḍiyāna and Pūrṇagiri. I infer, without a source, that Caryā stands for Jālandhara. If so, Uḍḍīśa, the lord of Oḍḍiyāna, is still named in Śrīvidyā worship today.
The claim sometimes made, that Śrīvidyā as a whole descends from Jñānanetra's three female disciples, is more than I can document. What is documented is narrower. Kashmirian Śākta authors wrote exegesis on the practice of Tripurasundarī from the standpoint of the Kālīkula and Trika. So the Krama and the Tripurā tradition shared a Kashmirian Śākta milieu, and the pīṭha geography is common ground.
What “Uttarapīṭha” means, and doesn't
The term needs handling with care. For the Krama it is straightforward: the Northern Pīṭha is Oḍḍiyāna. But the word is polysemous in the wider literature. In the Kubjikā corpus uttarapīṭha can be a ritual label for the left corner of a triangle, Pūrṇagiri, which says nothing about geography, since Pūrṇagiri is in the Deccan. Some modern writers also read Kashmir itself as the Uttarapīṭha (Rastogi, on the strength of a single mention in the Tantrālokaviveka). And one popular account calls the Krama “Auttarānaya" because it started in Kashmir, the Uttarapīṭha. Such loose usage shows how the term drifted. The safe conclusion is that the Krama texts themselves, which are the earliest, say “in the north" and name Oḍḍiyāna. That is enough to rule out the south-east.
VI. The Geography of Flight
A curious thing happened to Oḍḍiyāna in the eighth and ninth centuries. It stopped being only a place and became a device.
The lists. The great tantras enumerate sacred places. The Cakrasaṃvara list of twenty-four runs from Lampāka, Kaliṅga and Kośala through Kāmarūpa, Oḍra, Mālava, Devīkoṭa, Rāmeśvara and Godāvarī to Arbuda, Uḍḍiyāna, Jālandhara and Pullīramalaya. Note that Oḍra and Uḍḍiyāna appear as separate entries in the same list. I return to this below.
The body. The seats were then internalised. In Jigme Lingpa's tradition, the crown of the head is Jālandhara, the right ear is Oḍḍiyāna and the left ear Godāvarī. The Sampuṭodbhava does something similar (it puts Uḍḍiyāna in the right ear). The pilgrimage became a meditation.
The cosmic myth. In the Rudra-subjugation myth, Rudra's scattered body-parts sanctify the earth. His heart fell in Uḍḍiyāna, creating the conditions for tantra to arise there. The valley becomes the heart of a cosmic drama.
Davidson's three stages. Ronald Davidson summarises the process: an early collection of spells from the sixth century, the Indrabhūti myth in the eighth, and the extensive mythologisation of Oḍiyāna in the yoginī tantras from the ninth. Each stage adds a layer without erasing the last.
The other elsewheres. In the same idiom there are other far countries. The Ḍākārṇava has a hero named Camara who forms a pair with the ḍākinī Camarī, and Cāmara is also a country or division of the world in the Mahāvyutpatti. In Tibetan cosmology, Cāmara (Ngayab Ling) is one of two outlying subcontinents, to the southwest of Jambudvīpa. It is where Padmasambhava went to subdue the rākṣasas. The Tibetan biographies treat it, and Oḍḍiyāna, and Laṅkā, as near-interchangeable far lands. In the twelfth-century Zangling-ma, Padmasambhava's destination is defined at different points as Cāmaradvīpa and as Laṅkā, and in the last chapter the king mourns after the master has departed for Oḍḍiyāna. I take “Camara" to be this cosmological term. The point is that once Oḍḍiyāna is one of several sacred elsewheres, its location becomes a matter of literary and ritual convenience.
A guardian's name. The Manthānabhairava lists a guardian called Ḍāmara for Oṃkārapīṭha. The same text names Ḍāmara among the eight Bhairavas of Pūrṇagiri. Whether these Ḍāmaras are related to the ḍāmara tantras of magic is a question I leave open.
