Modern Human Variation: The Typological, Populational and Clinal Models, the Classifications of Blumenbach, Deniker, Hooton, Coon, Garn and Birdsell, the Indian Classifications of Risley, Guha and Sarkar, and the Linguistic Distribution of Indian Populations

The typological model, which governed anthropology from Linnaeus and Blumenbach to the 1940s, assumed that humankind consists of a few fixed "races", each with an ideal type defined by a bundle of traits — skin colour, hair form, head shape (the cephalic index), nose shape (the nasal index), stature — from which living individuals deviate through "mixture". Its method was to measure skulls and bodies and sort them into types. The populational model, which Washburn's "new physical anthropology" (1951), Dobzhansky and William Boyd's Genetics and the Races of Man (1950) brought in, replaced the type with the breeding population (the Mendelian population or deme): variation within a population is normal, not "impurity"; what differs between populations is the frequency of alleles, not the presence of a type; and the unit of study is the gene pool, described by blood-group and other allele frequencies. The clinal model goes one step further. Julian Huxley coined "cline" (1938) for a gradient in a character across geographical space; Frank Livingstone's 1962 paper "On the non-existence of human races" argued that human traits vary in independent clines — skin colour with latitude and ultraviolet light, body form with temperature (Bergmann and Allen), the sickle-cell allele with malaria, the B blood group increasing from west to east across Eurasia — that do not coincide, so that no boundary drawn on one trait matches a boundary drawn on another. "There are no races, there are only clines." Richard Lewontin (1972) showed with blood-group and protein data that about 85% of human genetic variation lies within local populations, some 8% between populations within a "race" and only about 6–7% between the classical races; later genome-wide studies confirm the proportions. Race is therefore a social classification with real consequences, not a biological subdivision of Homo sapiens, and the syllabus's classifications are taught as history.

Herbert Hope Risley, an ICS officer who directed the 1901 Census and wrote The People of India (1908; second edition 1915), measured heads and noses across northern India and made the nasal index his key: "the social position of a caste varies inversely as its nasal index". He proposed seven "physical types": Turko-Iranian (the frontier), Indo-Aryan (Punjab, Rajputana, Kashmir; tall, fair, long-headed, narrow-nosed), Scytho-Dravidian (western India — Marathas, Coorgs), Aryo-Dravidian or Hindustani (the Gangetic plain), Mongolo-Dravidian (Bengal, Odisha), Mongoloid (the Himalaya and the north-east) and Dravidian (the south and the tribal belt, described as the "original" population with a broad nose). His scheme is an artefact of the two-race Aryan-invasion theory of his day, tied caste to race and is of purely historical interest. Biraja Sankar Guha, trained under Hooton at Harvard, prepared the racial analysis for the 1931 Census (published 1935) from measurements of many thousands, and proposed six main races with nine sub-types: the Negrito (the Andamanese, and traces claimed among the Kadar and Irula); the Proto-Australoid (the tribal peoples of central and southern India — Bhil, Santal, Oraon, Munda, Chenchu — long-headed, dark, broad-nosed); the Mongoloid, divided into Palaeo-Mongoloid (long-headed and broad-headed sub-types, in the north-eastern hills) and Tibeto-Mongoloid (Sikkim, Bhutan); the Mediterranean, with three sub-types — Palaeo-Mediterranean (the Dravidian-speaking south; the Telugu and Tamil cultivators), Mediterranean (the north-western and Gangetic populations, credited with the Indus civilization) and Oriental (Punjab, Sindh, Rajputana); the Western Brachycephals, with three — Alpinoid (Gujarat, Bengal, Odisha), Dinaric (Bengal, Odisha, Coorg) and Armenoid (Parsis); and the Nordic (in the north-west, among Rajputs and Punjabis). Sasanka Sekhar Sarkar of Calcutta University, in The Aboriginal Races of India (1954) and his 1961 paper, criticised Guha for over-classification and reduced the picture to six: Australoid, Indo-Aryan, Irano-Scythian, Mundari-speaking (later "Mundari" for the Austroasiatic groups), Mongolian and Far-Eastern; he rejected a distinct Negrito element on the mainland and read Indian variation as the outcome of overlapping migrations rather than of surviving types.

