Chengdu to Yushu, written while we drive

From the grassland the peak reads as a pyramid: steep snow on three sides, a long ridge running north-east, and a summit that spends a great deal of its time inside cloud. That shape is the whole story. The faces are steep enough that snow does not settle on them evenly, which is why in late light the mountain looks striped rather than white.

The mountain has three widely quoted heights. The county and every Chinese tourism page use 5,820 m. The 2006 first-ascent report measured 5,833 m. A few older sources say 5,884 m. None of them has been re-surveyed against modern satellite data as far as we can tell, so treat the number as a name rather than a measurement.

The mountain also sits on an old road. The county describes it as the first snow mountain crossed by the Cha-ma trade route — the mule caravans carrying tea west from Ya’an into Tibet — and the pass on its southern shoulder still has the ruins of blockhouses built in the 1930s, when the road was a military problem rather than a trade one.

Before any of that, this was Muya country. The Muya 木雅 are the Tibetan-speaking people of the grassland between Kangding and Daofu, with their own dress, their own dialect and their own claim on the mountain. The monasteries on the plain below — Tagong, and the golden stupa beside it — are Muya institutions.

Morning is the better bet. The usual pattern is a clear early morning, cloud building on the summit by late morning, and the mountain gone by mid-afternoon. This is what local drivers and the photographs both suggest; it is a tendency, not a schedule.

The dry months are the better bet. Late autumn and early spring, when the air is clear and the snow is still on the faces. Midsummer gives green grassland and a mountain that may not appear at all.

Two places to look from. The grassland around Tagong and Lhagang, where the mountain rises behind the monastery roofs; and the Bamei side, where the northern valleys lead up towards the face. From the first you see a pyramid over a plain. From the second you see a wall.

The sequence is the one most of Kham went through. A township was set up here at the end of the Republic under the name Huiyuan; renamed Bamei in 1957; a commune in 1960; transferred to Daofu in 1978; a township again in 1984; a town in 1992. What it lost in 1978 was the county. Qianning had governed from here, and the buildings are still here — a grid of streets slightly too wide for the traffic, a theatre, a walled compound.

Today the town covers 281.3 square kilometres at an average of 3,502 metres, with one community and eight villages, and 5,398 registered residents at the end of 2019. Two highways meet in it, the G350 and the S303, which is why it has hotels, a hospital and a bus station at all. It calls itself the heart of Kham and the gateway to the eight northern counties; both are chamber-of-commerce phrases, and both are roughly true.

Its own list of eight sights is worth having, because it explains what else is here: the Yala snow mountain, the stone forest, Huiyuan monastery, the Longdeng grassland, a spring called the Dragon Pearl, the birth house of the eleventh Dalai Lama, Sangka monastery, and the village of Lhari. Three of them are within twenty minutes of each other; in 2019 the town was named a provincial cultural-tourism town, largely on the strength of the second.

What the park sits on. The forest occupies a slope above the Qingda, a tributary of the Yalong, and it is not a plateau of rock: it is a band. The mylonite runs along the fault for a hundred kilometres in either direction, but stands up as pinnacles only here, along a couple of kilometres of hillside — which is why there is a gate at this particular bend.

The Xianshuihe runs about 370 kilometres from north of Ganzi down past Luhuo, Daofu and Qianning to Kangding, trending north-west to south-east, and it moves left-laterally: stand on one side and the far side slides to your left. Cross-fault surveying in the 1970s and 80s measured up to six millimetres a year at the fastest stretch, near Xuxu and Xialatuo. It is the most seismically active fault inside the Chinese mainland, and the comparison everyone reaches for is the San Andreas — usually with the note that this one releases more.

The stretch that matters here is the Qianning segment, about sixty kilometres long, running between Daofu and the Huiyuan basin. It has produced five earthquakes above M6.5 on the record. The largest was on 29 August 1893, at Bamei itself: M7.0 in the current catalogue, intensity X at the epicentre, with the felt-and-damaged zone drawn out forty kilometres along the fault. Huiyuan monastery, ten kilometres to the north, was destroyed; 237 people were killed. The Sichuan earthquake compilation revised this event upward at one point, from 6¾ to 7½, which tells you how much of nineteenth-century Kham seismology is reconstruction rather than record.

