and the Performance of Dalit Rights in South India

Donor-funded development NGOs are sometimes portrayed as co-opting, privatizing or depoliticizing citizen action or social movements. This much is implied by the term ‘NGOization’. Alternatively, NGOs can be seen as bearers of rights-based work increasingly threatened by tighter regulation or substitution by corporate social responsibility models of development. This article engages critically with both perspectives. It traces the role of NGOs and their funders in agenda setting, speciﬁcally in bringing the previously excluded issue of caste discrimination into development policy discourse in the form of a Dalit-rights approach in Tamil Nadu, south India. The authors explore the institutional processes of policy making and NGO networking involved, the alliances, entanglements of NGOs and social movements, and the performativity of NGO Dalit rights. But at the same time, the article illustrates how NGO institutional systems have constrained or failed to sustain such identity-based claims to entitlement. In Nancy Fraser’s terms, the article explores success and failure in addressing ‘ﬁrst-order’ issues of justice, that is rights to resources (in this case, land), and in tackling ‘second-order’ injustices concerning the framing of who counts (who can make a claim as a rights holder) and how (by what procedures are claims and contests staged and resolved). This draws attention to the important but fragile achieve- ments of NGOs’ discursive framings that give Dalits the ‘right to have rights’.


INTRODUCTION
There has been a shift from the critique of development as an 'anti-politics' discourse (Ferguson, 1994) to ethnographies of development as the practice of politics in varied social-historical circumstances (Mosse, 2013). 'Development' articulates aspirations, connects to resource flows and is a site for struggles over the terms of recognition and citizenship (ibid.). But international development remains a discourse constituted by exclusions. One such exclusion (in India and internationally) is caste. Inherited caste identity is an important determinant of life opportunity in India's caste-networked economy, and a continuing aspect of poverty, inequality and social exclusion, especially for Dalits (those subordinated as 'untouchable') (Mosse, 2018). Caste has always been pivotal in India's cultural and electoral politics (Teltumbde, 2017) but remains marginal to mainstream development discourse, and absent from the social characteristics of concern, such as gender, race or age, in the Sustainable Development Goals.
Why caste is kept out of development is a question discussed elsewhere . This article examines efforts in the 2000s to draw caste into the frame of international development, reworked as the right of Dalits to development resources alongside political rights and social justice. We consider the role of NGOs here and ask three key questions. First, what made it possible for the hitherto excluded vernacular politics of caste to enter and become central to NGO development policy in the late 1990s? Second, what were the implications of this 'Dalitization' of development for the NGO field, especially given the confrontation of NGOs with caste power embedded in local state bureaucracy that a focus on Dalits and rights involved? Third, why was it difficult to sustain the NGO policy focus on caste discrimination and Dalit rights?
We use two fieldwork cases from Tamil Nadu, south India: one focuses on the effects of a higher-order advocacy network for Dalit rights; the other examines grassroots activism for the recovery of alienated Dalit land. These cases raise a fourth question: how do NGOs combine 'movement work' (Alvarez, 2009) -that is networking, coalition building, advocacy and popular protest -with the funding, reporting or projectized relationships of 'NGOing' (Hilhorst, 2003;Lewis and Schuller, 2017: 642)? We consider the relationship between the discursive and the organizational relations aspects of NGO networks; between their 'expressive' and their 'instrumental' dimensions (Knutsen, 2012). While the cases highlight instability and disarray alongside material and discursive achievements of NGO networks, they also challenge the preconception of donor-funded NGOs as co-opting, privatizing or depoliticizing citizen action, acting as contractors of state or capital rather than agents of the oppressed, as the notion of 'NGOization' often implies (Bornstein and Sharma, 2016;Ismail and Kamat, 2018;Kamat, 2002Kamat, , 2004. Finally, the article indicates the importance of NGOs in addressing second-order injustice, that is the 'framing' of what counts as an NGOs as Social Movements in South India 3 issue and who counts (who can make a claim as a rights holder), and how (by what procedures claims and contests are staged and resolved) (Fraser, 2005: 85).
We draw on research undertaken from two different perspectives within a broader project titled 'Caste Out of Development'. The first traces a policy process in the world of NGO donors, in part from participation in that world. David Mosse recalls policy work as a donor representative in southern India in the 1990s 1 and returns to trace a two-decade-long institutional sequel, finding long-term shifts in NGO partner organizations and networks. He discovers how caste became a central policy issue, mobilizing new forms of network and public action by and for Dalits; but also how, within little more than a decade, caste and 'Dalit rights' had been ousted from policy vision again and Dalit NGO networks were in disarray. Over the course of months between 2009 and 2012, as he sat, talked and lodged with key leaders and members of unravelling networks in Tamil Nadu, there was every reason for both Indian and European players in the drama to want to turn over events and reflect on what had happened and why.
As part of the same research project, Sundara Babu Nagappan was embedded in a Dalit NGO network deploying a 'rights-based' approach in struggles over Dalit land dispossession. He experienced the tactics of grassroots activists who successfully combined the political performances of a social movement with the proceduralism (legal/technical) of an NGO, distinctively targeting caste power networked in the local bureaucracy. He subsequently witnessed the failure of the coalition of activists to prevent the acquisition of Dalit village land for allocation to a multinational company.
The fieldwork on which this article is based took place in 2009-12. We summarize themes from a vast archive of NGO documentation 2 and in-depth individual and group interviews, including David's interviews with over 120 Dalit NGO staff, leaders and network members, church leaders, international NGO donors, Indian bureaucrats, activists and international organization officials (e.g., World Bank, UNICEF, DFID; see Mosse, 2012b), and Sundara Babu's 12-month (2010-11) multi-sited fieldwork with Dalit land rights NGO networks, interviews with NGO staff and participation in workshops and public action.
The article has two parts. The first examines international NGO (INGO) policy agenda setting and its transnational processes. It describes the rise of Dalit activism, the reframing of caste in the discourse of human rights, and the organizational imperatives that 'Dalitized' the agenda of one major NGO 1. David Mosse was Oxfam Regional Representative for South India in Bangalore (Bengaluru), 1987-91, remaining in contact in the 1990s and 2000s; NESA (see below) was a partner in the research project. 2. Some of this takes the form of unpublished reports from the NGOs, which are listed in the references as mimeos.

4
David Mosse and Sundara Babu Nagappan donor (Oxfam) in southern India. 3 It explores policy processes that manipulate categorical distinctions between INGO, NGO and social movements. The second part analyses two different NGO approaches to Dalit rights activism. In one, INGOs promoted higher-order network forms needed to promote caste justice; in the other, NGOs directly challenged bureaucratic power locally over the alienation of Dalit land, as well as regionally and internationally in mobilization against a state-supported multinational company. 4

NGOs and the Rise of Dalit Activism
The 1990s was a decade of renewed Dalit activism. While nationally public debate over the extension of affirmative action 'reservations' in public sector jobs and higher education made caste visible beyond the village, in south India an upsurge in anti-Dalit violence and killing fuelled social movements and Dalit political parties (HRW, 1999;Rawat and Satyanarayana, 2016). The birth-centenary of Dalit leader Dr B.R. Ambedkar increased his status as a political icon, and his writing (newly available in vernacular languages) crystallized a Dalit/caste critique of Marxian and feminist frameworks within student, youth and women's organizations. Dalit activism was led by a professional Dalit elite (government officials, doctors, lawyers, teachers, ex-servicemen), the Scheduled Caste (SC) associations of public sector employees in the railways, banks or food corporations, Dalit members of parliament in various parties, Left movement leaders, and the charismatic founders of new Dalit political parties (Gorringe, 2005). The political space for Dalits, closed off in the decades following Indian Independence, began to open. In Tamil Nadu, NGOs and churches found new connections to this heterogeneous Dalit politics (Mosse, 2012a: 193-97, 208-32) that had hitherto been avoided.
By the 1990s, there was a mutual entanglement of actor categories in which church-formed or trained NGOs produced cadres, who promoted movements, which accessed resources from NGOs, which provided skills for political leaders, whose parties NGO workers joined. A first generation of Dalits to become NGO leaders in the 1970s were those freed from agrarian subordination by the mobility of their fathers (and less commonly 3. We take Oxfam as an example, recognizing that there were parallel but distinct processes affecting other INGOs, as explained in Mosse (2012b). 4. An important caveat is necessary. This article is not an analysis of Dalit movements and politics in India (or Tamil Nadu) in general, their roots, trajectory and challenges (on which there is a now a large literature); its focus is on international development NGOs' interface with this Dalit activism.