VII. Women of the Valley
The most human evidence for Oḍḍiyāna is that it produced female teachers. The Kashmirian evidence has already appeared. The Buddhist evidence is now being recovered by Ulrich Timme Kragh, whose 2018 study is titled, in full, “Chronotopic Narratives of Seven Gurus and Eleven Texts: A Medieval Buddhist Community of Female Tāntrikas in the Swat Valley of Pakistan". Kragh has surveyed eleven Sanskrit texts from Swat ascribed to six female gurus.
Two of them stand out. The earliest known female guru from Swat was a local princess, Lalitā Devī, who studied with a forest recluse. The best-known is Lakṣmī: Princess Lakṣmī of Swat, who possibly lived in the tenth century, wrote a Sanskrit work on sahaja now preserved only in a Tibetan translation made in Nepal between 1070 and 1090. That same text also lists early transmission-lineages of male and female teachers. So the Buddhist and Śaiva traditions agree. In this valley, women were teachers, and women's names sit at the head of lineages.
The Tibetan notion of “the land of the ḍākinīs" now looks less like fantasy. It has a plausible historical basis in a real society of tantric women, later magnified and mythologised.
VIII. Tibet’s Orgyen
Tibetan tradition cannot serve as a primary witness for the location. By the time the hagiographies reached Tibet, the place was a pure land. But it preserved valuable information, and the Tibetan account of Padmasambhava is why so many readers first meet the name.
The standard story has Padmasambhava appearing on a lotus in Lake Dhanakosha in Oḍḍiyāna, which sources identify with the modern Swat valley. King Indrabhūti, sailing home from a quest for a wish-fulfilling jewel, finds the child in the petals of a multicoloured lotus and takes him to the capital as heir. Prahevajra or Garab Dorje, the first human Dzogchen master, is likewise said to be a native of Oḍḍiyāna, and the Mañjuśrīmūlakalpa, the Krama and the Nyingma all treat the north-west as the source of their deepest teachings. Mayer notes that Padmasambhava's Nyingma tradition and Kashmir's Śaiva tradition share a veneration of Uḍḍiyāna as a tantric holy site and source of scripture, which suggests real historical entanglement between the Tibetan and Kashmiri cultural zones.
One Tibetan detail survives with striking consistency: the north-west. The Seven-Line Prayer to Padmasambhava opens by placing his birthplace in the north-western part of the land of Orgyen. This agrees with the Kashmirian “northern quarter" and the Mañjuśrīmūlakalpa's “in the North". It may well be a Kashmirian inheritance, which is Tibet's real evidentiary contribution to the question.
Mayer also makes a subtle observation about the Tibetan side of the debate: many Indian traditions have treated Uḍḍiyāna as so sacred and magical that its geographical location itself has sometimes moved, and a number of Tibetan lamas today place it in Odisha, though most academic scholars since Tucci have centred it on Swat. The scholarly and devotional traditions can both be right about their own subject matter. The valley is one thing, the pure land another.
Orgyenpa's pilgrimage supports the identification from the Tibetan side. His itinerary, the Guide to Uddiyana, is mystical, but allows a reasonably accurate reconstruction of his route and the cities he visited, and Tucci translated it in 1940. The arrival of a Tibetan in Swat, seeking Padmasambhava's homeland, in the very valley where Chinese, Kashmirian and archaeological evidence converge, is a strong piece of independent confirmation.
IX. The Eastern Claim: The Orissa Hypothesis in Full
Now to the rival. The argument that Oḍḍiyāna was in Orissa (Odisha) is not a fringe notion. It has been defended by serious Indian scholars, and it appeals to real evidence. It deserves a fair hearing.
The case for Orissa
Names. N.K. Sahu argues that Oḍa, Oḍra, Uḍra, Oḍiviśa and Oḍiyāna are used interchangeably for Uḍḍiyāna. He notes that the Sādhanamālā spells it Oḍrayāna and the Kālikā Purāṇa gives Uḍḍiyāna or Oḍra.
