Indus Valley Civilization

← Encyclopedia

In 1921 a railway-ballast quarry on the upper Ravi exposed a mound of ancient fired brick at a Punjab village called Harappa, and the culture that archaeology had just discovered took its name from the first of its cities to be dug. The Indus Valley Civilization — Harappan, after that site — was the Bronze Age urban culture that flourished across the Indus basin and its tributaries, in what is now Pakistan and northwestern India, at its height roughly between 2600 and 1900 BCE. It was one of the three great river-valley civilizations of the third millennium, contemporary with the Old Kingdom of ancient Egypt and with Sumer in lower Mesopotamia, and far the largest of the three in the ground it covered. Where Egypt threaded a single valley and Sumer clustered on the southern alluvium, the Harappan world reached across more than a million square kilometers — from the Makran coast to the foothills of the Himalaya, from the Arabian Sea up the Indus and out along the now-dry bed of the Ghaggar-Hakra. Gregory Possehl’s survey counted well over two thousand settlements within that expanse. Most were villages; a handful were among the first true cities the world had made. Behind them lay the long preludes the whole Old World shared — the Paleolithic foragers of the subcontinent and the Neolithic farming villages, Mehrgarh on the Kacchi plain among the earliest, out of which the urban Indus grew over the preceding two millennia.

The Record in Brick

What excavation has recovered is substantial, and it is unusually coherent. The Harappans built planned cities, Mohenjo-daro on the Indus in Sindh and Harappa on the Ravi the greatest among them, laid out not as organic accretions of lane and courtyard but on something close to a grid: broad arterial streets running roughly cardinal, narrower lanes branching from them, the whole divided in the mature cities into a raised western “citadel” mound and a lower town below. The brick itself is the signature. Across hundreds of miles and several centuries, Harappan fired brick holds to a near-constant proportion — length, width, and thickness in a ratio close to 4:2:1 — so that bricks from one city would course true in the wall of another. Drains ran covered beneath the streets, fed by household chutes and sumps; wells, public and private, stood throughout the settlements, hundreds at Mohenjo-daro alone; and bathing platforms and latrines recur in the houses with a regularity that has made water, its supply and its disposal, look like a governing preoccupation of the culture.

The most discussed single structure embodies that preoccupation. On the citadel at Mohenjo-daro stands the tank that excavators named the Great Bath — a watertight basin roughly twelve meters by seven and some two and a half meters deep, sunk into a podium and reached by brick steps at either end. Its makers laid the floor and walls in finely fitted brick set in gypsum mortar, then sealed the whole behind a damp-proof course of bitumen, and ringed it with a colonnade and small rooms. Nothing about its scale or finish suggests ordinary washing, and it has been read, cautiously, as a place of collective immersion — a built form of the concern with purity that the domestic bathrooms already imply. What it was for, in the sense of what immersion meant, the brick does not say.

The economy that filled these cities was a craft and trade economy of high precision. Harappan smiths worked copper and bronze, alloying tin where they could get it; their lapidaries drilled long carnelian beads and etched them with alkali to leave white patterns on the red stone, a technique whose products traveled far. And the culture standardized its commerce to a degree no other Bronze Age society matched. Its weights are small cubes of chert, unmarked, graded in a binary series — 1, 2, 4, 8, 16, 32 — that shifts to a decimal step at the larger sizes, the system holding constant from city to city; the commonest weight runs near 13.7 grams, the smallest under a gram. Concentrations of weights near city gates have suggested to Jonathan Mark Kenoyer that they served the assessment of dues on goods entering the towns. A society that measures so uniformly is a society administering something across distance, though what authority stood behind the standard — palace, temple, guild, something with no later name — is not recoverable from the things themselves.

Meluhha

The reach of that trade is fixed, unusually, by the testimony of a literate neighbor. To the cuneiform scribes of Akkad and Sumer the lands east across the sea were Dilmun, Magan, and — furthest — Meluhha, which scholarship identifies with the Indus world. An inscription of Sargon of Akkad, who reigned about 2334 to 2279 BCE, boasts that ships of Meluhha, Magan, and Dilmun tied up at the quay of his capital; a later Akkadian tablet names a man, with an Akkadian name, as the holder of a Meluhha ship, and texts speak of a Meluhhan settlement and interpreters of the Meluhhan tongue resident in Mesopotamia. The material record answers the texts. Some twenty Indus seals have surfaced in Mesopotamian soil, several of them in contexts datable to the Akkadian period and a couple later, to the dynasties of Isin and Larsa; etched carnelian beads of plainly Harappan make lie in the Royal Cemetery of Ur, in graves of the mid-third millennium. Lapis from the mines of Badakhshan, carnelian, gold, tin, and timber moved west along these routes, and Mesopotamian things — fewer, but present — moved east. For a culture that left no readable word about itself, the Harappans are thus dated and placed in part by the words of Babylon’s ancestors. The intermediary land of Elam, on the Iranian plateau between the two, sat astride the overland leg of the same exchange.