NGOs as Social Movements in South India 5 mothers) as teachers, or in the army or railways; these NGO leaders were trained and mentored in Liberation Theology/Marxian thinking by Christian priests in student organizations (mostly Christian, see Arun, 2012). To these few were added, in the 1980s and 1990s, many Dalits with personal experience of village untouchability, and of being 'burned by the fire of discrimination' (as one put it) when working as community animators within non-Dalit NGOs and Left movements. They formed NGOs to work in their own localities for Dalit access to common property, public spaces, land or house-site titles, water, roads or electricity, and engaged in vernacular struggles over caste dignity at temples/churches or funeral sites. 5 At the time, such activity did not register as 'development' with the INGO funders to whom Dalit-led NGOs turned, and so had to be translated in their proposals and reports into prevailing policy categories. 6 In the case of Oxfam, the prevailing economization of poverty and 'culturalization of caste' (Natrajan, 2011; see also  meant that caste was regarded as part of the context of Oxfam's programme -deepening agrarian poverty and inequality -but not as an organizing principle behind it; it was 'best dealt with indirectly in ways which diffuse rather than amplify conflict'. 7 It was as landless labourers or marginal farmers that Dalits predominated among project beneficiaries, and caste was avoided in explaining the need for literacy, training or Freirean awareness raising. When demand came for more strategic coherence in grant-making policy from head office, as surely it did (for example in 1989-90), it was issues of gender, environment or drought vulnerability that provided salient categories, not caste inequality. Moreover, neoliberal institutionalist policy trends in the early 1990s brought the rhetoric of participation, local institution building, user associations, or microfinance Self-Help Groups as the substitute 'technology' of people's mobilization (Elyachar, 2005) 8 -the terms in which the Indian government's 8th Five-Year Plan sought to enrol NGOs in restructuring rural development -all of which evaded the caste question.
By the early 2000s, things had entirely changed. NGO activity in Tamil Nadu was now significantly funded in terms of a 'Dalit rights' agenda. A vernacular politics had entered the international NGO world. This was not, however, a globalization from the grassroots. Indeed, the 'Dalitization' of 5. Sometimes, being dependent on non-Dalit 'parent' NGOs for access to foreign funds, Dalit NGO leaders felt their work was appropriated, and complained of a caste-based division between the raisers of money and the performers of activities (Picherit, 2018: 358); as one Dalit objected, 'collecting news cuttings of our action, that is the "fieldwork" of non-Dalit NGOs' (interview

International Human Rights and 'Dalit Rights' Discourse
The realm of international human rights afforded a set of pre-existing institutional links for reframing caste as a development matter. From 1996, caste discrimination gained recognition as a violation of human rights at the United Nations (UN), as a consequence not of Dalit advocacy coalitions per se, but rather of an interaction between the agendas of UN experts and state delgates (Aranguren, 2011). India had been reporting on the state of its Scheduled Castes and Tribes (SCs/STs) or Dalits to CERD (the Committee on the Elimination of Racial Discrimination) -the expert monitoring committee for the International Convention on the Elimination of All Forms of Racial Discrimination (ICERD) -from the time of its formation in 1970; but in 1996, it ceased to do so on the grounds that caste did not fall under the provisions of CERD since racial discrimination of any kind did not exist in India's territory. Aranguren (2011) links this refusal to report to CERD to the consequences of India's 1989 Scheduled Caste/Scheduled Tribes (SC/ST) Prevention of Atrocities Act, or PoA, which, for the first time, produced numerical data on legally defined violations (or 'atrocities') against Dalits, revealing not only the high volume but also the increasing rate of what would be indicators of poor human rights performance under CERD.
In the face of this state refusal, Aranguren argues, CERD experts asserted an independent exegetic function insisting on a broadened interpretation of race, beyond a narrow biological/pheonotype concept to wider social discrimination; and (in the absence of Indian government reporting) for the first time accepted the report of a non-official (Canadian) Dalit diaspora NGO 9 on descent-based (caste) discrimination in India. This not only marked a significant shift in CERD in 1996, it also brought caste explicitly into the notion of 'descent' as a prohibited basis of discrimination under ICERD (formally so by 2002, despite the Indian goverment's refusal to accept this), so making the international human rights treaty apparatus available for Dalit activists (Waughray, 2009: 191-92). Meanwhile, following the success of campaigns against apartheid in South Africa by 1994, Human Rights Watch (among other organizations) turned its attention to India and discrimination against Dalits in its 1999 report, Broken People: Caste Violence against India's ' Untouchables' (HRW, 1999); and international church networks increasingly turned to the plight of their majority Dalit Christian populations in India (Bob, 2007;Steur, 2012). 9. The Toronto-based Ambedkar Center for Justice and Peace (Aranguren, 2011: 34) India's NGO Dalit activists could use the idea that theirs was part of a universal struggle against the violation of human rights, unbound to the particularities of India, Hinduism or Brahmanic domination. It involved a critique of the Indian state apparatus from outside, not for the abuse of citizens, or legislative failings but, as Mehta (2013) points out, for the violence of state neglect: the failure to meet obligations to Dalits, manifest in development disparities (poverty, illiteracy, infant mortality), dispossession of land, failure to spend the mandated proportional (population-equivalent) Dalit development budget allocations, and failure of protection from untouchability and anti-Dalit violence. 10 This was all set out in a 'Black Paper' on caste inequality and development released in December 1998 by the National Campaign on Dalit Human Rights (NCDHR, 1999), a body headed by Dalit NGO leaders from southern India that came out of the national coordination of Dalit organizations in 1997-98 around issues raised in the preparation of the HRW Broken People report, prompted and funded by the Ford Foundation (Bob, 2007: 178-80).
This campaign for Dalit human rights was also a challenge to liberals to see the bearers of rights not just as autonomous individuals, but as a distinct group, historical victims of exploitation and untouchability, to whom the Indian state has a duty under the Constitution (Mehta, 2013: 153). It forged a new alliance between NGOs and Dalit movements/parties (the Liberation Panthers, Putiya Tamilan, Republican Party of India) who participated in the launch of the Black Paper (Gorringe, 2005: 76). The campaign aimed to elicit state action by appeal to the Indian public as well as by a politics of shaming (Mehta, 2013: 157) through international fora such as, pre-eminently, the 2001 UN World Conference Against Racism (WCAR) in Durban, at which many hundreds of Dalit activists (NGO and political) campaigned using slogans such as 'caste is race' or 'India's hidden apartheid' to have caste recognized as a global form of discrimination.
While the Dalit campaign and advocacy coalition failed to get caste onto the multilateral agenda defining governmental policy making and monitoring because it was blocked by the Indian government and pushed out by competing interest coalitions (Aranguren, 2011), the new international discourse of Dalit rights did make caste available for other developmentoriented institutional agendas.

An INGO Policy Process
The transnationally shaped Dalit human rights discourse became embedded in Tamil Nadu NGO practice in the late 1990s, significantly because it 8 David Mosse and Sundara Babu Nagappan was adopted by INGO donors. This happened as caste/Dalit issues, hitherto marginal to development goals, became important in demonstrating the donors' commitment to the rising, if ill-defined, 'rights-based approach' focusing on discrimination, inequality and claims on the state (Cornwall and Nyamu-Musembi, 2004). 11 The idea of 'rights' implied that INGO donors and their partners would have to go beyond local projects and align themselves to goals of transformational change by means of supra-local action for global justice and against human rights violations (Anandhi, 2017). But organizations adopt new policies not on the basis of ideas alone, but because of the work ideas do to resolve their own dilemmas. In the case of Oxfam, to which we now turn, a new Dalit rights agenda in fact achieved two important things: it produced success out of failure; and continuity out of rupture.
First, success out of failure. In the 1990s, Oxfam, like many INGOs across south India, faced a problem. From work in the Oxfam field office at that time, David recalls that the approach to 'people's empowerment' through NGO-promoted village/neighbourhood associations (called sangams) was perceived as not working. Field reports arriving at Oxfam and other donor offices were reaching similar conclusions: a now standardized approach multiplied sangams which, rather than fostering autonomous grassroots action (for land, services etc.), became beneficiary clients expecting deliverable schemes and inputs from their NGO patrons who were, in turn, institutionalized through donor support. Adopting the self-critical discourse on NGOization, donor staff, in varied texts, policy workshops and strategy reviews, began to see NGOs as the problem and to distinguish the federated associations or 'people's organizations' (the worthy end) from donor-funded NGOs (the temporary means). While informed by a moral discourse on righteous social movements and unaccountable NGOs (Bornstein and Sharma, 2016: 82), this representation bore little relation to the reality of clientelist relationships in which donors were themselves embedded.
