Geoforum 57 (2014) 192–204

Contents lists available at SciVerse ScienceDirect

Geoforum

journal homepage: www.elsevier .com/locate /geoforum

Canal irrigation and the hydrosocial cycle

0016-7185/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.geoforum.2013.05.011

E-mail address: pm35@soas.ac.uk1 ‘Critical’ (irrigation studies) here refers to approaches explicitly addressing the

social relations of power that are part of irrigation and which have a normativeconcern about its often problematic equity/poverty, democracy, and sustainabilitydimensions.

2 In irrigation science ‘structures’ is the generic technical term for built dwater control systems (like discharge measurement structures, division stoutlet structures, escape structures, etc.). It needs to be distinguished fromas used in ‘structure-agency’, and the more general use of structure ascomposition and pattern of organisation of objects and processes (having strubeing structured). All three meanings are used in this paper.

The morphogenesis of contested water control in the TungabhadraLeft Bank Canal, South India

Peter P. MollingaDepartment of Development Studies, School of Oriental and African Studies (SOAS), Thornhaugh Street, Russell Square, London WC1H 0XG, UK

a r t i c l e i n f o a b s t r a c t

Article history:Available online 14 June 2013

Keywords:IrrigationHydrosocial cycleMorphogenesisTechnologySpaceTimeIndia

Using South Indian large-scale surface irrigation as a case, this paper combines emerging interdisciplinaryconceptualisation in resource geography of the hydrological cycle as a hydrosocial cycle with Archer’s the-orisation of society’s structure-agency dynamics as a morphogenetic cycle. Characteristic of large scalecanal irrigation are a pronounced spatiality of social process, and a strongly cyclical nature of social inter-action around water through seasonality and rotational supply, framed by irrigation infrastructure that isboth grid and subject of water resources management practices. This allows an investigation of howhuman agency as the animator of structural elaboration reproduces and transforms a hybrid andmulti-scale water control system, thus establishing a ‘hydromorphogenetic’ cycle of unequal irrigationwater distribution. The detailed account of irrigation practice provides caution against simplified inter-pretations of dam + canals based irrigation as abodes of green revolution capitalist farming, and of theobjectives of neoliberal irrigation reform policy. It is, lastly, suggested that the hydrosocial relations focusproduces new insights and questions for irrigation studies, but that complexity and emergence ratherthan hybridity are the key analytical challenges.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction: socio-technical systems and hydrosocialrelations

In irrigation studies, conceptualisation of irrigation systems ascombined physical and human socio-technical systems dates fromthe 1980s (Uphoff, 1986; Vincent, 1997). The interest in ‘hybrid’,socio-technical understanding of irrigation derived from the per-ceived poor performance record of irrigation interventions in thecontext of international development assistance and nationalplanned development – both in mainstream and critical observa-tion of the sector.1 For large-scale formally government managedirrigation, a well known illustration is Uphoff’s suggestion that thelevels of primary, secondary, and tertiary canals of surface irrigationsystems do not only have hydraulic significance for the physical con-veyance of water, but also constitute social spaces for irrigationmanagement activities as contested by irrigators and governmentofficials (Uphoff, 1991: 33). For smaller-scale farmer managed irriga-tion Coward has shown that the creation and upkeep of irrigation

infrastructure go hand in hand with the (transformation of the) so-cial relations: they co-evolve and are each other’s expression as‘hydraulic property’ (Coward, 1990).

Theorisation of the socio-technical nature of irrigation pro-cesses received a boost with the advent of the ‘social constructionof technology’ (SCOT) perspective (Pinch and Bijker, 1984). Theori-sations from this SCOT, and later ANT (Actor-Network Theory) lit-erature, mostly focusing on western societies, and without specificinterest in irrigation or water resources, could be usefully trans-posed to the study of irrigation infrastructure. The social construc-tion of irrigation artefacts, notably division structures,2 the devicesconnecting Uphoff’s levels and embodying Coward’s hydraulic prop-erty rights, has been a central theme (Mollinga, 2013). The conceptof ‘water control’ has posited that technical/physical, organisa-tional/managerial and socio-economic/political control of water areinternally related (Bolding et al., 1995). Methodologically, this

evices inructures,structureenduringcture, or

P.P. Mollinga / Geoforum 57 (2014) 192–204 193

current of work has articulated ‘technography’ as a method for inter-disciplinary irrigation studies (Bolding, 2004).

Conceiving the hydrological cycle as a hydrosocial cycle is an ef-fort to avoid the pitfalls of reductionist and depoliticised water re-sources management analysis.

‘‘In a sustained attempt to transcend the modernist nature –society binaries, hydro-social research envisions the circulationof water as a combined physical and social process, as a hybrid-ized socio-natural flow that fuses together nature and society ininseparable manners (...). It calls for revisiting traditional frag-mented and interdisciplinary approaches to the study of waterby insisting on the inseparability of the social and the physicalin the production of particular hydro-social configurations (...).’’(Swyngedouw, 2009: 56)

In water studies binarism is clearly visible in early conceptionsof the hydrosocial cycle like that of Falkenmark (1997), where thesocial and the material appear in conceptual models as separateboxes, linked with arrows.3 What such modelling is unable to cap-ture is exactly hybridity. In contrast, hydrosocial analysis conceivesof the relation as internal and infested with social power (Swynge-douw, 2009). The hydrosocial perspective also suggests that ‘scalarpolitics’ is a key element; scale is not given but politically con-structed (Swyngedouw, 2007).

The programmatic announcement of ‘hydrosocial research’ as anew perspective focusing on analysis of the ‘‘intricate and multidi-mensional relationships between the socio-technical organizationof the hydro-social cycle, the associated power geometries thatchoreograph access to and exclusion from water, as well as the un-even political power relations that affect flows of water’’ (Swynge-douw, 2009: 59) for many a critical irrigation scholar may soundlike sticking a new label on already existing research. However,much critical irrigation research has remained irrigation systemconfined, taking the boundaries of the infrastructural systemsand the communities using and managing them as defining the ob-ject of research.4 The emerging hydrosocial research perspective canbe used to bring together in a single framework the different scalesand dimensions of the socio-technicality and hydrosociality of irriga-tion. It resonates with the increased (largely policy-driven) interestin irrigation studies to ‘scale up’ analysis from the system level tothe level of the basin (Wester et al., 2003), and is able to provide apolitical economy and political ecology infusion into that research(cf. Lebel et al., 2005 on scalar politics in the Mekong basin). Simul-taneously the detailed socio-technical analysis of irrigation studiescan help to elaborate the general notion of hydrosocial relations.By unravelling the contestations ongoing within irrigation projects,it can add to the space and landscape focus of hydrosocial analysisan emphasis on time and technology. The latter is virtually absentin political ecology.5 It can also nuance all too sweeping analysesof the role of dams + canals for irrigation in the project of stateand/or market-led modernisation and assessments of neoliberal irri-gation reform.

This paper, thus, seeks to combine ‘hydrosocial analysis’ andthe socio-technical study of irrigation. It does so in three steps,and by investigating one particular case, unequal water distribu-tion in the Tungabhadra Left Bank Canal irrigation systemin South India (Mollinga, 2003). First it discusses in general

3 I thank Susanne Mauren for collecting conceptual models of the hydrosocial cycle.4 Theorisation of irrigation as a ‘large technological system’ in SCOT/ANT mode (cf.

Hughes, 1987) has, to the knowledge of this author, not been undertaken.5 Political ecology has focused on knowledge rather than technology, while water

has not been a particularly popular topic in such research (Budds, 2009; also seeLinton, 2008; Shah, 2008; Trottier and Fernandez, 2010).

theoretical terms how Archer’s (1995) morphogenetic approachresonates with the endeavour of hydrosocial analysis, providingthe general framework for investigation of the Tungabhadra case.In a second step the paper looks at irrigation ‘from without’,interpreting the meaning of the ‘slicing off’ of irrigation fromthe hydrosocial cycle. It is shown that the storage and diversionof river water for the productive purpose of irrigation is an act ofpower, a strategy of state rule, and an effort to singularise the va-lue and meaning of water to serve particular trajectories of polit-ical economic development. Third, the paper looks at theirrigation system ‘from within’ along the axes of technology, timeand space. It provides an analysis of the hydrosocial dynamicswithin the system that produce a recurrent pattern of unequalwater distribution,6 and shows that the project of state rule andpolitical economic development is far from accomplished andinherently contradictory. The paper concludes with reflecting onhow hydrosocial analysis can be elaborated beyond confirmationof the fact that, indeed, water resources management structuresand practices are ‘hydrosocial relations of power’.

