Of Cast Net and Caste Identity: Memetic Differentiation between Two Fishing Communities

of Karnataka
Author(s): Debal Deb
Source: Human Ecology, Vol. 24, No. 1 (Mar., 1996), pp. 109-123
Published by: Springer
Stable URL: http://www.jstor.org/stable/4603187 .
Accessed: 26/12/2010 10:12

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=springer. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Springer is collaborating with JSTOR to digitize, preserve and extend access to Human Ecology.

http://www.jstor.org
HumanEcology,Vol.24, No. 1, 1996

Of Cast Net and Caste Identity:Memetic

DifferentiationBetween Two Fishing
Communitiesof Karnataka
Debal Deb'

Twofishing castes,Ambiga and Harikantra,of Kamatakaemploydifferent

techniquesfor throwingcast nets in the same habitat.Ambigasthrowthe net
from above the head, as opposed to the far easier Harikantramethod of
throwingat the waist level. Thesedifferenttechniquesare describedhere as
memes, transmitted through learning. A logico-deductive hypothesis is
developedhereto explainthe originand maintenanceof thememeticdifference
between the castes, with the help of circumstantial semiotic and
anthropologicalevidence.
KEY WORDS: caste; caste marker; fishing; India; Karnataka, meme; social learning.

INTRODUCTION

Endogamous,occupationalcastes in India are unique in social history

in that they seem to be analogousto reproductivelyisolated populations
of biologicalspecies occupyingdifferentecologicalniches. Like the species
populationscoexistingin a habitatby diversifyingthe niche space, the caste
groups pertainingto any particularhereditaryoccupationdisplay certain
differencesin the mannerand extent of common resourceuse. The term
"niche space" is used here in a descriptivesense, and is restrictedto the
space of resourceuse as means of subsistence.Thus the niche space of a
caste is composed of the multipledimensionsof the resourcebase, equip-
ment for exploitingthe resource,methods of employingthe gear, and so
forth. While there exist large overlapsbetween the occupationalniches of
1WWF-IEasternRegion Office,Tata Centre,5th Floor,43 ChowringheeRoad, Calcutta700
071, India.

109

0300-7839/96/0300-0109$09.50/0 ? 1996 Plenum Publishing Corporation

110 Deb

the sympatriccaste populations,the differencesin resource use patterns

allow the caste groupsto avoid competitiveexclusion.Thus, two castes of
basket weavers of western Maharashtraexclusivelyuse different raw ma-
terialsfor their craft(Gadgiland Guha, 1992).Gadgiland Iyer (1988) have
also shown that HalakkiVakkal,Naik, and Holeya of Karnatakause three
different plant species for weavingmats, thus avoidingoverlap with each
other in raw materialrequirements.The divisionof the resourcespectrum
in such cases mighthave resultedfrom attemptsto avoidoccupationalcom-
petition between the castes (Gadgiland Malhotra,1983;Gadgil and Guha,
1992).
Castes were originallydefined by their distinct categories of occu-
pations (Datta, 1944; Klass, 1980). However, numerous castes have un-
dergone occupational shifts over the past few centuries through the
disintegration of traditional village crafts and services (Beteilles, 1965;
Owens and Nandy, 1977). Modernizationhas caused the disappearance
of numerous traditional caste occupations, such as toddy-tapping,pal-
anquin bearing, and shellac handicrafts.The growth of the techno-in-
dustrial sectors has also expanded the range of alternative occupations
for different caste guilds. This in turn has resulted in extensive overlap
of resource use patterns and occupations.However, in general the oblit-
eration of traditionalcultural barriersto caste exogamy has not yet oc-
curred. Thus, the Bestha of Andhra Pradesh, who originally were
palanquin bearers (Thurston and Rangachari,1909), have now adopted
fishing as a primaryoccupation, yet maintain their cultural distinction
from other fishing castes. Likewise,in the Sunderbansarea of West Ben-
gal, the cultivator castes Pod and Bagdi have invaded the occupational
niche of the fishing caste Kaibarta,while retaining their own traditional
caste identities. In particular,in the regions where the pace of industri-
alization is relatively slow, rules of caste endogamy are rigorously ob-
served, and the traditional patterns of niche diversification between
castes are still apparent.
The question of origin and maintenanceof the differencesin the tra-
ditional modes of resourceuse is of intrinsicinterestfor an understanding
of the differentiationof the niche space between castes. In this paper, I
present a probabilisticmodel describingthe origin and persistenceof the
difference between two traditional fishing castes of Karnatakain their
methods of throwingcast nets. I present an example of how traditional
castes sharinga particularniche space maintaina differencebetween each
other on semioticlevels. However,I do not intend here to mount any func-
tionalist argumentto explainthe originof the caste system in terms of re-
source use norms.
Cast Net and Caste Memes 111

RESEARCHSETTING

This study,conductedin six maritimevillagesof KumtaTaluk,Uttara

Kannadadistrict,Karnataka(Fig. 1) concernstwo fishingcastes-Ambiga
and Harikantra-who observe endogamy,clan exogamy,and preferential
cross-cousinmarriage.Both castes are distributedmainlyin estuarineand
coastal regions of Karnataka,and in large settlementsin Uttara Kannada