The Dalit rights politics of the 1990s allowed Oxfam to view NGO action and the sangam process rather differently. Reframing programme objectives in terms of Dalit assertion of social-political rights -as a 1997 report did (Parasuraman and Vimalanathan, 1997) -involved a reorganization of categories. The growing number of small Dalit-led NGOs whose staff worked in their own localities could be construed as really being people's organizations. Their caste-based activism -struggles over land, temples, social boycotts and anti-Dalit violence -locally initiated but regionally interconnected through expanding Dalit NGO networks and internationally 11. Recent disenchanted assessments of rights-based approaches describe impacts being limited by vague definitions, procedural rights (consultation, participation) being prioritized over substantial rights (e.g., to land or housing), and the hegemony of target results (not rights) (Nelson and Dorsey, 2018). 9 salient, was now evaluated as a kind of self-organizing movement for structural change. This offered a holistic (material, social, spiritual) perspective on dignity and development beyond projects, to which -as our interviews made clear -jaded European agencies could reconnect. It provided a way out of the failures of the sangam model and the elusive goal of 'empowerment'. It is not that a Dalit rights policy changed practice, but rather that it provided new interpretive possibilities (and thus new resource flows), bringing a range of existing but hitherto unrecognized social struggles into a rejuvenated framework of development goals. The second work of Dalit rights policy was to produce continuity out of rupture. Oxfam brought its new vision of Dalit-led development to the fore at the precise moment of a central decision to close its south India programme (the oldest, started in 1962), to rationalize, cut management costs and, like many INGOs, refocus on the relatively poorer northern Indian states. To put the matter simply, the notion of development as a selfsustaining struggle for Dalit rights offered an ethical narrative with which to clothe institutional abandonment, with all its material consequences, and handle the resentment and resistance of some 48 NGOs who pressed their claims to partnership. In practice, Oxfam negotiated exit by, in March 1998, transferring its programme and assets to a self-governing network of 42 mostly Dalit-led NGOs, ambiguously titled the New Entity for Social Action, or NESA -part donor (Oxfam, then Novib), part membership bodyfirmly aligned to a 'rights-based approach' that now articulated the culturalpolitical assertions of Dalits.
At the time, the first local council (Panchayat) elections were taking place after the 73rd Constitutional Amendment Act (1993) mandated reserving constituencies for Dalits and women. Many saw this as changing the field and putting development resources in the hands of Dalit women and men as Panchayat leaders -an interpretation only emphasized by the extreme violence unleashed against Dalits standing for election, including the notorious massacre of six Dalits of Melavalavu (near Madurai), including beheading the elected President in June 1997 (HRW, 1999). Never had NGOs more to contribute; continuity was essential. This was no time to abandon a rightsbased development programme.

THE CONSEQUENCES OF 'DALIT RIGHTS' IN NGO PRACTICE
Making Dalit struggle and the failure of the state to protect this constituency an explicit focus of rights-based development action had implications for NGOs. On the one hand, they newly targeted untouchability, caste exclusion from land and water, from commons, services and schools, public spaces and political participation, and used the PoA law otherwise out of reach for vulnerable Dalit communities. The Dalit rights focus validated talk of dignity, 'the moving force' behind all development, as one NGO leader put it to David, 12 such that debt-bonded Dalit labourers would mobilize first to have their own burial/cremation ground or to be served tea in the same glasses as others in the village teashop.
On the other hand, targeting caste-based injustice brought new obstacles and risks. Dalit NGOs faced inaction or resistance when they took up cases of caste-based violence, or retaliatory 'social boycotts' 13 such as that imposed by upper castes on Dalit youth in the village of Irunjirai when the latter asserted their statutory right to use the village barbering and laundry services (Arulnathan, 2012). Activists claimed the route to justice, including registration of cases under the PoA, was blocked by the local police, bureaucrats and politicians who were caste-connected to the perpetrators of injustice (Ramaiah, 2007), and who also diverted development resources intended for Dalits. 14 'If we file a case', said an NGO worker, 'there won't be any inquiry for evidence'. 15 Indeed, NGO activists told us that recourse to the police or legal action often provoked counter-cases against victims by the perpetrators of anti-Dalit injustice while Dalit rights activism was criminalized. As one head put it, 'you can work for Dalits but to work as Dalits is politically dangerous'. You may organize separate caste associations, but 'if you put "Dalit liberation" on your board they will attack you'. 16 One way or another, for the many small Dalit-led development NGOs, a focus on caste-based abuse made local action difficult. It strained relations with the local bureaucracy, restricting access to state-funded schemes or state-regulated foreign donor funding, increasing political and financial risk, and creating reliance on local sponsors and volunteer workforces. 17 The next sections illustrate two responses to this dilemma. The first focuses on a strategy of de-localizing claims by creating supra-local forums able to assemble knowledge and authoritative intervention to circumvent local caste/bureaucratic obstacles to justice. This involves what Latour calls 'scaling' (2005: 184-85), that is, actor strategies to recontextualize interactions so as to enrol the interests of distant others, to mobilize large-scale agendas while shifting back to local disputes. This created an extensive regional Dalit NGO network, the dynamics and varied effects of which we examine. The second response was to target the local state directly, especially by transposing those obstructing encounters with the bureaucracy to the public arena through street protest mobilized under the banner of social movements or fronts (iyakkam or pēravai) -the federated youth or women's groups having 'different rights in public space' (Picherit, 2018: 362) -set up in parallel to the registered (funded) NGO. In the first case, we find local activism gaining coherence as it is amplified and translated into a regional discourse on Dalit rights, while in the second, local activism on recovering Dalit land requires tactical camouflaging of the caste basis of claims.

Dalit NGO Networking and its Implications
In the late 1990s, Dalit NGOs in Tamil Nadu were part of a complex organizational field. In each district or region key NGO leaders had risen to prominence with ideological and material influence over clusters of small NGOs, some established by former members of their staff, some funded by channelled foreign contributions. These local NGO clusters were provided with political and legal support to oppose caste injustice, sometimes through alliance with Dalit movements or political parties active in the regions, which in turn gained from the spaces and resources opened up by NGOs. 18 Foreign donor INGOs played a part in forging such networks. From 1999, Oxfam -having endorsed a shift to Dalit rights as part of its exit policywas bound to increase rather than decrease support to vulnerable NGOs in need of training, legal backing and advocacy from higher-order networks. District-level Dalit NGO groupings were thus consolidated into named networks linked to the State-wide body, NESA (now funded by Dutch Oxfam-Novib). This was by no means a harmonious set of alliances around the agenda of 'Dalit rights'. The Dalit rights agenda was ambiguous and the new networks heterogeneous. They enrolled NGOs with different views and approaches, providing a new context for competitive NGO identity differentiation, while reshaping debates (see Mosse, 2012b). There were new points of contention: was the focus on caste -as Dalit or specific jātis (castes) -a virtue, a necessity, or a risk? Did we confront, or were we co-opted by, the state; and were Dalit political parties to be sought out as protective, or vilified for compromise, corruption and kat . t .ā panchayats (that is, conflict broker dealing)? NESA members focusing on microfinance might be judged for their caste-inclusive individualizing schemes, or for adopting a 12 David Mosse and Sundara Babu Nagappan 'Chettiyar' (dominant merchant 'banking' caste) cultural style, including religious gifting and public feeding -as witnessed by David, sitting in the charitable food hall of a particularly large Dalit microfinance NGO in central Tamil Nadu in 2009. Some NGO workers also criticized the phototaking, report-writing, fame-seeking, time-wasting, non-local human rights NGOs failing to follow through or grasp specific contexts and tactics of Dalit struggle. 19 The post-Oxfam new-entity NESA financed and organized information and training support (on legal rights or state schemes), public awareness and policy lobbying events across Dalit development 'sectors' including anti-atrocity human rights, Panchayati Raj (elected local government), Dalit women's issues, natural resources, and microcredit. These 'sectors' each centred on different member networks (see Mosse, 2012b (NESA, 2003: 12).