2. Hydrosociality, structure-agency and morphogenesis

The basic theoretical puzzle of hydrosocial analysis is to capturethe ontological complexity of water resource management situa-tions, as being structured, stratified and heterogeneous, and in crit-ical perspectives, contested, systems and processes, animated byconfigurations of actors networked in variety of social relationsof power that shape their individual and collective agency. Concep-tualisations of the circulation of water, as for example in models ofthe hydrological cycle, need to be combined or integrated withconceptualisations of social dynamics, as for example, and founda-tionally, in models of structure-agency dynamics.

The hydrological cycle as understood in hydrology is a circu-lation process in which water moves through different phasesand ‘compartments’. Details are too well known to bear repeat-ing – the intricacies of the circulation have been documentedand modelled in great detail. With the advent of GeographicalInformation Systems, spatially explicit modelling has becomepossible (Sakthivadivel, 2006). Combined with a river basin his-tory perspective, trajectories of hydrosocial evolution of basinstructure may be described (for the Krishna river in South India,see Venot, 2009). The ‘social (re)construction’ of the hydrologicalstructure and stratification of the water circulation system canthus be mapped, modelled and understood in relation to societaldynamics, mediated by technology and institutions. Time plays arole in such trajectories as the (short term) yearly climatic cycleand the (long term) gradual change of the hydrosocialconfiguration.

This imagery closely resonates with that of Archer’s (1995) pic-turing of the morphogenetic cycle. She uses the term morphogen-esis to refer to the way societal structuration and stratificationdevelops through the interaction of agency and structure.7 AgainstGiddens (1984), for whom structure and agency are inseparable andtwo sides of the same coin, she argues for analytical dualism in

6 The Tungabhadra irrigation system exhibits the classical head-tail pattern owater distribution, in which those located upstream along a canal (at its headappropriate water beyond their entitlement, depriving those located further down-stream along the canal (towards its tail). In the perspective of this paper ‘locationaadvantage’ (implying queuing for access) is an emergent property, constituted by acomplex hydrosocial structure, that needs to be explained, rather than a geographica‘given’.

7 And morphostasis in case of reproduction.

f)

l

l

194 P.P. Mollinga / Geoforum 57 (2014) 192–204

which structure precedes agency in time, and time is needed for‘structural elaboration’, which happens in cycles (Archer, 1995,chapters 4–6).8 In this critical realist ontology (Sayer, 1992) this evercontinuing process constitutes a stratified reality,9 in which, accord-ing to Archer, three types of structures have three kinds of emergentproperties: structural, cultural and agential (Archer, 1995, 175 ff.).Structural emergent properties are characterised by their primarydependence of material resources, physical or human. Culturalemergent properties are the items in a society’s ‘propositional regis-ter’, the cultural system of a society. The features of agential rela-tions represent the third type of emergent properties, people’semergent properties ‘‘with their two defining features – that is theymodify the capacities of component members (affecting their con-sciousness and commitments, affinities and animosities) and exertcausal powers proper to their relations themselves vis-à-vis otheragents or their groupings (such as association, organization, opposi-tion and articulation of interests).’’ (Archer, 1995, 184)

Archer thus pictures society as consisting of a heterogeneousset of evolving structures that is continuously reworked (elabo-rated) by human action, leading to cyclic (in the sense of episodic)change of these structures and their emergent properties. ThoughArcher’s morphogenetic perspective is not directly concernedabout material-social hybridity in the way hydrosocial analysisis, the analytical dualism of structure and agency linked over time,and the distinction of different types of structures and emergentproperties much more easily allows the inclusion of, for example,the features of canal infrastructure as a component of social anal-ysis than Giddens’ approach, where structures are only ‘instanti-ated’ when agency is actively deployed. The challenge is toexplain, by providing ‘analytical histories of emergence’, the con-figuration of heterogeneous objects, structures, and their emergentproperties (Archer, 1995, 324 ff.).

The circulation of water through the hydrological cycle almostseems an archetypical example for Archer’s framework: time cy-cles are ‘naturally’ given, the (spatial) features of the landscapeare the medium of the terrestrial part of the cycle, both of whichprovide the basis for building complex human societies by(re)shaping the time, space and other dimensions of the circulation(and value and meaning) of water through sets of technologies andinstitutions. If it wasn’t so cumbersome, the concept of hydrosocialcycle could be usefully rephrased as hydromorphogenetic cycle.

This formal and abstract conceptualisation is elaboratedthrough an analysis of the Tungabhadra Left Bank Canal irrigationsystem in the next two sections.

3. Canal irrigation from without: the political economy of‘slicing off’

The Tungabhadra Left Bank Canal irrigation system is located ininterior South India in the State of Karnataka (Fig. 1). The reservoir-fed irrigation system on the left bank of the Tungabhadra river,operational since 1953, has a planned irrigated area of about240,000 ha. The main canal running from the reservoir and ending

8 Archer (1995) critiques three types of conflation (one dimensional theorising) ofstructure and agency. In downward conflation structure determines agency, inupward conflation agency determines structure. Giddens’ (1984) theory of structur-ation is a case of central conflation. ‘‘[E]ndorsement of [structure and agency’s]mutual constitution precludes examination of their interplay.’’ (Archer, 1995: 14)‘‘Lack of ontological depth’’ is the central fallacy of Giddens’ type of ‘‘elisionistthinking about society’’. (Archer, 1995: 133)

9 The approach is committed to a framework that ‘‘incorporates (a) pre-existentstructures as generative mechanisms, (b) their interplay with other objects possessingcausal powers and liabilities proper to them in what is a stratified social world, and(c) non-predictable but none the less explicable outcomes arising from interactionsbetween the above, which take place in the open system that is society’’ (Archer,1995, 159)

close to the district capital of Raichur town has a length of about227 km. An approximately similarly sized irrigation area was cre-ated on the river’s right bank, having two main canals. These arenot discussed.

3.1. The establishment of the irrigation system

The construction of a dam across the Tungabhadra river createda reservoir that, with an accompanying system of canals, allowedelectricity generation and intensification of agriculture throughirrigation.10 The government agencies who built it thus pursued eco-nomic growth and development objectives, in combination withwelfare objectives – the latter related to the fact that the Tungabha-dra irrigation system was conceived as a protective irrigation sys-tem, as discussed below. The storing and diversion of river waterthat the grafting of the project on the landscape amounted to, im-plied not only a physical change in the hydrological cycle by alteringthe time and space contours of water availability, but also an inter-vention in the meaning and value of the river water. The project, likemany dam + canals projects, was an effort at ‘singularising’ themeaning of water: storage and diversion means to reserve the waterfor exclusively productive purposes. The socio-ecological meaning ofthat water in terms of supporting a river ecosystem was ignored, aswell as the livelihoods of people depending on the river eco-hydrol-ogy. Along the river many communities had livelihoods based on riv-er fisheries, which seems to have severely declined, to almost non-existence in the zone downstream of the Tungabhadhra dam.11

The effects on the flora and fauna of the Tungabhadra river valleycan only be guessed at, and similarly unknown are the ecological ef-fects of the subsequent intensification of agriculture, for instancethrough non-point pollution of water by agricultural chemicals (cf.Gooch et al., 2010, chapter 10).

This lack of knowledge on eco-hydrological effects, and the con-comitant livelihood impacts, reflects the ease with which this damand irrigation system could be constructed, as compared to manycontemporary dams. This is no doubt partly due to the centralisedimperial and feudal modes of governance of the two riparian statesin the colonial period (the directly ruled Madras Presidency andthe formally autonomous ‘princely state’ of the Nizam’s Domin-ions), and the unquestioned legitimacy of the first post-indepen-dence ‘planned development’ governments of India. It also has todo with the marginality of the region in which the project is lo-cated. For Madras Presidency the area was a remote and marginalarea upstream of the prominent agricultural area of the Krishnadelta (a reason why within the Madras Presidency governmentthere were always forces opposing construction on the ground thatthe project would reduce water supply to the delta). For the Ni-zam’s Dominions the Raichur District was not a core agriculturalor economic area either – the dam site chosen would even inun-date part of the feudal estate of the Salar Jung family, a prominentfamily in the Nizam Dominians’ ruling elite. This was one, thoughnot the main, reason why project construction was a political hotpotato for the 80 years between the 1850/1860s first formula-tion/design by Sir Arthur Cotton as part of his interlinking of Indianrivers plan, and the 1940 agreement of the riparian governments tobuild the system.12

State-driven canal irrigation development in India is based inthe contradictions of colonial rule, and the reworking of these after

10 The hydropower dimension of the dam + canals is left aside – the irrigationfunction was and is predominant.

11 There is virtually no research-based evidence on this. The conclusion is based oninterviews with members of a fisher(wo)man caste in one of the study villages.

