Witness the Indus - 4,000 Years Forgotten, Still Unread

A girl is playing on a low hill in the sin at dusk. She is barefoot. The earth is the color of old brick. She bends down, picks something out of the dust, turns it in her fingers, looks at it for a moment, and puts it back. She does not know she is standing on a city. This is the story of the A hand presses a small square of carved soapstone into wet clay. The hand is dark, calloused, certain. When the seal lifts, the impression of a one-horned animal appears, surrounded by four signs in a script no one alive can read. The year is around 2500 BC. The hand belongs to a craftsman in a city of 40,000 people. A second hand sets a brick of identical proportions into the bed of a railway being built across the ponjo. The brick is fired clay, four and a half thousand years old. The hand belongs to a worker in 1856, laying ballast for the Karachi to Lahore line. He does not know what the brick is. Neither does the British engineer supervising him. A third hand holds a magnifying glass over two small seals on a mahogany table in Simla. The hand belongs to John Marshall, Director General of the Archaeological Survey of India, and the date is September 1924. He is comparing one seal from Harappa with one seal from Mohenjo-Daro, 600 kilometers apart. The signs are the same. The proportions are the same. Something very large has just become visible. What Marshall is seeing in that room is the largest civilization of the ancient world. Bigger than Sumer, bigger than Egypt, more than a thousand cities planned on grids, with sewers running beneath the streets, with bricks fired to the exact same proportions across a thousand kilometers of land, contemporary with the pyramids, trading with Mesopotamia in Lapis Lazuli and Carnelian, home to perhaps five million people, and it had vanished from human memory so completely that for four thousand years no one knew it had ever existed. How does a civilization that big disappear? Who were these people? And what does it mean that they left behind four thousand inscriptions in a script that, even now, more than a century after Marshall set down his lens, no one can read? This is the story of a thread that may never be picked up, and of the people who have spent their lives reaching for it anyway. A farmer is walking out to his fields at first light. He is barefoot. The ground is cool. As he passes the western edge of the lower town, his ankle brushes against a small ceramic pipe set into the base of a brick wall, and a thin run of cold water from the drains above sluices across his foot. He does not flinch. He has walked this way every morning of his adult life. Behind him, the city is still mostly asleep. Forty thousand people are sleeping in houses laid out in a grid as exact as graph paper. Every house has a well. Every street has a covered drain running beneath it. The drains run into larger drains, and the larger drains run out beyond the walls into the floodplain of the Indus. No king has decreed any of this. No palace stands above it. No temple looks down from any height. The city simply works. He climbs a low ridge and looks back at it. The first sun strikes the upper terrace of what the archaeologists, four and a half thousand years from now, will call the citadel. From the ridge, the streets resolve into long parallel lines, and the lines run true. He has been in three other cities in his life. Harappa, to the north. Dholavira, to the southeast. A smaller place whose name no one will recover. They are all built the same way. The bricks are all the same size. The wells are sunk to the same depth. He has never thought to ask why. He has no word in his language for the territory he lives in. He knows the cities by name. He knows the river. He does not know that the cities to his north and his south and his east, more than a thousand of them, are all part of a single thing. He does not know that his civilization covers one and a quarter million square kilometers. He does not know that he is one of perhaps five million people, and that there is nothing else on earth at this hour of comparable scale. He turns toward the field. The wheat is shoulder high and green. A man is standing on the upper rim of a great rectangular pool. It has just been finished. The bricks of the basin have been sealed with bitumen, layer by layer, so that no water will escape into the earth below. He is the supervisor of the work. The morning is hot. He bends slightly forward and touches the surface of the still water with his palm. The reflection of the citadel breaks, settles, reassembles. He stays bent for a moment longer than is needed. The pool is around 12 meters long and seven meters wide. Steps lead down at either end. Around its rim, a colonnade rises. Behind the colonnade, a complex of small rooms. No one knows exactly what it was for. The archaeologists will call it the Great Bath. They will not be able to say whether it was sacred or civic or both. There is nothing else like it in any known city of its time. He looks out across the city. He has been a builder all his life. He has worked on house walls, on platforms, on drains, on the great granary at the northern end of the citadel. He has worked under no king. He has worked