That segment is now described by seismologists as a seismic gap: it has not broken in a hundred and thirty years, and the strain accumulated is reckoned to be enough for a magnitude 6.8. Satellite radar puts shallow creep at five to eight millimetres a year. Near Longdeng, the fault has offset a river terrace by 106 metres, give or take five, and left scarps up to 8.75 metres high. You drive across all of this at speed and notice nothing.

What the fault makes. In the zone itself the rock is not melted and not shattered into rubble; it is ground fine and recrystallised, so that it comes out as a hard, banded, flinty stone with the minerals smeared into parallel streaks. That is mylonite. Where it is exposed it looks like slate that has been through a mill.

How it becomes a forest. Three things in sequence. The fault grinds the mylonite and leaves it poorly cemented. The plateau rises, and the fault zone is lifted into the weather. Water, frost and gravity then work on the difference in hardness between the mylonite bands and their walls, and cut the softer, more shattered parts into gullies. What is left standing are the blades.

The dates people give. The forest was reported in 1981 and the weathering origin was worked out in 1994. Before that it was a local oddity that herders took for a coal seam — the rock is dark, layered and soft enough to dig, and it does not burn. The name has moved with the understanding: 八美土石林 on the old maps, 构造石林 or 变质岩石林 in the geological literature, 糜棱岩石林 when you are being precise, and 墨石公园 since the gate went up.

The rock’s own colour, the one it had when it formed, is a dark green-grey. What changes on the surface is a salt: the mylonite here carries calcium, and the calcium salts hold water of crystallisation that comes and goes with the humidity. Dry, the surface reads pale grey and faintly blue. Damp, it darkens to near-black, and in the wet months it goes black with a green cast.

That is why the same photograph can look fake. In a dry April the blades are the colour of a kerbstone; after a night of rain in July they are the colour of wet slate, and against green grass they look like something imported from another planet. Cloud cover matters as much as moisture: a flat white sky removes the shadows that give the blades their edges, and the forest goes soft and grey. Every guidebook photograph you have seen was taken in the twenty minutes after rain, in raking light, on a dark day.

The stream. Below the main group of pinnacles the water runs dark, and the county calls it the ink stream, 墨溪. Standing above it you would swear the water itself was black. Do not repeat the explanation that it is dissolved ink-stone — the rock is not a pigment, and the colour is in the bed and in what the water is carrying. It photographs beautifully and it is not a phenomenon.

Two things people tell you. One is that the forest is the petrifaction of five guards in five different-coloured robes — a local story, recorded as a story, which usefully explains why nobody agrees on the colour. The other is that the park was the birthplace of King Gesar, and that the cluster of blades at the entrance replays his soldiers holding their swords up to shield him. The first is folklore, the second is tourism copy. Both are told here as they are told there, and neither is geology.

A young mylonite outcrop weathers into gullies and troughs: the soft bands go first and cut channels, and what is left is a ridged, corrugated surface — the park calls this 槽状, trough-shaped. With time the ridges between the gullies are eaten back until they stand as columns; the columns lose their tops and become pyramids. At the end, the pyramid is reduced to a thin blade with a sharp edge, and then to nothing. The whole sequence is one process at different points, which is why a single hillside can show all of it at once.

What stands where. At the entrance, a hundred metres south of the gate, is the group called Ten Thousand Swords 万剑穿空: sharp-edged blades ten to twenty metres high. Further in, the Ink Painting group 水墨石林 rises thirty to fifty metres and reads, from a distance, as a brushed landscape. The Stone Wall 石长城 is the opposite — five or six metres high, twenty long, and about a metre thick at its thinnest, an immature blade that never got tall. The Buddha Stupa 八美佛塔 is a cone thirty metres high, hollowed by runoff into something that resembles a chorten; the Bell 八美神钟 is five or six metres of inverted bell on a base less than a metre thick.

A caution about numbers. You will see claims for the height of the tallest pinnacle. None of them is a survey, and several contradict each other. What the park does publish, and what is given here, are heights for named features — which is the only figure worth quoting.

Why it matters that it is soft. Mylonite is not granite. It is poorly cemented, it crumbles at the edges, and a boot on a blade takes a year off it. The handrails are not decoration, and staying on the boardwalk is not a suggestion about tidiness.

The usual direction. You enter at the main gate, take the shuttle up to the top of the path, and walk down through the forest to a second gate, where the shuttle picks you up again. That is the arrangement the park recommends, and it is the right one: the top of the slope gives you the whole group as a landscape before you are inside it, and the walk down puts the light behind you in the afternoon. Walking up from the bottom is possible and much quieter.

