See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/303825733

`Between number and knowledge: Career of caste in colonial census’, in Ishita

Dube, ed., Caste in History, New Delhi: Oxford University Press, 2008, pp. 46-
66.

Chapter · January 2008

CITATIONS READS

0 336

1 author:

Padmanabh Samarendra
Jamia Millia Islamia
8 PUBLICATIONS 7 CITATIONS

SEE PROFILE

All content following this page was uploaded by Padmanabh Samarendra on 12 October 2016.

The user has requested enhancement of the downloaded file.

 Between Number and Knowledge 47

4
Between Number and Knowledge
Career of Caste in Colonial Census

PADMANABH SAMARENDRA

From the second half of the nineteenth century, the colonial state in India
was involved in a series of projects aimed at surveying the subject popula-
tion. The decennial census operations that started in India in 1871Ð72
arguably represented the most comprehensive attempt made by the state in
this regard. Scholars differ in their assessments of the objectives that seem
to have inspired the colonial state. They also analyse differently the knowl-
edge about the colonised society that got produced through these projects.
The extensive literature dealing with colonial attempts to map Indian society
can be divided into two broad groups. The first, ÔpostcolonialÕ writings,
argue that colonialism was not merely about economic exploitation and
political domination. It was a system sustained through a reciprocal relation-
ship between knowledge and conquest. Dirks has shown in the case of caste
that colonial knowledge, from the second half of the nineteenth century, was
premised on western academic disciplines like ethnology. Indigenous infor-
mants, it has been argued, were mere providers of raw data; their role in the
production of this knowledge was, at best, marginal. The application of
western knowledge consolidated colonial control by suppressing and dis-
placing indigenous forms of knowing.
The ÔrevisionistÕ critique of the postcolonial position, on the other hand,
contends that colonial knowledge was not necessarily produced with the
objective of obtaining control over the indigenous population. More impor-
tantly, the foundation of this knowledge was not exclusively western. In
fact, ÔnativeÕ informants played an active role by introducing indigenous
concepts in the body of this knowledge, ensuring in the process continuity
between the pre-colonial and colonial situations.1
48 Caste in History

Yet, despite the given differences, even ÔrevisionistÕ scholars seem to

concede that colonial knowledge, from the second half of the nineteenth
century, was closely codified by western disciplines.2 It is revealing that the
instances of collaboration between the colonizers and the colonized in the
production of knowledge, that have been cited by ÔrevisionistÕ scholars,
often come from the first half of the nineteenth century. For example, the
survey of Colin Mackenzie that Wagoner discusses in his article, spanned
the first two decades of the nineteenth century; the census of Bengal under
Alexander Boileau, which Peabody deals with, was conducted in 1835.3
The insistence on an immediate and unvarying connection between
knowledge and power in the colonial context, I would like to argue, has
generated several misconceptions about the nature of colonial knowledge,
and the role of the colonial state. While knowledge is presented to be
western, homogenous and unhindered in its march to define the essence of
caste, the state has been invested with an ahistorical desire to ceaselessly
produce knowledge in order to secure control over the subject population.
By examining the processes of classification of caste in the census reports of
the late nineteenth and early twentieth centuries, I have tried to trace the
objectives that guided the colonial state in the holding of census operations,
analyse the knowledge relating to caste that was produced in the processes,
and locate the identities that got constituted in the body of this knowledge.
The decennial census operation when initiated in 1871Ð72 was deemed to be
a statistical project, and whether the subject of caste should at all find a place
in a statistical exercise was intensely debated at least till the 1881 census.
The unwillingness shown by the colonial officials in charge of the census
operations to include caste in the project arose from the difficulties in
finding uniform categories for classifying castes. The lack of uniformity had
left the statistical compilation of number relating to caste incomplete and
contradictory at the completion of both the 1871-72 as well as 1881 census.
This led the colonial state, in the subsequent decades, to sponsor detailed
ethnographic surveys of caste.
Two qualifications need to be placed before we start conjuring the image
of an Ôethnographic stateÕ. First, the investment in securing ethnographic
knowledge was constantly weighed in the government quarters against the
specific requirements of the state and the extent to which this knowledge
was expected to meet these. Secondly, the knowledge that got produced in
the course of the ethnographic surveys, I argue, was not, and in fact could
not have been configured by western academic disciplines alone. Census
reports, after all, were not meant to be academic documents. Western
 Between Number and Knowledge 49

anthropological theories, for instance, had to relate toÑby way of explain-

ing, reinterpreting and suppressingÑthe ÔnativeÕ views of caste in order to
make the census reports a register of the Indian people. This did not
represent, I should add here to clarify my differences with the ÔrevisionistÕ
writings, an instance of collaboration or fusion of western and indigenous
knowledge. Indigenous concepts could not retain their integrity or identity
in a corpus of knowledge produced by the colonisers. Finally, while other
conceptions of caste existed beyond the domain of state, census operations
in the process of selecting, comparing and classifying the characteristic
features of this institution, explicated and authorized a specific mode of
identifying and defining a caste, its status, customs and practices. In this
context, I will be focusing on the identification of caste with religionÑthe
designation of caste as ÔHindu casteÕ. The interconnection between caste
and religion, outside the pages of the census reports, was neither definitive
nor easily demonstrable. Yet, the interconnection got endorsed and
institutionalised in the course of census operations.
This essay is divided into three sections. The first illustrates the predomi-
nantly statistical nature of the census operations between 1871Ð2 and 1881;
the second section discusses the ethnographic turn that starts featuring in the
census reports after 1881, and the third deals with the interconnection
between caste and religion.