under no priest he can identify as such. There are men who direct the work. There is no city of this size without them. But there are no thrones, no royal tombs, no inscriptions praising any one man's reign. He cannot name his own ruler. He is not sure his own city has one. He turns and looks at the figures the artisans have been making in the lower town. There is a small bronze of a young woman, hand on her hip, body slightly turned, an attitude of complete unselfconsciousness. There is a small stone bust of a bearded man, eyes half-closed, a cloak trimmed with three-petaled motifs over his shoulder. There is a square soapstone seal showing a figure seated in a yogic posture surrounded by animals. The archaeologists will name these the Dancing Girl, the Priest-King, the Pashupati Seal. He does not know what they are called. He has seen them in workshops. He does not know who they represent. It may be that no one does. He straightens up. The reflection in the pool is steady again. A man is standing on the docks at Akkad. The afternoon sun is heavy on his shoulders. A merchant ship has just come up the river from the Persian Gulf, and the porters are unloading sacks of lapis lazuli, blocks of carnelian polished to a deep red, bundles of timber, and small carved seals of soapstone. The man's name is Shu'ilishu. Around his neck, on a leather cord, hangs a small cylinder seal of dark serpentine, with his own name and his profession cut into it in Akkadian cuneiform. The inscription says, Shu'ilishu interpreter of the Meluhan language. He is the only person on the docks who can speak to both sides. The merchant who came in with the ship is gesturing now, a stream of language that no one else at the harbor understands. The harbormaster in his Akkadian wool robe is waiting. Shu'ilishu listens to one, turns, speaks to the other. The cargo is named, the price agreed, the tax recorded. The merchant smiles briefly. Shu'ilishu touches the seal at his throat once, as if to check that it is still there. The Akkadians have a word for the land this ship has come from. They call it Meluha. They have been calling it that for the two centuries already, since the time of Sargon, who boasted on a clay tablet that ships from Meluha tied up at his key. The scholars who will study these tablets 4,000 years from now will argue about where Meluha was. Most will conclude that it was the Indus. Some will note that the word in a proto-Dravidian reconstruction may mean something like high country, and that it may have been what the Indus people called themselves. What matters on this afternoon in around 2020 BC is that Shu'ilishu is doing his job. He reads cuneiform. He almost certainly reads the script on the seals coming off the ship. The one-horned animals, the four signs, the small square impressions in the clay attached to bales of cloth. He is, as far as anyone now knows, the only named human being in history who could read the writing of the Indus. He does not leave a translation. He leaves only his seal, which will be found in the rubble of Akkad more than 4,000 years from now, and placed in a glass case in the Louvre. A man is sitting at a low bench in a workshop at the back of his house in Mohenjo-daro. The light through the small window is steady. On the bench in front of him, a square of white soapstone, three centimeters on a side, freshly polished. In his right hand, a thin copper graver. In his left, the seal. He has been cutting since the early morning. The animal, the bull with a single curved horn, head bent toward a small ritual stand, is almost finished. Above the animal's back, four signs. He cut the signs first, before the figure, working from right to left. He has been doing this work for 20 years. He knows the signs by touch now. The fish, the jar, the double stroke, the figure with raised arms. He sets down the graver. He picks up a small strip of fresh clay from the dish at his elbow, lays it flat on the bench, and presses the seal into it firmly with the heel of his hand. The soapstone resists. He lifts. The impression appears in reverse. The bull facing the other direction now. The four signs raised in relief. He examines it. The third sign is slightly shallow. He will deepen it in the afternoon. What he does not know, what he could not know, is that this gesture, the seal pressed into clay, is the gesture that will preserve almost everything we will ever have of his civilization's writing. Around 4,000 inscriptions will survive into the 21st century. Most of them will be on seals like this one, or on small clay impressions taken from seals like this one. A few will be on copper tablets, on pot sherds, on the walls of pots. The longest single inscription anyone will find will have 26 signs. The average will have five. There will be perhaps 400 distinct signs in the corpus, perhaps as many as 700, depending on how the scholars count. No bilingual inscription will ever be found. No long text will ever be found. No name of any king or god or city in any other ancient language will ever be matched against any sign in this one. The underlying language will not be known. There will not even be agreement 4,500 years from now on whether this is writing at all. He lifts the small clay impression to the light. The