What is on the way. The geology museum sits in the middle of the route, with an indoor display on the Quaternary uplift, the fault and the formation of mylonite, and an outdoor yard of specimens. At the bottom, near the entrance, a spring comes out of a crack in the rock and is called the ink-stone spring; people drink from it. The park has put a name on almost every cluster of blades, and the signs are worth reading once — they tell you what the Chinese visitor next to you is looking at.

The season to come. June to October is green: the grass is up, the wildflowers are out, and the contrast that makes the place famous is at its strongest. Winter is the other version — the blades under snow, blue-grey rather than black, with nobody there. Both are worth a half day, and they are not the same place.

The basin has changed its name more often than its shape. In the old Tibetan geography it was Gada 噶达. The Qing administration called it Taining 泰宁 and built a walled town here in 1724, at the order of Nian Gengyao, then governor-general of Sichuan and Shaanxi. From the 1940s the county that administered it was Qianning 乾宁; that county was abolished in 1978 and the valley passed to Daofu. In 2020 the township was upgraded and renamed Taining again — the old Qing name, back on the signboard after a hundred and fifty years.

Of the walled town, one stretch of rammed-earth rampart survives on the south side, thick enough to walk on. The plan was not square: it was drawn in the shape of the sole of a Qing official’s boot, wide at one end and narrow at the other, on the theory that it sat under the emperor’s feet. Inside there was a Guandi temple for the Shaanxi merchants who ran the caravan trade, and a street they still call the Old Shaanxi Street. In 1936 the Red Army passed through and set up a short-lived Tibetan soviet here, which is why the town appears in two entirely separate histories.

Today Taining is six villages and about 340 square kilometres, ninety kilometres from the county seat and eight from Bamei, with the S303 running straight through it. The fields grow spring rape — eight thousand mu of it, turned into cooking oil and, since 2019, into a photographable yellow plain in July under a 4A scenic-area label. The monastery is the reason the label exists.

The decision was to move him out. Kham was the obvious direction — close enough to watch, far enough from Lhasa — and Gada was chosen for reasons the county states plainly: the climate resembled Litang, where the boy had been born, and Lhasa, where he would eventually return; there was forest and pasture above the valley and irrigated fields in it; and the site sat on the Sichuan-to-Tibet road. It was a placement, not a foundation. Nobody asked whether a monastery was needed here. There was not one.

The bill was imperial. Four hundred thousand taels of silver were allocated, over five hundred mu of land was taken, and the works ran to more than a thousand rooms of hall and storey plus five hundred single-storey buildings, with a wall around the lot. A garrison of eighteen hundred men was posted in three camps — at Taining itself, at Daowu, and at Zhongdu on the road south — to hold a building that had not existed two years earlier.

The name came from the throne. The Yongzheng emperor fixed it as 惠远 and wrote the plaque himself; the county gives the sense as “blessing that endures far”. The Tibetan name went the other way, from the ground up: Gada Jampa Ling, the Maitreya enclosure at Gada. Two names, two authors, one building.

The name was not unique. In 1763 the court gave it again to the garrison town at Ili, in what is now Xinjiang, where 惠远 is still on the map. Frontier names came in sets in that century, and the set names the policy: not conquest, but favour extended to people at a distance.

In 1733, with earthquakes shaking the valley, the emperor ordered a separate residence built to the west of the main hall so that the Dalai Lama would not be sleeping inside a hall of assembly. It is called the Dougang lou 都纲楼 — the residence, in the court’s vocabulary. The year after that the Dzungars sued for peace, the strategic reason for holding him evaporated, and in 1735 he went back to Lhasa, escorted by the emperor’s brother, Prince Guo 果亲王, who had come to Taining to settle the arrangements.

The prince left two things behind. One is a stone inscription, dated the second month of 1735, confirming that some seventy of the Dalai Lama’s people would stay on in the monastery to keep it chanting — the first document of its life as a working institution. The other is a long poem, the Seven Hook 七笔勾, a courtier’s set of verses about the road; the stone it was cut on has never been found.

About the seven years. Construction began in 1728 or 1729, depending on which document you follow; the Dalai Lama was installed in 1730 and left in 1735. That is five years in residence. The seven-year figure is old and consistent, and it probably counts from the founding rather than from his arrival. The sort of small thing that gets copied from one account to the next.

