Content uploaded by Andrew Flachs
Author content
All content in this area was uploaded by Andrew Flachs on Sep 29, 2018
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=fjps20
Download by: [Universitaetsbibliothek Heidelberg] Date: 27 April 2017, At: 00:19
The Journal of Peasant Studies
ISSN: 0306-6150 (Print) 1743-9361 (Online) Journal homepage: http://www.tandfonline.com/loi/fjps20
The ox fall down: path-breaking and technology
treadmills in Indian cotton agriculture
Glenn Davis Stone & Andrew Flachs
To cite this article: Glenn Davis Stone & Andrew Flachs (2017): The ox fall down: path-breaking
and technology treadmills in Indian cotton agriculture, The Journal of Peasant Studies, DOI:
10.1080/03066150.2017.1291505
To link to this article: http://dx.doi.org/10.1080/03066150.2017.1291505
Published online: 26 Apr 2017.
Submit your article to this journal
View related articles
View Crossmark data
The ox fall down: path-breaking and technology treadmills in Indian
cotton agriculture
Glenn Davis Stone and Andrew Flachs
Although India’s cotton sector has been penetrated by various input- and capital-
intensive methods, penetration by herbicide has been largely stymied. In Telangana
State, the main obstacle has been the practice of ‘double-lining’, in which cotton
plants are spaced widely to allow weeding by ox-plow. Path dependency theory
primarily explains the persistence of sub-optimal practices, but double-lining is an
example of an advantageous path for cash-poor farmers. However, it is being actively
undermined by parties intent on expanding herbicide markets and opening a niche for
next-generation genetically modified cotton. We use the case to explicate the role of
treadmills in technology ‘lock-in’. We also examine how an adaptive locked-in path
may be broken by external interests, drawing on recent analyses of ‘didactic’learning
by farmers.
Keywords: agricultural industrialization; path dependency; knowledge production;
biotechnology
In India, Green Revolution wheat may have led the penetration of industrial agriculture in
the 1960s (Cullather 2010), but today cotton is the main driver of capital- and input-inten-
sive farming. Until late in the twentieth century, most of India’s cotton was grown with low
external inputs, often using seeds of indigenous cottons. But by the late 1980s a revolution
was underway, based on seeds of the more factory-friendly New World cotton. Unique in
world cotton production, these new seeds were hybrids, which locked adopters into the seed
market. They were also fertilizer- and water-intensive, and required heavy insecticide appli-
cation since they lacked resistance to Asian pests. The input-heavy cotton package was
expensive, but promoted by government policies (including rapidly climbing support
prices), encouraged by the agro-scientific establishment, and buoyed by favorable
markets, it spread rapidly. Many farmers saw early profits; the intractable problems with
agroecology and debt would not come until later (Mohanty 2005; Stone 2007; Kennedy
and King 2014,inter alia).
But while doors to purchased inputs –including hybrid seed, fertilizer, irrigation, insec-
ticide and, more recently, genetically modified (GM) traits –were being flung open on
cotton farms, the door to one key industrial input –herbicide –remained fastened. In Tel-
angana (Figure 1), a relatively poor state in India’s cotton belt, the main barrier to herbicide
penetration has been a set of ensconced practices we will term ‘ox-weeding’, based on the
trusty ox plow, human labor and functionally linked planting conventions. This cultivation
system is efficient and advantageous for cash-poor farmers. It relies on household labor and
© 2017 Informa UK Limited, trading as Taylor & Francis Group
The Journal of Peasant Studies, 2017
https://doi.org/10.1080/03066150.2017.1291505
work exchange; draft animals that may be owned, rented, or borrowed; and simple familiar
tools. With respect to cycles of innovation and change, ox-weeding is a stable system in
contrast to the technology treadmills discussed below.
Path dependency theory helps to explain how inefficient and sub-optimal technologies,
practices, and institutions may prevail and persist by being ‘locked in’(Arthur 1989). The
cotton ox-weeding system may also be seen as a technological path, locked in by local
knowledge and by social institutions with which the technology articulates. But despite dis-
tinct farm-level advantages, ox-weeding has come to be seen as increasingly inconvenient
to agricultural capital and the developmental state, which view it as a backward path
obstructing the penetration of herbicides in the cotton sector. This is a particular priority
to these external interests for two reasons. First is the herbicide itself: Indian farmers
plant the world’s largest area to cotton and buy over USD 2.5 billion worth of insecticides
yearly, but spend only USD 350 million on herbicides (Sapale and Malani 2015). The
potential for herbicide market growth is enormous, and industry looks for sales to reach
USD 800 million by 2019 (Sapale and Malani 2015). The second reason pertains to the
opening of a market for GM seeds with herbicide tolerance (HT) traits. HT traits are the
biotechnology industry’s biggest money maker by far, with 86 percent of the world’s
GM acres in 2015 containing plants resistant to glyphosate or glufosinate (James 2015),
and a new generation of crops resistant to 2,4-D and dicamba now arriving.
1
However,
the only GM crop now sold in India is Bt cotton, and biotechnologists face significant head-
winds in getting HT crops approved for sale, including a Supreme Court technical expert
committee that recommended a ban on HT crops. Opening weed management systems to
herbicide use improves the eventual market for HT crops and may even enlist farmers as
advocates for HT crop approval.
Figure 1. Map of India, Telangana, and sample villages in Warangal District.
1
This includes 95.9 million hectares of HT-only crops and 58.5 million ha of crops in which the HT
trait is stacked with a Bt trait (James 2015). Bt traits, which provide some protection against insects,
are discussed below.
2Glenn Davis Stone and Andrew Flachs
Longitudinal ethnographic research among cotton farmers in Telangana has allowed us
to observe the undermining of the established ox-weeding regime to promote herbicide-
intensive production.
2
This situation offers an instructive case study in how farmers may
be diverted from a seemingly ‘locked in’technological path to become more ‘modernized’
producers and better customers for products currently on the market (herbicides) and in the
pipeline (new GM seeds). We use the trajectory of agro-technological change in Telangana
cotton farming to make two significant contributions, both related to path dependency
theory.
The first contribution concerns technology treadmills, defined here as situations in
which use of a technology encourages or necessitates the use of that technology in increas-
ing and/or ever-changing forms (cf. Levins and Cochrane 1996). Agricultural technology is
known to offer examples of path dependence and lock-in mechanisms (e.g., Cowan and
Gunby 1996), but the treadmill is a key feature of industrial technology penetration in agri-
culture that has not been appreciated or theorized. Agricultural treadmills are distinctive in
that they actually reverse a basic mechanism of technology lock-in because of how agricul-
tural technology interacts with local knowledge and decision-making systems. Whereas
Arthur (1989) describes how the benefits of a technology increase by the learning that
accompanies its use, and David (1985) describes technology lock-ins where the costs of
learning skills prohibit change to more efficient practices, agricultural treadmills are
plagued by erosion of local knowledge that ironically encourages even more intensive
use of the technology.
Second, we use a longitudinal analysis of the ongoing undermining of the ox-plow
system to understand the process of instrumental path-breaking by outside interests. Path
dependency theory says relatively little about how and why paths are broken, focusing
more on how paths emerge and persist. When paths are broken, the cause is usually seen
as ‘exogenous innovations’or ‘external force or shocks’that transform existing relation-
ships (David 2001), but such innovations and shocks are generally not instrumental –
i.e., intentionally designed to break a dependent path. The few writers who have looked
seriously at mechanisms of path-breaking (Martin 2010; Martin and Sunley 2006) have
not attended to such instrumental path-breaking. Since in our case it is specifically agricul-
tural technologies being targeted by outside actors, we turn to theories of agricultural
decision-making to explain how such a path could be broken. Theories of agricultural
decision-making have had their own blind spot for instrumental interventions, with most
analyses focusing on the apolitical processes of ‘environmental learning’(basically exper-
imentation and observation) and ‘social learning’(basically emulation). But recent theoreti-
cal writing introduces the process of ‘didactic learning’whereby farmer decision-making is
instrumentally influenced by external parties (Stone 2016). As agricultural didacts charac-
teristically pursue their own interests from positions of relative power, this perspective is a
political ecology of agricultural practice. By focusing on the changing ecology of cotton
2
Fieldwork was conducted between 2000 and 2015 in the Telangana region of Andhra Pradesh State,
which became Telangana State in June 2014. During this time ethnography was conducted in over a
dozen villages in Warangal District, selected to represent variations in soil, caste and prosperity. Stat-
istics presented here are drawn from five of these villages, as shown in Figure 1, for which repeat agri-
cultural household surveys were administered to random samples stratified on land holdings. Sample
sizes ranged from 144 to 254 households, except for 2009–2010 when sample sizes were only 43–44,
and 2015 when sample size was 311. Because our sample includes farmers from diverse castes, irri-
gated and rainfed farms, and from villages both proximate to cities and in isolated rural areas, we
argue that our claims are widely applicable to cotton agriculture in semi-arid India.