bull is sharp. The signs are clean. He sets it aside to dry. A woman is closing the wooden door of her house for the last time. The hinges drag against the frame. The wood has swollen with damp and then dried unevenly, and the door no longer hangs true. She has lived in this house for 30 years. Her mother lived in it before her. Her grandmother built the upper room. In the street behind her, three sacks of grain are tied and waiting. Her sons have already gone ahead with the cattle. She is the last of the family. The street is quiet. It should not be. Two years ago, at this hour of the morning, she would have heard water running in the drains under her feet, neighbors calling across the courtyards, the slow creak of the cartwheels coming up from the river. Now the drains have silted up. No one has cleared them in months. The neighbors have gone. The cartwheels do not come. She does not understand all of why. The summer rains have been thinner for 10 years. The river, her grandmother knew, has a wide brown current, the one the later world will call Gagarhakra, and which the authors of the Vedic hymns will remember dimly as Sarasvati, has narrowed to a thread, then to a trickle, then to a dry bed where dust runs in the wind. The fields outside the city have turned white at the edges with salt. The harvests have come in smaller every year. The ships that used to come from the west have stopped coming. She lifts the bar of the door and sets it down inside, on the floor, where the next traveler may find it. She steps back into the street. There has been no battle here. No army has come. No fire. No invasion. The city is simply emptying, family by family, road by road. What is happening will later be argued about in many languages. The model of a chariot-driven Aryan invasion proposed in the 19th century will, in time, be set aside. The bones in the city will show no slaughter. What there will be instead is this, a woman closing a door and walking east. She does not look back. Hold the image of the woman in the empty street. In the centuries after she walks east, the cities she leaves behind do not fall. They simply hollow out. The upper buildings collapse first. The drains silt fully. The wells fill with dust. Wind moves over the mounds and rounds them. In 200 years, the outlines of the streets are gone from the surface. In 500 years, the mounds look like natural hills. In a thousand years, the people living within sight of them have no name for what they are. Speakers of an Indo-European language move into the region a few centuries after she has gone. They find the ruins already weathered, already nameless. The hymns they compose, the hymns that will become the Rig Veda. Do not remember the cities. The Buddhist chroniclers, a thousand years later, do not know. The Mauryan emperors do not know. The Greeks who came with Alexander do not know. The Arab conquerors who took the Sind in 712 do not know. The Mughals who ruled the subcontinent for three centuries do not know. Every empire that will rise on this land will rise without memory of the empire that was here first. For four thousand years, the largest civilization of the ancient world will be invisible to history. And then, on a dry afternoon in 1856, two British engineers in the Punjab will run out of ballast for a railway. John Brunton is on horseback at the edge of a dry plain north of Multan. It is the afternoon of a hot day in early 1856. In front of him, a low mound, perhaps 400 meters long, made of fired brick. Local workers with picks are already at the base of it, and empty wagons stand behind them on the new line. Brunton is the chief engineer of the Sindhi railway. He has 93 miles of track to lay between Karachi and the rest of the Punjab, and he needs ballast, a great deal of ballast, to set the rails on. He dismounts. He picks a brick out of the loose pile his workmen have brought down. He turns it in his hand. It is the size of a small loaf. Its proportions are exact. One part thick, two parts wide, four parts long. He has not seen a brick fired this well in any kiln in British India. He is not sure where it came from. The villagers of the nearby town, a place they call Harappa, have been quarrying these mounds for their own walls for as long as anyone alive can remember. There are, his foreman tells him, more mounds like this farther south. One is large enough to bury a city in. Brunton makes a decision that is not to him a difficult one. He has rails to lay. The mound has bricks. He gives the order. Over the months that follow, his men pull the upper city of Harappa apart for ballast. The bricks of houses that stood for 1,500 years and were abandoned 3,500 years before that are loaded into wagons, hauled south, and packed beneath the wooden sleepers of the Karachi to Lahore line. Roughly 90 miles of railway will sit on Harappan brick. The locomotives will pass over the unread inscriptions of a civilization whose name is not yet known in any language, and the inscriptions will hold the weight. Brunton's men recover in passing some small square objects with carvings on them, animals, strange marks. A few are sent to the Lahore Museum. A few find their way to London. In 1875, the head of the Archaeological Survey of India, Alexander