The earthquakes were not bad luck. The valley sits close to the Xianshuihe fault, and the monastery was wrecked three times — in 1748, in 1793 and again in 1893 — and rebuilt after each. The last of those rebuilds, at the end of the Qing, is the one you are looking at, and it is smaller than the original: the axis survives, the scale does not.

Inside the hall. Three bronze Buddhas on the altar, Shakyamuni in the middle. On the columns of the upper storey there is a painted narrative of how the monastery came to be built, with the figures dressed as characters from Chinese opera — a political history told in the visual language of the theatre. The hall faces east — not south, as a Chinese imperial building would, and not the direction the grid of the compound implies.

Five stones out of six. The imperial inscription of 1731 in Chinese and Mongolian; the prince’s decree of 1735; a local magistrate’s account of 1865. During the Cultural Revolution the steles were pulled down and used as flooring in a pigsty; five were dug out afterwards and re-erected. The sixth, the Seven Hook poem, is still missing.

Khedrub Gyatso was three when the lottery was held, in 1841. The golden urn had been instituted by the Qianlong emperor in 1793 precisely to stop great families from arranging the outcome, and this was its first use for a Dalai Lama. He was taken to Lhasa, enthroned the following year, and died in 1855 at eighteen, months after taking power — the shortest reigns in the line are all in that century.

His birthplace is now a small memorial hall in the village, and the story attached to it is the kind that Chinese travel writing loves: that when the family house was rebuilt, a stone was dug out of the foundations with the six-syllable mantra already formed on it, clearly and naturally, and that nobody has explained it. The stone is in the hall. You can look at it and decide for yourself.

What the urn was for. Before 1793 a reincarnation was settled between the monastery, the family and the Lhasa nobility, and the outcome was predictable. The Qianlong emperor’s regulation had the candidates’ names written on slips, sealed in a gold urn and one drawn before the image in the Jokhang — with imperial approval required either way. It did not take politics out of the process. It moved the last word to Peking, which was the intention.

Nine dragons and nine lions. The monastery carries the title jiulong jiushi 九龙九狮, and the carved door frame is supposed to hold the nine of each. Nine is the extreme yang number; the dragon stands for the emperor, the lion for the leadership of Tibetan Buddhism — some versions say for the Lhasa government instead. Either way the message is the same, and it is not a religious one: this house has the protection of two powers, and holds rank accordingly.

Why the reputation travelled. One Dalai Lama sheltered here, another was born down the road. That is the whole of it. A monastery can have a hundred years of teaching and no name outside its own valley, and this one has a name across the plateau because of two accidents of eighteenth- and nineteenth-century politics.

The town is essentially one street, and the street exists because of the courtyard at the end of it. Shops, guesthouses, a bus stop, a row of places selling yak yoghurt — all of it is arranged around a monastery that was here first. Behind the monastery the hill rises, and the hillside above the monks’ houses is covered in prayer flags, white and coloured, in lines that run up to the ridge.

The statistics are worth knowing because they explain the atmosphere. Nine monasteries and one monastic college stand inside the township boundary, with more than 1,200 monks and nuns between them. In a population of 9,400, that is roughly one person in eight. This is not a village that has a monastery; it is a place whose economy, calendar and skyline are all religious.

The livestock figures are equally unsentimental. About 50,500 head — yaks, sheep, horses — were counted at the end of 2021. The grassland is worked: summer grazing up the valleys, hay in autumn, milk coming into the village dairies to be turned into butter and yoghurt. When you photograph yaks on the plain, you are photographing someone’s herd.

There are two Tagong seasons. From June to September the street serves herders, pilgrims and visitors at once. From November the coaches thin out and what is left is a working pastoral town at 3,750 metres. Both are the real place; only one is on the internet.

And the name. A second story explains Tagong rather than the monastery. Songtsen Gampo is said to have ordered 108 temples with their doors facing China; Tagong was the hundred-and-eighth and the last completed. Generations later an Indian master re-consecrated it, the consecration went well, and voices were heard in the sky saying tagong, tagong — “the bodhisattva is pleased”. The name stuck to the place. It sits oddly beside the Tibetan Lhagang, which simply means the upland of the gods, and both are in use.

What is actually datable. The buildings you walk through are Ming and Qing. When the Sichuan government listed the monastery on 16 September 1996 — fourth batch of provincial heritage sites, number 30, classed as historic buildings — the period given was Ming–Qing, not Tang. So the honest sentence is: a seventh-century foundation, in Ming and Qing buildings, on a site people have used continuously for a very long time.