12 Cotton’s interlinking of rivers plan is discussed in the Indian Irrigation Committee1901-1903 report (IIC, 1903), which includes a map with the right bank ‘Bellary’ canaland the Kurnool-Cuddapah Canal, as part of the ‘Tungabhadra-Kistna Project’. Themap is reproduced in Mollinga (2003: 102). Also see MICC (1859, 1867).

Fig. 1. Location of Tungabhadra Left Bank Canal irrigation project.

13 Water allowance is the amount of water envisaged for irrigating a piece of land,usually expressed as the continuous flow (l/s ha) needed over the length of thegrowing season. The South Asian term ‘duty’ is the inverse of this, expressing theextent of land to be irrigated with a unit flow (usually expressed as acres/cusec).

14 The earliest comprehensive statement on the inherent problems of commodifi-cation in irrigation as caused by the character of water and water infrastructure isMoore (1989).

P.P. Mollinga / Geoforum 57 (2014) 192–204 195

Independence. The contradiction of the imperial pursuit of eco-nomic gain while maintaining political control and stability, in irri-gation translated in the articulation of two types of canal irrigationin the second half of the 19th century. ‘Productive’ irrigation sys-tems were systems that generated sufficient revenue (expressedas a percentage of total financial outlay for construction), while‘protective’ irrigation systems stayed below the revenue thresholdbut were still considered, and occasionally constructed, for protec-tion against crop failure, to avoid the social unrest and misery asso-ciated with famines, and to reduce the costs of famine relief (fordetailed discussion, see Mollinga, 2003, chapter 3). In the colonialperiod few protective systems were built; the revenue consider-ation tended to get preference. The terminology survived indepen-dence (GOI/MOIP, 1972); the logic of protective systems was nowargued on rural development and poverty alleviation grounds.Many protective irrigation systems were built in the first decadesafter independence, the Tungabhadra irrigation system was oneof them.

The productive/protective distinction was not only financial. Italso translated into specific agricultural and infrastructural charac-teristics. Productive systems mainly aimed at the cultivation ofcommercial crops – which is what made them remunerative. Pro-tective systems mostly aimed at irrigation of subsistence foodcrops, notably, in South India, sorghum and millet. Productive sys-tems were often designed for intensive irrigation, i.e. aimed at thesupply of full water requirements to crops, in South India oftenrice. Protective systems were often designed for supplementaryirrigation, i.e. for only a part of the full crop water requirements,of low-water consuming crops. Protective irrigation also aimed tospread water thinly over as large an area/number of villages, intune with its famine, social stability and poverty alleviation objec-tives. In protective irrigation systems water is ‘scarce by design’

(Jurriëns and Mollinga, 1996), with low water allowances (or in In-dian terms high irrigation duties13) for the planned irrigated area.

In protective irrigation systems like the Tungabhadra LBC at-tempts at commodification of and accumulation through irrigatedagriculture have a contradictory history. In several ways water hasbeen an ‘uncooperative’ commodity – to transplant Bakker’s (2003)phrase from UK urban water to Indian agricultural water.14 Welldocumented, notably for the Nira Left Bank Canal in present Maha-rashtra and the Kurnool-Cuddapah Canal in present Andhra Pradesh,is the lack of interest of South Indian farmers in utilising the irriga-tion services provided by the British rulers and engineers in the sec-ond half of the 19th century to secure local food production(Attwood, 1987; Bolding et al., 1995). The reasons were located inthe character of the soils (highly water-retentive vertisols that be-come waterlogged when irrigation is followed by rainfall, suffocatingcrops), and the response of local crop varieties to irrigation (mainlyvegetative growth without increased grain production). This issuepresented itself also in the early years of Tungabhadra LBC operation.Local farmers were hesitant to irrigate their sorghum, millet and cot-ton crops fearing they would lose them through over-watering, anddestroy the quality of the soil in the process, even when they had thefinancial means to do the land preparation and levelling required foreffective irrigation. It took the immigration of experienced ricefarmers from coastal Andhra Pradesh (Upadhya, 1988) to show that

196 P.P. Mollinga / Geoforum 57 (2014) 192–204

intensive irrigation of vertisols (‘black cotton soils’) was possible.15

Through a trial-and-error based innovation process animated by astrong pioneer spirit, the fact that rice can be grown in submergedconditions, the advent of the green revolution, and government pro-curement policy, rice cultivation became very productive and beforelong local farmers also started practising it. The government manag-ers of the Tungabhadra LBC allowed intensive rice cropping in the1960s and 1970s because full reservoir capacity was available foran only partly developed planned irrigated area16 and because thenational ‘Grow More Food’ campaignrequired intensification of agri-cultural production. Thistriggered an agricultural ‘rice’ boom in thedistrict, which counted more than 100 rice mills in 1991–1992.The uncooperativeness of water in terms of agricultural growthwas thus partly overcome – through a combination of concerted ef-fort and circumstance.

This agronomic and economic success greatly intensified un-equal water distribution, as that success was dependent on cropand space–time changes in the equitable design of the protectiveirrigation cycle. Inequality intensified because widespread cultiva-tion of rice was allowed, meaning a shift from design low waterconsuming crops to a high water consuming crop, and double sea-son irrigated cropping was allowed as against the protective designof a single irrigated crop per year on each individual piece of land.Instead of supplementary irrigation, irrigation to full crop waterrequirements came to be practised. As compared to an ‘average’design localised irrigated plot, the shift to rice double cropping im-plied a multiplication of water use in the order of 4–5 times. Thedramatic geographical concentration of water as a result of thisis immediately visible to any visitor to the region who makes theeffort to drive down, for instance, a secondary canal.

Based on this account, the design and construction of irrigationprojects like the Tungabhadra project could be interpreted withsome validity as top-down, and perhaps violent, state-led acts of‘modernisation’ that ‘freed up’ water as an input for new processesof accumulation of a class of (newly emerged) capitalist farmers(on the role of irrigation capitalist agricultural development in In-dia see Thorner and Thorner, 1962; Byres, 1981; Jairath, 1985; Gor-ter, 1989). Creation of irrigation systems by government thenbecomes understood as a ‘state simplification’ (Scott, 1997), an at-tempt to reduce the complexity of the meaning and embeddednessof water (use) in order to make it amenable for the pursuit of state-defined objectives by selected local actors. Critiques of dam-basedirrigation development have tended to adopt interpretations of thisnature (Mollinga, 2010).17 This interpretation, however, risks tooverlook the continued (p)relevance of the private/common goodcontradiction, and the complications involved in ’slicing off’ irriga-tion from the broader hydrosocial cycle.

3.2. The structural elaboration of a contradiction

The protective design and policy paradigm imposed the task toration water on colonial government managers, as well as on con-temporary government managers, that is, to discipline irrigators toaccept supplementary levels of irrigation water. From a govern-ment perspective the economic logic of spreading water thinly isthat it maximises total agricultural output, and thus makes the

15 Debate on the irrigability of these soils ranged from the late 19th century to atleast the 1970s (Venkata Ramiah, 1937; UAS, 1973). The debate seems to have dieddown with the practice of effective intensive irrigation of these soils gettingestablished.

16 The last secondary canal was constructed in 1968, 15 years after the first waterreleases from the dam. Land preparation to make individual plots suitable forirrigation was an even much longer-drawn process (see below).

17 For instance, Morrison (2010: 182) quotes Goldsmith (1998) as stating ‘‘Modernirrigation systems in tropical areas are, almost without exception, social, ecological,and economic disasters.’’

largest contribution to ‘national development’. It has also been ar-gued that protective irrigation generates more agricultural work-ing days, that is, employment (Dhawan, 1988, 1989; Mitra, 1986,1987). Further, welfare/equity considerations have carried politicalforce in both colonial and post-independence periods, even withthat logic being partly utterly pragmatic, deriving from politicalstability and constituency based politics considerations. However,this differentially constituted spreading logic for the common goodcontradicts the individual farmer logic of maximising of agricul-tural output per unit area, that is, his/her farm, which has equallybeen carrying considerable political force in both the colonial andpost-independence period.