CASTE AND THE CLASSIFICATORY CHAOS

The decennial census of the population had been conceived not as an
ethnographic project but as a statistical exercise.4 In the beginning, the
primary function of census operations was to ascertain, for administrative
purposes, the number of the subject population.5 Several defects marked the
first decennial survey of population: it was not conducted on the same day;
it did not even cover all the regions of British India. But the most crucial
defect that rendered a compilation of caste-data unattainable, and thus
marred the very statistical value of the project, was the lack of uniformity in
the classificatory schema of caste. This forced Henry Waterfield of the
Statistics and Commerce Department in the imperial government to admit in
a memorandum presented before the British parliament, that the outcome of
census operations in India was Ônot satisfactory, owing partly to the intrinsic
difficulties of the subject, and partly to the absence of a uniform plan of
classification, each writer adopting that which seemed to him best suited for
the purpose.Õ6
50 Caste in History

The priority therefore, as is evident from the deliberations of the Census

Committee, set up in 1877 by the Government of India in order to frame the
guidelines for the census of 1881, was to ensure uniformity in the arrange-
ment of statistical data. At this juncture however an ethnographic inquiry
into the structure of society in general and the various divisions and subdi-
visions of caste in particular was beyond the agenda of the census operations.
Questioning the wisdom of including caste as a subject in a statistical survey
of population, the Committee wrote that there was Ôno part of the work of
compilation which presents so many difficulties, involves so much labour,
and at the same time is so unsatisfactory when completed, as the working up
of the caste tables.Õ7 Hence, the Committee continued, ÔWe do not É
recommendÑindeed we are completely opposed toÑany classification of
caste of the Hindu population of British India.Õ8 The data concerning caste,
the Committee explained, Ôform the subject of a special compilation to be
undertaken by an antiquarian rather than a statistical authority.Õ9
The census tables compiled for1881 continued to enlist castes.10 Paving
the way for the constitution of a pan-Indian image of caste, the Census
Commissioner proposed that the castes in the ongoing census be grouped
under the following headings: ÔBrahmans, Rajputs, Castes of Good Social
Position, Inferior Castes and Non-Hindus or Aboriginal Castes.Õ11 The
necessity of sorting out the castes anew had been forced upon the officers
following the rejection by the Census Committee of the four-varna
modelÑthe only model that could have any claim of having a pan-Indian
presenceÑas Ôprimaeval and obsoleteÕ.12
The administrative aspiration of securing uniformity in caste classifica-
tion along the new lines in what happened to be Ôthe first synchronous
enumerationÉ attempted for all IndiaÕ, is clearly indicated from the follow-
ing statement of Bourdillon, the Census Superintendent for Bengal.
Bourdillon, in his Report on the Census of Bengal, 1881, wanted to catego-
rize castes on the lines attempted by his predecessor in 1872. But, he
informed his readers, Ôthe paramount necessity of maintaining uniformity in
the system of caste classification throughout India compelled the Census
Commissioner [for India in 1881] to negat[e]É [his] request.Õ13 Despite the
best attempts made by the census establishment, the outcomes of caste
arrangement, it was discovered at the conclusion of the operations, were as
diverse as before. In Bengal the central scheme was modified to divide all
the castes into Ôfive great classes, Brahmins, Rajputs, Other Hindu castes,
Aboriginal castes, and Hindus not recognising caste; and the sixth indefinite
category of Hindus, caste not stated.Õ14 Referring to the lack of uniformity,
 Between Number and Knowledge 51

Bourdillon wrote, Ôthe practical difficulty of carrying out on any uniform

principleÉ [the] grading of castesÕ as suggested by the Census Commis-
sioner, Ôwas found to be so greatÕ that this had to be Ôabandoned.Õ15
The commencement of the next census in 1891 was preceded by critical
changes in the approach of the colonial state relating to the survey of caste.
The census operations, at inception, were aimed at ascertaining the number
of the subject population. However, the variation in the scheme of classifi-
cation employed during the earlier surveys left the caste-data non-comparable.
While compiling the caste tables for Bengal in 1881, Bourdillon admitted
that, Ôconsiderable differences exist between the caste totals of 1872 and
1881, which it is now impossible to reconcile.Õ16 These errors could be
avoided, he suggested, Ôby the preparation of a report on the castes of
Bengal which shall show their origin and classification... and will give an
exhaustive list of the subdivisions of each caste and their numerous syn-
onymsÕ.17
The conflict in number had rendered the authenticity of data suspect and
the very purpose of the census operations unrealized. It was being recog-
nized that to count, it was first necessary to know. In 1882, Plowden wrote
in a note, forwarded by the Government of India to the provincial govern-
ments, that while efforts had been made Òto obtain, on a uniform method,
information of the occupations and castes found throughout British IndiaÓ,
the enquiries concerning these subjects needed to be Òpushed much furtherÓ.
In every province, he elaborated, Òsome officer who has a taste for, and a
knowledge of, archaeological researchÓ should be deputed to compile infor-
mation about caste. The Òadvantage of having such information at hand at
the next censusÓ, needed Òno comment.Ó18 The proposal met with a favourable
response from the Government of Bengal. The Lieutenant Governor, Rivers
Thompson, fully endorsed the relevance of such an inquiry and also its
urgency. He also believed that the proposed survey could not be satisfacto-
rily carried out along with the ordinary duties assigned to an officer. A
separate enquiry was required for the purpose to be conducted by an officer
Òof special taste and capacity of this description of research.Ó19 The Govern-
ment of Bengal appointed H. H. Risley in 1885, initially for a period of two
years, to conduct an ethnographic survey of the population of Bengal.
RisleyÕs proposed survey underscored a shift in the policy of the colonial
state not only because of the ethnographic nature of the inquiry but on
account of the stateÕs desire to gather information about the indigenous
society in the second half of the nineteenth century. Around the time of the
commencement of the decennial census, the state was also engaged in
52 Caste in History

compiling an ethnological list of races in Bengal20 and conducting a statis-

tical survey of the country.21 The inquiries of Risley however carried a
specific agenda. The census surveys had failed, in the wake of the rejection
of the varna schema, to come up with a pan-Indian classification of caste.
Risley, under the circumstances, was expected to find out the universal
principles that could help define caste, identify the sources of status and
explain the criteria of hierarchy. The Tribes and Castes of Bengal, published
in 1891 at the conclusion of his ethnographic survey, purported to present an
authentic knowledge of caste. The findings informed the explanation of
caste in the census report of 1901 when Risley became the Census Commis-
sioner, and from there these found their way into the gazetteers and other
tracts dealing with the tribes and castes of India.22

