Was 'Hinduism' invented by the British? Conceding the word, contesting the claim

The word is borrowed. "Hindu" entered Indian languages from Persian, which took it from a river: Old Persian hindu-, from Indo-Iranian sindhu, the Indus. It appears in the inscriptions of Darius I as the name of a province. The English abstraction "Hinduism" is far younger, with its earliest recorded evidence in 1786, and it stayed rare for decades after that. Then the colonial census took a field of overlapping affiliations and turned it into a single box on a form.

All of that is true. None of it is conceded reluctantly.

What does not follow is the sentence usually attached to it. The serious constructionist case is not that Hindus did not exist. It is that Europeans imposed one category on sects whose members did not recognise themselves as sharing anything essential. That is a claim about evidence, and the evidence runs against it.

The critics' case at full strength

Wilfred Cantwell Smith set the terms in The Meaning and End of Religion in 1962. His objection was not primarily historical. He held that there is no Hinduism either in the minds of Hindus or in empirical reality, and that outside observers should stop imposing their own categories on religious phenomena. That is a philosophical argument about the concept of religion itself, and Smith applied it to Christianity too.

Robert Frykenberg, writing in 1989, put it as a flat denial of referent: unless "Hindu" means nothing more than "Indian", there has never been any such thing as a single Hinduism or a single Hindu community for all of India, nor for any one region of it. Heinrich von Stietencron argued a related point, that the word covers several religions rather than one, Vaishnava, Shaiva, Shakta and others, each with its own theology, canon and community, and that the British made a category error in taking an administrator's term for a religious one. Vasudha Dalmia's work on late nineteenth-century Vaishnava self-representation showed a tradition being repackaged as the real Hinduism in print.

These are careful scholars making a checkable argument, and it should be met at full strength or not at all.

Two claims are often run together. One is that the English word is a late coinage, a fact about English. The other is that no corresponding concept existed among Hindus before Europeans supplied one. Only the second is contested here.

Start with the word, because the usual dates are wrong

The figure repeated everywhere is 1829, the first citation in the unrevised Oxford English Dictionary entry that Smith used, pointing to Henry Barkley Henderson's book on Bengal. It is superseded. The revised entry now records evidence from 1786, in the Asiatick Miscellany. Geoffrey Oddie traced "Hindooism" to a letter written from Calcutta in 1787 by Charles Grant, an evangelical who wanted converts out of it; Grant used it again in his Observations of 1792. David Lorenzen, searching early journals rather than trusting the dictionary, found the word in the Asiatic Journal for 1818 and seven times in John Crawfurd's 1820 article on the Hindu religion in Bali.

The date that matters most is 1816. In that year Rammohan Roy used "Hindooism" in English, complaining that the chief part of its theory and practice had been made to consist in a peculiar mode of diet. In 1817 he used it again, to claim that the doctrines of the unity of God were the real Hinduism as practised by his ancestors. He probably picked the word up from the Serampore missionaries, one of whom he had met the year before. A Bengali Hindu is therefore among the earliest known users of the label for a supposedly British construct, and he used it to argue about his own tradition.

The word spread mostly after that, helped by Alexander Duff's book of 1839 and Monier Monier-Williams' handbook Hinduism in 1877.

The census, which is the strongest evidence against us

The census anchor is usually given as 1871-72, and that is too narrow. That exercise was not synchronous, did not cover all territory, and was not centrally compiled. The first synchronous all-India enumeration came in 1881 under W. C. Plowden, and the machinery hardened over the four decades after it. The most damaging material is not in the scholarship. It is in the reports themselves.

Edward Gait's 1911 report opens its religion chapter by admitting that outside the exotic religions, Christianity and Muhammadanism, there is no such thing as a definite creed; that dharma, the word closest to the English "religion", connotes conduct more than creed; and that in India the line of cleavage is social rather than religious. Gait calls what is left "a most heterogeneous mixture", a category sheltering monotheists, polytheists and pantheists, worshippers of Shiva and Vishnu and of the spirits of trees, rocks and streams, people who propitiate their deity by bloody sacrifice and people who will not use the word "cut". On the boundary with the category called Animist, he concedes that the decision would often depend on the idiosyncrasy of the enumerator: a high-caste Hindu might record as Animists those a Christian enumerator would enter as Hindus.