The Journal of Peasant Studies 3
cultivation, we are able to analyze the path being broken, the economics of why it is being
broken, and the process by which it is being broken through didactic learning.
Beyond these theoretical aims, this study has practical importance. Indian cotton
farmers have struggled with treadmills involving numerous technologies including seeds,
insecticides and, most recently, genetic traits. (These separate technology treadmills have
their agroecological peculiarities, but share key interactions with farmer decision-making
as summarized below.) The future of GM cotton in India is unsure, but if HT seeds
make it to market, the result of the path diversion described here is likely to be an herbicide
treadmill as well. We believe that such ramifications of the adoption of GM crops needs to
be considered in a debate that is too often constructed in myopic terms of ‘success’or
‘failure’in the field (see Stone 2011a on shifting focus from the field to larger questions
of the farm).
We first provide a brief history of the spread of input-intensive commodity cotton pro-
duction and the resulting seed and insecticide treadmills. We show that this production
regime is marked by deleterious effects on local knowledge and decision-making, as the
conditions of technological change have generated a dysfunctional learning environment.
We then explore the effects of GM seed and industry’s aspirations for HT seed and herbi-
cide. This section reveals how the ox-weeding system that is being depicted as an impedi-
ment to agricultural success is effective and inexpensive for many farmers (recent increases
in labor costs notwithstanding).
We then review the situation with herbicide in India, including recent developments in
law, labor and agricultural technology that various parties are capitalizing on to make
inroads for herbicide penetration. We look at the interventions to subvert the ox-weeding
system, beginning with the seemingly obscure but actually critical detail of seed placement.
Seeding arrangements are functionally linked to other moving parts in the cultivation system,
and by altering those arrangements didactic learning interventions are leading farmers toward
a dependence on a new path toward increasingly capital-intensive agriculture.
Finally, we offer comments on the effects of the path-breaking on cotton farmers.
Cotton input treadmills and local knowledge
Hybrid cotton seed, which is planted on a large scale only in India, was invented at a public
agricultural university in Gujarat in the 1970s. Once commercialized, the technology gained
a foothold in southern and central India in the 1980s and then spread rapidly in the 1990s
(Lalitha, Ramaswami, and Viswanathan 2009; Stone 2011a). The early adopted seeds were
from public agricultural institutions, but during the 1990s these were largely replaced by
seeds from private corporations. Reflecting corporate interests, these seeds were input-
intensive by design, optimized for response to chemical fertilizer and irrigation. They
were largely based on the New World species Gossypium hirsutum, more industry friendly
both in production (with better response to inputs) and in output (with a longer staple pre-
ferred in thread production) (Prasad 1999). These seeds were also highly vulnerable
to India’s abundant insect pests, lacking the natural defenses of the indigenous species
Gossypium arboreum, and so they were tightly bound to the heavy use of insecticides.
The seeds-inputs package comprised a technological path that was quick to be locked
in, particularly by knowledge-related mechanisms. The interplay between technology and
local knowledge has been described and theorized elsewhere (Stone 2007,2011a,2016;
Stone, Flachs, and Diepenbrock 2014) but a summary will be helpful. In agriculture,
skilling is the ongoing process of farmers developing the ability to perform with a technol-
ogy (Stone 2007). Skilling results from both environmental learning (assessment of
4Glenn Davis Stone and Andrew Flachs
empirical payoff information on specific technologies and practices) and social learning
(emulation of others chosen on social criteria). Writers in anthropology and related fields
have long stressed the power of environmental learning, with the ‘experimenting, innova-
tive, adaptive peasant farmer’being commonly taken as the norm (Richards 1989, 20). But
Stone (2016) argues that this perspective obscures how difficult experimentation actually is
for practicing farmers and how payoff information may be inaccessible or unreliable, result-
ing in agricultural deskilling.
Such deskilling occurred in India during the 1990s and early 2000s as the explosive pro-
liferation of private seed companies and seed brands left farmers facing a bewildering array
of rapidly changing and often deceptively labeled seeds (Stone 2007,2011a). Environ-
mental learning was further hampered because yield variation within each seed exceeded
variation between seeds (Stone 2007, 83). Farmers’response to this fraught learning
environment served to lock in the high-inputs package, albeit by a different mechanism
than described by Arthur (1989) and David (1985). Unable to ‘skill on’specific seeds,
farmers relied on social learning to the point of herd behavior. Detailed longitudinal data
on agricultural decision-making show remarkable seed fads with scant links to the proper-
ties of the seeds (Stone, Flachs, and Diepenbrock 2014). Farmers also exhibited a desperate
penchant for new seeds, which further encouraged seed producers to bring increasing
numbers of brands to market. Farmers often expressed the belief that new seeds were
likely superior and that it was advisable to switch to new seeds at least every three years,
a demand that seed companies have been happy to oblige, with the number of cotton
seed brands in India climbing into the hundreds (GEAC 2012). Even before the introduction
of transgenic traits in seeds (below), this was a treadmill, driven by the intersection of insti-
tutional aspects of the cotton seed market and farmer responses to deskilling.
The high-input package has also been locked in by an insecticide treadmill, which
works somewhat differently than with seeds. With insecticide sprays, short-term payoff
information is readily available: the farmer is immediately gratified by the sight of insect
corpses. What is inaccessible is payoff information on the hazards of long-term use. Insec-
ticide sprays kill not only insect pests but the insect predators that keep pest populations in
check, and phytophagous (plant-eating) pests usually develop resistance much faster than
entomophagic (insect-eating) predators (Kranthi and Russell 2009). By the late 1990s,
most cotton pests showed resistance to insecticides (Kranthi et al. 2001,2002). But the
insecticide treadmill is not simply an entomological phenomenon, even if it is often charac-
terized as such:
3
deleterious effects on the skilling process are key to it. Farmers wind up
with not only new chemically induced pest problems, but problems for which they have no
store of local knowledge and practice from which to find remedies. There are two reactions.
One is to escalate sprayings, which both exacerbates the ecological problem and runs up
debt. Indian cotton fields that had been receiving 2–3 sprayings in the 1970s were receiving
over 30 by the early 2000s, with insecticide costs rising to over 50 percent of all inputs in
some cases (Peshin et al. 2008).
4
The second reaction is to seek sprays with new modes of
action, which farmers urgently do, providing a strong incentive for industry to develop new
insecticides to perpetuate the treadmill. Indian cotton farmers have progressed from
3
For example, ‘cotton follows the classic “pesticide treadmill”…namely, once the pests become
resistant, even more pesticides will need to be used’(Jin, Bluemling, and Mol 2015, 24).
4
In this environment, regular pesticide sprays also take on a social value associated with responsible
farming –in a 2013 interview, one farmer explained, ‘you should always seek to produce more than
your neighbors. If they spray four times, you have to spray five. That way, you’ll always have the best
yield’.
The Journal of Peasant Studies 5
organochlorines and organophosphates to carbamates, synthetic pyrethroids, sodium
channel blockers, fungal naturalytes, and the transgenic Bt seeds discussed below.
Again, constantly operating with sprays with which they have little experience leads to agri-
cultural deskilling and exacerbated treadmill effects: numerous studies have shown that
ignorance of agricultural insecticides leads to excessive spraying (Vandeman 1995;
Bentley 1989; Tripp, Wijeratne, and Piyadasa 2005; Chen, Huang, and Qiao 2013). In a
sense this is the opposite of integrated pest management (IPM), the knowledge-intensive
alternative to over-reliance on insecticides, in which increased knowledge leads to
reduced spraying (Cowan and Gunby 1996, 526–27). Such an insecticide treadmill reverses
the classic lock-in mechanisms in which skills and knowledge accrue to the use of a tech-
nology (Arthur 1989; Cowan 1990).
Both the seeds and the insecticides show us the importance of the knowledge economy
–especially farmer modes of learning and decision-making –in agricultural treadmills. The
loss of skill and knowledge that accompanies the replacement of a non-treadmill technology
is usually benign. The replacement of draft animals with mechanized vehicles rendered the
disappearing skills and knowledge of teamsters and plowmen obsolete and unneeded. Not
so with the seeds and sprays on the farm: farmers badly need to know the performance
characteristics of their seeds and effectiveness of their sprays, but they wind up continually
(and often desperately) seeking new seeds and sprays with which they have no experience
and for which there is little accumulated local wisdom. The trap of decreased skilling,
exacerbated insect problems, and increased spraying has been a major factor in India emer-
ging as a global outlier in insecticide use (Figure 2).