Cunningham, will examine one of these seals at his desk in Calcutta. He will note that the script does not match any known Indian writing. He will conclude that the seal is foreign to India. He will set it aside. The seals will sit in drawers for almost 50 years. A man is kneeling at the edge of a trench on a December morning in 1922. The wind off the Larkana Plateau is cutting. It gets through wool and into the bone. His name is Rakhal Das Banerjee. He is 37 years old, Bengali, an officer of the Archaeological Survey of India, and he has come to the sin to excavate what he believes is a Kushan period Buddhist stupa from the 2nd century A.D. He is, by trade, careful. He is, on this morning, mostly cold. One of the workmen pauses, bends, picks something out of the dust at the base of the stupa, and holds it up. Banerjee takes it. He turns it in his hand. It is a small piece of worked flint, a scraper. The edge has been retouched. He has seen flint tools in many places. He has not seen one come up from beneath a Kushan stupa. He sets it on a piece of cloth at his side. He does not say anything yet. A few minutes later, another comes up, then a small carved seal of soapstone with an animal on it, and signs above the animal that he cannot read. He has not been the first to find such a seal. A year earlier, in 1921, Daya Ramsani, another Indian officer of the ASY, has begun excavations at a place called Harappa, 600 kilometers to the north in the Punjab. Sani has found seals like this one, animals like these. The reports between the two excavations have begun slowly to pass back and forth through the headquarters in Simla. In April of 1924, Madho Sarup Vats, a junior officer, will write to the director general of the survey, suggesting that the two sites are somehow related. But that letter is still more than a year away. On this morning, Banerjee only knows that he has come for a stupa, and that the stupa is sitting on top of something older. He does not yet know how much older. He brushes the dust from the flint scraper. He looks up at the mound rising behind the trench, the mound that the locals call Mohenjo Daro, the mound of the dead, and thinks about what he will write in his report. The report, when he writes it in 1926, will not be published in his lifetime. The discovery, when it is finally announced, will be announced by someone else. For now, he closes his hand around the flint, and the wind keeps coming off the plateau. Sir John Hubert Marshall is sitting at a mahogany table in the summer residence of the Government of India at Simla. It is a high-ceilinged room with whitewashed walls and a tall window that opens onto the foothills of the Himalaya. The afternoon light has the particular thin quality of a Himalayan September. He is 48 years old. His beard has grown gray. He has been Director General of the Archaeological Survey of India for 22 years, and he is a man not easily astonished. On the table in front of him, six small square seals of white soapstone. Three came in by post a week ago from Daya Ramsani, who has been working at Harappa. Three came up by rail from Rakul Das Banerjee, who has been working at Mohenjo-Daro. The two sites lie 600 kilometers apart. Marshall has set the seals in two parallel rows. On the table beside them lies a map of British India, with the two excavation sites marked in red ink. He has a magnifying glass in his right hand. He presses the lens to his eye until it leaves a small ring on the skin around the socket. He examines the first seal from Harappa. He sets it down. He examines the first seal from Mohenjo-Daro. He sets it down. He examines the second seal from Harappa. He pauses. The signs are the same. Not similar. The same. The fish sign with the diagonal stroke through it. The jar sign. The figure with raised arms. The same characters cut in the same order. He picks up a brick fragment that has come in with the Mohenjo-Daro material. The proportions. One part thick, two parts wide, four parts long. He picks up a brick fragment from the Harappa shipment. The proportions are identical. He sets them side by side. He sits back. He puts the lens down on the table. He looks at the map. What he is realizing in this room, in this hour, is not that two ancient sites are related. He has known that for some months. What he is realizing is that what lies between them, the entire plain of the Indus with its hundreds of mounds noted by surveyors and never excavated, is a single civilization. A civilization with standardized weights, a standardized script, a planned urbanism repeated at the scale of a continent. A civilization not mentioned by name in any document of any literate culture he knows. A civilization that has been entirely outside the history of the world as he was taught it. He picks up his pen. He uncaps the ink. He begins to write the article that will appear in the Illustrated London News on the 20th of September, 1924. He writes the title across the top of the page. First light on a long forgotten civilization. He underlines it once. He sets down the pen and looks at the row of seals again. He is not exhilarated. He is sober. He understands more clearly than perhaps anyone else in the world at that moment what the next century of his discipline