Indrabhūti. The Caturaśīti-siddha-pravṛtti calls him king of Oḍiviśa, and his Jñānasiddhi opens with an invocation to Jagannātha, a deity intimately associated with Odisha.
Literary sources. The Hevajra Tantra, pīṭha-nirṇaya literature and the Kālikā Purāṇa have been read as indicating a location in Orissa, even though Tibetan sources hint at the north-west.
A late inscription. A 1473 inscription under King Virūpākṣa, from the Jambai Śiva temple, is said to refer to Orissa as Oḍḍiyāna.
Archaeology. Orissa has an abundance of early Vajrayāna material: mandalas, wrathful deities, Tārā forms. And it has two yoginī temples.
East-Indian Siddhas. Bhattacharya tied Indrabhūti's Sambhala and the rival Laṅkāpurī to East Bengal or Assam, and noted that Luipa was a Bengali; Barber argues that Vairotsana's teacher lived near Dhanya-kota in the south-east, not at Dhanakosha.
The case against
First, the names are not one name. The Cakrasaṃvara list has both Oḍra and Uḍḍiyāna, as noted. Bagchi separates two Tibetan series: O rgyan, U rgyan, Oḍiyāna tied to Indrabhūti, and Oḍiviśa tied to Uḍra (Orissa). And the Chinese called Orissa Wu-tu, distinct from Wu-chang. Sahu's merging of the two series assumes what needs to be shown.
Second, the chronology runs the wrong way. Joshi points out that Odisha became a centre of Tantric Buddhism after the seventh century, while Swat was a centre of tantric practice long before 700. The Odisha yoginī temples are similarly a little later. Hirapur is traditionally attributed to a queen and dated to 864, and Ranipur-Jharial appears to date from soon after 900. Both are later than, or at best contemporary with, Jñānanetra's revelation.
Third, yoginī temples are pan-Indian. There are around thirteen 64-yoginī temples in India, at Khajuraho, Bhedaghat, Mitaoli, Dudhai, Badoh, Hinglajgarh, Shahdol and Naresar in Madhya Pradesh, Hirapur and Ranipur in Odisha, Lakheri and Rikhiyan in Uttar Pradesh, and Kanchipuram in Tamil Nadu. Another count gives eleven extant, two of them in Odisha. Odisha is one region in a broad early-medieval Śākta culture. A yoginī cult shows a region was tantric, not that it was Oḍḍiyāna.
Fourth, the Kashmirian texts say north. No Krama source suggests the south-east, and those are the earliest and most explicit witnesses.
Fifth, the 1473 inscription is very late. By the fifteenth century “Oḍiyā" (Oriya) and “Oḍḍiyāna" were phonetically close, so the usage more plausibly reflects later confusion than an original identity. That is my interpretation, and it is not itself proved.
Sixth, the Oḍi ethnonym is attested in Swat in the first century. That is the only place where a people called Oḍi are epigraphically documented at that date.
An assessment
The Orissa argument answers a different question. It shows persuasively that Orissa and Bengal were major centres of Buddhist and Śākta tantra, and I concede that fully. The Siddha movement, the Sahajiyā songs and the great monasteries of Ratnagiri and Lalitagiri are real. What it does not show is that the toponym Oḍḍiyāna belonged there. Its own advocates half-concede this. One recent paper states that literary references in Tibetan sources hint at a north-western location, while the Hevajra Tantra, the pīṭha-nirṇaya and the Kālikā Purāṇa indicate a possible location in Orissa. That is a fair statement of a conflict between sources that the hypothesis does not resolve.
The confusion is also historically understandable. Once Oḍḍiyāna had become a symbolic seat of the yoginīs, Orissa, which had its own Oḍra name and its own yoginī cult, could naturally be drawn into the same category. That is the floating of the sacred geography described in section VI, and it may be the origin of the Orissa claim.
X. Other Candidates
For completeness, the other proposals.
Ghazni and Bactria. The sixteenth-century Buddhaguptanātha identified Oḍḍiyāna with Ghazni. Herbert Günther, repeating this, placed it in Bactria and its neighbourhood, exposed to Gnostic and Manichaean ideas.