Traveling the insecticide-treadmill path has been especially hazardous for farmers in
India, where education levels are low, farms and bank accounts are small, and government
safety nets are largely non-existent. In Andhra Pradesh, matters came to a head in the 1997–
1998 season when the intersection of several factors led to tragedy. There had been con-
siderable recent expansion of area planted to cotton and also a steep decline in cotton
prices, leaving many poor farmers in an unusually precarious position. That year a pertur-
bation in rainfall and farmers’unfortunate response to it led to a severe outbreak of an
erstwhile minor (but increasingly resistant) pest, the cotton leafworm (Spodoptera litura)
(Reddy and Rao 1998; Armes et al. 1997). Cotton yields in Andhra Pradesh plummeted
and several hundred farmers in Warangal District alone killed themselves, commonly by
Figure 2. Comparison of pesticide use in India and the world.
Source: Aktar, Sengupta, and Chowdhury (2009).
6Glenn Davis Stone and Andrew Flachs
drinking insecticide (Reddy and Rao 1998; Revathi 1998). Although the 1998 calamity was
well covered in major US newspapers (Karp 1998; Lambrecht 1998), it was seemingly for-
gotten later when genetically modified organism opponents began to blame the continuing
suicides on the GM seeds that were not approved until 2002.
We have observed many organized attempts to break the path of insecticide-intensive
cotton production in India, including Food and Agriculture Organization field schools pro-
moting IPM; the National Agricultural Technology Project, a grant-sponsored program for
individual farmer advising (Stone 2011b); smaller schemes by numerous rural non-govern-
mental organizations; and even commercial schemes offering a premium on organically
grown cotton (Peshin et al. 2008; Flachs 2016). But as Cowan and Gunby’s(1996) com-
parison shows, didactic interventions to wean farmers off insecticides can fail spectacularly
unless key pieces are in place, such as a successful history of non-chemical management to
fall back on. As our brief history showed, the Indian cotton farmers had nothing of the kind,
and it is not surprising that none of these interventions has had any sustained success in low-
ering insecticide use.
5
GM seeds, treadmills and aspirations
The wreckage of the insecticide treadmill provided the niche for the technology Monsanto
(and their Indian partner Mahyco) were trying to bring to market: seeds genetically modi-
fied for caterpillar resistance, conferred by a Cry gene from the soil bacterium Bacillus thur-
ingiensis (hence ‘Bt cotton’). Bt cotton began field tests around the same time that farmer
suicides peaked in 1998, and was approved and released for the 2002 season. At first the
technology was available in only a few unpopular seed brands and was taken up by very
few farmers, and adoption did not reach 5 percent until 2005. Meanwhile the insecticide
treadmill continued to spin. In Warangal, the most discussed technology in 2002 was not
Bt cotton (which almost no one bought), but the new insecticide Tracer® (spinosad).
Tracer® was the most expensive spray yet, but reasonably effective against bollworms,
which had not yet had a chance to develop resistance. Before Warangal farmers accepted
Bt cotton they also began to adopt the spray Avaunt® (indoxacarb) as well as new insecti-
cidal seed treatments.
6
Bt seeds began to gain traction in 2005, after which they spread quickly, claiming 73
percent of the Indian market by 2008 (Cotton Corporation of India Ltd 2009; Stone
2015, 43). Until 2005, all Bt cotton in India contained the same transformation event:
Monsanto’s Bollgard® based on the Cry1a gene. This technology was only partly effective,
and was particularly ineffective against the Spodoptera leafworms that wreaked havoc in
1997–1998 (Showalter et al. 2009). In 2009 Monsanto announced that another serious
5
A pesticide treadmill intervention in one village in the early 2000s temporarily weaned farmers away
from excessive sprays and encouraged the use of neem powders, pheromone traps and other IPM
methods. But after a few years most participants abandoned the IPM methods, as the IPM team
never checked back with the villagers and then newly introduced Bt cotton seeds controlled pests
just as effectively (at first) and with less trouble. However, during that same time period the same
farmers began working with university extension services. This time the intervention was mediated
through a cooperative agriculture supply store rather than an outside group, meaning that the
program would feature a more permanent investment in the community. With this additional social
commitment, the cooperative store has had a long-term impact on farmer management decisions
that short-term interventions fail to replicate.
6
Recent studies show cotton leafworms have developed resistance to four insecticides (Nandakumar
2016).
The Journal of Peasant Studies 7
caterpillar pest, pink bollworm, was also showing resistance to Bollgard®, in the same
breath applauding its new Bollgard II® technology, which combined two Cry genes (Mon-
santo 2009b). By 2014 four additional Bt transformation events had been approved and
there were over 1300 Bt seed brands in India. Today Bt resistance in pink bollworm is wide-
spread (Kurmanath 2015; Buradikatti 2015). Thus GM cotton has emerged as the latest
agricultural technology treadmill, made possible by the insecticide treadmill, which was
in turn made possible by the seed treadmill. However, it is increasingly looking like the
already commercialized Bt traits were uniquely low-hanging fruit. There are a few other
Cry genes under development, but much of the discussion on dealing with the spread of
resistant pests now revolves around pyramided Bt genes and tinkering with refuges
(Carrière, Fabrick, and Tabashnik 2016).
7
Monsanto faces other headwinds in generating revenue with Bt crops, including con-
stant struggles with state authorities seeking to limit its licensing fees. In early May
2015, the Hyderabad High Court, acting under pressure from the Telangana state govern-
ment, issued an order capping Monsanto’s royalties. This move initiated a year of argu-
ments and threats as Telangana, India and Monsanto fought legal battles over the right
of governments to limit licensing fees on GM technology. When the Indian central govern-
ment slashed Monsanto’s royalty fee by 70 percent in March 2016 (Mulvany 2016), Mon-
santo threatened to leave India and end new research and development programs.
Researchers affiliated with the Central Institute for Cotton Research (CICR) responded
that Monsanto’s technology was no longer necessary to the Indian cotton industry (Sally
2016), and a Warangal newspaper opined that farmers would have been better off if Bt
cotton had never been approved (Sakshi 2016).
The vastly more attractive route to profits is offered by the linked revenue streams of
herbicide and HT seeds, where the potential for revenue growth is enormous. For instance,
Brazil, the world’s second biggest GM crop planter after the US (James 2015), has emerged
as the world’s second biggest consumer of pesticides (again, after the US; Bajak 2016). In
the US, HT crops likely increased herbicide usage by 527 million pounds between 1996 and
2011 (Benbrook 2012).
Having established its Bt trait as virtually universal in India’s cotton fields, Monsanto’s
stated aim is to open the market for seeds with stacked Bt and HT traits (Monsanto 2014). In
2014 Monsanto got approval to begin field testing of Roundup Ready Flex® (RRF) cotton,
which combines a Bt trait with glyphosate tolerance. This approval followed years of frus-
trating regulatory impasse after the Genetic Engineering Approval Committee was dis-
banded in 2012 during a temporary moratorium on new GM crops. But the timing of
the trials of the new seeds awkwardly coincided with a wave of farmer outrage over the
spread of insecticide-resistant bollworms, causing political backlash against the
company. On the heels of its regulatory victory, Monsanto suffered a new round of criticism
as it celebrated herbicide tolerance, a trait that would have no impact on the bollworms.
8
7
To reduce the selective evolutionary pressure for Bt resistance in target pests, farmers are required to
plant a pest refuge of approximately 20 percent of the planted area to non-Bt cotton. Some entomol-
ogists advocate mandating larger refuges (Carrière, Fabrick, and Tabashnik 2016), but such mandates
are universally ignored in India. Monsanto has proposed mixing non-Bt refuge seeds in with the Bt
seeds, which would effectively reduce the refuge area to 5 percent of the field, but has been prevented
by approval process mandated by the recently reconstituted Genetic Engineering Appraisal
Committee.
8
In August 2016, Monsanto withdrew its application for approval for RRF seed (Bhardwaj 2016),
although since India is their second largest market this was understood to be a bargaining position.
8Glenn Davis Stone and Andrew Flachs
Still, herbicide resistance remains atop Monsanto’s (and, more generally, industry’s) wish
list and figures prominently in efforts to influence farmer decisions. As with the first GM Bt
cotton, HT seeds are already being planted illegally by farmers in anticipation of their
official, legal release. The problem for industry and state agrocapitalism is how to break
the path on which most farmers in Warangal are dependent: the interconnected systems
of planting and weeding that warrant a closer look.