will have to do. He picks up the pen and continues. Up to the present, he writes, no other civilization, even remotely contemporary, has been found which can rival it in extent or in the splendor of its achievement. He sets the pen down. He looks out the window. The light is going. Father Henry Harris is sitting at a desk in a Jesuit college in Bombay on a hot night in 1941. He is 53 years old. He wears a black cassock. The single electric bulb above him is dim and he has set up two oil lamps for additional light. Across the desk, a wide spread of glossy photographic plates of seals from Mohenjo-Daro, ordered and reordered into long rows. He has been working on this for seven years. He believes, and he has begun to publish in volumes that will be largely ignored in his lifetime and largely set aside after his death, that the language of the seals is an early form of Dravidian, the family of languages still spoken across the southern half of the subcontinent. He has assigned tentative readings to many of the signs. The fish sign, he proposes, is mean. The Dravidian word for fish, which is also the word for star. The jar sign, he proposes, stands for a particular ritual offering. Most of his readings will not survive scrutiny, but the underlying hypothesis that the language was Dravidian will turn out to be the position that the majority of serious scholars over the next 80 years will, with refinements and qualifications, come to share. He will not live to see this. It is 1941. In six years, the subcontinent will be partitioned. Harappa and Mohenjo-Daro, the two great cities of the civilization Marshall announced in 1924, will fall on the Pakistani side of the new border. Rakigarhi, Dolavira, Lothal, and most of the other excavated sites will fall on the Indian side. The archaeologists who had worked across the same plain will become citizens of two different nations and will work from then on in two largely separate institutional traditions. In 1953, Sir Mortimer Wheeler, the British archaeologist who succeeds Marshall as the senior figure in the field, will publish a synthesis of the evidence and propose that the cities fell to an invasion by Indo-European speakers, what he will call, in a phrase that will haunt the discipline for generations, an Aryan invasion. The phrase will be repeated in textbooks for decades. It will be slowly dismantled by the next generation of archaeologists who will find no evidence of slaughter in the bones, no destruction layers consistent with conquest, no chariots in the ruins. The cities, they will conclude, were not conquered. They emptied. Harris turns over another plate. The lamp flickers. He makes a small note in the margin of his notebook. Outside the open window, the city of Bombay goes on with its evening. Asko Parpola is sitting in a winter armchair in his study in Helsinki. It is a morning in the present. He is 84 years old. The window beside him looks out onto snow. He has not turned on the overhead light because the light off the snow is enough. On the small table at his elbow is a cup of coffee that has gone cool. On his lap, a large photographic enlargement of a single Indus seal. He has been looking at seals like this one since 1964. He came to the work as a young man, just out of his doctoral studies in Indology at the University of Helsinki, and he came to it in the middle of a peculiar moment in the history of the discipline. In the early 1960s, a Soviet team in led by Yuri Knorozov, the same Knorozov who had broken open the Maya script a decade earlier, had begun applying computational methods to the Indus signs. A Finnish team with Parpola among them had begun the same work in parallel. The two teams competed for years. Neither read the script. Both produced catalogs and statistical analyses every later researcher would have to use. In 1977, in Madras, a Tamil epigrapher named Iravatha Mahadevan published the Indus script, texts, concordance, and tables. It became the standard reference for the corpus. Mahadevan, like Parpola, argued throughout his life that the underlying language was an early Dravidian. He worked on it until his death in 2018. He did not read it. In 1994, Parpola published Deciphering the Indus Script with Cambridge University Press. It is, to date, the most systematic argument that has ever been made for a Dravidian reading. It does not, despite its title, decipher the script. Parpola has been clear about this in interviews for 30 years. The book sets out the state of the question. It does not close it. He turns the photograph in his hands. The seal shows a one-horned animal, what the archaeologists call a unicorn, though no one knows what the animal was meant to represent. Above its back, four signs. He has read these particular four signs more times than he can count. He picks up a pencil. He makes a small note in the margin. He sets the photograph aside. He picks up the next one. Outside the window, the snow continues to fall on Helsinki. Somewhere, in another country, a man in the third millennium B.C. once knew exactly what these four signs said. His name was Xu Ili Xu. He left no translation. A man is sitting in front of a computer screen in a quiet office somewhere in California. It is late at night in 2004. The cursor