RACE AND CASTE

The census of 1891 revealed changes: it was not marked by the apprehen-
sion of the earlier censuses with regard to the classification of caste; nor did
it claim to be an exclusively statistical project. Further, an attempt was made
to bring greater uniformity in the classification of caste by restricting the
multiple criteria employed the preceding operations. J.A. Baines, the Cen-
sus Commissioner of India in 1891, wrote that the caste groups placed
within the Ôimperial tablesÕ were Ôbased mainly upon function.Õ23 However,
as the Census Superintendent of Bombay Presidency in 1881, Baines felt
that both descent and occupation decided the status of a caste. Stating the
reason for the modification in the criteria, the Census Commissioner pointed
out that what was Ôaimed at in prescribing the classification was as much
uniformity as the nature of statistics will show, so that the return of each
province might be dealt with on the same basis.Õ24
An elaborate explanation of the caste system was attempted in the census
report of 1891. After tracing the Ôdistinctly racialÕ25 origin of caste, Baines
went on to attribute the subsequent changes within the institution to the
Ôdevelopment of function.Õ26 However, notwithstanding the claims made by
the Census Commissioner, the chief defect in this explanation of caste, as
Risley was to explicitly state later, was seen in its lack of consistency: race/
descent, function, history and tradition, all of these were invoked, as and
when required, in the body of the report. The goal that Herbert Hope Risley
was to set for himself, as the Census Commissioner in 1901, was to
overcome these defects in an overtly academic and comprehensive exposi-
tion of caste.
 Between Number and Knowledge 53

Risley began his discussion of caste with a critique of the preceding

census. Rejecting the functional mode of classification that had been fol-
lowed in the census report of 1891, Risley argued that the preceding census
had provided only Ôa patchwork classification in which occupation pre-
dominates, varied here and there by considerations of caste, history, tradition,
ethnical affinity, and geographical positionÕ.27 Because of this inconsis-
tency, he concluded, the classification Ôaccords neither with native tradition
and practice, nor with any theory of caste that has ever been propounded by
students of the subjectÕ.28 The critique of the census reports of 1891 reveals
the two objectives that Risley had in mind: to propound a theory of caste,
and to show perfect correspondence between his theory and ÔnativeÕ tradi-
tion and practices.
RisleyÕs explication had all the bearings of a scientific treatise. Review
of previous researches and recommendations from western societies of
learning were incorporated, scientific tools were applied and laws were
formulated, non-scholastic interest or the agenda of the Empire were masked,
and the advancement of knowledge was proclaimed to be the primary
consideration behind the exercise. Risley began with a definition of caste. A
caste, he wrote, was known by a common name, it had a common occupa-
tion and claimed a common totemic origin. But most importantly, a caste
was Ôalmost invariably endogamousÕ.29 The Ôgrowth of caste sentimentÕ or
the formation of endogamous groups, Risley suggested, was rooted in Ôa
basis of fact and a superstructure of fictionÕ. The basis of ÔfactÕ obtained
widely if not universally; Ôthe superstructure of fictionÕ was Ôpeculiar to
IndiaÕ.30 The ÔfactÕ referred to the concern of the invading Aryans to main-
tain their racial purity, as was ÔcommonÕ for the dominant races in other
parts of the world. The Aryans, however, were forced to take women from
indigenous sections since Aryan women could not accompany their men in
the conquering expeditions. A total amalgamation with inferior races could
be avoided only because the Aryans Ôonly took women and did not give
themÕ. These less than pure Aryans became Ôthe founders of Rajput and
pseudo-Rajput houses all over IndiaÕ. Race was thus held to be Ôthe ultimate
basis of casteÕ.31
The ÔfactÕ of racial distinction however was subverted and fictionalised
by the ÔOrientalsÕ, and this resulted into the proliferation of caste. In the
Census CommissionerÕs own words:
Once started in India the principle [of distinction of race] was strengthened, perpetu-
ated, and extended to all ranks of society by the fiction that people who speak a
different language, dwell in different districts, worship different godsÉmust be so
54 Caste in History

unmistakably aliens by blood that intermarriage with them is a thing not to be

thought of.32
Further elaborating the fictional nature of caste division, Risley added
that the varna schema, though Ôaccepted as an article of faith by all orthodox
HindusÕ, was little more than a Ôgrotesque scheme of social evolutionÕ.33
Why did the people choose to confine themselves to endogamous groups
which proliferated on the pretence of fictional differences, or, if one were to
ask in the vein of the Census Commissioner, why did they fictionalise their
lives in the way that they did? What science could not explain to Risley,
ÔOrientalismÕ did. The fiction of caste or the growth of caste sentiment,
Risley confidently asserted:
must have been greatly promoted and stimulated by certain characteristic peculiari-
ties of the Indian intellectÑits lax hold of facts, its indifference to action, its
absorption in dreams, its exaggerated reverence for tradition, its passion for endless
division and sub-divisionÉ It is through this imitative faculty the myth of the four
castes, evolved in the first instance by some speculative Brahman, and reproduced
in the popular versions of the epicsÉhas attained its wide currency as the model to
which Hindu society ought to conform. That it bears no relation to the actual facts of
life is in the view of its adherents an irrelevant detail. It descends from remote
antiquity, it has the sanction of the Brahmans, it is an article of faith, and every one
seeks to bring his own caste within one or other of the traditional classes.34
RisleyÕs caste did not reiterate the image of a fixed and immutable
ÔOrientÕ. On the contrary, the author highlighted the processes of creation of
new castes through change of customs, migration, adoption of new function,
etc. It was the processes of change in caste society, and not its fixity, that
convinced the Census Commissioner of the inherent and immutable irratio-
nality of the ÔOrientÕ, of its ÔothernessÕ.
Risley was a man of science and science, called for the demonstration of
the ÔfactÕ of Ôcaste sentimentÕ. Accordingly, he proposed that the racial types
in the indigenous population should be discovered through measurements
of head, nose and bodily stature in accordance Ôwith a scheme approved by
the late Sir William Flower of the British Museum and Professor Topinard
of ParisÕ.35 Risley adopted anthropometry as the tool for uncovering the
ÔfactÕ relating to the organization of caste in India where Ôhistorical evidence
can hardly be said to existÕ.36 Further, the distinct cultural characteristics of
social groups were lost due to Ôthe wholesale borrowing of customs and
ceremonies which goes onÉin India.Õ37 When norms obscured the truth,
one could recover it only through natural or physical indices. Anthropom-
 Between Number and Knowledge 55