35

RIVER
20*N AGHANASHINI

UTTARA
KANNADA

ghnhini

Gudk t 80-
T4E E
30'
i
Kagat odkan

03 Hollonagadde

Manki

Kumta
14 25' -

? 1 2 3 4 5
L I I I - I I

KM

7 4?25' 30'
Fig. 1. Map showingthe fishingvillagesat the Aghanashiniestuary,KumtaTaluk,
UttaraKannadadistrict,Karnataka.
112 Deb

district.The Ambigacaste name refers to "water"(Sanskritambu and am-

bha). The Ambiga people identify themselves as the "Water Children"
(Gangemakkalu)and their traditionalmode of subsistencehas alwaysbeen
associatedwith riverineand maritimehabitats (as reflected by their por-
trayalin folkloreas ferrymenand fishermen).The wide range of traditional
fishing equipmentand techniquesemployedby the Ambiga attest to their
long specializationin fishing.
The Harikantraemploya similarrangeof fishingcraftsand gear. How-
ever, they use considerablyfewer traditionalfishing items and techniques
than the Ambiga. Their name is probablyderived from an older name,
"Hurikartaru,"implying that originally they were "coir rope-makers"
(Campbell,1883). In contrastto the Ambiga, Harikantrasare generalists
insofar as a large segment of the populationhas numeroussecondaryoc-
cupations,such as farming,small business,sailoring,school teaching, and
other services.
In recentyears,consequentupon the introductionof mechanizedcrafts
and correspondinglymodernizedtechniques,the frequencyof individualas
well as groupownershipof modernfishingequipmenthas increasedamong
the wealthier section of the fishing communities.Even non-fishingcastes
have begun investingin the fishing business. Thus, castes hitherto tradi-
tionallyseparatedby occupationalniche are apparentlyundergoinga proc-
ess of social homogenizationbased on resourceuse. Therefore,this study
focuses on the traditionalcrafts, gear, and techniquesstill in use, as well
as those obsolescentthroughmodernization.
In terms of these fishingtechniquesand equipment,the overall niche
space occupiedby Ambigasis thus inclusiveof, and largerthan that of the
Harikantra(Fig. 2). The old-fashionedsingle-mancanoe,pati, the sail boat
with rudder,pongai, the enclosurenet, chachu-bale6, the drag net, elu-bale,
and a large varietyof hooks and lines are used exclusivelyby Ambiga fish-
ermen. Except for the drag net, goru-bale,and the stake net, bubsi-bale,
which in principleis a largervariationof the Ambiga'sbag net gandi-baleh,
all kinds of fishinggear used by the Harikantraare also traditionallyused
by the Ambiga. Even the stake net, which is now a prerogativeof Hari-
kantras,seems to have been used earlier by Ambigas (Campbell, 1883).
How its use was taken over by Harikantrasis, however,unknown.

THE CAST-NET THROWING MEMES

In additionto these differences,there is also a considerabledifference

between the Ambigaand Harikantramethodsof throwingthe circularcast
net, bisu-bale,commonly used by both the castes. The Harikantra(like
Cast Net and Caste Memes 113

tlo

10 - / ,g~~~~~~~~~~~1

o-

r llr ikai b-a

7~~~~~~~~~~~~~
0 S6015 2

No. of nets

Fig. 2. A portionof the niche space composedof fishingequipmentas

sharedby Ambigaand Harikantra.

other fishingcastes in the district-Bhoi, Kharvi,and Gabit) throwthe net

from the waist level, swingingfrom the left side, whereasthe Ambiga em-
ploy an apparentlymore difficult method of shooting it from above the
head. The techniquesare invariantand caste-specific,and one can identify
a fisherman'scaste merelyby observinghis mannerof throwingthe net.
While a Harikantrafishermancan throw a cast net most efficiently
while standing in shallow waters (<0.6 m), an Ambiga can throw it effi-
ciently also in deeper waters (ca. 1.2 m). When it is used from aboard a
canoe, the efficiencyin catchingfish becomesequalfor both the techniques.
However,the bisu-baleis used more often for artisinalfishingby the fish-
ermen while standing in shallow river water than from aboard a canoe,
and the net is used more frequently by Ambiga than Harikantra.The
greater dependence of the Ambiga on the cast net is related to their in-
ferior economic status (Fig. 3). Whereasa large section of the Harikantra
populationowns modern craft (includingtrawlersand purse seine boats),
most Ambiga cannot afford craft bigger than a pati or great costlier than
hooks and bisu-ba'le(ISST, 1986,personalobservation).Almost all Ambiga
householdsown Bisu-ba'16 nets (Fig. 3).
Bisu-badlenets are used in beppu,a fishing method in which fish are
scared into bisu-ba'16by glisteningtender coconut leaves fastened to lines
leading to the gape of the net. Relatively recently, the Harikantrahave
114 Deb