is blinking at the end of a manuscript he has just finished editing. His name is Steve Farmer. He is a comparativist working independently. Two co-authors are listed at the top of the paper, Richard Sprott, a computational linguist, and Michael Witzel, the Wales professor of Sanskrit at Harvard. The title of the paper reads, The Collapse of the Indus Script Thesis, The Myth of a Literate Harappan Civilization. He moves the cursor to the send button. He pauses for a moment. He clicks. The paper, when it appears in the Electronic Journal of Vedic Studies, will detonate quietly in a small field. Farmer, Sprott, and Witzel argue with statistical evidence that the Indus inscriptions are too short, the average inscription too small, the patterns of sign repetition too unusual for the corpus to represent writing in the linguistic sense. There is no syntax, they say. There is no grammar. There is no text. What look like signs are political and religious symbols, closer to coats of arms than to written language. The civilization on the reading was non-literate. Parpola rejects it. Mahadevan rejects it. The mainstream of the field rejects it. But the paper plants a doubt that does not go away. Five years later, in May of 2009, a different team publishes a different answer. The lead author is Rajesh Rao, a computer scientist at the University of Washington. The co-authors include Iravatham Mahadevan. The paper appears in Science. Its title is Entropic Evidence for Linguistic Structure in the Indus Script. Rao and his colleagues run a statistical analysis on the sign sequences. They compute the conditional entropy of the corpus. They compare it to known writing systems, Sumerian, Old Tamil, Sanskrit, English, and to known non-linguistic sign systems, like DNA sequences, like the symbolic vocabularies of Mesopotamian deities, like computer code. The Indus corpus, they show, falls cleanly inside the range of known writing systems. It is not random. It is not formulaic. It has the entropy profile of language. This is not decipherment. It is the answer to a different question. It says, whatever this is, it is structured like writing. The argument is, as of 2026, still open. There are scholars in both camps. What is not in doubt is that the question itself, whether the Indus people wrote at all, became, a hundred years after Marshall's announcement, a serious open question in the field he founded. A researcher is sitting in a university lab at three in the morning. Three monitors are on. The overhead lights are off. The only sound is the steady hum of the cooling fans. On the first monitor, Mahadevan's 1977 catalog, digitized. Four thousand inscriptions in high resolution, arranged in rows. The signs tagged and indexed. On the second monitor, a transformer-based language model, mid-run, processing co-occurrence patterns across the corpus. Numbers flicker. The model is looking for what comes before a given sign, what comes after, what does not come at all. On the third monitor, a terminal window, a long paper in PDF, and an email she has not yet answered. She has been working on the Indus script with machine learning methods for three years. She is not the first. Since the 2000s and 10s, teams in Mumbai, Helsinki, Cambridge, Seattle, and Tokyo have applied increasingly sophisticated computational tools to the corpus. Markov models, n-gram analyses, neural networks, attention-based architectures. The methods have refined. The results have refined. The decipherment has not arrived. The fundamental problem is the one parpola named clearly 40 years ago. To decipher an unknown script, you need either a long text, long enough to let the structure show itself, or a bilingual inscription, a fragment of the same words in a known language. The Indus corpus offers neither. The longest single inscription has 26 signs. There is no Indus circadian seal, no Indus Sumerian dedication, no Indus inscription whose meaning can be checked against an outside reference. Modern language models are trained on languages that someone, somewhere, knows how to read. The Indus language has no living readers, and possibly no living relatives. On the 5th of January, 2025, exactly one century to the day after Marshall's announcement of the civilization, the chief minister of Tamil Nadu, M.K. Stalin, stood at a podium in Chennai and announced a prize of 1 million United States dollars for anyone who could decipher the Indus script. As of this filming, the prize has not been claimed. The researcher takes a sip of cold coffee. She moves a window aside. The model continues its run. She does not know whether the answer is in this technique, or in a technique that has not yet been invented, or whether there is no answer that any technique will ever recover. She does the next part of the work. A family of three generations is sitting on the floor for dinner in a low house in the sin. The grandmother is serving dal from a clay pot. The father is speaking Sindhi, a language descended from Sanskrit by way of 2,000 years of branching and reshaping, to his eight-year-old son. The son is answering in the same language. Eighty kilometers to the north on a low mound under a sky of the same dusk, the excavated streets of Mohenjo-Daro are emptying of tourists for