etry was particularly suited to the Indian situation because of the existence
of the caste system. Risley wrote:
Nowhere else in the world do we find the population of a large continent broken up
into an infinite number of mutually exclusive aggregatesÉthis absolute prohibition
of mixed marriages stands forth now as its [caste structureÕs] essential and most
prominent characteristicÉ. In a society thus organized, a society putting an extrava-
gant value on pride of blood and the idea of ceremonial purity, difference of physical
type, however produced in the first instance, may be expected to manifest a high
degree of persistence, while methods which seek to trace and express such differ-
ences find a peculiarly favourable field for this operation.38
India was thus Orientalized, disengaged from the (European) world it
was converted into a ÔfieldÕ. Anthropometry was the wedge that would
distinguish the ÔOrientÕ from the ÔOccidentÕ; the distinction formed the basis
to vindicate other colonial pronouncements about the ÔOrientÕ.
Risley identified three main races, seven in all, with the help of anthro-
pometry to be living in India: the Aryan, the Dravidian, and the Mongoloid.
Leading European scholarsÑFlower, Beddoe and Haddon in England,
Topinard in France, and Virchow, Schmidt and Lollmann in GermanyÑ
approved of RisleyÕs findings, he informed his readers.39 If RisleyÕs objec-
tive was to provide a scientific exposition of caste, his project was now
complete. He had defined the field of investigation, the Indian / caste
society; selected the tools, race theory and anthropometry; and cited the
findings, three main races forming the bedrock of the society. However, the
Census CommissionerÕs endeavours would have been of little value for the
colonial state if he only demonstrated the racial typology present in the
indigenous population. Notwithstanding his scientific engagement, he had
to classify the ÔnativeÕ population. The biological races therefore had to
explain the social profile of castes. The circumstances led Risley to tread
beyond the jurisdiction of science. I illustrate the enmeshing of scientific
knowledge and imperial concerns in the production of colonial knowledge
with reference to the question of hierarchy of castes.
Discussing the attributes of the races that he had identified, Risley had
stated that there was nothing in their profile to merit a gradation: Ôpeople
with long heads cannot be said to be cleverer or more advanced in culture
than people with short headsÕ.40 At the same time, he had also proposed to
spell out in his report the rank of the various castes vis-ˆ-vis one another.
The scientific premise that Risley had outlined and the agenda of the census
thus were in conflict. This forced the Census Commissioner to embark on a
56 Caste in History

series of calibrating manoeuvres between race and caste that ultimately saw
the races in India getting invested with hierarchical status.
Comparing race with caste, Risley expressed surprise at Ôthe curiously
close correspondence between the gradations of racial type indicated by the
nasal index and certain social data ascertained by independent inquiryÕ.
Elaborating upon the phenomenon, he wrote, if one took
a series of castes in Bengal, Bihar, the United Provinces of Agra and Oudh, or
Madras, and arrange them in order of the average nasal index, so that the caste with
the finest nose shall be at the top, and that with the contrast at the bottom of the list,
it will be found that this order substantially corresponds with the accepted order of
social precedence.41
He concluded his scientific inquiry, which now had ÔcuriouslyÕ assumed
cultural connotations, by laying down a ÔlawÕ: Ôfor those parts of India
where there is an appreciable strain of Dravidian blood it is scarcely a
paradox to lay down as a law of the caste organizations that the social status
of the members of a particular group varies in inverse ratio to the mean
relative width of their nosesÕ.42 The method of affixing status to race, as
should be clear from the passages quoted above, was ironical in its simplic-
ity: high caste was same as the Ôhigh raceÕ. Risley had so far investigated
race to explain caste; he now reversed his approach: the status of caste
determined the status of race. The following example illustrates the super-
imposition of caste status on racial division. The head measurement in
Bihar, the Census Commissioner wrote, corresponded Ôsubstantially with
the scale of social precedence independently ascertained. At the top of the
list are the Bhuinhars who rank high among the territorial aristocracy of
Hindustan and Bihar; then come the Brahmans, followed at a slight but yet
appreciable interval by the clerkly KayasthsÕ.43 Despite the given head
index, it was not the Bhuinhars or the Bhumihars who were placed at the top
of the caste hierarchy; they came after the Brahmans!
The head or nose measurements could not remove uncertainties concern-
ing the hierarchy of castes. A caste enjoyed differing social status over
different regions; different castes originating from any one of the races also
did not share a common rank. There was thus an asymmetry between the
rigidity of the nasal index/identity of a race and the diversity in the position
of a caste. The former could not decide the latter. Further, once the knowl-
edge of caste society was linked explicitly with effective governance, the
number of castes enlisted in the report had increased considerably, making
the exercise of hierarchization even more intractable. The task could not
satisfactorily be left to the discipline of science; a direct intervention by the
 Between Number and Knowledge 57

state was required to solve the problem. A new sub-section made its
appearance in the chapter on caste in the census reports: a genealogy of
stateÕs prerogative in matters of caste was traced beginning with the rights
of the king in the ÔHindu periodÕ to decide the status of a caste. Edward Gait,
the Census Superintendent of Bengal in 1901, wrote:
Under the Hindu regime the social precedence of different castes was settled by the
monarch himselfÉ. There are numerous stories regarding the interference of Ballala
Sena in caste matters, how he degraded the Subarnabaniks and jugis andÉhow he
settled the grades of several high castes including that of the Brahmans themselves.44