The Slow Withdrawal

Around 1900 BCE the cities unmade themselves. The civic order that had held for seven centuries loosened: the grid decayed, the standardized brick and the weights fell out of use, the great sites shrank and were left, and the script disappeared with them. For a long time the favored explanation was catastrophe — an Aryan invasion sweeping down to sack Mohenjo-daro, a reading built partly on a scatter of unburied skeletons in its upper levels and partly on the martial imagery of the later Vedic hymns gathered in the Rigveda. That account has not survived scrutiny. The skeletons are too few and too various to mark a massacre, and no destruction horizon answers to a conquest.

What has displaced it is climate and water. The Harappan cities depended on the rivers and on a summer monsoon that, around the close of the third millennium, weakened and shifted east. The Ghaggar-Hakra — the great seasonal channel some identify with the Sarasvati of later tradition — declined from a perennial river toward a chain of seasonal pools, drying its settled corridor; the work of Liviu Giosan and colleagues, mapping the floodplains, reconstructs a long aridification under which the dense urban population dispersed eastward toward the more reliable rains of the Ganges basin. The end, on this reading, was not a fall but a draining away — a gradual de-urbanization, the cities devolving into the smaller farming villages of the Late Harappan phase, the integrated culture coming apart into regional successors. No siege ended it; the water did.

The Undeciphered Seals

Everything to this point is, by the standards of prehistory, firm. What lies beyond it is not, and the boundary is sharp, because the Harappans left writing that cannot be read. Their script survives almost entirely in miniature — a few signs each, incised most often on small square steatite seals carved with an animal above a line of characters, and stamped onto sealings, copper tablets, and pottery. The inscriptions are short, averaging only four or five signs and rarely exceeding a dozen; the sign list runs to several hundred symbols. After a century of effort the script remains undeciphered, with no bilingual key and no agreement on the language beneath it; the corpus is short, repetitive, and lacks any long text against which a reading could be checked. The Dravidian hypothesis, developed at length by Asko Parpola, holds that the seals record an early Dravidian tongue and remains the strongest linguistic candidate. Against the whole enterprise, Steve Farmer, Richard Sproat, and Michael Witzel argued in 2004 that the corpus is too short, too repetitive, and too lacking in long texts to be writing at all, and that the signs may be nonlinguistic emblems — heraldic, religious, or political marks rather than speech set down. Statistical work has answered them: a 2009 study in Science by Rajesh Rao and colleagues found the conditional entropy of Indus sign sequences closer to that of natural languages than to nonlinguistic symbol systems, evidence the script has the structure of a writing system even if its sounds are lost. Scholars contest the particulars fiercely; the governing fact is plain. Without readable texts, the inner life of the civilization — its names, its gods, its order, its account of itself — is closed to direct approach.

The God That May Not Be There

Religion is therefore reached only sideways, through objects, and the inference is correspondingly fragile. Excavators have pointed to terracotta female figurines, some heavily ornamented, as possible images of a goddess or of fertility; to a recurring motif of a figure within a pipal tree, perhaps a sacred tree or its spirit; to composite and horned animals, and to the unicorn-like beast that dominates the seals. The single most consequential object is one small steatite seal from Mohenjo-daro, recovered from the DK-G area and numbered 420 in the excavation catalog. It shows a figure seated with knees splayed and heels together, wearing a tall horned headdress, possibly multi-faced, the body marked with bangles, and flanked by wild beasts — an elephant and a tiger to one side, a rhinoceros and a water buffalo to the other, with smaller animals below.

John Marshall, who directed the early excavations, read this figure in his foundational report of 1931 as a “proto-Śiva”: a third-millennium forerunner of the Hindu god Shiva in his aspect as Paśupati, lord of creatures, the horned crown, the surrounding animals, and the cross-legged posture all taken as anticipations of the later deity and of the discipline of yoga. The reading was vivid and it has had a long life. It has also been contested almost from the start, and the objections cut at every joint. The “faces” may be a misread headdress; the horns and the prominence of the buffalo have suggested to several scholars — Doris Meth Srinivasan most influentially, in an iconological assessment published in the mid-1970s — that the figure is better understood as a divine bovine or buffalo-man than as a proto-Śiva, and that the Vedic Rudra from whom Śiva descends was not, in the early texts, a protector of wild animals at all. To call the posture “yoga” is to read a thousand years of subsequent practice back into a seated pose; the god Shiva and the codified yoga of Patanjali alike are attested only much later, in texts whose own relation to the Indus world is exactly what is in question.