Why the hybrid. This edge of Kham is close enough to the Chinese lowlands that carpenters, tiles and pigments came up the road with the tea. The result is a genuine hybrid rather than a pastiche: Chinese roof structure carried on Tibetan load-bearing walls, with the joins painted over so thoroughly that you have to look for them.

The Jowo chapel. The image from the legend sits here, and it is covered in the offerings of centuries — gold, silver, turquoise, coral, coins, silk scarves — until almost nothing of the original surface shows. That accumulation is the object now — and the reason photography inside is restricted.

The assembly hall. Shakyamuni on the central throne, with Sakya Pandita and Sachen Kunga Nyingpo (1092–1158), the founding figures of the Sakya school, flanking. Seating for several hundred monks, and butter lamps that are lit by the bucket.

How many. Most Chinese sources say 108, and some say “a hundred or more”. Nobody publishes a count made by walking the row. 108 is the number of beads on a mala and the number of afflictions in Buddhist lists, so the figure is both a measurement and a statement. Treat it the way you would treat “a thousand yaks”.

The four colours. Local account gives the stupas at the four corners of the group colours — white in the east, yellow in the south, red in the west, green in the north. The colours map onto the four enlightened activities. This is how it is explained on site; we pass it on as an explanation, not as doctrine.

How to walk it. Clockwise, keeping the wall on your right. Pilgrims do the row and then the long corridor of prayer wheels on the west side of the compound, and the two together are the standard circuit. Early morning and late afternoon are when it is being used rather than photographed.

The summer on the grass. From mid-July into early August the grassland fills with tents for shua bazi 耍坝子 — families camping out, cooking, drinking, racing horses, dancing in the evenings. It is a holiday, not a performance, and it is the one time of year when the plain in front of the monastery is as busy as the monastery itself.

How the dates are given. Chinese sources disagree on whether these follow the Tibetan or the Chinese lunar calendar; the two are close but not identical, and both drift against the Gregorian year. We give the sequence rather than a false precision — ask in the town for the week. For the masked dance, ask at the gate a few days ahead.

Two names, two different things. 各日马 is the unit that appears in the gazetteer, with a boundary, a committee and a budget. 甲谷弄 is a gully, and the houses in it — a cluster that has never had an administrative existence. If you search for the second name on a Chinese map app you will probably get nothing, or you will get a village of the same name in another county. The monastery entrance is what most people actually navigate to.

The red houses are not the monastery’s. They were built by the people who live in them, or by their families, on land the village allocated. Many belong to nuns and monks attached to the four institutions down the hill; a large number belong to lay practitioners who have taken vows but not ordination. The colour is not a rule — it is what everyone does. Up close the rows are uneven, improvised, patched with sheet metal and plastic, and each one has a small yard with a water barrel and, often, a cow.

It is worth saying plainly that this is not a secret. The gully has a milk station, a visitor centre and a collective income that went from 260,000 yuan in 2019 to 3.36 million in 2024, mostly on wedding photography, guesthouses and a car park. The village was listed among China’s traditional villages in 2022. It is beautiful and it is a product; both are true, and the second is why the road is good.

The prefecture’s own cultural office defines the territory carefully: 木雅热岗, the Minyag ridge-country, runs from the far side of Zheduo Pass west to Daofu, east to Yajiang, north to Jiulong. The Tibetans inside that polygon are 木雅娃. The monastery stands at its northern edge, which is why it carries the name and the monastery eight kilometres away at Tagong does not.

The old Tibetan word for them was 弭药, mi nyag, and that word is where the argument starts. It was applied to the Tangut, who built the Western Xia state in what is now Ningxia and were dispersed by the Mongols between 1205 and 1227. From the 1880s, Western travellers in this valley concluded that the Minyag must be Xia survivors who fled south — a conclusion Chinese scholars took up and never quite disproved. The prefecture’s own text says what should be said: the origin is still a knot, and the Xia theory is the common answer, not the established one.

The strongest evidence is linguistic and it is not romantic. Minyag people speak a language that is neither Tibetan nor Chinese, called rongke by its speakers, in two dialect areas: an eastern one in Jiulong, whose speakers call themselves 木勒, and a western one across Shade and Pengbuxi, whose speakers call themselves 木雅. Its structure has been described as close to Tangut — a claim about grammar, not a family tree. If you hear a language in the shop that is neither Mandarin nor the Tibetan you have learned to recognise, that is what it is.