the evening. The television in the corner of the room is on, low. A news anchor in Karachi is reading a brief story about an international conference at the National Museum about the preservation of certain seals against ongoing salt damage at the site. The word Mohenjo-Daro passes once. Nobody looks up. The grandmother has lived in this house for 40 years. Her parents lived nearby. Her parents' parents lived nearby. So did her parents' parents' parents. Genetic studies that have been published in this decade were coming out of David Reich's laboratory at Harvard in collaboration with Indian and Pakistani archaeologists have made something very clear. The population that lived in the cities of the Indus did not vanish at the collapse. They dispersed. They mixed. They moved south and east and stayed where they were. The people living today in the southern and western parts of the subcontinent carry a substantial fraction of their ancestry from the Indus people. The grandmother, the father, the son, biologically, they are the descendants. The language they speak at dinner is not the language those people spoke. No one knows what language those people spoke. If it was an early form of Dravidian, as Parpola and Mahadevan argued, then its closest surviving relative is Brahui, a small Dravidian language spoken by perhaps two million people in the highlands of Pakistani Balochistan, surrounded on every side by Indo-European tongues. It is the last possible echo of the Indus voice. It has perhaps a few thousand cognate words with Tamil. It has no written tradition older than the colonial era. The son finishes his daal. He asks his father something. His father answers. None of it is a language any Indus craftsman, any Indus farmer, any Indus supervisor of bricks would have understood. The seals in the National Museum in Karachi, 80 kilometers north, do not speak to this family in their own tongue. The genetic thread came through. The linguistic thread did not. Asko Parpola is at his desk in Helsinki. The morning is the same morning. The light from the snow is the same light. He has picked up the next photograph in the stack. He is examining it in the same way he examined the last one, and the one before that, and the thousands before that. He brings the magnifying lens close to the image. He looks. He sets the lens down. He picks up the pencil. He writes a small note in the margin, three letters and a number, a reference to a comparable sign on a seal from Lothal, which he will check this afternoon. He picks up the cup. The coffee is cold. He drinks it anyway. He sets the cup down. He sets the photograph aside. He picks up the next one. This is what the work looks like. It is not a final battle. It is not a man on the brink of revelation. It is a man of 84 doing on a Tuesday morning what he did on a Tuesday morning in 1974, and in 1991, and in 2008. The work has not arrived. The work continues. The arrival is not promised. He has said in interviews that he expects no breakthrough in his lifetime. He has also said that the work must continue. Both things are true at the same time, and he has made his peace with that. There are others coming up behind him. There are computational teams in five countries. There are doctoral students in Tamil, and Sindhi, and Finnish, and English, who are reading his book and writing their own. The thread, he believes, will be picked up by someone. He does not know when. He does not know by whom. He looks at the photograph in his hand. Four signs above the back of a one-horned animal. He has read these signs more times than he has read any other group of marks on earth. Then the camera moves. It is the sin. It is the same dusk. It is the same low mound. It is the same girl. She is still barefoot. The light has dropped another notch. The earth is darker now, almost the color of dried blood. Off to her left, her brother is calling her name. She is not listening. She bends down. She picks something out of the dust. She turns it in her fingers. It is small. It is fired clay. It has been in the earth for four and a half thousand years. She does not know this. She looks at it for a moment longer. Then she opens her hand and lets it fall back into the dust. And she turns and runs across the mound toward her brother. In Helsinki, the snow keeps falling. In the sin, the night is coming. Somewhere in between, four signs above the back of a one-horned animal wait to be read. Perhaps someone will read them. Perhaps no one will. The thread waits. We wait with it. Six films, six civilizations, six ways of trusting or failing to trust, a memory to time. The Polynesians trusted the body. They put the knowledge of 2,000 islands into the muscles of a few men and almost lost it. Mau Pialug on Sarawal in the second half of the 20th century taught what he had been taught. And the thread was picked up by the Hawaiians and the Maori and the Tahitians. The Sasanians trusted an entire empire, a court, an army, a state religion, a literature in three scripts. The empire fell in a single generation. And what survived survived because a small community of priests in the village of Udvada in Gujarat kept a flame burning in silence for a thousand years. The Armenians trusted a script. Mesrop Mashtots in the