The implication of this allusion is obvious: if kings decided caste related

issues in the ÔHinduÕ period, why could the colonial state not do so? The
colonial state however made a very reluctant king; it had professed not to
intervene in the social and religious affairs of the people. Hence, an indirect
method to hierarchize caste was adopted by the Census Commissioner that
decisively ushered this institution in the arena of public debate.
Risley proposed to accept in the census reports the Ôclassification [of
caste] by social precedence as recognized by the native public opinion at the
present dayÕ.45 It was clear from the very beginning that the Ôpublic opinionÕ
was going to unite rarely if ever, on the question of social precedence. This
did not deter the census authorities and Gait confidently asserted that there
were Ôcertain well-recognized tests of social position, by the consideration
of which a fairly accurate scale of social precedence can be drawn upÕ.46 The
Ôfirst great test is whether good Brahmans will serve as priest, and if not,
whether the caste is served by any Brahmans at allÕ.47 The critical role of the
Brahmans in defining the status of a caste was discovered because of a
selective take on indigenous traditions. The section dealing with the ÔSocial
Precedence of CastesÕ in the Bengal provincial census report reveals how
Gait read into the past of the society:
At first each class of the community had a variety of occupations open to it, but by
degrees the process of differentiation spread further and particular occupations were
gradually restricted to particular groupsÉ. As the Brahmans and Kshatriyas
Égradually grew into different groups a long struggle for mastery arose which is
reflected in the legends that cluster round the name of Parasurama, the great
protagonist of the Brahmans. The result, as we know, was that the priest triumphed
over the warrior, and from that time to the present day the supremacy of the
Brahmans had become one of the cardinal doctrines of Hinduism, and is the main
test by which we decide whether members of the non-Aryan tribes are to be classed
as Hindus or Animist.48
58 Caste in History

The status of Brahmans and the authority of the Brahmanical traditions

were reproduced in the census reports. A comparison between different
provincial reports convinced Risley of Ôthe predominance throughout India
of the influence of the traditional system of four original castesÕ. In every
scheme, he continued, Ôthe Brahman heads the list. Then come the castes
whom popular opinion accepts as the modern representative of the Kshatriyas,
and these are followed by the mercantile groups supposed to be akin to the
VaisyasÕ.49
The irony of RisleyÕs scientific schema was now complete: the ghost of
the varna classification repeatedly banished since the time of the Census
Committee of 1877 and again by Risley in 1901, had reappeared through a
privileging of the Brahmanical traditions in the pages of the census report.
While reviewing Ôthe Indian theory of casteÕ before he proposed his own,
Risley had concluded that the principle of Manu had Ôno foundation in factÕ,
these only represented a Ôgrotesque scheme of social evolutionÕ. He, none-
theless, believed that due to the Ôpeculiarities of the Indian intellectÕ, ManuÕs
arrangement was Ôaccepted as an article of faith by all orthodox HindusÉis
indeed a fact in itself, a belief which has played and continues to play a large
part in the shaping of Indian societyÕ.50
Once anthropometry failed to deliver a design of caste hierarchy and
Risley decided to invite public opinion for guidance, the reinstatement of the
varna model was only to be expected. For Risley could hear public opinion
as only echoing what was Ôaccepted as an article of faith by all orthodox
HindusÕ.51 However, what Risley accommodated in the census reports was
only a ghost of the varna classificatory framework. The colonial state could
not have relinquished its authority to classify caste at a juncture when its
knowledge of the subject was both comprehensive and ÔscientificÕ.
Brahmanical traditions were subverted, redefined and selectively appropri-
ated; Ônative opinionÕ was prevailed over in the exercise of classification in
order to meet the requirements of the state. The category of Brahman in the
census report included such castes whose names contained the word ÔBrah-
manÕ coming as a suffix irrespective of whether their touch was polluting to
the Ôtwice-bornÕ castes. Referring to such discrepancies, Risley wrote:
As everyone knows, there are Brahmans and Brahmans [in Bengal], of status
varying from Rarhi, who claim to have been imported by Adisura from Kanauj, to
the Barna Brahmans who serve the lower castes, from whose hands pure Brahmans
will not take water. No attempt has been made to deal with these multifarious
distinctions in the Table. It would be a thankless task to attempt to determine the
precise degree [of such differences].52
 Between Number and Knowledge 59

In this essay so far I have not discussed the status of the indigenous
informants in the production of knowledge on caste. In scholarly writings
the local informants were either denied any role, with the privilege being
accorded to the western academic disciplines, or had been elevated to the
rank of collaborators. The inconsistencies in RisleyÕs report as detailed
above, I briefly submit, reveal a more complicated picture. While presenting
his theory of caste Risley firmly located himself within the western aca-
demic complex. His premises were drawn from race theory, his interactions
embraced European societies of learning, and he mentioned his intellectual
debts to only western scholars of anthropology. Yet, when he proceeded to
classify caste, a situation where his racial theory was not of much help, he
started referring to ManuÕs formulations, varna schema, local history and
public opinion. Could it then be suggested, and I choose to be tentative at
present, that colonial knowledge did not have a homogenous corpus, and
that the ÔnativeÕ informants could not have the same status at various levels
of knowledge production?

HINDU CASTE
What is not often recognized in academic researches is that census surveys
instituted a link between the Hindu and caste identities. This was most
clearly visible in the pages of the census reports of 1911. Let me clarify at
the outset that I do not suggest that caste was an exclusive entity and that it
got tied to religion only in the course of census surveys. It is probable that
people drew upon, not quite consciously, a range of symbols and practices
from the extant traditions to corroborate their caste or religious selves. In
such a context where the boundary of the Hindu was not defined, the
inclusion or exclusion of castes from its fold could not be determined either.
The census surveys altered this situation. The census officials selected
certain symbols and practices to define the Hindu, and then cited the
presence of these in the conduct of a caste to pronounce it as a ÔHindu casteÕ.
Thus a definitive, uniform and visible link was constituted to adjudicate and
allocate a caste to the Hindu fold.
The status of caste vis-ˆ-vis the Hindu, at least during the early editions
of the census surveys was far from fixed. Till the 1891 census, caste was
intermittently subsumed in the census tables within the broader brackets of
race or religion. And even when castes were addressed in the census reports
as the ÔHindu castesÕ, what constituted their Hinduness was not explained.
No comprehensive inquiry into the social organization or Ôcaste structureÕ of
60 Caste in History