Language, unlike "race", is a real and countable classification, and it is the axis along which Indian population history is now read. George Grierson's Linguistic Survey of India (1903–28) mapped 179 languages and 544 dialects; the 2011 Census returned 121 languages (with over 10,000 speakers each) and 19,569 raw mother-tongue names, grouped in four major families and two minor ones. The Indo-Aryan branch of Indo-European, with about 78% of the population, covers the north, west and east and Sri Lanka: Hindi and its associated tongues, Bengali, Marathi, Gujarati, Punjabi, Odia, Assamese, Maithili, Sindhi, Kashmiri, Nepali, Konkani, Sinhala. The Dravidian family, about 20%, covers the south — Tamil, Telugu, Kannada, Malayalam, Tulu — with outliers Brahui in Balochistan, Kurukh (Oraon) and Malto in the Chotanagpur–Rajmahal region and Gondi in central India, which are read as relics of a once wider distribution. The Austroasiatic family, about 1.1%, has two Indian branches: Munda (Santali, Mundari, Ho, Korku, Sora, Kharia) in the Chotanagpur plateau and central India, and Mon-Khmer (Khasi in Meghalaya, Nicobarese). The Tibeto-Burman branch of Sino-Tibetan, about 1%, covers the Himalaya and the north-east: Bodo, Garo, Manipuri (Meitei), Mizo, the Naga languages, Lepcha, Ladakhi, Bhotia. The Andamanese languages (Great Andamanese, Ongan: Onge and Jarawa) and Burushaski in Hunza (now in Pakistan) are isolates; Tai-Kadai (Khamti, Ahom) is represented on the Assam border. The Eighth Schedule lists 22 languages: 15 Indo-Aryan (Assamese, Bengali, Dogri, Gujarati, Hindi, Kashmiri, Konkani, Maithili, Marathi, Nepali, Odia, Punjabi, Sanskrit, Sindhi, Urdu), 4 Dravidian (Kannada, Malayalam, Tamil, Telugu), 2 Tibeto-Burman (Bodo, Manipuri) and 1 Austroasiatic (Santali, added with Bodo, Dogri and Maithili by the 92nd Amendment in 2003). Guha's "races" and these families do not coincide: Munda speakers and Dravidian-speaking Oraons share the same physical environment and largely the same ancestry, and the north-east's Tibeto-Burman speakers include long- and broad-headed groups alike.

The cephalic index, introduced by Anders Retzius (1842), is the maximum head breadth divided by the maximum head length × 100. On the skull (the cranial index) the classes are dolichocranial below 75, mesocranial 75–79.9 and brachycranial 80 and above; on the living head, where soft tissue adds about a unit, Martin's classes sit one unit higher: dolichocephalic (long-headed) below 76, mesocephalic 76–80.9, brachycephalic (broad-headed) 81 and above. The nasal index is nasal breadth ÷ nasal height × 100: leptorrhine (narrow) below 70, mesorrhine 70–84.9, platyrrhine (broad) 85 and above, on the living. Stature was classed as short (below 160 cm), medium, tall (above 170 cm); skin colour on the von Luschan scale; hair as straight (leiotrichous), wavy (cymotrichous) and woolly or frizzy (ulotrichous). Worked: a head 190 mm long and 145 mm broad has a cephalic index of 145 ÷ 190 × 100 = 76.3, mesocephalic; a nose 38 mm broad and 48 mm high has a nasal index of 79.2, mesorrhine. These are legitimate measurements of variation, and they are still used in growth, forensic and ergonomic work; what is abandoned is their use to sort people into types. The populational and clinal models replaced them with allele frequencies — first the ABO, MN and Rh blood groups (Hirszfeld 1919; Boyd 1950; the Indian surveys of Majumdar and Sarkar), haemoglobin variants, PTC tasting and G6PD deficiency, then HLA, then mitochondrial and Y-chromosome haplogroups and genome-wide SNP data. Every one of these shows the same picture: clines across the subcontinent, more variation within than between groups, and a history of migration and admixture rather than of preserved types.