Different rationing approaches were followed in different re-gions of India, involving different institutional and infrastructuralarrangements (Wade, 1976; Attwood, 1987; Bolding et al., 1995).The northern part of India adopted the warabandi system of area-based time-shares, with a semi-modular distribution technology,the so called Crump outlet.18 In the present day Maharashtra partof Western India the introduction of the so called ‘block system’was attempted in the early 20th century. It involved permission togrow sugarcane (a water intensive commercial crop) on one-thirdof the land, with the other two-thirds protectively cultivated withfood crops like sorghum, the main subsistence food crop of interiorSouth India. The block system design involved this new croppingpattern, packaged with institutional elements (bulk delivery of wateragainst volumetric payment to groups of users) and a technical inno-vation (a modular outlet structure that could measure the volumesdelivered and would be tamper proof), for which design competi-tions were held (for details see Bolding et al., 1995).

The present South Indian states of Karnataka, Andhra Pradeshand Tamil Nadu have harboured the strongest state attempt to reg-ulate irrigation water use for maximising aggregate production inthe form of ‘localisation’. Localisation is a form of land use planningavant la lettre. It was designed in the 1930s and 1940s as a mirrorof canal irrigation design practice, which has to assume croppingpatterns on certain extents of land to calculate necessary canalcapacities (and thereby construction costs). In reverse this be-comes prescriptive land use planning.19 In the Tungabhadra LBCthis took the form of the publication in the State Gazette of lists ofsurvey numbers (cadastral units) with irrigation entitlements, de-fined as permission to irrigate in either the kharif (monsoon) orthe rabi (post-monsoon) season, with the type of crop allowedspecified.

The assumption, apparently, was that state agencies would beable to implement this, and distribute water according to the local-isation pattern in both space (survey number) and time (season).Non-adherence to this prescribed pattern was made a violation,with fine levels defined, under the Irrigation Act as UnauthorisedIrrigation (irrigating outside the prescribed area) and Violation ofCropping Pattern (irrigation of other, notably more water consum-ing, crops than prescribed). When intensive irrigation (double ricecropping) won the day, as explained above, many farmers went tocourt – till the early 1980s thousands of writ petitions were regis-tered at the Karnataka High Court. After that the belief in the pros-pects of legal action seems to have waned.

Government of Karnataka committees deliberated on how tobetter implement localisation well into the 1970s (GOMYS/DOA,1968, GOKAR/PD, 1976). However, in the late 1970s/early 1980slocalisation practically became a dead letter for day-to-day

18 For modular outlet structures neither upstream nor downstream canal waterlevels determine discharge; for semi-modular outlets only the upstream water leveldoes; for non-modular outlets both upstream and downstream do (for hydraulicdetails see Mahbub and Gulhati, 1951).

19 For discussion see Mollinga, 2003, chapter 3; the 1956 Hyderabad State Rules forlocalisation are reproduced there.

20 See footnote 2.

P.P. Mollinga / Geoforum 57 (2014) 192–204 197

irrigation management, even when, till today, water release sched-ules (for different levels of canals) are calculated based on the offi-cial localisation pattern, and unauthorised irrigation and violationsof cropping patterns continue to be administratively recorded, andthe ensuing fines calculated. Efforts to implement the ‘spreadinglogic’ of localisation shifted to other policy instruments.

The 1980s saw a shift to water management improvement byorganisation of water users in associations, first through the Com-mand Area Development (CAD) programme, later under the um-brella of Participatory Irrigation Management (PIM), includingpiloting by Non-Governmental Organisations. These efforts havebeen extensively researched (Joshi and Hooja, 2000), basicallyshowing their lack of effectiveness (for a summary statement, seeMollinga et al., 2007). The two main sticking points are (a) theunwillingness of government (including both the IrrigationDepartment and elected parliamentarians) to devolve power overbudget allocation and water allocation to irrigator associations,and (b) diverse interests in (un)equal water distribution amongthe farming community. Perhaps counter-intuitively, the adventof (neo)liberalisation seems to have brought new dynamism to‘irrigation reform’. Under the Chief Ministership of Chandra BabuNaidu, seen and projected as a neoliberal ‘champion’ (Mooij,2007) Andhra Pradesh adopted and implemented the Andhra Pra-desh Farmer Management of Irrigation Systems (APFMIS) Act in1996–1997. This is the most far reaching effort in India so far atlegislating irrigation reform through devolution of power to irriga-tor organisations. It has served as a model Act for several otherstates. Reforms aim at achieving financial sustainability of the gov-ernment irrigation management enterprise through ‘cost recovery’as well as at more equitable water distribution, which would en-hance the revenue base of the irrigation system and its financialsustainability. As another instance of neoliberal thinking, in recentyears the states of Maharashtra and Andhra Pradesh have been atthe forefront of establishing Regulatory Authorities for the watersector, the concrete effects of which on irrigation management re-main to become manifest.

The situation is thus more complex and contradictory than asingular ‘victory’ of the rich/larger farmers’ class power interpreta-tion suggests. The post-construction story of Tungabhadra LBC eco-nomic and socio-political transformation harbours a complexdynamic of capitalist accumulation in agriculture, with irrigationmanagement getting ensnarled in the post-independence Indianpolitics of ‘competitive populism’, both underpinning a changingrole and image of large-scale surface irrigation systems as instru-ments of development. This is a story of the changing fortunes offarmers and farming in the post green revolution area, includingthe rise and decline of middle and large farmers’ class power(Brass, 1995; Nadkarni, 1987), of a series of institutional interven-tions (partly internationally supported) attempts at enhancing irri-gation system performance through ‘water user participation’, andof the logic of their half-heartedness in India’s competitively pop-ulist democracy and system of ‘political and administrative corrup-tion’ characteristic of the public works bureaucracy and the polityand administration in general (Wade, 1982), and a story of new ef-forts at institutional reform under neoliberalism. The story also in-cludes elements such as the impacts of economic growth in theregion, and the contestation and partial renegotiation of the pro-ductive singularisation of the meaning of diverted water.

For the theoretical purposes of this paper a narrower focus thanthis monograph-wide canvas suffices. By ‘zooming in’ on the con-crete water distribution dynamics in the Tungabhadra LBC, a spec-ification of the hydrosocial conceptual apparatus is undertaken.This more limited focus will turn out to be more than complexand empirically rich enough to suggest how analysis of water re-sources management in terms of a ‘hydrosocial cycle’, and ‘hydro-social relations of power’ can be usefully linked with Archer’s

(1995) ‘morphogenetic approach’ to structure-agency dynamics,while giving materiality its due. In the process the political eco-nomic and socio-political dynamics of recent decades will be illus-trated at case-level.

4. Canal irrigation from within: technology, time, and space inunequal water distribution

To show how the practice of concentrated unequal rather thanthinly spread equal water distribution is produced, and contested,on a day-to-day basis, I analyse water management practices in theTungabhadra Left Bank Canal along three axes. First I show how thenature of the technical infrastructure configures unequal out-comes. Second I look at time: the social processes of (unequal)water distribution derive their institutional specificity from thetime cycles of irrigation and thus constitute hydrosocial cyclesand relations. Third, (unequal) water distribution has strong spatialspecificity, suggesting that the social differentiation of agriculturalproducers that shapes and is shaped by irrigation practices has anirreducible spatial component.

4.1. The technical configuration of unequal water distribution

The low allowance/high duty ‘scarcity by design’ feature of pro-tective irrigation that embodies the objective to spread waterthinly and widely, results in long canals and large spatial extentof the planned irrigated area. In addition to the ‘hard to police’characteristic inherent to spatial spread, technical features thatconfigure water management behaviour in the Tungabhadra LBCirrigation system include the levelled structure of the canal hierar-chy, non-modular water division structures at canal bifurcations,and the absence of flow regulation facilities in the canal system.

4.1.1. Canal levels and outlet structures20

The process of water distribution that determines the practicalfate of localisation and participatory management efforts is mate-rially configured by the organisation of the canals in a hierarchy oflevels. The levels are a single main canal with a length of 227 km,over 80 secondary canals with lengths up to several tens of kilome-tres called distributaries (regularly with sub-distributaries branch-ing off) and the level of the tertiary (or farmer field) canals. Themain and distributary canals are formally the domain of the Irriga-tion Department managers, the field level canals the domain ofgroups of farmers (the local irrigation units measure several tensup to 100 ha, with several tens of farmers having land within sucha unit). The different canal levels are connected by outlet or divi-sion structures. At each bifurcation point a masonry or concretestructure can be found with steel gates, at least as per design.These structures serve to determine how much water goes where.The outlet structure that links the (sub-)distributary with the localirrigation unit is particularly important. Its operation directlydetermines the quantity and timing of irrigation water for groupsof farmers (and thereby the pattern of (in)equality), and it is thephysical interface between the domain of government manage-ment and the domain of farmer management. For the governmentit is the final point of control for supply and rationing, for farmersthe point of access to a government controlled resource.