ÔHinduismÕ had taken place before the census operations of 1901. A few
reasons might be suggested here for this lacuna. First, Hinduism, like other
religions, was seen to be an undifferentiated faith that could be identified
with reference to specific gods and rituals alone. Secondly, the concern of
the early enumerators was to separate Ôthe aboriginalsÕ from the Hindus.
Hence, they attempted to demarcate the Hindu community from without,
and not from within. Finally, there was no definition of Hinduism available
to the census officials that could help them investigate the social constitu-
ents of the faith.
The discussion over Ôwho are and who are not HindoosÕ53 had featured as
early as in 1872 in the Report on the Census of Bengal. Responding to the
dilemma, the early enumerators invariably slid into asking what was Hindu-
ism: ÔThe problem can only be satisfactorily solved by a clear definition of
what we mean by HinduismÕ,54 declared H. Beverley in the report of 1872.
As an ism the term Hindu was explicated with reference to gods, rituals,
beliefs, etc. In the definition given by Alfred Lyall, cited by Bourdillon in
the Report on the Census of Bengal, 1881, Hinduism stood for Ôa tangled
jungle of disorderly superstitions, ghost and demons, demi-gods and deified
saints; household gods, tribal gods, local gods, universal god.Õ55 These
imprecise religious motifs had to be supplanted with definitive social con-
tents before a link could be established between Hinduism and the caste
structure. The process of construction of the social dimension of Hinduism
in the subsequent census reports can be best illustrated in the way Lyall was
selectively quoted by the latter census officials.
Risley invoked Lyall, who for him was Ôthe first living authorityÕ on the
subject of Hinduism, not to repeat the point of Ôreligious chaosÕ that
Hinduism was thought to be. On the contrary, he quoted him as having
described Hinduism to be Ôthe religion of all the people who accepted the
Brahmanic Scriptures.Õ56 The cognition of the Brahmans as the supreme
expositors of scripture seemed to meet RisleyÕs own conclusions about the
position of this caste in the society. From this place, it was not difficult for
Risley to select and recommend the Ôpractical criterionÕ to identify a Hindu:
ÔThe most obvious characteristics of the ordinary Hindu are his acceptance
of the Brahmanical supremacy and of the caste system.Õ57 The social aspect
of ÔHinduismÕ was further elaborated in the general report on census in India
in 1911. Lyall, wrote Gait, used the term to denote Ônot exclusively a
religious denomination, but...also a country and, to a certain extent, a
race.Õ58 The Ôtangled jungleÕ had disappeared by now; Hinduism was en-
dowed with three denotationsÑreligion, race and country. To these Gait
 Between Number and Knowledge 61

added, following Risley, the criterion that rendered the linkage between the
Hindu and a caste more direct and definitive: Ôa man who does not belong
to a recognized Hindu caste, cannot be a Hindu.Õ59
Kenneth Jones, in one of the earliest writings on the subject, takes note
of the role census played in the Ônew conceptualization of religion as
communityÕ. Thus Hindu could now be conceived as Ôan aggregate of
individualsÕ, as a community that could be Ômapped, counted, and above all
compared with other religious communities.Õ60 Yet what were the changes
that had to be introduced in the conception of the Hindu to render it as a
community of individuals? JonesÕ summation that census Ôcreated a concept
of religious communityÕ that was merely Ômore detailed and more exact than
any existing prior to the creation of the censusÕ61 does not take into account
the alterations, as evident from Edward GaitÕs pronouncements, that mark
the mode of identifying a Hindu. An individual could be a Hindu only if he
belonged to a caste which, given its observed customs and practices, quali-
fied as a recognized Hindu caste. Thus caste came to substitute for those
myriad gods and godlings that only seemed to create confusion about the
identity of the Hindu, and the sacred texts were superseded, as explained
below, by the observed practices.62
The new definition proposed by Gait did not prove entirely satisfactory.
It still left unresolved the task of identifying the ÔHindu casteÕ. Further, this
definition also excluded the dimension of religious belief from the identity
of the Hindu. The exclusion appeared particularly anomalous since the
Ôgeneral tendency of the Hindu gentlemanÕ who were consulted by the
census officers Ôwas to regard Hinduism as a matter of belief rather than of
social or even religious practice.Õ63 In the first decade of the twentieth
century, any doubt about the authenticity of knowledge regarding the
religious identity of the people was hardly permissible. The provision of
Ôseparate electorateÕ and the proceedings of the Morley-Minto reform had
sensitised the Hindu and Muslim population of their religious identity and
also linked their number with a share in political power. The primary
responsibility lay with the state to define and delimit these religious commu-
nities. A circular was issued by Gait that contained a set of questions in order
to inquire what could disqualify a caste from entering the Hindu fold. The
indices of deviation from the ÔstandardÕ Hindu norm, and the ÔdisabilitiesÕ
that might be present in the social interaction of a caste were outlined in the
questionnaire:
deny the supremacy of the Brahmans; 2. do not receive the mantra [sacred formulae]
from a Brahman or other recognized Hindu guru [teacher]; 3. deny the authority of
62 Caste in History