And then there is the stone. In Minyag country, white stones are set on the four corners of houses, and at mountain passes a cairn of white stones stands where other Tibetans would build a mani pile: a heap put there by hand, added to by the next person. The wall in the next valley is the same gesture at a different scale.

They are spread out. The four sit along more than a kilometre of grass, connected by a vehicle track and very little else. This is the single most useful thing to know before you arrive: there is no gate you walk through into a courtyard, no single ticket, no obvious beginning. You drive to one, then to the next. The nunnery and its wall are at the far end, past the college.

The main hall is new. Gilded roof, red walls, painted beams, and a floor you take your shoes off to stand on. It is large enough for several hundred and on an ordinary afternoon it holds a dozen people, most of them elderly, walking clockwise with a hand on the rail. Photography inside is permitted, which is unusual, and is one of the reasons the place appears so often online.

The population is real. The village profile counts a thousand-plus monks and nuns in the group, including the nuns. On a weekday you will see almost none of them — they are in the college or in their houses — and the emptiness reads as abandonment when it is only a timetable.

Which school? The monastery is described as Nyingma, the old school, in most accounts and by the monks themselves when asked. A number of Chinese listicles label it Gelug, the yellow-hat school, which would be a different institution with a different curriculum and a different dress code. The two are not close. If it matters to you, ask; the answer may depend on which of the four institutions you are standing in.

The animals. Deer, argali, macaques, some of them released here after injury. They are used to people and this is exactly the problem: they will approach a car. Do not feed them, and do not let a child chase them for a photograph. There is a small herd of white-lipped deer that has learned where the car park is.

It stands in front of the nunnery, at the head of the gully, with the prayer-wheel circuit around its foot, a row of white stupas above it and the big wheels below. You will reach it before you understand what you are looking at: a long, low, grey mass the height of a person, walled off from the grass, with a path worn around it in the dust.

It is slate, split thin, laid dry. No mortar, no frame, no interior. Each slab is a page: the six syllables cut into the face, sometimes a figure, sometimes a whole line of scripture. The slabs are stacked flat, overlapping like roof tiles, and the flat ones on top are weighted with stones so the wind does not take them. The paint — white, blue, yellow, green, red — is renewed by whoever feels like doing it.

There are two walls, not one. They run parallel with a narrow gap between them, and the gap is not a passage: it is where the structure has been extended, wall against wall, as the pile grew. Most visitors walk the circuit without noticing. Once you have seen it you cannot unsee it.

The age is a puzzle. Two hundred years is the figure in every account. A monk at the main hall, asked the same question in 2019, said twenty. Both may be right: the practice of depositing stones here is old, the standing structure is recent, and the accumulation has been rebuilt more than once. What nobody disputes is that it grows.

The word matters. 觉姆 is not a translation of “nun”. It is a Tibetan honorific, historically used for a ruler’s mother and sisters, and in Kham it is the ordinary respectful term for a woman who has taken vows. 尼姑, the Chinese word for a Buddhist nun, is what outsiders say. Using the wrong one is not an insult so much as a small signal that you have not been paying attention.

There are more of them than there are monks — unusual at this size, and the reason the place has the shape it has: the nunnery is the old core and the monks’ institutions grew up around it. A large share of the residents are not ordained at all but lay practitioners who have taken vows and live here permanently. The distinction matters to them and is invisible to you.

The nunnery is closed to men. A monk at the main hall says so cheerfully, and adds that a man who goes in will be shown out. Women may be admitted with permission, but it is not a site with visiting hours — it is a house with several hundred residents who did not choose to be an exhibit. The wall in front of it is public. The hall is not.

The red houses belong to them. One slope of the gully is nuns’ houses, the other monks’: self-built, family-funded, one room and a yard, and finished according to what the family could afford rather than the occupant’s rank. Some keep a cow. Loudspeakers carry the teaching down the valley all day, at what is simply the ambient volume of the place.

Photography. The wall, the stupas, the wheels, the valley: yes. A nun’s face, without asking, no — and asking is difficult when the answer is usually no and the asking itself is the intrusion. If you want a person in the frame, wait for a moment when they are part of the scene rather than the subject.

Giving. Money handed to a person on the path creates a transaction where there was none, and the next visitor inherits it. If you want to contribute, buy a carved stone at the wall: the money goes to the carver.