The structures at the bifurcations of canals are not only instru-ments for water distribution activities, but their features are alsothe subject of that interaction. The latter is illustrated in Fig. 2for a three kilometres long subdistributary in the D24 canal area(see Fig. 1). Fig. 2 shows how the technical features of the outletstructures systematically change going from the upstream to the

Fig. 2. Outlet structures along a subdistributary canal.

198 P.P. Mollinga / Geoforum 57 (2014) 192–204

downstream side, from efforts to control excess abstraction by hea-vy, not easily damaged structures, to abandonment of damagedstructures, with the original standard design seen in Section 2.21

For South Indian protective irrigation I have not found evidencethat the technical design of the division structure was activelythought about in relation to the rationing principle, unlike in theother two regions referred to above. Non-modular outlets that donot allow any measurement or assessment of actual water flowswere chosen, possibly as a South Indian path dependent choicefrom the historically dominating rice irrigation schemes (in deltasand otherwise) where there is water abundance by design, notrequiring rationing in the way the upland schemes designed asprotective irrigation schemes do. Moreover, in the 1930s/1940sprotective irrigation design was a relatively new concept for thisregion, certainly for the Nizam’s Dominions in which the Tung-abhadra Left Bank Canal area then fell. For the right bank canalsof the Tungabhadra system, coming under the Madras Presidencybefore independence, a ‘melons on a vine’ (Nickum, 1977) systemwas designed in which whole local units of irrigation were eitherfor kharif irrigation or for rabi irrigation, so that the governmentcould close units off for irrigation at a single point, the outlet struc-ture – by cementing these in the off season. In the TungabhadraLeft Bank Canal the survey numbers are spread – one irrigationunit can have cadastral units permitting irrigation in both seasons.It is unclear how the government anticipated technically managingthe season-wise distribution thus prescribed, a lack of clarity thatfacilitates excess appropriation of water.

4.1.2. Absence of flow regulation structures and intermediate storageAnother relevant technical design feature is that the system has

no facilities for flow regulation and storage within the canal sys-tem. Once water has entered the canal system at the reservoir, ithas to flow through the system. It cannot be slowed down orstored, it can only be directed to different places. The size of thesystem and its spread over a very large area combined with this

21 The stability of this particular configuration has been documented for a periodexceeding 15 years. The ‘structural elaboration’ from the original uniform designhappened before the first fieldwork in 1991.

lack of regulation facilities means that the possibilities for flexibleforms of management responding to local demands and needs arehighly circumscribed; the system is designed for stable and contin-uous flow. This is in tune with the protective objective of thinlyand widely spread supplementary irrigation, but not with the ac-tual use of the system.

These infrastructure design features configure a series of arenasand locations for water distribution interaction with large spatialextent that is difficult to police, and a ‘top-down’ system with lim-ited options for water flow regulation and flexible management. Asa totality, the irrigation system infrastructure creates a structuredpattern of dependency among individuals, groups of farmers, vil-lage communities and administrative sections of the IrrigationDepartment that manage different parts of the system. The struc-ture is that of a complex set of queues along canals, in which waterflows in one direction and those located upstream having a strate-gic, locational advantage, making skewed water distribution highlylikely. Since the start of ‘participatory approaches’ under the CADprogramme, irrigation reform, as the contemporary state effort todiscipline irrigator behaviour, has been conceived as institutionalinnovation primarily or only: there is no conscious considerationof the infrastructure requirements for new governance and man-agement regimes.

4.2. The rhythms of irrigation: contestations and institutional forms

Rainfed agriculture in this region has two seasons, the monsoonor rainy season (kharif), starting from June and lasting into Septem-ber–October, and the post-monsoon season (rabi), starting fromSeptember–October lasting till January–February. Exact seasontimings depend on the crop grown and the timeliness of rainfall.The two seasons thus overlap in time, but usually not in space. Inrainfed agriculture, a particular piece of land would normally beplanted either with a kharif or with a rabi crop. Given that averageyearly rainfall is around 600 mm, only a single crop can be grownon a piece of land.

Canal irrigation changed all this. Though localisation envisagedirrigation of part of the area in kharif and part in rabi like in rainfedagriculture, and of similar crops, this could not be implemented. In

22 The force of the original localisation ruling shows in the settlement pattern of themigrant farmers. In the 1950s and 1960s they preferentially settled in areas localisedfor rice – of which there was a small percentage of 9% in the official, localisedcropping pattern. Rice areas were localised mostly in low lying, valley areas, on thereasoning of clayey soil prevalence in such locations and better water availability.This explains early migrant settlement in what are now tail-end areas. Later settlers,observing the lack of force of the localisation policy, purchased land in upstreamlocations directly.

P.P. Mollinga / Geoforum 57 (2014) 192–204 199

the early years of the canal’s operation in the 1950s and 1960swater was abundant for those willing and ready to irrigate, asthe canal system and land development for irrigation were ongo-ing. In the early years the main canal supplied water for elevenout of twelve months, with one month closure for repairs in thehot summer period. Availability of canal water allowed two con-secutive crops on a piece of land, provided access to the canalwater could be obtained. Because in the 1960s India had high lev-els of food insecurity, maximum use of water was allowed andintensive irrigation spread rapidly. The irrigation system createda new seasonality, that of two consecutive irrigation seasons. Acomplex mix ensued of the rainfed and canal seasons as manycrops grown had desirable planting dates in relation to other cli-mate related factors (for example temperature influencing yields).The issue gained increasing importance with the completion of thecanal system and increasing acreages coming under irrigation. Thecanal opening became delayed through upstream water use inother systems and slower filling of the reservoir; the canal closurecame earlier and earlier in the year because of exhaustion of thestored monsoon water. It became difficult to fit two consecutiveseasons into the irrigation year.

This created at least three periods in the year with intense socialinteraction around water distribution. The first is irrigator lobbyfor canal opening to allow timely planting. The release date waspartly a direct product of available water in the reservoir, but alsobecame related to the yearly maintenance cycle – repairs to bedone in the closure period. As budget allocations for these oftencame late, there was often not enough time to do repairs beforere-opening the canal; particularly the main canal regularly brea-ched, which requires time consuming structural repairs in the offseason.

More intensive interaction is found in the other two periods.The first is the overlap of the end of the kharif/first irrigation seasonand the preparation and start of the rabi/second irrigation season,which is a period of peak water demand. September and Octoberare usually months with high intensity of water distribution con-flicts. The second conflict period is towards the end of the irrigationyear in February–March. Temperatures start rising as summer isapproaching, and the canal closure is usually scheduled for some-where in March. There is a scramble for water in this period and alot of irrigator pressure on the Irrigation Department to extend thecanal opening period to allow crops to mature. This need is un-evenly spread over the system as upstream parts are able to plantearlier than downstream parts through delays in the arrival ofwater at the start of the season. In irrigation, time is clearly a re-source that is scarce (Carlstein, 1982).

As a response to these constraints, managers and irrigators haveattempted more efficient management of the scarce time resourceby introducing and negotiating rotation schedules at the differentlevels of the canal system. Rotation involves the concentration ofwater flow and supplying areas in turns rather than continuously.This increases the efficiency of water use. At main system levelsupply is rotated over secondary canals; particularly the shorterones with less planned irrigated area may get water only a fewdays per week. Within secondary (distributary) canals detailedrotation schedules exist. These are often formally announced andintroduced by the government managers, but in fact the result ofrepeated negotiation processes between government managersand irrigators, and among irrigators located along the same canal.At the level of local irrigation units a wide variety of rotationschedules established by irrigators was documented (Mollinga,2003). The rotation schedules at the different levels are ‘sleeping’,that is, not implemented when water is not scarce, and mobilisedwhen water does becomes scarce in the ‘peak periods’ described.