early 5th century sat down and made one. A church and a people held it through everything that came afterward. And when there was no state and no army and no land of their own, there was still the script and the script was enough. The Sumerians trusted clay. They had no idea they were doing it. Their fired tablets sat in the mounds of southern Iraq for 5,000 years. And then a young Englishman named George Smith in a room in the British Museum in 1872 read a flood story on one of them and stood up and walked around the room and could not speak. The Maya trusted books and the books were burned by a Spanish bishop named Diego de Landa who then sat down and wrote a manual of the script he had just destroyed. Four centuries later, a Russian named Yurik Norozov used the work of the destroyer to read what the destroyer had not been able to destroy. In five of those films, the thread held. By chance, by craft, by sacrifice, by the work of one obsessed person in a quiet room, the thread held. This film is different. The Indus people left more material than any other ancient civilization of their scale. More than a thousand cities. More than 4,000 inscriptions. Bricks fired so precisely that British engineers used them 3,500 years later as the foundation of a railway and the railway held. They left no bilingual inscription. They left no long text. They left no name of themselves in any language that survived. There was one man in Akkad in the third millennium BC who could read what they wrote. He did not leave a translation. There was a Bengali archaeologist who pulled their first seal out of the dust under a Buddhist stupa in 1922. There was a British director general who saw in a room in Simla in 1924 what the world had forgotten. There has been for 60 years a Finnish scholar in Helsinki turning over photographs one by one. There has been a Tamil scholar in Madras who worked until his death and did not read it. The arc of the other five films suggested that every lost civilization can be recovered, given enough time, given enough patience. The Indus people make us admit this is not quite true. Recovery depends on four things. Material that survives. A bilingual key. A decipherer and acceptance by the field. The Indus have the first. They have decipherers. They are missing the second and possibly always will be. Without a Rosetta Stone, even with all the material in the world, even with artificial intelligence, the decipherment may simply never come. And that is also part of human history. Not every thread was held. Not every voice was heard. There is a quietness in admitting that. The other five films of this arc ended in affirmation. This one needs to end in something less confident. A form of patience that does not expect an answer. The Indus people may exist for as long as we exist only as form, as cities laid out on a grid, as bricks in a precise proportion, as seals with signs that refuse to speak. Perhaps that is all the memory any civilization needs. They showed that it was possible to build cities without kings, without palaces, without monuments of war, without visible hierarchy. That is itself a form of transmission. A strange one. A wordless one. But a form. To survive, a civilization has to choose a medium. Some choose well, like the Armenians. Some choose by accident, like the Sumerians with their clay. Some are saved by strangers, like the Polynesians and the Sassanians. Some need one person who heard in silence, like Smith in London in 1872 or Norozov in Leningrad in 1952. And some, no matter how many cities they planned, no matter how many seals they carved, may never be fully recovered. The thread was not always held. The thread was not always released. Sometimes it was simply buried in a form we do not know how to read. And we learn to live with that. The thread waits. We wait with it. And perhaps that is also a form of survival. You have watched six films on six civilizations. Polynesians, Sassanians, Armenians, Sumerians, Maya, the Indus. They covered 5,000 years of human history. They covered six continents. They covered six different ways a civilization can trust or fail to trust its memory to the long, slow wait of time. Each of them found its own ending. Some survived in other forms. Some came back through strangers. One, the one we just watched, is still waiting for a voice. The next films will belong to another axis. For six civilizations, we asked how memory survives. For the next three, we will ask a different question. What happens when memory does not stay on land? What happens when an entire civilization or an entire country is taken not by fire and not by forgetting, but by water? A palace on a Greek island found in 1900 with a script that does not speak. A landmass beneath the North Sea, larger than a small nation, drowned 8,000 years ago, whose people left their tools on the seabed for trawlers to pull up by accident. A kingdom on the Atlantic coast of Iberia, hunted for 18 years by a German archaeologist in the marshes of the Guadalquivir that may or may not be the city the Greeks called Atlantis. But the sea swallowed. The next civilization is being prepared. Subscribe and you will be there when it arrives. Thank you for listening. Every empire ends, the thread does not.