Of course, the networks connected through HRFDL were heterogeneous with distinct regional histories, types of NGOs, big-personality leaders and political connections. As fields of power and reputation, network interrelations were characterized by a mixture of collaboration, competition and antagonism. They had different contexts and prioritized different concerns. For example, NGOs in two southern Tamil Nadu networks worked in the context of assertive Dalit (Pallar caste) politics and retaliatory violence; those in the western region responded to persistent forms of debt bondage (agrarian and industrial), or illegal 'manual scavenging'; in other districts NGOs came together over the exploitation of female migrant workers, diverted state housing or credit schemes, threats to Dalit officers, alienation of Dalit land or commons resources, or caste-based sexual violence. Linkage to HRFDL allowed the injustices they confronted to be 'scaled', becoming legally visible and actionable through supra-local formal mechanisms, monitoring and fact-finding missions. Among the large number of cases 'scaled' through HRFDL's Fact Finding Team and Dalit Lawyers' Forum, included in just one annual report (HRFDL, 2007) were a case of the eviction of Dalits from land assigned for post-tsunami housing for fisherfolk in a coastal village, the barring of Arunthathiyars (Dalits) from 13 fire-walking rituals and temple entry by dominant castes, and (in another village) attacks by upper-caste Reddiyar youth when Dalits conducted their own festival procession accompanied by honorific mēl . am temple drums and nākacuvaram wind instruments. Police complaints led to a 'social boycott' imposed on the Dalits: refusal to sell them goods from village shops, cutting their cable TV connections, blocking their drinking water supply. The Dalit sub-postmaster in this caste-divided village was prevented from conducting his office.
Also reported from other villages were: Dalit exclusion from common property fishing and the right to cut fodder grass; a teacher telling Dalit students they had no reason to register themselves for a Republic Day sports event; several cases of insults and assaults of Dalit men and women in public spaces including on buses; the rape of a Dalit woman and a young woman's suicide believed to have been induced by upper-caste harassment; and two murders of Dalit Panchayat Presidents in the same (Tirunelveli) district just three months apart (in 2006-07) (HRFDL, 2007). HRFDL responded to petitions, gave 'moral support', provided legal advice on filing cases under the SC/ST Prevention of Atrocities Act, pressed abandoned cases with the police and revenue officers, organized public agitations, and arbitrated through village-level 'peace talks'.
The network's judicialized activism (Bornstein and Sharma, 2016) raised and recontextualized such issues in regional, national and international oversight bodies (State or National Human Rights or Scheduled Caste Commissions, the UN Commission on Human Rights) and used instruments such as State-wide consultations and public hearings with jury panels. For example, a national public hearing was held in June 2005, co-organized with the Liberation Panthers Party (Vit . utalai Ciruttaikal . Kat . ci) and NCDHR, on the 10-year-long exclusion of Dalits from Panchayat elections by locally dominant Thevar castes in certain constituencies reserved for Dalits. 20 As Lama-Rewal (2018) explains, although not formal judicial procedures, such public hearings prepare the ground for trials, and point to national or international legal frameworks. They articulate the specific instance so as to point to the general issue; and the evidence/documentation of Dalit rights abuse, and the emotionally laden performances by victims of atrocities who are selected and rehearsed as witnesses, serve to educate and sensitize their audiences from affected communities, the public, media and state officials (ibid.). While it is true that the state's unwillingness to act on the findings of public hearings and judicial commissions shows the limits of such NGO activism, 21 when performed as 'a representation in the theatrical sense 14 David Mosse and Sundara Babu Nagappan of the term' (ibid.: 13) these public hearings nonetheless had important discursive effects.
Affiliation to the network as a scaling effect also provided small Dalit NGOs with a kind of institutional capital. 'Power contagion' was the NESA Executive Director's phrase for entry into imagined collectivities at State and national levels. Then, the network could be scaled horizontally, imagined (especially by its INGO donors) as a mass organization, a State-wide federation of NGO-promoted community-based organizations (CBOs)the Nayaya Sangama -to which the network would, in time, be handed over (Barton et al., 2000: 16;NESA, 1998;Oxfam Novib, n.d.). Lower-order NGO networks and a grassroots base were needed as a legitimizing constituency, such as to engage the attention of Dalit politicians, who could then be advertised in events, rallies and mass poster and signature campaigns. These included the campaigns preceding the 2001 Durban anti-racism conference, or the 2003 Asian Social Forum (Hyderabad) and 2004 World Social Forum (Mumbai) which articulated caste issues into other global agendas such as anti-globalization and claimed -questionably, according to Jaoul (2017) -to represent ordinary Dalits to international publics. 22 By the end of 2004, when David visited, NESA-HRFDL comprised a powerful representational machinery. Through its vast array of trainings, events, fellowships, research, films, publications and public hearings, this NGO super-network had established a Dalit political-cultural optic for viewing everything from microfinance to HIV/AIDS, and pointed to the interlinked capacities of some 216 NGOs and a mass constituency of 2.5 million Dalit women and men. HRFDL worked because it created a 'discursive field of action' that interlinked diverse Dalit rights actors (Alvarez, 2009: 177) while engaging local and regional publics, State and donor aid institutions, including the UK's DFID. In fact, former expatriate DFID advisors interviewed during the research indicated that they had sought connections to varied Dalit NGO networks to support different positions on caste in aid programming in the early 2000s. 23 The NGO network not only sustained a discursive field of Dalit rights, it also produced itself as a parallel to -even a simulation of -Dalit social movements. It was an unregistered, flexible, overlapping and shape-shifting entity that reached across regions, social (caste) groups and party-political affiliations. Goal ambiguity allowed enrolment onto, and the shaping of, the network by diverse members and purposes (see Youngblood, 2016). It was activated around particular events/issues and, like a social movement, found focus in the personalized fields of key leaders with followings. These were the convening nodal NGO heads -one was the focus of activities on human rights violations, another on Panchayati Raj, a third on Dalit women's 22. Anandhi (2017)  issues -through whom, as a matter of pragmatic necessity, HRFDL was aid-funded as a programme (of individual NGOs) who then further dispersed donor funding as grants to smaller NGOs in their networks. Power and position came from being an intersection in the channelling and circulating of resources -a cultural style of leadership familiar to scholars of southern India, from Dirks's (1987) work on kingship to Bate's (2009) on Tamil party politics. Absent was any centralized control or accounting. The function of the NESA-HRFDL secretariat was itself incorporative rather than administrative: enrolling more and more people, displaying widening connections in larger-scale public conventions, rallies etc., such that those involved would, in recollecting, stress the scale and effervescence of a given event, the reach of the network, over particular agendas. Meanwhile, committed to promoting a movement, the key donors emphasized non-interference, decentralization and participative informality. 24 As Riles notes, such network activity has a 'dual quality as both a means to an end and an end in itself ' (2001: 51). Applied to NESA-HRFDL, the 'network' was a powerful mobilizing metaphor (such as 'community' might be), not a descriptive label for a set of actual inter-NGO relations or community associations. As a 'storied network', a construct that did not exist before or outside of its enactment in happenings and events, HRFDL came into existence in what the NESA convenor later referred to as 'bouts of visibility '. 25 There was, as Edelman (2005: 39) observes of Central American peasant organizations, 'a proclivity for demonstrating the effectiveness of the network with reference to its own self-description and activities'. NESA-HRFDL demonstrated a remarkable capacity for reflective self-documentation. Watkins et al. (2012) argue that NGOs with goals of long-term transformation, and lacking technologies with known input-output relations, typically encourage mobilization rather than doing, and 'behave in ways that resemble culture-producing organizations, such as those that create and market films or women's fashions ' (ibid.: 289). But to view HRFDL as cultureproducing in this sense is not to diminish its powerful discursive and mobilizing effects, underpinning the 'human rights agenda' as an imagined global polity and producing its local constituency (ibid.: 286). Certainly, between 1998 and 2006, NESA-HRFDL helped create a discursive space that attracted international donors back to south India; donors such as the French association of Catholic movements, or CCFD, 26 which would directly fund HRFDL (sidelining NESA) for the next decade, as an investment in what was conceived as a 'self-organizing system' of Dalit capacity building and 24. The NESA Executive Director tried deliberately to weaken his control over the process, and a donor Programme Officer threatened to resign rather than intervene in a critical HRFDL leadership struggle (

Network Instability: Discursive Effects and Organizational Relations
NESA-HRFDL can be considered as two quite distinct phenomena. At one level, 'the network' was, as explained, an end in itself, above all a discursive effect that intentionally blurred distinctions between people's organizations, NGOs, networks and donors, masking financial-managerial power relations and strategies. At a second level, NESA-HRFDL was a set of organizational relationships. It was a long-standing inter-NGO order of reputation and status, patronage or factionalism, expressed in idioms of personal ties, ideological affiliation, programme territoriality or caste membership. In practice, donor field offices (with Oxfam and CCFD variations) developed fund-mediated relationships within this institutional order, as part of the donor-NGO-sangam clientelist system (Picherit, 2018: 375). All those things that disappear at the level of 'the network' reappear in the realm of organizational relations. As HRFDL grew in reach from 2004, CCFD invested in the strategic value of a vast network of Dalit-rights NGOs whose movement work interconnected caste politics and Dalit human rights, creating scale and discursive fields of action that enhanced capacities (at many levels) to respond to anti-Dalit injustice. Actors at all levels, not just donors, invested in the network in this form, collaboratively producing and enacting its self-representation. Success relied on a disjuncture between this and the network as a structure of NGO relations, interests, routes to funds and power, and input-output audits (see Mosse, 2005). This disjuncture was actively maintained as INGO donor reports constantly bifurcated HRFDL, locating all the 'positive gains' ('upwardly connecting', 'grounding Dalit advocacy' etc.) at the level of the network, and all the 'problem areas' (leadership, ownership, financial or output monitoring etc.) in the realm of organizational relations (Chellathurai et al., 2006: 4).