Path-breaking and the political ecology of crop spacing
Until recently, almost all Warangal cotton planting used a system that mixed some ancient
technology with modern adjustments and innovation: a hybrid system like so much in
Indian agriculture (Gupta 1998). First the field is tilled by an ox-drawn plow. Then the
farmer attaches a spacing rake (Telugu: achchumudu) with large tines 90–100 cm apart
(Figure 3), and drives the ox and rake across the entire field to etch a set of parallel
lines. The farmer then drives in the perpendicular direction, leaving a crosshatched or
tic-tac-toe pattern. A single cotton seed is planted at each intersection (Figure 4). Since
this involves marking lines in perpendicular directions, this practice is termed double-
lining, or in Telugu dabba pathi (literally ‘box cotton’, referring to the square pattern).
As the cotton grows, weeding quickly becomes a serious concern. Weeds compete for
both water and nutrients and Indian Bt cotton is a nutrient-intensive crop growing where
rainfall (and sometimes even irrigation) is unpredictable. Double-lined fields are weeded
by two methods. Manual weeding using sickles is done by either hired dayworkers
(Telugu: kuli) or household members; this method is relatively common early in the
season because the young plants are small and vulnerable. But later in the growing
season, the preferred method is to attach a weeding rake to the ox and drive it in both (per-
pendicular) directions, as in the field etching (Figure 5). This ox-weeding is a key feature of
double-line cultivation.
Double-line cultivation is effective and time-tested. The technologies for weeding
(oxen and implements) are dependable and supported by a body of shared local technical
knowledge. Oxen provide dependable traction (even responding to simple commands),
add fertility to the field, and make little oxen. Ox-weeding is well adapted to farmers
with low and/or unreliable cash flows, and many of the poorest farmers in our sample
villages regularly shared animals and field management labor. Ox weeding is cheap:
farmers often weed their own field, or hire a single driver and ox to weed an acre field
for around Rs 500 (∼USD 10). The cost of hired kuli labor for early weeding was also
very low until recently: only Rs 50 (∼USD 1) per person/day in 2005 in Warangal district.
Double-line cultivation is thus locked in by local knowledge, social institutions and econ-
omics. It also locks out reliance on herbicide by rendering it unnecessary and relatively
expensive. Until recently, only a handful of farmers in our Warangal study villages
used any herbicide on cotton.
The challenge for agrocapital is how to break the dependence on double-lining and
ox-weeding to open the door to herbicide-based management. Of course this is a perverse
reversal of the expected direction of path-breaking (Cowan and Gunby 1996). As noted,
didactic attempts to push farmers off an insecticide-intensive path failed; how could
farmers be pushed onto an herbicide-intensive path? Over the last five years we have
witnessed this process, and it begins with the details of seed placement.
Seed spacing may seem an obscure technical detail in farming, but it has often been a
pivotal feature in manipulations of agriculture by external parties. As these manipulations
are virtually always driven by political, economic or bureaucratic interests that diverge from
The Journal of Peasant Studies 9
those of the farmers, seed spacing is best seen through the lens of political ecology (e.g.,
Zimmerer and Bassett 2003; McCann 2011; Stone 2016). Seed spacing played a key role
in the post-World War II shift to hyper-industrialized farming in the US (Kloppenburg
2004, 118–19), in Mao’s micromanagement of agriculture in the Great Leap Forward
(Becker 1996,68–102), and in various colonial schemes for ordering indigenous agriculture
(Scott 1998). The change in Warangal involves a new method called single-lining (salla
pathi) that has been rapidly adopted in recent years. In single-lined fields the conventional
grid seeding is replaced by dense seeding in parallel lines. The achchumudu spacing rake
only etches parallel stripes in one direction, spaced between 40 and 70 cm apart; seeds are
then planted more closely along the lines, like a row crop. Whereas a one-acre double-lined
Figure 3. Two different types of achchumudu (spacing rakes). In both types, tine spacing is adjus-
table. Copyright (A. Flachs).
10 Glenn Davis Stone and Andrew Flachs
field can be planted almost perfectly with one 450-g pack of seed (approximately 5000
plants), a single-lined field can take several times that much seed; in an extreme version
called the ‘high-density planting system’(HDPS) now being considered by leading
cotton scientists, one acre may take up to 12 packets (nearly 60,000 plants) (Venugopalan
et al. 2014; Kranthi 2013).
What makes this innovation important to path-breaking is that a single-lined field
cannot be properly ox-weeded: the ox can be driven along but not across the rows
because the plants are too dense (Figure 6). This leaves too many weeds along the row.
Hiring more kuli labor late in the season is expensive enough that farmers are turning
increasingly to herbicides for salla pathi fields, carefully spraying to avoid hitting the
cotton plants. Our surveys show that in 2015, 31 percent of farmers used herbicides on
their cotton. The cost to spray an acre was as much as Rs 1000, equivalent to hiring
three weeding laborers. Spraying requires more time on the part of the farmer, incurs
risk of accidentally spraying one’s cotton plants, and delivers less immediately visible
effects, but given rising labor costs it is cheaper for farmers who hire labor external to
their households.
We first encountered single-lining in 2001, when it was an oddity reported on a single
field in our sample villages. By 2005 it was still only used on an estimated one percent of
cotton fields. We began collecting systematic data on single-lining in 2012, at which point
it still accounted for a minority of fields but was replacing double-lining very rapidly,
topping 90 percent within three years. (The adoption curve is significant, as addressed
below.)
Figure 4. ‘Dabba pathi’double-lined field ready for planting, 2004. Copyright (G. Stone).
The Journal of Peasant Studies 11
One factor in the sea change in cultivation strategy pertains to legislation affecting
agricultural labor costs. A guaranteed workfare program called MGNREGA
9
took effect
nationwide in 2008, guaranteeing at least 100 days of work on government-sponsored
village improvement projects during the summer agricultural lull. In Telangana, the law
mandated a daily wage of Rs 180 (USD 3). The wage increase does not directly affect agri-
cultural labor as the program employs laborers for public works, but the new pay baseline
has had spillover effects; by 2015 agricultural labor (including transportation) was typically
Rs 200/day in our sample villages. How long these wage effects will last is unclear; some
recent writing sees the MGNREGA as ‘in crisis’(Aggarwal 2016). But higher labor costs
have made herbicide use more attractive to some farmers (Kulkarni 2011), and have given
the herbicide industry new grounds for touting their products (Saripalli 2014; Monsanto
2014), particularly as a component of the shift to input-intensive single-lining.
But, interestingly, evidence for higher yield and/or profits is equivocal. Controlled
research plots show yields to vary significantly depending on the variety planted and its
care in the field (Venugopalan et al. 2014; Arunvenkatesh and Rajendran 2013). In any
case there is often a huge gulf between yields achieved under controlled research conditions
and actual farm use, and no data were heretofore available on farm conditions. Our own
sample of 1007 plantings over four years in five villages allows us to compare actual on-
farm yields for double-lined and single-lined fields, finding the frequency distribution of
Figure 5. Weeding a ‘dabba pathi’double-lined cotton field, 2005. The grid plant spacing allows the
farmer to later plow in a perpendicular direction. Copyright (G. Stone).
9
The Mahatma Gandhi National Rural Employment Guarantee Act.
12 Glenn Davis Stone and Andrew Flachs
yields to be nearly identical for the two systems (Figure 7). The difference in average yields
was not significant at the .05 level.
Whether single-lining can offer overall economic advantages is currently unknown. No
rigorous comparisons of partial budgets in the two systems have been published and we
have not conducted our own systematic study; our experience with small farm research
in India also leaves us skeptical of the accuracy of many partial budget analyses in
other contexts.
10
But over the course of more than 300 farmer interviews we have obtained
much anecdotal information on typical input costs which can at least provide a rough but
realistic estimate of partial budgets within our sample (Table 1).
We underscore the imprecision in these estimates, but note that they support our skepti-
cism that empirically observed returns to single-lining are so superior as to explain the
sweeping change. It is also important to know that for these farmers, both yields and
bottom lines often fluctuate erratically, and the modest but statistically meaningful differ-
ences in means touted in many analyses are drowned out by the variation in farmers’
fields and thus are invisible to farmers.
Our earlier findings on the problems of skilling in Telangana cotton production there-
fore loom large. Longitudinal research has documented the challenges in the information
environment, producing over-reliance on social learning with little grounding in
Figure 6. A ‘salla pathi’single-lined field, 2013. The farmer can only ox-weed in one direction,
leaving some weeds along the rows. Copyright (A. Flachs).