The 1921 report goes further. Its commissioner, J. T. Marten, wrote flatly that Hindu is an unsatisfactory category in the census classification of religion. The term implies country, race and a social organisation as well as a creed, and no test covering all of that could be applied by census staff. Its correct application, he concluded, must always be a matter of sentiment and opinion, and his own classification rested on a vague and almost negative conception: anyone not definitely assignable somewhere else.

Bernard Cohn's essay on the census and objectification in South Asia is the standard account. Counting requires categories that do not overlap, and a counted population starts to think in the categories used to count it. That happened, and it is not small.

The box on the form is real, and it changed something. The question is whether it created the thing it counted or hardened something already there. Only the first is refuted by what follows.

What the tradition called itself

Here the honest answer costs something.

"Hindu dharma" never entered the premodern Sanskrit lexicon. Lorenzen, the leading opponent of the constructionist case, says so himself. Sanskrit intellectual culture had a habit of not naming outsiders and not naming itself in contrast to them.

Sanatana dharma is older than the English word, but it did not mean what it now means. The phrase runs through the Mahabharata as a formula closing a statement about right conduct, and Manusmriti 4.138 uses it the same way.

"Let him say what is true, let him say what is pleasing, let him utter no disagreeable truth, and let him utter no agreeable falsehood; that is the eternal law." Manusmriti 4.138 in Georg Bühler's 1886 rendering. The Sanskrit closes eṣa dharmaḥ sanātanaḥ.

That is a rule about speech, sealed with a formula meaning roughly "this is the enduring norm". In the Bhagavad Gita the compound appears at 1.40, where Arjuna fears that the destruction of a family will destroy its kula-dharmāḥ sanātanāḥ, its immemorial family observances. Wilhelm Halbfass concluded that the term's precise sense was always unstable. Its use as a proper name for the religion as a whole belongs to the late nineteenth century, to the sanatana dharma associations founded from the 1880s and to the Bharat Dharma Mahamandala assembled from them in 1902. Audrey Truschke places it alongside "Hinduism" and "Brahminism" as a term whose current meaning is a colonial-era product. Anyone using it to prove the tradition always had its own name has picked the wrong exhibit.

The exhibit that does work is duller. Vaidika and avaidika, and the paired astika and nastika, marked a boundary Sanskrit thinkers used and argued about for centuries: those who accept the Veda as an authority and those who do not. That line put Buddhists, Jains and materialists outside and a quarrelling set of schools inside. A boundary term people fight over is evidence of a field.

The doxographers were unifying before anyone arrived to do it

Andrew Nicholson's Unifying Hinduism is the technical core of the counter-argument. Between roughly the twelfth and sixteenth centuries, Indian thinkers writing surveys of philosophical positions began treating the Upanishads, the epics, the Puranas and the schools later called the six systems as parts of a single whole.

The texts are nameable. The Sarvadarsanasamgraha, from fourteenth-century Vijayanagara and attributed to Madhava, arranges sixteen positions in a scale starting with the materialists at the bottom. Madhusudana Sarasvati's Prasthanabheda, from around 1600, sets out the systems and then presses their underlying agreement. Vijnanabhikshu, in the sixteenth century, read Samkhya, Yoga and Vedanta as one teaching rather than three.

Nicholson's sharper observation is about the vocabulary. Astika and nastika had been loose terms, attached at various times to those who affirmed an afterlife, or another world, or the authority of scripture. Over this period they hardened into a doctrinal test, assent to the Veda, separating the schools that counted from those that did not, in Sanskrit, centuries before an English word existed.

The unification that constructionists date to the nineteenth century has a paper trail running back to the fourteenth, written in Sanskrit by men arguing with each other about who counted as inside.

Two objections deserve stating rather than hiding. Reviewers of Nicholson's book, Vishwa Adluri among them, argue that the synthesising work he describes began well before the twelfth century, which would loosen his chronology and shrink the weight he gives to the encounter with Islam. The heavier objection is that a unity worked out by philosophers reading each other's Sanskrit is not the unity felt by someone making an offering at a village shrine. That one is answered only partly, by the next body of evidence.

People using the word about themselves

Lorenzen's evidence is vernacular rather than Sanskrit, which is why it survives.

Jain authors were using hindua and hinduka in Sanskrit and Prakrit by the 1330s. The earliest datable case is the Vividhatirthakalpa of Jinaprabhasuri and Vidyatilaka, compiled between 1330 and 1333, which sorts rulers into earlier Hindu ones and later Muslim ones and judges both about equally tolerable for Jain monks. From 1347 the Vijayanagara kings carried the title hinduraya-suratrana, sultan over Hindu kings, used first of Marappa and then in inscription after inscription for two and a half centuries. Cynthia Talbot, who worked through the Andhra material, reads the sense as political and geographical more than religious, and Truschke adds that the title boasts of superiority over lowly Hindu rulers rather than membership among them.