The inherent contradictions between representations and power relations were not, of course, unnoticed by actors. As one Dalit NGO head said: 'So you [donor] support only networking, so I come with my own network because I want to be leader of that network; and somebody else wants to have a network to lead. So, who is bigger; who has more groups [NGOs]'? (paraphrased). And, he went on to ask, is the NGO an element of the 'network' or the 'network' a programme of the NGO? 27 The 'common good' of the superordinate network was often, in practice, subordinated to the interests of 27. Interview, Dalit NGO head, Tamil Nadu, October 2009. the individual NGOs, especially when NGO leaders experienced networks as a constraint on their agency, rather than extending its reach.
Correspondingly, community-level groups did not prioritize or produce supra-local networks in the way NGOs did (Wichterich et al., 2006: 22, 26-28). While NGOs emphasized their role as agents of grassroots capacitybuilding, CBOs regarded them as a source of economic benefits tapping into donor resources, such that one donor evaluation asks 'are NGOs empowering autonomous community-based organisations, or are CBOs being created in the form of NGOs?' (Barton et al., 2000: 10). And while donor staff emphasize process, devolution and non-interference, they work concretely to ensure influence.
So long as the goal was promoting a Dalit rights discourse, HRFDL's concealed contradictions mattered little. But when treated as an instrument for other purposes, the fragile articulation of network idea and organizational process faced a crisis (Knox et al., 2006: 131). The problems began with a physical devastation to the Tamil Nadu shoreline by the Southeast Asian tsunami of 26 December 2004, particularly affecting Nagapattnam and Cuddalore districts and Karaikkal. The international fundraising and emergency response was huge and rapid. However, even by January 2005, it was clear that caste discrimination pervaded the relief effort (Gill, 2006). Dalit communities (the 58 hamlets affected) were barred from relief supplies, officials did not visit their hamlets, non-Dalit survivors avoided food touched by them. They were sheltered separately since upper-caste marriage halls and the like refused them. Their land was used for the burial or cremation of victims of the disaster, while it was Dalits who were called upon to handle the corpses. Later it was apparent that Dalit women were subject to routine sexual exploitation and HIV infection by post-tsunami construction workers (Dalton et al., 2009). This systematic discrimination is now well known, thanks largely to the coordinated campaign of NESA-HRFDL (and NCDHR). But when HRFDL was turned into a relief-rehabilitation agency, responding to public opinion pressure from Europe, handling large multi-donor budgets with short time frames and weakly imposed accounting, this taṅka-tsunami ('gold tsunami') amplified existing tensions. The things that allowed the NGO network to appear as a social movement -role ambiguity, overlapping network territories, person-centred spheres of influence -were incompatible with the requirements of executing large-scale rehabilitation works (especially house building after 2006).
The multiplication of donors incentivized bilateral NGO-donor relationships, weakening the network collective; and the propensity of fundchannelling member NGOs to act as if their own office was the Secretariat of the entire network intensified discord. Organizational inequalities were amplified by uneven access to rehabilitation funds (between coastal and inland networks) and rendered permanent through the institution building enabled by post-tsunami finance. The mutual scepticism of NGOs facing 18 David Mosse and Sundara Babu Nagappan characteristic input-to-output uncertainties, reliant on talk-based technologies (meetings, workshops, trainings, etc.) yet claiming specific impacts like empowered Dalit panchayat presidents or land acquisitions (Watkins et al., 2012), could, in the context of funded rehabilitation works, turn into more concrete accusations of misappropriation or fraud. In this climate, when the NESA Secretariat organized mass meetings at which the network-wide people's Nyaya Sangama members asked their own NGOs to open up proposals and budgets to public scrutiny, it prompted indignation. 28 NGO donors began to treat HRFDL, the network ideal, as if it were a structure of organizational relationships, failing to perceive the ontological disjuncture between the two phenomena. Post-tsunami, CCFD directed attention to the network as having problems that required structural solutions -restructuring, rationalizing district-level networks, centralizing coordination, governance and audit systems -the very things evaded in promoting the network as a movement. Although these changes never took place, CCFD's new structural view of the network was taken forward through a European Union-funded programme (2007-09) that involved the further step of 'projectizing' the network: complying with agency-standard Logical Framework Analysis, identifying a beneficiary population of 2 million Dalit people and a monitored set of activities under five themed areas. A process of 'NGOing' decisively replaced the collapsed movement work of HRFDL which was now registered as an organization, defined as a project implementation agency, with selected network leaders being theme coordinators. Quantified achievements (activities/participants) linked to 'outputs' were the measures of 'success'. Failure was construed (by evaluations in 2008 and 2009) as due to weak 'network management practices' or the 'limited understanding of network members regarding the overall project objectives', while addressing local-level Dalit rights abuses was a constraint on HRFDL's implementing agency role (Rugmini, 2008: 7, 21;also ADER, 2009). The new French co-funding donor intermediary, Association pour le Développement Economique Régional (ADER), eventually pulled back control and appointed its own experts and resource people (e.g., for training, for a District Panchayat forum) (ADER, 2009).
As the HRFDL network was instrumentalized as an implementation machinery, a new distinction was made between means and ends such that the Dalit NGO network processes were no longer coterminous with Dalit empowerment, but rather readable as defective project implementation, entirely reversing Oxfam's original reinterpretation. Meanwhile, the State Secretary of the newly registered HRFDL told David (in November 2009) that the concept of 'the network' was to be done away with. As often in development, failure was produced by policy -or at least co-produced by network actors and the donor framework.
28. Interviews, NESA, November 2009. Despite the conspiracy of success in donor reporting formats, by the end of 2009 it was evident to anyone involved that the HRFDL had lost coherence and credibility; the network receded into the files and spreadsheets of the agency computers. A few regional leaders tried to float inheritor or alternative fronts; but by this time key network players had redirected their energies by joining Dalit political parties; for instance, the HRFDL State Convenor became Youth State Secretary of the Liberation Panther Party, and another NESA sector head its State Organizing Secretary.
Among donors, there was stoicism. 'We had a beautiful organic plan', said a CCFD officer, an 'architectural plan': through NGO networks and Dalit social movement, the grassroots would be mobilized to create political influence on the state. 'We started to build the four walls and the roof collapsed. One of our components -the networks -had not responded'. And with hindsight she added, 'If this [HRFDL] had remained as a forum for Dalit rights it might still be there … [but] there was no one else to cover the Dalit victims of the disaster'. 29 Interestingly, when -in 2010, under new staff -CCFD reviewed HRFDL, the focus was not on organizational problems, or even the failed network concept, but on a crisis of the narrative of Dalit rights itself, triggering a 'de-Dalitization' of NGO policy. The 2010 programme review concluded that the politics of recognition (the focus on Dalit identity-based claims) had become divorced from the politics of redistribution (class-based struggle for resources), resulting in a fragmenting 'caste-communitarian' development. 30 A 'new' approach -territory-rather than identity-focused (which was really an old approach) -would bring together 'trans-casteist' coalitions of NGOs and social movements (Dalit, Adivasi, fisher, farmer) in mass action on livelihood issues such as land (see below). While the distinction between Dalit and non-Dalit was blurred, a revalorizing of 'mass action' against clientelist NGO networks reinstated the moral-conceptual distinction between social movements and NGOs.
CCFD's 'de-Dalitization' policy occurred within an increasingly authoritarian national political environment in which NGO Dalit-focused interventions were curbed and losing legitimacy. External funding had reduced or disappeared with donor withdrawal or loss of registration for foreign contributions following punitive goverment audits. Various other Dalit federations, unions and NGO networks were disbanded, amidst conflicts, compromises, recriminations and betrayals (as David Pitcherit, 2018 notes in Andhra Pradesh). This was not helped by the fissiparous effects of separate Dalit jāti-based strategies, and dispersal of the Dalit NGO leadership 29. Interview, former CCFD representative, Geneva, June 2011. 30. Interviews with CCFD regional Project Officer, London, August 2011. This coincided with a significantly broadened focus of the National Campaign on Dalit Human Rights and other rights organizations, from caste atrocities to socio-economic rights and development budgeting.
as some moved closer to political parties, others to national or international fora. Each of these factors requires further investigation, but together they have undoubtedly made it more difficult to sustain a development NGO focus on caste in southern India. 31 We have explained how 'scaling' through higher-order NGO networks was an effective response by Dalit activism locally blocked by dominant caste threats and bureaucratic inaction. But we also found that NGO movement work and its discursive effects were inherently unstable and, in this case, unsustainable because the network implied organizational relations (INGO and NGO) that were disruptive. But scaling was not the only response to entrenched local caste and bureaucratic power. We turn now to our second case which differently combined social movement and Dalit NGO activism. Here tactics of vernacular protest were deployed alongside NGO procedural activism, concretely targeting the local state for recovery of alienated Dalit land.