the Vedas; 4. do not worship the great Hindu gods; 5. are not served by good
Brahmans as family priests; 6. have no Brahman priests at all; 7. are denied access
to the interiors of ordinary temples; 8. cause pollution a. by touch b. within a certain
distance; 9. bury their dead; 10. eat beef and do not reverence the cow.64
The response of the people to the questionnaire was varied. Gait noted in
his report that in the Central Provinces and Berar Ôa quarter of the persons
classed as Hindus deny the supremacy of the Brahmans and the authority of
the VedasÉa third are denied access to templesÉand two-fifths eat beef.Õ65
The custom of beef eating did not always cause degradation in status, and
even when it did the extent was not uniform. Thus Gait continued, ÔOf the
thirteen castes whose touch causes pollution, nine do not eat beef, while of
the eight who eat beef, four are not regarded as polluting, and two are
allowed access to temple.Õ66 Such ambiguities did not prevent certain sym-
bols and practices from emerging as the markers of Hinduism. While Gait in
his general report on the census of India in 1911, left it to the readers to
decide which castes could be called as the Hindu castes, given the latterÕs
conformity to or deviation from the standard norms cited in the circular,
many among the census officials did not hesitate from defining Hinduism
and the Hinduness of a caste. Thus OÕMalley, the officer in charge of census
operations in Bihar and Orissa, wrote, ÔIn spite of their divergences Hindus
have a common religion of which there are two salient features, viz. 1.
religious objection to the slaughter of cows, and 2. veneration, or at least
acknowledgement of the supremacy of Brahmans.Õ67
Thus cow was officially proclaimed as the cardinal symbol of the
Hindus. Other indices mentioned in the questionnaire also assumed signifi-
cance through the process of official deliberations. A caste claiming noble
Hindu status endeavoured to demonstrate the presence of Brahmans in the
ceremonies performed, or the observance by its members of the ÔVedicÕ
injunctions.

CONCLUSION
The decennial census operations had started in India in 1871Ð2 with the
object of ascertaining the number of the subject population. Yet, without
clearly identifying a caste it was not feasible to know its numerical strength.
Given the statistical bent of the project, the census operations till 1881
(barring a few exceptions like the census report of Panjab for 1881 by
Denzil Ibbetson) devoted little space to analyse the institution of casteÑtrace
its origin, note the principle of caste segregation and hierarchization, under-
 Between Number and Knowledge 63

line the changes that had come in the institution over time, etc. The absence
of any foundational knowledge led to disparities in the classification of
caste, within and across the census surveys. The problem was further
compounded when the four-varna model of caste classification, which
alone carried the pretence of a pan-Indian presence, was rejected during the
census of 1881. The consequent uncertainties surrounding the identity and
population of various castes put the very statistical project that census
operations were in jeopardy.
Consistency in enumeration and classification of caste depended on the
certitude of knowledge of the subject. Hence, in the decades following the
census of 1881, caste was subjected to meticulous academic analysis by the
census officials. The theories of materialist evolution and racial origin, as in
vogue in contemporary Europe, were deployed for the purpose. However,
the census operations were not meant to be an academic exercise. The
knowledge of caste produced in the course of the surveys of population had
to meet the agenda of the state; it also had to relate to the perception shared
by the people. Hence, an ethnographic explanation drawing upon the idea of
biological race laboured hard to ensure a tabular arrangement of caste; it
also accommodated Manu or the four-varna system in the flow of its
ÔscientificÕ elaboration. These constraints and contradictions did not pre-
clude identities from getting constituted and corroborated in the body of
colonial knowledge of caste. One such identity was that of the ÔHindu casteÕ.
The large number of petitions submitted to the census authorities requesting
alterations in name, status or classificatory location suggests that images
produced during the census operations did not remain confined to the pages
of the reports.

NOTES
1. For an elaboration of these views, see, Susan Bayly, ÔCaste and ÒraceÓ in the
colonial ethnography of IndiaÕ, in Peter Robb, (ed.), The Concept of Race in
South Asia, Delhi: Oxford University Press, 1995, Susan Bayly, Caste, Soci-
ety and Politics in India from the Eighteenth Century to the Modern Age,
Cambridge: Cambridge University Press, 1999, Norbert Peabody, ÔCents,
sense, census: Human inventories in late precolonial and early colonial IndiaÕ,
Comparative Studies in Society and History, Vol. 43, No. 4, October 2001.
Phillip B. Wagoner, `Precolonial intellectuals and the production of colonial
knowledgeÕ, Comparative Studies in Society and History, Vol. 45, No. 4,
October 2003.
2. Ibid., pp. 166-214.
3. Wagoner, ÔPrecolonial intellectualsÕ, and Peabody, ÔCents, senseÕ.
64 Caste in History

4. The attempts at enumeration and classification of population preceding the

decennial census operations have been detailed both by Bernard Cohn and
Sekhar Bandyopadhyay. See, Cohn, this volume; Sekhar Bandyopadhyay,
ÔCaste in the perception of the Raj: A note of the evolution of colonial
sociology of BengalÕ, Bengal Past and Present, Vol. CIV, Parts IÐII, Nos.
198-99, JanuaryÐDecember 1985, pp. 58-61.
5. Indicating the purpose of the census operations in 1872, Beverley wrote, ÔThe
want of more precise information regarding the numbers of the people has
always been felt to be a serious inconvenience in the administration of Bengal.
Without information on this head, the basis is wanting on which to found
accurate opinions on such important subjects as the growth and rate of
increase of the population, the sufficiency of food supplies, the incidence of
local and imperial taxation.Õ H. Beverley, Report on the Census of Bengal,
1872, Calcutta: Bengal Secretariat Press, 1872, p. 1.
6. H. Waterfield, Memorandum on the Census of British India of 1871-72,
presented to both Houses of Parliament by Command of Her Majesty, Lon-
don: George Edward Eyre and William Spottiswoode, 1875, p. 20. Expressing
a similar view, W.C. Plowden, the Census Commissioner for India in 1881,
wrote that the earlier population surveys in various provinces had been
undertaken Ôat different times and by independent agencies. There had been
no attempt to secure uniformity in the arrangement of the statistics then
obtained.Õ W.C. Plowden, Report on the Census of British India taken on the
17th of February 1881, Vol. I, London: Eyre and Spottiswoode, 1883, p. 1.
7. Oriental and India Office Collection, British Library, London (hereafter,
OIOC), Statistics and Commerce Department, 1877, L/E/2/84, Register No.
5393: ÔNext Census of IndiaÕ, Letter dated 29 January 1878, Place not
mentioned, Ôfrom the President and Members of the Census Committee, to the
Secretary, Department of Revenue, Agriculture, and Commerce, Government
of IndiaÕ.
8. Ibid.
9. Ibid.
10. Several reasons can be suggested for the inclusion and continuance of the
category of caste in the census reports. Caste and religion, Cohn has argued,
were identified by many British officials, from the middle of the nineteenth
century, as Ôthe sociological keys to understanding the Indian
peopleÕ. See, Cohn, Ibid., p. 242. The category of caste was also used by the
enumerators to separate the ÔauthenticÕ Hindus from those who claimed a
Hindu identity under Brahmanising influence.
11. Cited in Cohn, ÔThe censusÕ, p. 245.
12. Cited in J.A. Bourdillon, Report on the Census of Bengal, 1881, vol. I,
Calcutta: Bengal Secretariat Press, 1883, p. 139.
13. Ibid., p. 140.
14. Ibid., p. 133.
15. Bourdillon, Report on the Census of Bengal, p. 140.
16. Ibid., p. 142.
 Between Number and Knowledge 65