The evolution of rotation schedules is a typical example ofArcher’s ‘structural elaboration’. The repeated seasonal and yearly

cycles of negotiating water distribution produce sets of rules vary-ing with local physical conditions and social relationships. Theyoften consolidate, and stabilise to a considerable extent, butsometimes they do not. In all cases their enactment in water scarceperiods signifies high drama on the canals, including farmerssleeping on outlet structures at night to avoid manipulation,nightly and daily canal patrols by government managers, some-times together with groups of farmers, the blocking of canals andgates, if not their demolishment, the blocking of roads, demonstra-tions in front of Irrigation Department offices, and the mobilisationof local politicians to exert pressure on the administration to sup-ply water. Though intense and seemingly chaotic to casual observ-ers, the interactions are highly patterned, and their structures andoutcomes quite stable. Detailed discussion of the structure of thesewater control relationships can be found in Mollinga (2003).

The institutional rhythms of water distribution in this canal sys-tem are thus shaped by the rainfall and surface flow patterns of thehydrological cycle, the latter being influenced by human interven-tions in the upstream part of the basin influencing reservoir wateravailability. These ‘macro’ factors translate into opening and clo-sure dates of the canal system, for which rules have been designed,and release schedules based on estimated water availability. ‘With-in system’ elaborate sets of rules have been negotiated for rota-tional water supply at all levels. They all work on the principle oftime shares of concentrated flow, but how exactly varies greatlywith physical conditions and social relationships. Time is a contin-uously contested resource, the structure of its use definitive of ca-nal irrigation management.

4.3. Water distribution, social differentiation and spatial relations

This section discusses how the social differentiation (of differ-ent categories of farmer-irrigators) associated with unequal waterdistribution takes spatial form.

The Tungabhadra LBC is one of several South Indian upland pro-tective irrigation systems with a history of migrant farmer settle-ment (Anjaneya Swamy, 1988). From the 1950s, farmers withsmall holdings in the coastal deltas of the Krishna and Godavaririvers sold their intensively used, mostly rice, land dearly andbought much larger extents of unirrigated land in the new plannedirrigation area. These purchases sometimes took place before ca-nals in the area were built. Settler farmers were interested tobuy land near canals and roads. Because irrigation canals are con-structed on the ridges in the landscape, while the villages in thissemi-arid rainfed region were located in the lower parts of thelandscape – the valleys where water could still be found in thedry season – settlers were able to buy land very cheaply: their pre-ferred locations were far away from the villages around whichrainfed cultivation was concentrated, in the ‘jungle’ as local farm-ers put it. Local farmers, inexperienced with irrigation, sold suchfar-away land on a large scale.22

When canal water started flowing, and the migrant farmersstarted to develop the land for (rice) irrigation, it became clear thatthis former ‘jungle’ land could be very profitably utilised. An inver-sion of the landscape took place. Water availability was now con-centrated in the higher part of the landscape because of the canalsupply. This allowed much more intensive cultivation than rainfedfarming, and two crop seasons, meaning that the higher parts also

Fig. 3. Head-middle-tail zones in a secondary canal.

23 This conclusion is based on unpublished fieldwork.

200 P.P. Mollinga / Geoforum 57 (2014) 192–204

became the core agricultural areas. This was even more so becausethe canals constructed served as roads, and thus new commerciali-sation routes passed through the ‘camps’ that the settler farmersestablished, often at the crossroads of canals and main roads.Moreover, intensive rice irrigation with the attendant seepagelosses could cause waterlogging problems for villages located inthe valleys, and make access to them more difficult. However, afterthe initial ‘surprise takeover’ local farmers also adopted the inten-sive irrigation practices and a less skewed irrigation developmentpattern ensued. The migrant/local spatial distribution pattern oflandholding has, however, remained distinct.

To illustrate that binary head/tail descriptions as common inirrigation studies can be too simplistic for capturing actual pat-terns, the canal depicted in Fig. 3 can be taken as an example. Itis the D93 secondary canal as indicated in Fig. 1. Fig. 3 shows thatthe migrant ‘camps’ are located along the canal, while local villagesare located along the natural drains. In contrast to many othercases, there is no ‘camp’ along the main road, but the settler hab-itations are at some distance from the main road. This has to dowith (a) the relatively late settlement of this canal area (from1979 to 1980) and unwillingness of local large landowners owninglarge tracts in the head end area, to part with their land havingseen its potential profitability elsewhere, and (b) the brokeragenetworks that facilitated the land deals were most accessible tosettler farmers for land located in the middle part of the canal,where the main ‘camps’ were thus established. During the 1991–1992 fieldwork in this canal irrigation was concentrated in themiddle part, with the head part yet hardly developed for irrigation,

and the tail part already struggling to secure access to sufficientirrigation water for even ‘light’ crops like sorghum and millet.

Fifteen years later when the canal was revisited, the scenariohad changed. Land development for irrigation in the head end re-gion by local farmers had significantly expanded, implying watersupply problems for the middle part, and irrigation having beenabandoned in the tail end part.

This example suggests two general points. The first is that loca-tional advantage is definitely an important factor in explaining pat-terns of irrigation distribution, but several other factors may be atplay simultaneously that generate other spatial patterns than sim-ple head–tail sequences. Secondly, though a particular pattern ofunequal distribution is apparently reproduced almost identicallyfrom season to season, and year to year, there seems to be a longuedurée of locational advantage ‘coming through’. Qualitative obser-vations of longer term shifting of head and tail locations in the1991–2007 period in a number of canals strongly suggests thatthere is a slow water/irrigation concentration process happeningat all levels of the canal system in the sense of movement towardsgeographical head ends.23

The third and last instance of spatial patterning is at the level oflocal units of irrigation and the spread of ‘large’ and ‘small’ farmers.An example of the typical pattern is given in Fig. 4 (the unit de-picted in Fig. 4 is the Bhatta outlet in Fig. 3).

Fig. 4. Cropping pattern (A), categories of farmers (B) and year of land development (C) in a Tungabhadra LBC local irrigation unit (pipe outlet command area) kharif season1991.

P.P. Mollinga / Geoforum 57 (2014) 192–204 201

202 P.P. Mollinga / Geoforum 57 (2014) 192–204

In these three maps, the canal water source is the secondary ca-nal, with the settler ‘camp’ located in the head end (with the outletstructure right in between the houses) and the local village in thetail end of this 64 ha unit. Rice and cotton are the most remunera-tive commercial crops, with rice being a water intensive crop. Theyare grown mostly in the upstream part of the unit. Lighter crops(notably sorghum and sunflower) requiring less water are primar-ily grown in the downstream part of the unit.

A typology of farming household-enterprises was preparedusing a qualitative version of Patnaik’s labour exploitation ratio.24

In the second map it can be clearly seen that the upstream rice andcotton crops are mostly grown by rich and middle peasants, whilethe downstream lighter crops are mostly grown by small and poorpeasants. Most of the upstream farmers are settler farmers; mostof the downstream farmers are local farmers. A glimpse of the pro-cess that produced this differentiation can be seen in the thirdmap, which gives the years in which the different plots were devel-oped for irrigation (which involved land levelling and constructing offield bunds and field channels). Land development in this unit beganwith the arrival of settler farmers in 1979–1980, in the head end ofthe unit, and gradually moved downstream, up to a point and mo-ment that water availability in the tail end portion became too con-strained to warrant land development investment. The first and thirdmap also show that in places where water could be picked up fromneighbouring units or drainage channels, land development alsotook place.

5. Conclusion

The analysis above has established, firstly, that large-scale sur-face irrigation processes are hydrosocial in character indeed, withphysical and human aspects internally related. By building large-scale irrigation systems as part of state projects of economic devel-opment, governments attempt to ‘singularise’ the meaning of riverwater to its value for agricultural production, by storage in reser-voirs and diversion in hierarchically ordered canals, thus rearrang-ing the hydrological cycle in time and space, making it an explicithydrosocial cycle. Crucial in this attempt is the deployment oftechnology, as dam and canal infrastructure. The paper has shownhow different infrastructural design characteristics of so calledprotective irrigation in south India configure a pattern of water dis-tribution that is equal in principle but unequal in practice. Thestarkly unequal pattern of water distribution in protective irriga-tion systems is produced in social practices configured by therhythms of the climate and the agricultural seasons, while the so-cial differentiation of peasant farmers associated with unequalwater distribution takes spatial forms, structured by the grid ofthe different levels of canals. The different materialities of watermanagement involved in the production of unequal distributiondo not just constitute the stage and context of social process, butthey are the subject of social interaction and reshaping too: irriga-tion devices like outlets are remodelled in the episodic distributionstruggles between and among irrigators and government manag-ers; the agricultural seasons of the rainfall cycle are reconfiguredby the definition of irrigation seasons through scheduled canalwater releases and the choice of crop varieties with differentlengths of their growing periods; the spatial grid of canals consti-tuting locational advantage and hydraulically defined queues is re-shaped by realigning canals, and re-use of water in drainagechannels by diversion or lifting. The general point for hydrosocialanalysis is that conceptualising hydrosocial relations not onlyinvolves the materialities of water as substance and of thelandscapes that water flows through, but significantly also the

24 The ratio is the balance between net labour hired in (labour hired in minus labourhired out) and family in self-employment (see Patnaik, 1987, chapter 3)

technical infrastructure that facilitates flow, in all its technicalspecificity. Approaches that theorise the social dimensions of tech-nology can usefully be added to and integrated with the politicalecology inspired hydrosocial conceptual repertoire.