10
Many analyses of partial budgets are based on eliciting farmer recall of specific expenses that are not
recorded and often remembered neither accurately nor precisely, and farmer responses are rarely
checked by any form of triangulation. Most ethnographers know the veracity of such studies to be
questionable.
The Journal of Peasant Studies 13
environmental learning, as described above. Just as this pattern of learning has been a major
driver of seed adoptions, it appears to be a key factor in the spread of single-lining. Classic
studies of innovation adoption found that adoption often follows a distinctive S-curve
which is an ogive of cumulative frequencies of a normal distribution (Rogers 2003);
Henrich (2001) shows this to be the expected pattern resulting from social learning, distinct
from the different r-curves produced by adoption based on environmental learning.
11
As
Figure 8 shows, the adoption of single-lining is quite close to the S-curve for the years
for which we have data, and the early part of the curve could hardly deviate from the S-
curve by much. Joined with our analysis of yields and estimates of partial budget effects,
it seems a sound inference that single-lining is spreading by emulation rather than environ-
mental learning, just as seed choices have done.
However, it is much harder to understand the emulative adoption of such a major trans-
formation of cropping strategy than other agricultural decisions. With seed choices, which
have been the most striking indicator of herding, at least there is little demonstrable differ-
ence in performance and more variation in yield within a seed than between seeds (Stone
2007, 83; Flachs 2016). In contrast, double-lining is highly advantageous and ‘locked in’by
various factors including local knowledge and resource-sharing institutions. To understand
how single-lining would spread rapidly via social learning, let us consider a new synthesis
of farmer decision-making.
Herbicides and didactic learning
A recent synthesis of agricultural decision-making adds to environmental and social learn-
ing the process of didactic learning, whereby farmers take cues from off-farm parties (Stone
Figure 7. Frequency distributions for cotton yields in double-lined and single-lined fields. Cases
from 2012–2015 in five villages are lumped. Difference in means is not significant at the .05 level.
11
This is a distillation of Henrich’s argument. He is particularly interested in a form of social learning
called biased cultural transmission (BCT). He shows that S-curves of adoption are produced by BCT,
or by mixed types of learning where BCT is predominant.
14 Glenn Davis Stone and Andrew Flachs
2016). Analysts have long been well aware of outside interventions in farmer learning but
have given these influences short shrift in explanations of farmer behavior. Indeed, in the
seminal study on technology adoption in agriculture –Ryan and Gross’s(1943) classic
study of hybrid corn in Iowa –the seed salesmen and state extension agents who promoted
Table 1. Estimated partial budgets for double-lined and single-lined
cotton cultuvation in sample Warangal villages. Figures are given in
rupees per acre.
Double lining Single lining
Seed 930 1800
Fertilizer 800 1200
Sowing 300 300
Weeding (labor) 800 1000
Weeding (ox plow) 500 500
Weeding (herbicide) 0 1000
Insecticide/fungicide 1000 1000
Harvesting 1800 3000
Total costs 6130 9800
Yield (quintal/acre) 7.81 7.96
Proceeds 30,459 31,044
Net 24,329 21,244
On yield figures, see Figure 7. Input costs are estimated from anecdotal interview
information as discussed in text. Figures on proceeds use a price of Rs 3900, which
was typical in our sample villages; it is Rs 100 below the average minimum support
price because most farmers sell at the lower market price to avoid the delay in
government payment if sold at minimum support price.
Figure 8. Adoption of single-lining (and abandonment of double-lining) in sample villages. The data
point for 2005 is an estimate. The S-curve is a cumulative frequency graph of a normal distribution,
the pattern expected when adoption is based on social learning.
The Journal of Peasant Studies 15
hybrid corn were acknowledged as key sources of information but then ignored in the
theory building. Off-farm parties that intentionally endeavor to influence farmer practice
are variable, but they ‘comprise a conceptually coherent category in agricultural knowledge
production, distinct from environmental and social learning because they introduce off-
farm interests’(Stone 2016, 6).
Agricultural didacts’own interests may, but often don’t, align with farmers’interests;
but didacts dependably claim for their interests to align. Didactic learning almost always
operates where there are significant power differentials, with farmers being lobbied to
adopt particular technologies or practices by parties with greater wealth, prestige and/or
authority; therefore there is a political dimension to farmer decision-making that is
missed by conventional models of farmer learning and decision-making (e.g., Stone,
Flachs, and Diepenbrock 2014; Henrich 2001; McElreath 2004).
Didactic learning can have a surprisingly strong influence on farmers, especially where
environmental learning is difficult and where farmers are unable to foresee long-term
impacts of technological paths (Stone and Flachs 2014). A classic strategy is for didactic
interventions to target farmers likely to be emulated: in other words, didactic learning oper-
ates through social learning.
Path-dependency analysts have considered didactic interventions that were intended to
break farmers away from chemical-dependent paths, which have often met with very
limited success (Cowan and Gunby 1996). But in Telangana the challenge for agro-
capital is to break farmers’dependence on a non-chemical path, and there has been signifi-
cant headway as a result of multiple channels of activity. One major effort has aimed at
general perceptions, putting a positive face on herbicide-intensive cropping. Many in the
agricultural establishment have stressed increased herbicide use as a step toward a
modern India (Panchal and Kapoor 2013). Agrochemical industry publications hailing a
new Green Revolution argue that backward Indian agriculture needs chemical salvation;
the fact that herbicide use accounts for only 16 percent of chemical inputs is cited as a
key area for growth (Sapale and Malani 2015), and cotton industry reports point to herbi-
cide as the way forward for cotton farmers (Saripalli 2014) (see Figure 2). Writers for US-
based industry organizations add: ‘The centuries-old practice of hand weeding fields is no
longer sustainable in the modern world. The human drudgery of hand weeding is one of the
first tasks readily given up by workers as countries industrialize’(Gianessi 2013, 1103).
But no entity stands to gain more than Monsanto, with its leading position on Roundup
sales in India and its intent to bring RRF cotton to market. Monsanto explains low herbicide
use as a failure in Indian farmer knowledge, claiming that weeds
are responsible for 30–60% of the damage to our agriculture yields. But there is very little
awareness of this among Indian farmers, while those who are aware, lack knowledge of the
appropriate solutions. The penetration level of chemical herbicides is also very low, ranging
from 17% in rice to less than 1% in maize.
(Monsanto India 2000)
Thus Indian farmers, hailed as astute evaluators of technology when they adopted GM
seeds (Stone 2007), are cast as slack-jawed cretins oblivious to weeds and stumped by
weed removal. Monsanto’s endeavors to remedy this lack of knowledge have included
farmer field schools and demonstrations of herbicide use. Beginning with 500 farmer pro-
grams in 2007, Monsanto India targeted a range of farmers through an herbicide research
program (Monsanto 2009a). They also conducted more than 10,000 farm demonstrations
directed at small and large farmers in 2012 to raise awareness of Roundup® and discourage
16 Glenn Davis Stone and Andrew Flachs
knockoff products (Monsanto 2012, 15). These efforts build on Monsanto’s didactic
activities since the late 1990s. For instance, in Andhra Pradesh the Meekosam Project
placed Monsanto employees in villages to demonstrate products and promote hybrid
seeds and chemical inputs (Glover 2007). At the same time that these efforts attempt to
reach wide numbers of farmers, they also work through social learning by their explicit
policy of recruiting ‘progressive’farmers who might then be emulated by ‘slow adopters’
(Glover 2007, 68).
Various other parts of the Indian agricultural establishment have been active in striving
to break the double-lining/ox-weeding path. The government planning commission lauds
the increasing use of chemical inputs as a means to grow industrial production and moder-
nize agriculture (Government of India Planning Commission 2013). The lead cotton
researchers in the ANGRAU agricultural university
12
extension office in Warangal,
which serves as a resource to local farmers, not only recommend single-lining to
farmers, but extol the virtues of the extreme HDPS version of it. Extension services con-
nected to the major agricultural university in Hyderabad (where a small contingent promot-
ing sustainable methods is overshadowed by a large majority promoting industrial inputs)
enthusiastically promote single-lining. In Figure 9, a poster at the Warangal Agricultural
Research Station (the key ANGRAU extension office for Warangal district farmers)
entices farmers to switch to single-lining with a promise of 25–40 percent higher yield.
13
These efforts again show didactic learning initiatives designed to manipulate social learn-
ing: extension agents are much more inclined to interact with prosperous and well-educated
farmers who are admired and often emulated by less well-off farmers.
Input shop owners, who have become prominent agents of didactic learning in cotton
production, also play a role in the rise of single-lining. Many farmers readily accept
advice from vendors on cotton, although not on other crops like rice, which lacks
cotton’s treadmills. Farmers accept this advice, even though they know perfectly well
that the shops stand to profit from seed and herbicide sales. Shopkeepers have little or
no technical training, but they are accessible and they always make time for farmers.
