What do you understand by discrete castes and muddled hierarchies? (2018)
The phrase is Dipankar Gupta‘s, from Interrogating Caste: Understanding Hierarchy and Difference in Indian Society (2000). Against Louis Dumont‘s Homo Hierarchicus (1966), which derived the whole caste order from one encompassing value, Gupta argues that caste is constituted by difference rather than by an agreed ladder, and that the ladder sociology inherited is an artefact of reading Brahmanical texts as India’s self-description.
Discrete castes
A caste, for Gupta, is not a rung but a bounded, self-contained entity. Each jati has its own origin myth — a purana narrating descent from a sage or a king — which places it high and explains its present condition as degradation from earlier glory. Three consequences follow.
- Castes are discrete, not continuous. One cannot slide from one jati into the next; the boundary is absolute, policed by endogamy and commensality. Caste is a system of separations, and hierarchy is only one possible ordering of separated units.
- Each caste is the author of its own ideology: there are as many caste ideologies as castes, each ranking itself high.
- Difference logically precedes hierarchy: to rank, one must first distinguish.
Muddled hierarchies
If every jati ranks from its own vantage, the rankings do not compose into one consistent scale. Muddled hierarchies names that condition — hierarchies multiple, locally variant, contested and mutually incompatible.
- The extremes are broadly agreed; the middle is a field of dispute, and most castes sit in the middle.
- Gerald Berreman‘s Sirkanda work (‘Caste as Social Process’, 1967) found not consensus but dissensus: from below the order is experienced as coercion, and low castes do not endorse the ideology that ranks them. McKim Marriott‘s interactional ranking exercises produced different orders for the same castes in different villages.
- Owen Lynch‘s The Politics of Untouchability (1969), on the Jatavs of Agra, shows a caste abandoning Sanskritising claims for political and Buddhist assertion — refusing the scale rather than climbing it.
- Dumont’s own concession, substantialisation — caste becoming competing self-contained blocks asserting identity — admits the point for modern India.
Illustrations
- Competing origin myths are routine: artisan castes claim Vishwakarma descent and a Brahmanical standing no one concedes; pastoral and cultivating jatis assert Kshatriya lineage.
- State enumeration exposes the muddle. Bihar’s 2023 caste survey counted more than two hundred castes and sub-castes, with Extremely Backward Classes alone at about 36 per cent — a category created because backwardness could not be ranked on one gradient. Telangana’s socio-economic and caste survey, reported in 2025, recorded 242 castes, and the disputes over which jati belongs in which list are muddled hierarchy in administrative form.
The critique
- B. R. Ambedkar called caste a system of graded inequality in Annihilation of Caste (1936): it coheres not through consensus but through the interest each stratum has in those below it. Difference-talk loses that.
- The muddle is in the middle. At the poles ranking is neither disputed nor merely symbolic: Telangana’s backwardness index ran from the most deprived Scheduled Caste groups to a dominant landed caste at the opposite extreme, with roughly seventy per cent of measured inequality lying between the broad social groups rather than within them.
- Satish Deshpande‘s warning applies: when the privileged declare themselves casteless, difference becomes an alibi and hierarchy vanishes from view exactly where it operates.
Conclusion
Gupta is right that India never possessed one agreed ladder, and that treating the Brahmanical account as the Indian account was a methodological error. But discreteness explains caste’s form while hierarchy explains its consequences: castes are discrete units whose rankings are muddled in the middle and brutally clear at the edges.