NGO Procedural Activism and Dalit Land Dispossession
The Dalit Man . urimai Kūt . t . amaippu or Dalit Land Rights Federation (DLRF) brings together (independently of HRFDL 32 ) 16 NGOs across six northern districts of Tamil Nadu. Its principal focus is recovery of what is called Panchami or Panchama land, that is 'wasteland' which had been allocated inalienably through the colonial Depressed Classes Land Act of 1892 to Dalits (then known as Pariahs or Panchamas), thus, not to be mortgaged or sold to any non-Dalit. At the time, the assignment of some 1.2 million acres represented a significant change in colonial government policy which had resolutely supported landlords in resisting claims to property from landless Dalit labourers over whom they asserted almost total control, and who were excluded as 'outer-people' in relation to the rights-bearing village community, the Tamilūr (Viswanath, 2014: 104).
So when, in the wake of the 1891 famine, Chingleput District Collector J.H.A. Tremenheere proposed an allocation of lands to Panchamas (unheard of beyond missionary farm settlements), he not only opened an avenue for economic autonomy, but also gave unprecedented recognition to Panchamas as resident villagers (ibid.: 96, 108). The persisting disquiet this provoked among elites is indicated by a comment printed in The Hindu newspaper in February 1923, that giving land to the Panchamas would be 'disastrous to the ancient religion of the country and its ancient social system … bringing ruin and disaster upon the landowning classes'. 33 Over subsequent decades, most of the assigned Panchami land was illegally acquired by non-Dalits for lack of the means to cultivate, in lieu of crisis debt, or on migration; some was transferred to temple trusts or diverted to farm forestry schemes -all enabled by the poor understanding Dalit owners had of the conditions of assignment (Moses, 2003). 34 Activists regard such systematic failure of state land allocations, including other land gift or land ceiling redistributions, as indicating a Dalit condition of 'chronic dispossession'. 35 The DLRF network is led by the Dalit movement (Ambedkar Pēravai) leader C. Nicholas, one of the first-generation 1970s liberation theology trained Dalit NGO leaders whose Integrated Rural Development Society (IRDS) turned to the issue of land recovery in the 1990s. This was partly to revive collective action among Dalits which had been attenuated by their increasingly dispersed and precarious livelihoods resulting from changing rural/urban labour markets (see Shah et al., 2018).
The turn to land brought contention. Land is awkwardly political in Tamil Nadu amidst marketization, intensified demand for industrial sites, corporate land grabs, land banks, and a land-broker state. Secondary processes of speculation mean that 'land appears to be more a store of value than a source of livelihood' (Vijayabaskar, 2010: 42;also Levien, 2013), and given land's lost agricultural value, pressured debt-related sales result. But land is also revived as a store of social value for Dalits. Indeed, Nicholas was not just leading a 'Polanyian countermovement' (Goodwin, 2018) against land commodification, but invoking the importance of land as the basis of social belonging, a share in the village wealth, sovereignty, power and status, coded as the legal right to land and historically denied to Dalits. Land was viewed as an end in this sense, and not just a means: 'I have to have land to be a person', a Dalit woman claiming Panchami land told Sundara Babu; 'to have peace (ātmā cānti)', said another. Landlessness is like 'refugee status' said an activist at a workshop where it was agreed that land is about the life within it, and land rights (man . urimai) a matter of 'life-soil' (uyir man . ). 36 While Panchami land was emphatically a Dalit issue (repeatedly raised by Dalit movements), the DLRF sought to take the issue out of 'substantive caste politics' and pursue an approach of 'procedural advocacy' (to use Waghmore's 2013 distinction), not least because of the violent and lethal caste confrontation over Panchami land in the 1990s (notably Dalit deaths 33. The Hindu, 23 February 1923, quoted in Viswanath (2014.

22
David Mosse and Sundara Babu Nagappan from police gunfire in Karanai village in 1994) which had largely failed to sustain either land acquisition or the Dalit land movement itself. 37 Nicholas, who himself witnessed bloody caste riots in Villupuram district in the late 1970s, recognized that as a matter of rights the critical battle over land was to be fought within the bureaucracy not within the sights of police rifles, and that the state apparatus itself (not caste communities) was the critical site of networked caste power. Moreover, contrasting with other Panchami struggles, Dalit women would be centred as the principal public actors and claimants. Control over knowledge, and what was revealed or concealed in land records, was the necessary first battle.
In simple terms (although far from simple), land records were used to identify Panchami land, to cancel existing titles and allocate land to landless Dalit families, registered in the name of (mostly married) women, the secure custodians and managers of assets. Until Chief Minister M.G. Ramachandran's 1981 Tamil Nadu Abolition of Village Officers Act, records were in the possession of hereditary upper-caste village officers, the karn . ams, and it was virtually impossible to identify Panchami land. Sundara Babu remembers as a boy nervously crossing the agraharam (Brahman street) past the third house where he would look up with awe to see the karn . am sitting on a high porch, like a high-ranking IAS (Indian Administrative Service) officer, with texts on his writing board that could only be described as sacred scriptures, innate to his office.
With the transfer of the office of karn . am to Village Administrative Officers (VAOs), some being Dalit through reservations, and the passing of the Right to Information Act in 2005 (giving access to court judgments and weakening bureaucratic deniability), the sacred-texts-become-government documents were now, for the activists, 'liberation documents ' (vit . ivu pattiram). Dalit NGO workers, invariably seen carrying a heavy load of land record files, ready to pull out for reference the relevant numbered file with details on the first petition, etc., instantiate Hull's view of the bureaucratic file as a technology (an actant) that makes a phenomenon real, a case actionable (2012: 116). Activists spoke to Sundara Babu of the sense of pride at knowing the land survey maps and records better than the Revenue Officer, at having command of those things previously denied (knowledge, records, property) 'which made us Dalit'; 'removing one by one the chains of slavery'. Bureaucrats, in their turn, trivialize such self-esteem, mocking Dalits who come to the office to pursue a case, saying 'file number such and such has come'. 'So we are reduced to a file', says a Dalit woman; and when she turns to object, saying that, 'what is fundamental to us is being ridiculed', the officer quickly replies, 'Oh, no, no, I didn't mean that, I was just referring to…'. 37. The Dalit Panchami Land Restoration Movement followed commissions of enquiry into the violence, empowered by the Madras High Court ruling in 1992 that all 'DC' (Depressed Classes, i.e. Dalit) lands should be returned to their rightful Dalit owners (see Samraj, 2006).
However, the bureaucracy is not moved to action by documents themselves. The 'bureaucratic ground for favorable decisions' has to be prepared (Hull, 2012: 155) by the popular support built in villages in the form of the Ambedkar Pēravai (or 'Front'), an aggregate of youth/women's associations in whose name (rather than the NGO's) mass submissions were made by Dalit women at the sub-district Revenue Officer's jamabanthi-, or land tribunal, to assign Panchami land. For example, in 2008, 600 people submitted petitions at one jamabanthi for around 220 acres of land in eight villages of Villupuram district. The petitioners' written demands to officials for their legal rights were unclothed in conventional idioms of praise and prayer (Cody, 2009). When the response to the demands was delayed, street protests followed, as well as poster campaigns in local towns, naming officials who were blocking the process. When the case continued to drag on, the Pēravai organized a roadblock, tying up cattle and goats at a highway junction; impromptu shelters were erected, and women created hearths and started preparing food on the road -it was an occupation. The Revenue Division Office was gheraoed (encircled/picketed); a memorandum was sent to the Chief Minister in Chennai and Union Ministers in Delhi.