Farmers have been taking shopkeeper recommendations on cotton seed choices for many
years (Stone 2007, 97), and farmers even bring samples of their plants to Warangal city
shopkeepers to inspect and diagnose (Figure 10). By 2013, shopkeepers were mentioning
(and farmers were asking about) seed spacing, although their advice was inconsistent.
However, by 2016, shopkeepers were consistently advising farmers to follow the dense
cropping strategy. ‘The suggestion comes directly from the scientists’, one shopkeeper
told us and his customers; double-lining was a ‘dying trend’, another told us, adding that
he instructs farmers to plant cotton at higher densities in accordance with the recommen-
dations of sales representatives from seed companies.
To observers of Indian cotton farmers and the challenging environments in which they
work, the traction these didactic projects gain on farmer decisions is not overly surprising.
Despite the claims that input-adopting farmers are ‘excellent businessmen who aren’t per-
suaded by anything or anybody’(Fumento 2003, 277), decision-making in Telangana
cotton cultivation is informed by dangerously little environmental learning. The signs
12
Acharya N.G. Ranga Agricultural University.
13
Agricultural extension reports also strike a defensive, nationalistic tone when stressing India’s need
to increase yield, pointing out relatively low yields per acre (tied in these reports to inefficient non-
chemical, low-tech agriculture), especially compared to China (Bagla and Stone 2012).
The Journal of Peasant Studies 17
indicate fairly strongly that the spread of single-lining is driven mainly by didactic learning,
operating both directly on farmers and through manipulation of other modes of learning.
The path ahead
One of the benefits of the path-dependency perspective is that it can help to explain not only
specific developments, but trajectories of change. As we observe the ongoing campaigns to
re-route the path of cotton cultivation, it is important to consider what future trajectories
may hold. Of course, this is a question of keen interest to the farmers themselves, but as
our survey of the history of treadmills indicates, Indian cotton farmers have an unfortunate
record of struggling to see around the next bend. As Krueger (1996, 198) points out, even
when it comes to industry leaders, it may be questionable ‘how well the representatives of
the various interest groups [know] their own self-interests’. Several future developments
seem likely enough to us to warrant mention.
Due to their sheer numbers, farmers are a potent political constituency in India. By cla-
moring (or being made to appear to clamor) for a technology or policy, farmers can exert
considerable pressure on regulators. Bt cotton was in regulatory limbo in 2001 when the
story broke that Guajarati farmers were planting illegal Bt seeds; this led to professed
fury by Monsanto’s Indian partner from whom the trait had been lifted (Jayaraman
2001), but Bt cotton was also approved for release soon after. As farmers continue to
shift to single-lining and so increasingly need herbicides that are awkward to use with cur-
rently available cottons, they are being transformed into advocates for the HT seeds that
Monsanto wants to bring to market. One prediction therefore is that the pendulum will
Figure 9. Telugu poster at Warangal Agricultural Research station. Point 3 translates to: ‘When
farmers use the crop spacer in Bt cotton to maintain a distance of (90 × 60 cm) they can get 25–
40% greater production per acre. By using this method many farmers are getting a good yield’.
18 Glenn Davis Stone and Andrew Flachs
swing toward approval of a technology that the farmers appear to want and need, largely as
a result of successful didactic learning initiatives.
As our discussion of the MGNREGA indicates, farmers who have already adopted
single-lining will likely find that adoption of herbicide will offer some savings in labor
costs, at least with the current configuration of the MGNREGA. This in turn is likely to
produce positive feedback to a seeding system that would otherwise not be clearly advan-
tageous, helping to institute the new high-input path and ensuring ongoing revenue flows
from herbicide sales. The benefits to state and agrocapitalism are clear: farmers producing
more overall commodity products with a higher investment in seeds and chemicals create an
industrial demand for displaced farm labor and a more industrialized economy.
The benefits to the farmers themselves are more mixed. The double-lining/ox-weeding
system is being replaced by a much less stable system that is almost certain to lead to new
technology treadmills. Mid-twentieth-century US cotton farmers switched to chemical-
intensive methods due to initial profits and government encouragement, only to encounter
disaster as sprays lost effectiveness (Cowan and Gunby 1996, 534–35). Herbicide resist-
ance has emerged wherever HT crops have been introduced (Powles 2008; Benbrook
2012), and problems have been particularly intractable with cotton. In the US, the advanced
Figure 10. Didactic learning from shopkeeper. A farmer shows samples of his crop to a Warangal
City shopkeeper and asks advice on products. Copyright (A. Flachs).
The Journal of Peasant Studies 19
industrial agricultural system often cited as a goal by Indian agrocapitalists (Sapale and
Malani 2015; Saripalli 2014) has created superweeds in its cotton fields after 20 years of
planting HT seeds. Herbicide-resistant Palmer amaranth, for instance, has been spreading
in the US cotton belt for over a decade (Culpepper et al. 2006) and is putting cotton
farmers out of business (Gallant 2013). The treadmill there has turned to multi-herbicide
resistant seeds, with Monsanto insisting that ‘these weed management solutions will
provide farmers with more consistent, flexible control of tough-to-manage broadleaf
weeds’.But‘others think the benefits will at best be short-lived. Weeds may soon become
resistant to the new herbicide mixtures, resulting in new generations of ever-more-intractable
weeds that will need to be controlled with yet more herbicides’(Keim 2015).
The herbicide treadmill will likely encourage a new transgenic treadmill (Binimelis,
Pengue, and Monterroso 2009), as farmers become obligate customers for a procession
of traits not only for insects but for herbicide tolerance. However, whereas the treadmill
for insecticidal transgenic technologies can make many revolutions by using various Cry
genes, the development of new herbicidal transgenics comes to an end much more
quickly. Only one gene is known to confer glyphosate resistance, which is why the new
herbicidal transgenics simply add resistance to different herbicides like 2,4-D and
dicamba (Benbrook 2012). What the farmer will do when the ox plow has been retired
and the weeds are resistant to herbicide is unknown.
Treadmills have strong implications for local knowledge. Condescending Monsanto lit-
erature notwithstanding, farmers for whom environmental learning is not obstructed have
considerable knowledge about weeds and their management (Tripp 2006); the various
forms of ox weeding were in use centuries before they were adapted for use in the
double-lining cotton system. But just as cotton farmers are ill-equipped to manage the
problem of pest-prone seeds, spray-resistant pests, and decimated predator populations,
they will be severely challenged by the nexus of herbicide-resistant weeds, expensive
kuli labor, and single-lined fields where ox-plowing does not work. This has to be a
serious cause of concern. Industrial cotton farming has managed to splutter along in
India for close to three decades despite the pernicious effects of its treadmills, but the
history of cotton farming is littered with catastrophic collapses (see for instance Cowan
and Gunby 1996, on Mexico; Castella et al. 1999, on Thailand), and an understanding of
the underlying dynamics is vital.
Acknowledgements
For comments and discussion we are grateful to Dominic Glover and to some particularly thoughtful
anonymous peer reviewers. For assistance during fieldwork we are grateful to the Rural Development
Foundation and to N. Ranjith Kumar. For the paper’s title we thank Robert Hunter.
Disclosure statement
No potential conflict of interest was reported by the authors.
Funding
Primary funding for this research was from the John Templeton Foundation initiative “Can GM crops
help to feed the world?”The opinions expressed in this publication are those of the authors and do not
necessarily reflect the views of the John Templeton Foundation. This paper also draws on fieldwork
supported by the Cultural Anthropology program of the National Science Foundation, the Wenner-
Gren Foundation, and the Jacob K. Javits Fellowship Program.
20 Glenn Davis Stone and Andrew Flachs
References
Aggarwal, A. 2016. The MGNREGA crisis. Economic and Political Weekly 51, no. 22: 38–43.
Aktar, M.W., D. Sengupta, and A. Chowdhury. 2009. Impact of pesticides use in agriculture: Their
benefits and hazards. Interdisciplinary Toxicology 2, no. 1: 1–12. doi:10.2478/v10102-009-0001-7.
Armes, N.J., J.A. Wightman, D.R. Jadhav, and G.V. Ranga Rao. 1997. Status of insecticide resistance
in spodoptera litura in Andhra Pradesh, India. Pesticide Science 50, no. 3: 240–8.
Arthur, W.B. 1989. Competing technologies, increasing returns, and lock-in by historical events. The
Economic Journal 99, no. 394: 116–31. doi:10.2307/2234208.