17. Ibid., p. 141.

18. OIOC, Revenue, Statistics and Commerce Department, 1884, L/E/7/73, Reg-
ister No. 521: ÔH H RisleyÕs employment on statistics relating to the castes
and occupations of the people of BengalÕ, Letter No. 1840, dated Simla, 17
August 1882, ÔFrom W. Chichele Plowden, Census Commissioner for India,
to the Secretary to the Government of India, Home DepartmentÕ.
19. OIOC, Revenue, Statistics and Commerce Department, 1884, L/E/7/73, Reg-
ister No. 521: ÔH H RisleyÕs employment on statistics relating to the castes
and occupations of the people of BengalÕ, Letter No. 91 ũ, dated Calcutta, 6
January 1884, Finance Department, Statistics, ÔFrom Colman Macaulay,
Secretary to the Government of Bengal, to the Secretary to the Government of
India, Home DepartmentÕ.
20. The compilation undertaken by Edward Tuite Dalton led to the publication in
1872, of the Descriptive Ethnology of Bengal.
21. In 1869, W.W. Hunter, on the direction of Lord Mayo, prepared a scheme for
organizing a statistical survey of the country. The outcome was the publica-
tion in 1881of the nine volumes of The Imperial Gazetteer of India.
22. The section on ÔEthnology and CasteÕ in the revised edition of The Imperial
Gazetteer of India, published between 1907-09, was an abridged version of
the chapter on ÔCaste, Tribe, and RaceÕ in the census report of 1901.
23. J.A. Baines, Census of India, 1891, General Report, London: Eyre and
Spottiswoode, 1893, p. 188.
The caste-population was divided into the following groups in the Ôimperial
tablesÕ: ÔA. Agriculture, B. Professional, C. Commercial, D., Artisans and
Village Menials, E. Vagrants, F. Races and Indefinite Titles.Õ Ibid.
24. Ibid., p. 189. The arrangement proposed by Baines was accepted with minor
modifications to classify caste in Madras Presidency.
25. Baines, Census of India, 1891, p. 183.
26. Ibid.
27. H.H. Risley, Census of India, 1901, Vol.I: India, part I: The Report, Calcutta:
Bengal Secretariat Press, 1903, p. 538. This report was written with the
assistance of E.A. Gait.
28. Ibid.
29. Ibid., p. 517.
30. Ibid., p. 555.
31. Risley, Census of India, 1901, p. 556. The initial version of the theory of racial
origin of caste was formulated by Risley in the course of his ÔEnquiry into
Castes and Occupations of the People of BengalÕ in 1885.
32. Ibid..
33. Ibid., p. 547.
34. Ibid., p. 556.
35. Ibid., p. 494.
36. Ibid., p. 489.
37. Ibid., p. 493.
38. Ibid., p. 496.
 66 Caste in History

39. Ibid., p. 494.

40. Risley, Census of India, 1901, p. 497.
41. Risley, Census of India, 1901, p. 498.
42. Ibid..
43. Ibid., p. 504.
44. E.A. Gait, Census of India, 1901, Vol.VI: The Lower Provinces of Bengal and
their Feudatories, part I: Report, Calcutta: Bengal Secretariat Press, 1903, p.
365.
45. Risley, Census of India, 1901, p. 538.
46. Gait, Census of India 1901, Vol. VI, p. 367.
47. Ibid., p. 368Ð9.
48. Ibid., p. 365.
49. Risley, Census of India, 1901, p. 539.
50. Ibid., p. 546.
51. Both Inden and Dirks have demonstrated the functioning of empirical ap-
proach in the study of caste undertaken by the colonial scholars from the
closing decades of the nineteenth century. Inden, Imagining India, pp.58Ð9;
Dirks, ÔCastes of mindÕ, Representations, No. 37, Winter 1992, p. 67. The
indirect admission of Manu and the varna division by Risley in his classifica-
tion of caste, I have tried to illustrate, reveals the limits of empiricism in the
formation of colonial knowledge on caste.
52. Risley, Census of India, 1901, p. 540.
53. Beverley, Report on the Census of Bengal, 1872, p. 129.
54. Ibid..
55. Quoted in Bourdillon, Report on the Census of Bengal, 1881, p. 71.
56. Risley, Census of India, 1901, p. 357.
57. Ibid., p. 360.
58. Gait, Census of India, 1911, pp. 115Ð16.
59. Ibid., p. 116.
60. Kenneth W. Jones, ÔReligious identity and the Indian censusÕ, in N. Gerald
Barrier, ed., The Census in British India: New Perspectives, New Delhi:
Manohar, 1981, p. 84.
61. Ibid..
62. Caste was not always described in conjunction with the Hindu. For a discus-
sion of the status of caste in the European accounts of the pre-colonial period,
see, Gita Dharampal-Frick, ÔShifting categories in the discourse on caste:
Some historical observationsÕ, in Vasudha Dalmia and Heinrich von
Stietencron, (eds.), Representing Hindusism: The Construction of Religious
Traditions and National Identity, New Delhi: Sage, 1995.
63. L.S.S. OÕMalley, Census of India, 1911, Vol. V, Part I, Report on Bengal,
Bihar and Orissa and Sikkim, Calcutta, 1913, p. 229.
64. Gait, Census of India, 1911, p. 117.
65. Ibid..
66. Ibid..
67. OÕMalley, Census of India, 1911, 1913, p. 234.

View publication stats