Secondly, it has been shown that the process of the transforma-tion of these hydrosocial relations is a process of cyclical (in thesense of episodic) structural elaboration in Archer’s (1995) sense:structure-agency dynamics is animating the hydrosocial irrigationcycle as a morphogenetic cycle. Notwithstanding the detailed artic-ulation and legalisation of a form of land-use planning and waterrationing called localisation, the effort at state rule to ‘disciplineits subjects’ into irrigation practices that maximise the commongood, was unsuccessful. It failed because it contradicted individualfarmers’ intensification and maximisation strategies on given sizesof farm land, because in the 1960s government allowed intensifica-tion under the ‘grow more food’ logic in an incompletely developedirrigation system, and because under competitive populism the lar-ger farmers appropriating excess water are the local leaders thatcontrol the ‘vote banks’ of a parliamentarian’s constituency. Giventhe technical design characteristics of protective irrigation, appro-priation of water for intensive rice production, as happened in theTungabhadra LBC, led to a dramatically skewed pattern of distribu-tion, with the ‘favoured area’ of intensive water use slowly concen-trating in the geographical ‘head ends’ of the different canal systemlevels, a clear example, in Archer’s (1995) terms, of the structuralelaboration that is part of morphogenesis. However, the rise of apowerful class of larger farmers colluding with Irrigation Depart-ment managers and local politicians, has not meant the end of gov-ernment efforts at ‘irrigation reform’. These may even intensifyunder neoliberal policy conditions. In the context of this paper’stheoretical argument, this suggests that the structural elaborationof the institutional arrangements of the hydrosocial cycle is nevercompleted. Hydrosocial relations of power within irrigation remaincontested, more so than critical analysis of large-scale irrigation asthe abode of green revolution capitalist farming tends to suggest.Though there is no reason for excessive optimism as regards irriga-tion reform leading to more equitable water distribution, there aredefinitely entry points for enhancing such efforts.

This analysis, finally, suggests the limitations of general, encom-passing concepts like ‘hydrosocial relations’, ‘socio-technical sys-tems’, and for that matter, ‘waterscapes’ (cf. Budds and Hinojosa,2012). These conceptual hybrids do well to establish the point ofthe need to look at the material and human aspect of natural re-source management as the co-evolution of a single object, and to‘reposition water as inherently political’ (Linton and Budds,2014). Though neither of these insights is new, they definitely bearrepeating and elaboration. Once these points are accepted, needarises for more specific conceptualisation to capture the differenthydrosocial mechanisms at work as emergent properties in com-plex systems like irrigation.

The different forms of queuing that are socio-technically/hydro-socially established as rotation schedules in ongoing, seasonal andyearly negotiation (structural elaboration) processes, are the keyemergent property in explaining the inequality of water distribu-tion. Other emergent properties ‘at work’ include the capacity toproduce high irrigated rice (and cotton) yields on vertisols – notonly requiring the establishment of economic incentives and net-works, labour markets and other ‘social’ structures, but also thestructural elaboration of the landscape, soil and crop itself, byknowledgeable, skilled and entrepreneurial agents. Additionalemergent properties are the spatially specific mechanism of socialdifferentiation of agricultural producers, the mechanisms forreproducing political legitimacy/credibility of elected politiciansas shaped by the intersection of political constituencies andhydraulic units, the interplay of caste hierarchy and local/migrantdistinctions in cultural notions of dominance, the (poorly

P.P. Mollinga / Geoforum 57 (2014) 192–204 203

documented and understood) new ecological dynamics constitutedby the embedding of the irrigation system in the river basin andecosystem, and many more.

Understanding complexity and emergence requires a concep-tual vocabulary that captures specific instances of hydrosociality,or, in an older vocabulary, that captures the ‘concentration of themany determinations’ of the concrete (Marx, 1973). Without this,analysis will not be able to move beyond the important but basicpoint of showing that hydrosociality exists. It has been argued thata combination of the emerging hydrosocial relations perspectivewith Archer’s (1995) theorisation of structural elaboration in mor-phogenetic cycles and a social construction of technology ap-proach, can form the basis of such specific and concrete analysisof the dynamics of the hydrosocial cycle. Emphasis on morphogen-esis/structural elaboration and mechanisms/emergent propertiesin the theorisation of socio-technical/hydro-social irrigation pro-cesses as open, complex systems (or configurations) is a choicefor a critical realist ‘deep ontology’ (cf. Sayer, 1992). Other theori-sations of irrigation have chosen, inspired by actor-network theory,the analysis of hydro-social networks as their entry point (cf.Wester, 2008). The ANT understanding of network is in my viewa version of a ‘flat’ ontology, and therefore flawed and not takenup is this paper. However, the proof of such foundational puddingslies in the practical adequacy of the concrete analyses they are ableto produce as part of the further development of hydrosocialanalysis.

References

Anjaneya Swamy, G., 1988. Agricultural Entrepreneurship in India. ChughPublications, Allahabad.

Archer, Margaret S., 1995. Realist Social theory: The Morphogenetic Approach.Cambridge University Press, Cambridge.

Attwood, D.W., 1987. Irrigation and imperialism: the causes and consequences of ashift from subsistence to cash cropping. Journal of Development Studies 23 (3),341–366.

Bakker, K.J., 2003. An Uncooperative Commodity: Privatizing Water in England andWales. Oxford University Press, Oxford.

Bolding, A., 2004. In Hot water: A Study on Sociotechnical InterventionModels and Practices of Water Use in Smallholder Agriculture,Nyanyadzi Catchment, Zimbabwe. PhD Thesis. Wageningen University, theNetherlands.

Bolding, A., Mollinga, P.P., van Straaten, K., 1995. Modules for modernisation:colonial irrigation in India and the technological dimension of agrarian change.Journal of Development Studies 31 (6), 805–844.

Brass, T. (Ed.), 1995. New Farmers’ Movements in India. Frank Cass, Essex andPortland.

Budds, J., 2009. Contested H2O: science, policy and politics in water resourcesmanagement in Chile. Geoforum 40 (3), 418–430.

Budds, J., Hinojosa, L., 2012. Restructuring and rescaling water governance inmining contexts: the co-production of waterscapes in Peru. Water Alternatives5 (1), 119–137.

Byres, T.J., 1981. The new technology, class formation and class action in the Indiancountryside. Journal of Peasant Studies 8 (4), 403–454.

Carlstein, T., 1982. Time Resources, Society and Ecology: On the Capacity for HumanInteraction in Space and Time. Preindustrial Societies, vol. 1. George Allen andUnwin, London.

Coward, E.W., 1990. Property rights and network order: the case of irrigation worksin the Western Himalayas. Human Organization 49 (1), 78–88.

Dhawan, B.D., 1988. Irrigation in India’s Agricultural Development. Productivity,Stability, Equity. Sage, New Delhi.

Dhawan, B.D., 1989. Enhancing production through crop pattern changes. ArthaVijnana 31 (2), 153–175.

Falkenmark, M., 1997. Society’s interaction with the water cycle: a conceptualframework for a more holistic approach. Hydrological Sciences Journal 42 (4),451–466.

Giddens, A., 1984. The Constitution of Society: Outline of the Theory ofStructuration. Polity Press, Cambridge.

GOI/MOIP Government of India/Ministry of Irrigation and Power), 1972. Report ofthe Irrigation Commission, vol. I, New Delhi.

GOKAR/PD (Government of Karnataka/Planning Department), 1976. Report on Re-examination of Cropping Pattern under T.B.P. Left Bank Canal by the TechnicalCommittee. Tungabhadra Project.