Arunvenkatesh, S., and K. Rajendran. 2013. Evaluation of plant density and cotton genotypes
(Gossypium hirsutum L.) on cotton yield and fibre quality. International Journal of Forestry
and Crop Improvement 40, no. 1: 1–5.
Bagla, P., and R. Stone. 2012. India’s scholar-prime minister aims for inclusive development. Science
335, no. 6071: 907–8. doi:10.1126/science.335.6071.907.
Bajak, A. 2016. The developing world is awash in pesticides. There may be a better way. Ensia,July3.
http://ensia.com/features/the-developing-world-is-awash-in-pesticides-does-it-have-to-be/.
Becker, J. 1996.Hungry ghosts: Mao’s secret famine. New York: The Free Press.
Benbrook, C.M. 2012. Impacts of genetically engineered crops on pesticide use in the US –The first
sixteen years. Environmental Sciences Europe 24, no. 1: 1–13. doi:10.1186/2190-4715-24-24.
Bentley, J.W. 1989. What farmers don’t know can’t help them: The strengths and weaknesses of indi-
genous technical knowledge in Honduras. Agriculture and Human Values 6: 25–31.
Bhardwaj, M. 2016.Exclusive: Monsanto pulls new GM cotton seed from India in protest. Reuters,
August 25. http://www.reuters.com/article/us-india-monsanto-idUSKCN10Z1OX.
Binimelis, R., W. Pengue, and I. Monterroso. 2009.“Transgenic treadmill”: Responses to the emer-
gence and spread of glyphosate-resistant Johnsongrass in Argentina. Geoforum 40, no. 4: 623–33.
doi:10.1016/j.geoforum.2009.03.009.
Buradikatti, K. 2015. Pink bollworm a nightmare for Bt cotton growers. The Hindu, December 5.
Online edition. http://www.thehindu.com/news/national/karnataka/pink-bollworm-a-nightmare-
for-bt-cotton-growers/article7950687.ece.
Carrière, Y., J.A. Fabrick, and B.E. Tabashnik. 2016. Can pyramids and seed mixtures delay resist-
ance to Bt crops? Trends in Biotechnology 34, no. 4: 291–302. doi:10.1016/j.tibtech.2015.12.011.
Castella, J.-C., D. Jourdain, G. Trébuil, and B. Napompeth. 1999. A systems approach to understand-
ing obstacles to effective implementation of IPM in Thailand: Key issues for the cotton industry.
Agriculture, Ecosystems & Environment 72, no. 1: 17–34. doi:10.1016/S0167-8809(98)00159-5.
Chen, R., J. Huang, and F. Qiao. 2013. Farmers’knowledge on pest management and pesticide use in
Bt cotton production in China. China Economic Review 27: 15–24.
Cotton Corporation of India Ltd. 2009.Annual report 2008–2009 (Annual Report 39). Cotton
Corporation of India Ltd., Mumbai.
Cowan, R. 1990. Nuclear power reactors: A study in technological lock-in. The Journal of Economic
History 50: 541–67.
Cowan, R., and P. Gunby. 1996. Sprayed to death: Path dependence, lock-in and pest control strat-
egies. The Economic Journal 106, no. 436: 521–42. doi:10.2307/2235561.
Cullather, N. 2010.The hungry world: America’s cold war battle against poverty in Asia. Cambridge,
MA: Harvard University Press.
Culpepper, A.S., T.L. Grey, W.K. Vencill, J.M. Kichler, T.M. Webster, S.M. Brown, A.C. York, J.W.
Davis, and W.W. Hanna. 2006. Glyphosate-resistant Palmer amaranth (Amaranthus palmeri)
confirmed in Georgia. Weed Science 54, no. 4: 620–6. doi:10.1614/WS-06-001R.1.
David, P.A. 1985. Clio and the economics of qwerty. The American Economic Review 75, no. 2:
332–7.
David, P.A. 2001. Path dependence, its critics and the quest for ‘historical economics’.InEvolution
and path dependence in economic ideas, ed. P. Garrouste and S. Ioannide, 15–40. Cheltenham:
Edward Elgar.
Flachs, A. 2016. Cultivating knowledge: The production and adaptation of knowledge on organic and
GM cotton farms in Telangana, India. PhD thesis, Anthropology, Washington University.
Fumento, M. 2003.BioEvolution: How biotechnology is changing our world. San Francisco, CA:
Encounter Books.
Gallant, A. 2013. Pigweed in the cotton: A ‘superweed’invades Georgia. Modern Farmer, July 18.
http://modernfarmer.com/2013/07/superweeds-2/.
The Journal of Peasant Studies 21
GEAC. 2012. Decision taken in the 115th meeting of the Genetic Engineering Appraisal Committee
(GEAC) held on 08.02.2012. http://dbtbiosafety.nic.in/Standing_Committee/YearWise_
List2002_May2012.pdf.
Gianessi, L.P. 2013. The increasing importance of herbicides in worldwide crop production. Pest
Management Science 69, no. 10: 1099–105.
Glover, D. 2007. Farmer participation in private sector agricultural extension. IDS Bulletin 38, no. 5:
61–73. doi:10.1111/j.1759-5436.2005.tb00409.x.
Government of India Planning Commission. 2013.Twelfth five year plan (2012–2017): Faster, more
inclusive and sustainable growth. Five year plans,12. New Delhi: Planning Commission,
Government of India.
Gupta, A. 1998.Postcolonial developments: Agriculture in the making of modern India. Durham, NC:
Duke University Press.
Henrich, J. 2001. Cultural transmission and the diffusion of innovations: Adoption dynamics indicate
that biased cultural transmission is the predominate force in behavioral change. American
Anthropologist 103, no. 4: 992–1013.
James, C. 2015.Global status of commercialized biotech/GM crops: 2015. ISAAA Brief No. 51.
Ithaca, NY: ISAAA (International Service for the Acquisition of Agri-Biotech Applications).
Jayaraman, K.S. 2001. Illicit GM cotton sparks corporate fury. Nature 413: 555.
Jin, S., B. Bluemling, and A.P.J. Mol. 2015. Information, trust and pesticide overuse: Interactions
between retailers and cotton farmers in China. NJAS –Wageningen Journal of Life Sciences
72–73: 23–32. doi:10.1016/j.njas.2014.10.003.
Karp, J. 1998. Deadly crop: Difficult times drive India’s cotton farmers to desperate actions. Wall
Street Journal, February 18.
Keim, B. 2015. Monsanto’s newest GM crops may create more problems than they solve. Wired,
February 2. https://www.wired.com/2015/02/new-gmo-crop-controversy/.
Kennedy, J., and L. King. 2014. The political economy of farmers’suicides in India: Indebted cash-
crop farmers with marginal landholdings explain state-level variation in suicide rates.
Globalization and Health 10: 16. doi:10.1186/1744-8603-10-16.
Kloppenburg, J.R., Jr. 2004.First the seed: The political economy of plant biotechnology, 1492–
2000. 2nd ed. Madison: University of Wisconsin Press.
Kranthi, K.R. 2013. How India can double the national average cotton yield. Cotton Statistics & News
2013, no. 24: 1–3.
Kranthi, K.R., D.R. Jadhav, S. Kranthi, R.R. Wanjari, S.S. Ali, and D.A. Russell. 2002. Insecticide
resistance in five major insect pests of cotton in India. Crop Protection 21, no. 6: 449–60.
doi:10.1016/S0261-2194(01)00131-4.
Kranthi, K.R., D.R. Jadhav, R.R. Wanjari, S. Shakir Ali, and D. Russell. 2001. Carbamate and orga-
nophosphate resistance in cotton pests in India, 1995 to 1999. Bulletin of Entomological Research
91: 37–46.
Kranthi, K.R., and D.A. Russell. 2009. Changing trends in cotton pest management. Integrated Pest
Management Reviews 1 (April): 499–541.
Krueger, A. 1996. The political economy of controls: American sugar. In Empirical studies in insti-
tutional change, ed. L. Alston, T. Eggertsson, and D. North, 169–218. Cambridge: Cambridge
University Press.
Kulkarni, V. 2011. Farm labour shortage triggers herbicide usage. The Hindu Business Line, December
20. http://www.thehindubusinessline.com/economy/agri-business/farm-labour-shortage-triggers-
herbicide-usage/article2732319.ece.
Kurmanath, K.V. 2015. Wily pink bollworm survives Monsanto’s Bollgard-II. The Hindu, October
28. http://www.thehindubusinessline.com/economy/agri-business/wily-pink-bollworm-survives-
monsantos-bollgardii/article7814810.ece.