The clearer material is a century later. Vidyapati's Kirtilata, written in Apabhramsha around 1400 to 1410, describes Jaunpur as a city where Hindus and Turks live together, hindū turuke milala vāsa, each finding the other's dhamma funny: one calls the faithful to prayer, the other recites the Vedas; some eat from copper plates, others from pottery; some perform namaz, others puja. Truschke reads the passage as setting the two dhammas in parallel rather than ranking them; Lorenzen points to the poem's hostility to Turkish power elsewhere. On either reading, dharma is doing the work of "religion", in an Indian language, four and a half centuries before the first census.

Eknath, a Brahmin poet in Maharashtra who died in 1599, wrote a dialogue in which a Hindu and a Turk mock each other's rituals and are reconciled at the end. Kabir, in Banaras around 1450 to 1520, rejects both, which requires both to exist as targets. The Kashmiri continuators of the Rajatarangini, Jonaraja in 1459 and Shrivara in 1486, use hinduka for Brahminical customs.

Truschke's caveat matters. These usages are few, they coexist with the older geographical sense, and several of these authors use "hindu" to mean Brahmins rather than everyone a modern census would count. Eknath's Hindu and his Brahmin are the same person.

Two outside observers close the loop. Al-Biruni, writing in Arabic around 1030, produced an account of the Hindus whose chapters run through their belief in God, the soul and rebirth, liberation, the varnas, the source of their religious law, idol worship, the Veda and the Puranas, the yugas, sacrifices and pilgrimage. That is very nearly the table of contents of a modern textbook. The Persian Dabistan-i Mazahib, from the mid-seventeenth century, sets out the Hindu schools at length alongside the other religions it surveys. Neither author was British and neither had a census to fill in.

Lorenzen's own limit on his case is the concession that makes the rest credible. Without the Muslim, or some other non-Hindu, he writes, Hindus can only be "Vaishnavas, Saivas, Smartas or the like." The identity he documents is relational. It sharpened against something.

Naming is not making

Hieronymus Wolf coined "Byzantine" in 1557 for an empire that had fallen a century earlier and had called itself Roman throughout. The name is an invention. The empire is not. "Buddhism" is an English coinage of the same colonial decades, produced by the same scholars for the same reasons, and nobody argues that the British invented Buddhism.

A name is not nothing. It changes what people can say about themselves, and it lets a state count them. But the things that carry a tradition across centuries are practices, texts, calendars and places, which can be tracked without a label.

The pilgrimage map is the clearest case. The tirtha-yatra section of the Mahabharata's Vana Parva already sends its listener around the subcontinent. The Puranic mahatmyas built it out, and Diana Eck's work on that landscape shows sites echoing each other, linked by pilgrims who walked between regions sharing no ruler and no language. A Bengali at Rameswaram and a Tamil at Kedarnath were not consulting a shared theology. They were using a shared map, and the map assumed a shared field long before there was a word for it.

Where the argument should stop

The constructionists are right that the English word is late, that the census category was arbitrary at the edges and its own compilers said so, and that much of what now presents itself as ancient unity was assembled in the nineteenth century. They are right that pre-modern Shaivas, Vaishnavas and Shaktas frequently did not think of themselves as one religion, that Sanskrit had no name for the whole, and that where "hindu" appears in a religious sense before 1600 it often means Brahmins.

Three claims should therefore be dropped. That "Hinduism" is a purely foreign imposition, when a Bengali Hindu was using it in 1816 to argue about his own tradition. That sanatana dharma has always been the tradition's name for itself, when in older texts it is a formula about right conduct and its use as a proper name is about a hundred and forty years old. That Hindus before the colonial period saw themselves as one people, when the record shows primary sectarian identities and a shared identity surfacing mainly on contact with an outsider.

What survives is smaller and holds. A body of texts treated as authoritative by schools that disagreed about everything else. A boundary term, astika, that people defended. Doxographers unifying the systems in Sanskrit from the fourteenth century. Poets around 1400 using hindū dhamma as a contrast term in the vernacular. An Arabic account from around 1030 describing Hindu belief as one system. A pilgrimage geography that presupposed a common field.

The British named it. They did not build it, and the record of what they found is detailed enough to check.