Practices of the state are used to put pressure on the state. For example, the erection of notice boards on plots of land declared as Panchami becomes officialized and ceremonialized with photographs of the Tahsildar (sub-district revenue officer) in situ, so that when upper-caste landowners rip them out, it is the Revenue Department that has to respond to the acts of trespass and vandalism. Pēravai notices carry maps, statistics and warnings of legal action, including calls for the Tahsildar to be arrested under the PoA law for inaction on an atrocity -the appropriation of SC land. Here the PoA signals the legal as much as the caste basis of the claim, and it is a warning to the state as much as to landowning castes; but it is also able to implicate both potential opponents. So, when a group of Vanniyar (non-Dalit) women arrived to reoccupy a plot of Panchami land (mirroring Dalit street tactics) claiming rights from having cultivated the land for many years, the NGO workers insisted that the police note it down as evidence of a violation against Dalits, with the accompanying duty/threat of arrest. In fact, here the PoA allows conflict over Panchami land to be individualized so as to avoid caste conflict. Disputes, as Nicholas insists, are 'only between the illegal occupants or encroachers of "condition land" and landless Dalits'. 38 In practice, use of this law is a device to draw police attention, and attention to the police response.
Everything is done by the NGO activists to prevent the land issue triggering a caste backlash because turning this into a 'law and order' problem strengthens the hand of the bureaucracy. Potential conflicts with Vanniyar caste landowners (mostly small farmers) are repeatedly deflected to the 38. Interview, November 2010. In other contexts, the PoA has triggered intensified caste politics, cementing dominant caste party political agendas (Carswell and De Neve, 2015).
bureaucracy, the unseen site of caste power. But Dalit informal social links to the Revenue Office are weak, so transactions there -the documents placed on the officer's desk -are transposed to the public arena of the pre-announced street protest. Here the legal basis of the claim is advertised with printed posters, naming Pēravai movement leaders and lawyers, accompanied by their photographs and phone numbers, to publicly invoke the backing of a popular Dalit movement. The NGO leaders themselves parade political power to the bureaucracy. Eschewing the belted, tucked-in pant-shirt of the bureaucrat in favour of politician's white dhoti, male leaders like Nicholas signal power that sits astride the oft-criticized party politics and NGO anti-politics; and when Nicholas's wife, elected Counsellor and Ambedkar Pēravai women's leader, arrives at the Revenue Office with an entourage of community animators and DLRF activists on a courtesy visit, ahead of land petitions, and is guided swiftly to the Tahsildar's cabin and there greeted with respect and nervousness on the face of the senior bureaucrat, those in the office are in no doubt that influence lies behind the file-clutching animators.
But the animators, like their leaders, leave the caste basis of the claims unstated and emphasize the civil justice base. Their self-presentation in the Revenue Office carefully diverts attention from their own caste identity, for example, by signalling religious identities. So one presents himself as Christian, another arrives with forehead marked with sandal paste in a typical middle-class Hindu manner; the point being: when you go with a file on a caste question you emphasize religion. Quite different, then, from the caste-invoking erection of Ambedkar statues or cooking beef on occupied Panchami land which characterizes the cultural politics of other Dalit movements (Mosse, 2012a: 223). A reflexive awareness is brought to dealings with a socially heterogeneous bureaucracy. Animators build relationships with senior bureaucrats, often outsiders to the district -for instance, the female Brahman revenue official who they address as akkā, 'elder sister', but who with deft evasion and a polite bureaucratic claim to powerlessness, moves the file quickly to the Collector lest her name and family get implicated in poster campaigns in the town. Their real aim is to bring the power of a popular movement to dealings with those junior officials who participate in the same networks of caste power as landowners in the villages, and whose politician kinsmen will understand the significance of a potential vote-bank behind the NGO workers.
The recovery of extensive tracts of Panchami land re-registered for landless Dalit women (but evading caste-based contention) was a significant achievement. 39 This has not been without challenges, for example, retaining a floating body of animators, the erosion of villagers' interest by bureaucratic attrition, varied capacity across a land rights NGO network. But 39. According to C. Nicholas (pers. comm. June 2020) around 3,000 acres have been recovered from 'caste Hindus' in the northern districts of Tamil Nadu in the past two decades.
the principal theme here is not the unstable combination of 'movement work' and donor-linked 'NGOing' that we saw with NESA-HRFDL, nor the much-criticized 'absorption' of social movement activism 'into the bureaucratic structures of NGOs' (Ismail and Kamat, 2018: 572). Rather, the case demonstrates the performativity of Dalit rights work by what is simultaneously (and here non-contradictorily) an NGO and a social movement with political force, skilfully negotiating antagonistic relations of caste and state power through vernacular tactics of protest, well-honed to target local revenue officers. Of course, the external funding is not incidental, but the characteristic NGO 'disjuncture between funding streams and the social spheres from which they draw their support' (ibid.) is not the critical point. Indeed, as DLRF shows and Jaoul (2017: 618-20) notes, such funding allows autonomy from clientelist relations with the bureaucrats and politicians. Overall, we are led to question attempts to dichotomize the 'technical proceduralism' of the professional NGO and the sociability and 'theatricality' of Dalit political movements (pace Jaoul, 2017; see also Bornstein and Sharma, 2016). Put another way, this is a demonstration of the extraordinary skill, continuity of purpose, strong and savvy leadership, and resources required for Dalits to organize access to their legal entitlements locally, rather than by means of 'higher-order' platforms. But when this NGO network was expanded and re-purposed to the wider ambition of resisting the new policy (from 2007) of the Tamil Nadu government creating land banks and industrial parks for investors in new industry, enabled by the Special Economic Zones (SEZ) Act of 2005 (Bommier, 2016: 113), the NGO coalition came up against far more intractable challenges.

A Failed Claim against Corporate Land Acquisition of Dalit Village Land
The case in question was opposition to the takeover of common lands in the Dalit village of Thervoy Kandigai (henceforth, Thervoy) which fell within a regional hub for automotive industries around Chennai, by the State Industries Promotion Corporation of Tamilnadu (SIPCOT) for lease to the French multinational tyre company Michelin in 2007. This is not the place to describe what is an extraordinarily complex and protracted (seven-year) case in which Sundara Babu was a close participant-observer. But the salient elements of the case are that a DLRF-formed coalition led by C. Nicholas sought to resist the acquisition of Dalit village common land, held as critical to livelihoods. A familiar repertoire of local public demonstrations (sit-ins, roadblocks, hunger strikes fronted by women) was deployed in dealing with the State, and the campaign widened nationally, with representations to national commissions on Scheduled Castes and human rights.
Dalit identity and caste were, again, de-emphasized in order to widen the campaign. At the time, the Dalit Land Rights Federation generalized itself as the State-level Tamil Nadu Land Rights Federation (TNLRF) with an anti-SEZ, anti-globalization, anti-neoliberal agenda. Locally, the campaign tried to attract the support of adjacent villages by broadening the issue from land to industrial pollution and the threat to water and environment from a rubber factory on the doorstep, mobilizing an anti-pollution padayatra (on-foot protest), invoking the Bhopal disaster and circulating scare stories of babies born without eyes (Bommier, 2016: 146). The issue entered the agenda of national advocacy organizations, student bodies, a concerned urban elite and political party manifestos. Public interest litigation was a key weapon to enforce the duty of the Indian state to protect villagers' rights. Meanwhile the advocacy coalition extended to France, through INGO CCFD's mediation, where (from 2012) grievances were pressed against Michelin through appeal against violation of OECD guidelines for corporate responsibility in relation to human rights and social and environmental impacts.
In the end, the coalition failed to halt the factory. From this we learnand there is no better account than Bommier's (2016) PhD thesis -about the failure to establish Dalit entitlements around this land. Tactics combining local protests, judicialized activism and NGO advocacy, effective in local Panchami land claims, were now outdone by the interlinked interests of the corporations, the State-level bureaucracy, political parties, the Indian courts and the international regulatory apparatus.
It became clear that in this claim against land dispossession, DLRF lacked control over the terms of dispute. The high-level formal processes to which social activists turned not only upscaled or de-localized the issue, they also took the contention out of their hands. For one thing, the separation of public (state) and private (company) responsibilities protected industry from the legal claims of communities by allowing a global corporation like Michelin to disown its responsibility, which was discarded to mechanisms of the democratic local State who delivered it land as 'free from all encumbrances' 40 as it was stripped of vegetation and protected by high wire -and legalfences. Moreover, in this case the land was 'common' rather than privately owned land, marking the government's policy turn from 'a market logic of bargaining [for compensation of private owners] … towards a[n internal] bureaucratic process of land conversion within the state apparatus, without perceived need to engage with land owners' (Bommier, 2016: 80). Footholds for resistance to land loss were scarce.