Lalitha, N., B. Ramaswami, and P.K. Viswanathan. 2009. India’s experience with Bt cotton:
Case studies from Gujarat and Maharashtra. In Biotechnology and agricultural development:
Transgenic cotton, rural institutions and resource-poor farmers, ed. Robert Tripp, 135–67.
London: Routledge.
Lambrecht, B. 1998. India gives Monsanto an unstable lab for genetics in farming. St. Louis Post-
Dispatch, November 22.
Levins, R.A., and W.W. Cochrane. 1996.The treadmill revisited. Land Economics 72, no. 4: 550–3.
Martin, R. 2010. Roepke lecture in economic geography–rethinking regional path dependence:
Beyond lock-in to evolution. Economic Geography 86, no. 1: 1–27.
22 Glenn Davis Stone and Andrew Flachs
Martin, R., and P. Sunley. 2006. Path dependence and regional economic evolution. Journal of
Economic Geography 6, no. 4: 395–437. doi:10.1093/jeg/lbl012.
McCann, J.C. 2011. The political ecology of cereal seed development in Africa: A history of selec-
tion. IDS Bulletin 42, no. 4: 24–35. doi:10.1111/j.1759-5436.2011.00233.x.
McElreath, R. 2004. Social learning and the maintenance of cultural variation: An evolutionary model
and data from East Africa. American Anthropologist 106: 308–21.
Mohanty, B.B. 2005.‘We are like the living dead’: Farmer suicides in Maharashtra, Western India.
The Journal of Peasant Studies 32, no. 2: 243–76. doi:10.1080/03066150500094485.
Monsanto. 2009a.Monsanto India limited annual report. St. Louis, MO: Monsanto.
Monsanto. 2009b. Pink bollworm resistance to GM cotton in India. http://www.monsanto.com/
newsviews/pages/india-pink-bollworm.aspx.
Monsanto. 2012.Monsanto India limited annual report (Annual Report 2012). St. Louis, MO:
Monsanto.
Monsanto. 2014. Indian cotton farmers’weeding challenges and the potential role for technology infu-
sion. http://www.monsanto.com/global/in/newsroom/pages/weeding-challenges-and-the-potential-
for-technology-infusion-punjab.aspx (accessed October 14, 2014).
Monsanto India. 2000. Investors annual report. http://www.monsantoindia.com/monsanto/layout/
financials/annualreports/annual19992000/annualreport.asp.
Mulvany, L. 2016. India cuts Monsanto modified cotton-seed royalty fees by 70%. Bloomberg News
March 9. Online edition. http://www.bloomberg.com/news/articles/2016-03-09/india-cuts-
monsanto-modified-cotton-seed-royalty-fees-by-70.
Nandakumar, T. 2016. Pesticide resistance poses new threat to farmers. The Hindu, August 7. http://
www.thehindu.com/todays-paper/tp-national/tp-kerala/pesticide-resistance-poses-new-threat-to-
farmers/article8954936.ece.
Panchal, M., and C. Kapoor. 2013. Indian agrochemicals market to reach $6.8 Bn by FY17: Tata stra-
tegic management group. The Business Standard, August 19. http://www.business-standard.com/
content/b2b-chemicals/indian-agrochemicals-market-to-reach-6-8-bn-by-fy17-tata-strategic-
management-group-113081200449_1.html.
Peshin, R., A.K. Dhawan, K. Vatta, and K. Singh. 2008. Attributes and socio-economic dynamics of
adopting BT cotton. Economic and Political Weekly 42, no. 52: 73–80.
Powles, S.B. 2008. Evolved glyphosate-resistant weeds around the world: Lessons to be learnt. Pest
Management Science 64, no. 4: 360–5. doi:10.1002/ps.1525.
Prasad, C.S. 1999. Suicide deaths and quality of Indian cotton: Perspectives from history and technol-
ogy and Khadi movement. Economic and Political Weekly 34: 12–21.
Reddy, A.S., and B.V. Rao. 1998.The gathering agrarian crisis: Farmers’suicides in Warangal
District (A.P.) India. Hanamkonda: Centre for Environmental Studies.
Revathi, E. 1998. Farmers’suicide: Missing issues. Economic and Political Weekly 33: 1207.
Richards, P. 1989. Farmers also experiment: A neglected intellectual resource in African science.
Discovery and Innovation 1, no. 1: 18–25.
Rogers, E.M. 2003.Diffusion of innovations. 5th ed. New York: Free Press.
Ryan, B., and N.C. Gross. 1943. The diffusion of hybrid seed corn in two Iowa communities. Rural
Sociology 8: 15–24.
Sakshi. 2016. End of the seed monopoly. Sakshi Print Edition, March 8. http://epaper.sakshi.com/
742533/Warangal-Main/08-03-2016#page/1.
Sally, M. 2016. Monsanto faces challenge from CICR’s Bt cotton varieties. The Economic Times,
February 23. http://economictimes.indiatimes.com/news/economy/agriculture/monsanto-faces-
challenge-from-cicrs-bt-cotton-varieties/articleshow/51099198.cms.
Sapale, S., and N. Malani. 2015.Ushering in the 2nd green revolution: Role of crop protection chemi-
cals. New Delhi: Federation of Indian Chambers of Commerce & Industry and TATA Strategic
Management Group.
Saripalli, D. 2014.Spurting the growth of Indian chemical industry: Industry report. New Delhi: Tata
Strategic Management Group.
Scott, J.C. 1998.Seeing like a state: How certain schemes to improve the human condition have
failed.New Haven, CT: Yale University Press.
Showalter, A.M., S. Heuberger, B.E. Tabashnik, and Y. Carrièred. 2009. A primer for using trans-
genic insecticidal cotton in developing countries. Journal of Insect Science 9: 22.
Stone, G.D. 2007. Agricultural deskilling and the spread of genetically modified cotton in Warangal.
Current Anthropology 48: 67–103.
The Journal of Peasant Studies 23
Stone, G.D. 2011a. Field versus farm in Warangal: Bt cotton, higher yields, and larger questions.
World Development 39, no. 3: 387–98. doi:10.1016/j.worlddev.2010.09.008.
Stone, G.D. 2011b. Contradictions in the last mile: Suicide, culture, and e-agriculture in Rural India.
Science, Technology and Human Values 36: 759–90.
Stone, G.D. 2015. Biotechnology, schismogenesis, and the demise of uncertainty. Journal of Law &
Policy 47: 29–49.
Stone, G.D. 2016. Towards a general theory of agricultural knowledge production: Environmental,
social and didactic learning. Culture, Agriculture, Food and Environment 38, no. 1: 5–17.
doi:10.1111/cuag.12061.
Stone, G.D., and A. Flachs. 2014. The problem with the farmer’s voice. Agriculture and Human
Values 31, no. 4: 649–53. doi:10.1007/s10460-014-9535-1.
Stone, G.D., A. Flachs, and C. Diepenbrock. 2014. Rhythms of the herd: Long term dynamics in seed
choice by Indian farmers. Technology in Society 36: 26–38. doi:10.1016/j.techsoc.2013.10.003.
Tripp, R. 2006.Self-sufficient agriculture: Labour and knowledge in small-scale farming. London:
Earthscan.
Tripp, R., M. Wijeratne, and V.H. Piyadasa. 2005. What should we expect from farmer field schools?
A Sri Lanka case study. World Development 33, no. 10: 1705–20.
Vandeman, A.M. 1995. Management in a bottle: Pesticides and the deskilling of agriculture. Review
of Radical Political Economics 27: 49–59.
Venugopalan, M.V., K.R. Kranthi, D. Blaise, S. Lakde, and K. Sankaranarayana. 2014. High density
planting system in cotton: The Brazil experience and Indian initiatives. Cotton Research Journal
5, no. 2: 172–85.
Zimmerer, K.S., and T.J. Bassett, eds. 2003.Political ecology: An integrative approach to geography
and environment-development studies. New York: Guilford Press.
Glenn Davis Stone is anthropologist whose research focuses mainly on ecological, political and cul-
tural aspects of agriculture. In recent years he has been particularly interested in issues relating to indi-
genous knowledge, technology change and genetically modified crops in developing countries. His
recent fieldwork has been in the Philippines, India and rural Appalachia. He is currently a professor
of anthropology and environmental studies at Washington University in St. Louis, on leave as a Gug-
genheim fellow. Corresponding author: stone@wustl.edu
Andrew Flachs recently completed his PhD in anthropology at Washington University, and currently
is a Volkswagen Exchange Postdoctoral Fellow at the Karl Jaspers Centre for Advanced Transcultural
Studies at Heidelberg University. His research interests include indigenous knowledge and agricul-
tural change in both North America and South Asia.
24 Glenn Davis Stone and Andrew Flachs