South India. Lokesh Chandra argued for a location further in South India.
A greater Uḍḍiyāna. John Reynolds suggests the name may cover a much wider area including parts of Pakistan, Afghanistan and even western Tibet (Zhang-zhung). Waddell noted that Xuanzang's extent implies a name covering much of the hill region south of the Hindu Kush, from Chitral to the Indus, and Patrul Rinpoche places Garab Dorje's birthplace near Lake Kutra, in the Dhanakosha region between Chitral, Gilgit and Swat. Xuanzang's mention of Daril gives this view some textual support.
East Bengal and Assam. As discussed above (Bhattacharya).
Of these, the “greater Uḍḍiyāna" idea is the most useful. It does not conflict with Swat as the core. It simply allows the name to have had a penumbra in the upper Indus hills.
XI. Verdict: A Valley in Five Layers
I read the evidence as five layers, stacked on one valley.
A people and a place. The Oḍi of the Swat valley had kings by the first century CE. Their towns were Ora (Udegram) and Bazira (Barikot).
A Buddhist holy land. The valley was rich in Jātaka sites, held relics of the Buddha, and once had 500, and later 1,400, monasteries. It gained significance as a place of Buddhist practice already under Aśoka.
A mantra land. By the sixth and seventh centuries it was famed for charms, and Buddhists and even Pamir kings went there to learn them.
A yoginī-tantric seat. From the eighth and ninth centuries a Śākta and Buddhist sacred geography grew up around it, with cremation grounds, sixty-four yoginīs, and women gurus such as Maṅgalā, Keyūravatī, Madanikā, Kalyāṇikā, Lalitā Devī and Lakṣmī.
A floating symbol. Tibetans (and their Secret Mantra masters from India) made it a pure land and mapped it onto the body. Others relocated it, a move that partly explains the later Orissa claim.
The verdict follows from the layers. Oḍḍiyāna, in the sense in which Kashmirian Śaivas, Chinese pilgrims and Tibetan translators used the word, was the Swat valley, centred on the Manglawar–Mingora–Butkara area, with a penumbra reaching up the Indus hills. Later devotion made it larger and stranger.
What remains unresolved
I would not want to overstate the case, so here are the real open questions.
The Vajrayāna gap. Swat has yielded few overtly tantric artefacts. The reasons offered are plausible but not proven.
The Gardez reading. It supports the identification but rests on emendation.
The Maṅgalā convergence. It is suggestive, but the name is a common one.
The lineage to Śrīvidyā. The shared milieu is clear, but I know of no documented line from the Krama women.
The Buddhist end. Traditional accounts blame Mahmud of Ghazni's invasions for ending Buddhism in the eleventh century, though Dalton suggests it lingered in rural areas near the Tibetan border. The final chapter of the valley's tantric history is still poorly documented.
What would change my mind? A securely dated tantric assemblage in Orissa carrying the Oḍḍiyāna toponym before the ninth century would be the evidence needed. A Krama or Kubjikā text placing Oḍḍiyāna in the east would be another. Neither exists in the sources I have seen.
Epilogue: The Valley That Kept Moving
Return for a moment to the three travellers. Song Yun, ill and homesick in a spring valley, was healed by charms. The Kashmiri pilgrim Jñānanetra, in the Karavīra cremation ground, was taught by women who may or may not have been human. Orgyenpa, weeping at Dhumathala in a valley of ruined monasteries, was looking for a country that had already become a legend.
The Chinese pilgrim saw a kingdom. The Kashmiri saw a seat of power and a lineage. The Tibetan saw a paradise. All three were right, because the valley had been all of these things in turn. The lakes are still there, and locals still say the fairies bathe in them. The name, meanwhile, has flown on: into Tibetan prayers, into Dzogchen Guruyoga, into Śrīvidyā worship, into the ears and hearts of the tantric body, and into the modern discussion about where it was. That flight is the reason for the discussion. It is also, in the end, a more honest testimony to what Oḍḍiyāna meant and continues to mean than any single map.