Beside land category and procedural loopholes, what this case illustrates is the importance -and in this case the failure -of activism in the realm of agenda setting, or the meta-politics of 'framing' (Fraser, 2005 interest, for example jobs in industry not protection of village resources (Bommier, 2016). By the same token, when the activist coalition sought to politicize the land-grab issue (following this failure to get redress), they found space in the public arena for villagers' land rights diminished by the now-dominant narrative of industrialization and jobs (ibid.). As activists withdrew from processes within which they found themselves voiceless, and the space to politicize the land rights issue narrowed, the coalition itself lost coherence and people started to negotiate for compensation and jobs. As we have already described, there is a proneness to 'splitting' of coalitions tying together different agendas and constituencies into 'spectacular but temporary displays of connection across the continents, largely mediated by specialized organizations, technologies and entrepreneurs' (Tilly andWood, 2015: 123, quoted in Bommier, 2016: 335). The common cause, fabricated out of villages divided by faction, class, religious denominations and historical conflicts over water, was easily unravelled into individual bargaining.
The contests over agenda setting that enable or block claim making return us to where this article began, with efforts to recognize a field of Dalit rights. This is to say, the meta-politics in the Thervoy case concerned the framing of not only what is at issue (rights over land) but also who are the relevant subjects. In this case, Dalits -or specifically Dalit womenwere not constituted as rights-bearing subjects. Unlike with Panchami land, there was no underpinning legal basis to the land claim as Dalits. This was 'wasteland' under State title following reclamation from landlords through post-Independence land-ceiling legislation, claims to which resulted from de facto cultivation (under 'land to the tiller' legislation), not Dalit identity. In Thervoy, the land was uncultivated and the campaign unable to specify Dalit women as the deprived and properly entitled vulnerable group. Moreover, locally this was perceived as a Thervoy village not a Dalit issue. If anything, Dalits in neighbouring villages had competing interests over such land, especially as water catchments feeding irrigation systems.
Although, formally, the Dalit basis of the Thervoy claim was unstuck, land and identity were not so easily separated for those resisting alienation of their common land. As with Panchami, the claim to land was also a claim to the citizenship that, for them, land represented. Once the tyre company project was inescapable, those representing the village set out demands to Michelin in terms of permanent jobs, gas connections to replace lost fuelwood supply, reassigned grazing land, houses, etc. and, significantly, a 2 per cent share in the Michelin factory profits. The demands were such as to claim what land represented for Dalits, namely recognition and social inclusion as a historical rights-bearing community, and shareholders in the ongoing productive value of the land, hence in the corporation and its governance (Bommier, 2016: 169-70).
To Michelin, such claims were as barren of rationality as the unencumbered wasteland they industrialized (ibid.: 323-24). Dalit villagers could 28 David Mosse and Sundara Babu Nagappan only be treated as generic clients, a population of customers or welfare beneficiaries of CSR gifts (health camps, schools, water tanks). These were not entitlements, but philanthropic allocations, which also fuelled division and selective sell-out in already politically factionalized villages (ibid.: 207-08). Meanwhile, the company appointed mostly upper-caste contractors to supply trucks and bulldozers, and outsiders to permanent skilled jobs. Locals, regarded as unsuited to core roles of a 'technologically-advanced and efficient company', could at best access temporary unskilled work in construction, as guards or gardeners (ibid.: 323). While stripped of the ability to assert claim-bearing identities, when it came to job recruitment it was clear that Dalits continued to be subject to direct or indirect identity/categorybased discrimination as 'costs to the company' were reduced by 'delocalizing' the labour force and avoiding the social claims against work-time from local family events (marriages, funerals) or festivals.

SUMMARY AND CONCLUSIONS
We learn from this article how NGOs in south India brought caste into development discourse, and centralized Dalits as internationally recognized sufferers of injustice and bearers of rights. The vibrant Dalit politics of caste in the 1990s did not itself bring caste into international NGO development discourse as a kind of 'grassroots globalization', certainly not independent of processes internal to international human rights institutions and the organizational logics of INGOs like Oxfam producing strategic responses to goals of structural change and a rights-based approach.
In the first case examined, these responses opened new space and resources for a Dalit rights agenda, promoted through judicialized activism and region-wide network forms, in alliance with Dalit movements. These mobilized ideas as much as people (Alvarez, 2009: 178), reorganizing categories of INGO/NGO/CBO as means and ends. The 'scale making' articulated local caste struggles into the discourse and instruments of Dalit human rights, essential for small Dalit NGOs whose activism was stymied by local caste and bureaucratic power. But the wide-ranging network forms implied an amalgam of NGOs' largely discursive 'movement work' and their organizational relations, rendered critically unstable by repurposing the network to deliver post-tsunami rehabilitation. The resulting collapse of the network also marked an INGO turn from caste and Dalit rights in strategic policy goals.
The second case involved successful claims over appropriated Dalit land through procedural activism and vernacular tactics of protest that targeted the local state itself. While claimed rights as Dalits were embedded in the historic Panchami land category, contests were intentionally removed from the context of inter-caste conflict. But when the Dalits of Thervoy village sought to retain their village common land against State-mediated corporate takeover they failed in their claims both to land and to rights-bearing identities. NGO 'scaling' was rendered ineffective by external as well as internal effects. The coalition and campaign that appealed to international fora was unable (a) to pursue a 'first-order' issue of justice, that is rights to resources (land) in face of State-corporate interests; and (b) to contest the 'meta-level' injustice concerning 'who counts as a subject of justice' (Fraser, 2005: 80) that lay behind legal decisions: what counted (a claim on common land) and who counted (Dalits as shareholding citizens).
We have shown the importance of international policy and the agendasetting work of NGOs and social movements especially to the 'secondorder' framing of who counts (what identities) as subjects of justice and development. But we have also revealed the contingency of discursive framing that gives Dalits the 'right to have rights', and the significance of the inherent instability of the NGO-donor institutional space brought into play (Lewis and Schuller, 2017). Actor categories (INGO/NGO/CBO, networks, movements, fronts) and their moral evaluations, and definitions of means and ends, turn out to be open to manipulation, especially in India's 'crowded … market for social and political entrepreneurs' (Jenkins, 2010: 423). How action is represented is consequential, especially for donor-mediated investments and resource flows, and hence for those whose struggles for justice depend on them.
Despite such fragility, Dalit organizations (NGOs, movements, parties) did make significant advances in realizing rights, reducing untouchability or debt bondage, improving resource access, dignity and self-respect in the 2000s. Today's nationalist politics brings authoritarian restrictions to civil society activism and is notably more hostile to a focus on caste and mobilization for rights. By 2011 in Tamil Nadu use of the phrase 'human rights' in the name of any NGO had been banned. 41 Discussing such trends with Picherit (2018: 361), Ramappa, a Dalit NGO leader in Andhra Pradesh, comments, 'We used to do dharna, now we have to hide in secret places'. Pitcherit describes a sense of disenchantment (ibid.). Others describe the rise of clientelist relationships with dominant caste (mainstream) businesses, corporations, NGOs or political parties (Gorringe, 2017), often on unfavourable terms for Dalits who enter competitive pursuit of individual interests. The casualized, migratory work of the poorest limits the likelihood of local collective action among Dalit women and men who now face less visible and actionable forms of exclusion in the economy (see Shah et al., 2018). As caste discrimination is reproduced in the informal relations of the economy (Mosse, 2018), market inclusion is the context for caste 41. On 17 December 2011, the Times of India reported that 'the Madurai bench of the Madras high court ordered notice to authorities to ban the usage of the word "human rights" in the name by any society, NGO or association', upholding a 2009 formal resolution from the Tamil Nadu government. See: http://timesofindia.indiatimes.com/articleshow/11140093. cms?utm_source=contentofinterest%26utm_medium=text%26utm_campaign=cppst (accessed 3 April 2020). discrimination, just as in the political field democratic politics is the condition for anti-Dalit reaction: 'high democracy' with 'low civility' as Waghmore (2013: 199) aptly puts it. Behind this weakened Dalit civil society activism may be a global trend (Ismail and Kamat, 2018) towards tighter regulation of NGOs' rights-based work and its substitution by corporate social responsibility (CSR). Here, Dalits are 'misrecognized' (Fraser, 2005), refused recognition as historical subjects and allowed only to be consumers, clients or welfare beneficiaries of state/corporate programming. For all the instability of the brokering middle ground, NGOs did enable a kind of claim making and social recognition which is impossible within the new individualized state anti-poverty programmes or CSR welfare taking their place. By comparison, the NGOs earlier vilified as capturing activism or as2009 agents of capitalism stabilizing the neoliberal order (Kamat, 2002), seem positively benign. Meanwhile, it is suggested, polarized and less negotiable conditions of social humiliation produce new forms of Dalit activism, not only on the street but through film and especially social media; and new questions about the contemporary forms of social exclusion and assertion invite further ethnographic engagement.