For the seal sits at the center of a larger and politically charged argument about continuity. One position holds that Harappan religion flows without major break into historical Hinduism, the seals its earliest surviving images, the modern tradition rooted in the soil of the Indus. The other holds that the Indo-Aryan languages and the Vedic cult arrived from the northwest in the second millennium, after the cities had already faded, and were laid over an Indus substrate whose own religion is largely lost to us. The question carries weight far beyond the seminar, touching modern accounts of who was first in the land, and the script’s silence denies it the one thing that could settle it: a sentence in the Harappans’ own voice. Both readings are arguments from a picture and a posture across a gap that no text bridges.

The newest evidence brought to that argument is genetic. In 2019 a team led by Vasant Shinde and Vagheesh Narasimhan recovered ancient DNA from a single Harappan-era burial at Rakhigarhi, one of the largest sites, and found an ancestry compounded of an Iranian-related lineage — one that had branched off before farming began — and a deep indigenous South Asian one, with no detectable trace of the steppe pastoralist ancestry that later spread across northern South Asia. A companion population study placed the arrival of that steppe-derived ancestry in the centuries after about 2000 BCE, in the window usually assigned to the spread of Indo-European speech, after the Harappan cities had already declined. The mainstream linguistic and genetic reading takes this as support for a later Indo-Aryan migration into a subcontinent the Harappans already occupied; a smaller body of scholarship reads the same absence of steppe ancestry in the Indus sample as consistent with an indigenous origin for the tradition. Either way the genome reports an ancestry, not a creed. It can say who the woman at Rakhigarhi descended from; it cannot say what she or her city held sacred.

The Texts and the Scholarship

Because no Harappan word can be read, the primary documents of this field are not literary but material — the seals, the brick, the weights, the stratigraphy — and the foundational corpus is the published excavation record: Marshall’s Mohenjo-Daro and the Indus Civilization (1931) and the later volumes from Harappa, against which every interpretation is still measured. The standard modern syntheses are Gregory Possehl’s The Indus Civilization: A Contemporary Perspective (2002), which assembles the settlement evidence and weighs the climatic decline, and Jonathan Mark Kenoyer’s Ancient Cities of the Indus Valley Civilization (1998), which treats the craft and urban order from long fieldwork at Harappa.

On the script, Asko Parpola’s Deciphering the Indus Script (Cambridge, 1994) remains the fullest statement of the linguistic — and Dravidian — case; the opposing nonlinguistic thesis was set out by Steve Farmer, Richard Sproat, and Michael Witzel in “The Collapse of the Indus-Script Thesis” in the Electronic Journal of Vedic Studies (2004); and the statistical rejoinder, Rao and colleagues’ “Entropic Evidence for Linguistic Structure in the Indus Script,” appeared in Science in 2009 (10.1126/science.1170391). The reinterpretation of the “proto-Śiva” image rests above all on Doris Meth Srinivasan’s iconological assessment of the so-called Paśupati seal, published in Archives of Asian Art (Srinivasan 1975–76). On the decline, the integration of river-history and archaeology by Liviu Giosan and colleagues, “Fluvial landscapes of the Harappan civilization” (PNAS, 2012), made the case for monsoon weakening over invasion; and on the continuity and migration questions the decisive recent evidence is the ancient genome reported in Vasant Shinde, Vagheesh Narasimhan, and colleagues’ “An Ancient Harappan Genome Lacks Ancestry from Steppe Pastoralists or Iranian Farmers” (Cell, 2019), read alongside the broader population history in Narasimhan and colleagues’ Science paper of the same year. The later Indian texts that the continuity argument reaches for — the Yoga Sutras attributed to Patanjali and the Upanishads — belong to a world a thousand years downstream of the cities, and the question of what, if anything, runs between is precisely the unanswered one.

The Harappans, then, are known the way few ancient peoples are: their cities planned to the brick, their commerce measured to the gram, their reach charted by the carnelian in a Sumerian grave. No Bronze Age people left a material record more exactly taken — fired brick in a fixed 4:2:1 ratio, weights graded in one system from a few grams up, streets laid to a grid across a thousand kilometers. And none left less of what it believed. The seated horned figure was cut into a hard stone meant to take an impression and to keep it; the impression survives, and the sense it was cut to carry does not. The same hands cut both seal and brick — the brick laid four-by-two-by-one in every city from the mountains to the sea, the seal in signs no reader has yet cracked. An etched carnelian bead from those workshops lies today among the royal graves of Ur, carried there by a trade that ran for the better part of a thousand years: the reach of this people measured in a single bead, half a continent from the river that made it.

→ In the library: The Yoga Sutras of Patanjali (Johnston, 1912) — the codified yoga read back onto the seals by critics · The Upanishads (Müller, Sacred Books of the East, 1884) — the later Vedic stratum at issue in the continuity debate

→ Related: Mesopotamia · Sumer · Elam · Babylon · Hinduism · Shiva · Yoga · Patanjali · Ancient Egypt · Rigveda · Paleolithic · Neolithic · Yajna