GOMYS/DOA (Government of Mysore/Department of Agriculture), 1968. Report ofthe Expert Committee on the Development of Light Irrigation on the Left BankArea of the Tungabhadra Project 1965, Bangalore.

Gooch, G.D., Rieu-Clarke, A., Stalnacke, P. (Eds.), 2010. Integrating Water ResourcesManagement: Interdisciplinary Methodologies and Strategies in Practice. IWAPublishing, London.

Gorter, P., 1989. Canal irrigation and agrarian transformation: the case of Kesala.Economic and Political Weekly (243), A94–A105.

Hughes, T.P., 1987. The evolution of large technological systems. In: Bijker, W.E.,Hughes, T.P., Pinch, T.J. (Eds.), The Social Construction of Technological Systems.MIT Press, Cambridge, MA, pp. 51–82.

IIC (Indian Irrigation Commission), 1903. Report of the Indian IrrigationCommission 1901–1903. Her Majesty’s Stationery Office, London.

Jairath, J., 1985. Technical and institutional factors in utilisation of irrigation: a casestudy of public canals in Punjab. Economic and Political Weekly (201), A2–A10.

Joshi, L.K., Hooja, R. (Eds.), 2000. Participatory Irrigation Management. Paradigm forthe 21st Century, vols. 1 and 2, Rawat, Jaipur and New Delhi.

Jurriëns, R., Mollinga, P.P., 1996. Scarcity by design: protective irrigation in Indiaand Pakistan. ICID Journal 45 (2), 31–53.

Lebel, L., Garden, P., Imamura, M., 2005. The politics of scale, position, and place inthe governance of water resources in the Mekong region. Ecology and Society10 (2).

Linton, J., 2008. Is the hydrologic cycle sustainable? A historical–geographicalcritique of a modern concept. Annals of the Association of AmericanGeographers 98 (3), 630–649.

Linton, J., Budds, J., 2014. The hydrosocial cycle: Defining and mobilizing arelational-dialectical approach to water. Geoforum 57, 170–180.

Mahbub, S.I., Gulhati, N.D., 1951. Irrigation Outlets (Revised and Enlarged by N.D.Gulhati). Atma Rau & Son, Delhi.

Marx, K., 1973. Grundrisse: Foundations of the Critique of Political Economy.Penguin Books, Harmondsworth (first published in 1939 in German).

MICC (Madras Irrigation and Canal Company), 1859. Report of the Directors.MICC (Madras Irrigation and Canal Company), 1867. Report of Proceedings at the

Sixteenth Ordinary General Meeting.Mitra, A.K., 1986. Underutilisation revisited: surface irrigation in drought prone

areas of western Maharashtra. Economic and Political Weekly 26 (17), 752–756.Mitra, A.K., 1987. Planning and Management of Surface Irrigation in Drought Prone

Areas. Artha Vijnana Reprint Series No. 14. Gokhale Institute of Politics andEconomics, Pune, pp. 305–422.

Mollinga, P.P., 2003. On the Waterfront: Water Distribution, Technology andAgrarian Change in a South Indian Canal Irrigation System. WageningenUniversity Water Resources Series. Orient Longman, Hyderabad, India.

Mollinga, P.P., Meinzen-Dick, R.S., Merrey, D.J., 2007. Politics, plurality andproblemsheds: a strategic approach for reform of agricultural water resourcesmanagement. Development Policy Review 25 (6), 699–719.

Mollinga, Peter P., 2010. The material conditions for a polarised discourse: clamoursand silences in critical analysis of agricultural water use in India. Journal ofAgrarian Change 10 (3), 414–436.

Mooij, J., 2007. Hype, skill and class: the politics of reform in Andhra Pradesh, India.Commonwealth & Comparative Politics 45 (1), 34–56.

Mollinga, P.P. 2013. Boundary concepts for interdisciplinary analysis of waterresources management in South Asia. In: Dik Roth, D., Vincent, L., (eds.),Controlling the water: Matching technology and institutions in irrigation andwater management in India and Nepal, Oxford University Press, pp.342-365.

Moore, Mick., 1989. The fruits and fallacies of neoliberalism: the case of irrigationpolicy. World Development 17 (11), 1733–1750.

Morrison, K.D., 2010. Dharmic projects, imperial reservoirs, and new temples ofIndia: an historical perspective on dams in India. Conservation & Society 8 (3),182–195.

Nadkarni, M.V., 1987. Farmers’ Movements in India. Allied Publishers, Ahmedabad.Nickum, J.E., 1977. An instance of local irrigation management in China. Economic

and Political Weekly 12 (40), 1714–1716.Patnaik, U., 1987. Peasant Class Differentiation. A Study in Method with Reference

to Haryana. Oxford University Press, Delhi.Pinch, T.J., Bijker, W.E., 1984. The social construction of facts and artefacts: or how

the sociology of science and the sociology of technology might benefit eachother. Social Studies of Science 14 (3), 399–441.

Sakthivadivel, R., 2006. Water balance studies and hydrological modelling forIWRM. In: Mollinga, P.P., Dixit, A., Athukorala, K. (Eds.), Integrated WaterResources Management in South Asia. Water in South Asia series No1. Sage,New Delhi, pp. 189–218.

Sayer, A., 1992. Method in Social Science. A Realist Approach, second ed. Routledge,London and New York.

Scott, J.C., 1997. Seeing Like a State: How Certain Schemes to Improve the HumanCondition Have Failed. Yale University Press.

Shah, E., 2008. Telling otherwise: a historical anthropology of tank irrigationtechnology in South India. Technology and Culture 49 (3), 652–674.

Swyngedouw, E., 2007. Technonatural revolutions: the scalar politics of Franco’shydro-social dream for Spain, 1939–1975. Transactions of the Institute ofBritish Geographers NS 32 (1), 9–28.

Swyngedouw, E., 2009. The political economy and political ecology of the hydro-social cycle. Journal of Contemporary Water Research and Education 142, 56–60.

Thorner, D., Thorner, A., 1962. The weak and the strong on the Sarda canal. In: Landand Labour in India. Asia Publishing House, London, pp. 14–20.

Trottier, J., Fernandez, S., 2010. Canal spawn dams? Exploring the filiation ofhydraulic infrastructure. Environment and History 16 (1), 97–123.

204 P.P. Mollinga / Geoforum 57 (2014) 192–204

UAS (University of Agricultural Sciences), 1973. Seminar on problems and prospectsof soil and water management in Tungabhadra Project area. In: Seminarproceedings. Held at Agricultural Research Station Siruguppa, January 29, 1973.

Upadhya, C.B., 1988. The farmer-capitalists of coastal Andhra Pradesh. Economicand Political Weekly 23 (27), 1376–1382 (and 23 (28), 1433–1442).

Uphoff, N., 1986. Getting the Process Right: Improving Water Management withFarmer Organisation and Participation. Working Paper. Cornell University,Ithaca.

Uphoff, N.(with Priti Ramamurthy and Roy Steiner), 1991. Managing Irrigation.Analyzing and Improving the Performance of Bureaucracies. Sage, New Delhi.

Venkata Ramiah, P., 1937. Report on the Soil Survey of the Tungabhadra Project.Department of Agriculture Madras. Government Press, Madras.

Venot, J.P., 2009. Rural dynamics and new challenges in the Indian water sector:trajectory of the Krishna basin, South India. In: Molle, F., Wester, P. (Eds.), RiverBasins: Trajectories, Societies, Environments. IWMI/CABI Publishing, Colombo/Wallingford, UK, pp. 214–237.

Vincent, L.F., 1997. Irrigation as a Technology, Irrigation as a Resource: ASociotechnical Approach to Irrigation. Inaugurale rede/Inaugural lecture.Wageningen Agricultural University.

Wade, R., 1976. Rationing water: principle and practice in South India. ODIworkshop on choices in irrigation management. In: ODI Workshop on Choicesin Irrigation Management 27 September–1 October 1976. Paper C.4. Institute ofDevelopment Studies, Sussex (mimeograph).

Wade, R., 1982. The system of administrative and political corruption: canalirrigation in South India. Journal of Development Studies 18 (3), 287–328.

Wester, P., 2008. Shedding the Waters: Institutional Change and Water Control inthe Lerma–Chapala Basin, Mexico (PhD Thesis). Wageningen University,Wageningen.

Wester, P., Merrey, D.J., de Lange, M., 2003. Boundaries of consent: stakeholderrepresentation in river basin management in Mexico and South Africa. WorldDevelopment 31 (5), 797–812.