75 75

Rmbiga Harikantra
0i~e j8 I

25 ~-25

R B C
Fig. 3. Comparisonbetweenpercentagesof Ambigaand Harikan-
tra householdsin termsof fishingas primaryoccupation(A), pos-
session of bisu-bale(B), and ownershipof cultivablelands (C).

introducedanotherkind of beppu,in which the cast net is replacedby the

drag net, goru-balek.The differentuses of the nets may be seen as distinct,
caste-specificmemes. I use the term "meme"in Dawkins'(1990) sense of
a "unit of culturalinheritance"or "unitof imitation."
The evidence discussedabove seems to indicatethat the Harikantra
came to the fishingprofessionlater than the Ambiga,and thereforehad to
learn many of the latter'sfishingtechniques.In what follows I attemptto
trace,with the help of Boydand Richerson(1985)models,(a) the earlyproc-
ess of the origin,and (b) subsequentestablishmentof the Harikantrameme.

ORIGIN OF THE VARIANT MEME

Since the different hereditaryoccupationsof castes served to reduce

competition for a given resource (Gadgil and Malhotra, 1983), any en-
croachmentinto the occupationalniche of a caste would be undesirable
from the point of view of the original occupant. It is therefore unlikely
that the Ambigawouldwillinglyteach their fishingtechniquesto any group
encroachinginto their niche.Thereforethe only plausibleway for the Hari-
kantrato learn the cast net throwingtechniquewould be by imitatingthe
Ambigause of the net. In other words,the meme can only be transmitted
via imitativeor observationallearning,which involvessome cognitive op-
erations and covert repetitions, regardless of reinforcements(Bandura,
1977). In the absence of any help or instructionsfrom the Ambiga, such
imitativelearning requiresthat the learningindividualsmust supplement
observationwith trial-and-errorto acquirethe skill.
Cast Net and Caste Memes 115

It is realistichere to presumethe existenceof n > 1 models for imi-

tation by each Harikantraindividual.If Aj is the performance(throwing
the net overhead)by the jth Ambiga model, and Xj is the learner'sper-
formance,then

Xi=A 1+ej

where ej are normallydistributedrandomvariableswith mean 0 and vari-

ance Ej. The learner'sestimatemay divergefrom the model'sperformance
Aj because of randomvariationsin the models' performance,and/or cog-
nitive errors,and/or errorsin his performancewhile imitating.The intro-
duction of the finite errorvalue, ej, is based on the fact that it is possible
to throw the net from a continuumof heights rangingfrom overhead to
knee level, and thereforeerrorsin imitatingthe overheadtechniquewould
range over that continuum.
Althoughthe net can be thrownfrom a continuumof throwingposi-
tions from overheadto knee level, most of the errorvariantsare likely to
be useless and subsequentlyeliminated.At this point, it is reasonableto
supposethat at least one of the errorvariantsprovedto be almost as effi-
cient in terms of coveringthe water area, capturingfish, etc., in shallow
waters as the model Ambiga meme (MA,while its efficiencydecreasedin
deeperwaters.Once this particularerrorvariant,sayMH,provesto be nearly
as efficient as MAin the shallowerwaters,it is likelyto become fixed as a
novel meme by instructionallearningin subsequentgenerationsof learners,
if and only if the variantMHrequiresless skill than MAto perform.
This importantproviso gives an opportunityto test the model: after
its invention, the error variantMH,that is, the Harikantratechnique of
throwingthe net from waist level, might have been maintainedif it proved
to be somewhateasier to perform,in additionto being efficaciousin shal-
lower waters. If MHis no less difficultthan MA,then it should not be fa-
vored as an alternativeto MA;the more so because it is inefficaciousin
deeper waters.Thus, the meme would be subjectto selection on the basis
of its relativeoperationalfacility.

DIFFERENTIAL DIFFICULTY: AN EMPIRICAL TEST

Student volunteers from non-fishingcastes were selected based on

their inexperiencein throwingnets, and dividedinto two groups, and as-
signed to either an Ambigaor to a Harikantrafisherman.Each groupwas
given ten demonstrationsof cast-netthrowing,and was then asked to imi-
116 Deb

tate their respectivemodel. The numberof trials made by each individual

was scored. In order to keep the varianceof error (Ej narrow,the indi-
vidualswithineach groupwere allowedto verballyaid each other to elimi-
nate mistakes in imitatingthe model. The procedurewas repeated over
successivedays until the membersof each groupwere certifiedby experts
to have satisfactorilylearned the technique.
The groupswere now reversedand the experimentrepeateduntil each
group had learned both techniques.The mean scores of each group for
learning the first and the second techniquewere calculated,and the sig-
nificanceof the differencewas ascertainedby a t-test. The resultsgiven in
Table I clearlyshow that while experienceor skill in employingeither tech-
nique facilitates learning the other, the Ambiga technique is about four
times more difficultto learn than the Harikantratechnique.

SELECTION AND SPREAD OF THE NEW MEME IN THE

POPULATION

Once a groupof Harikantraindividualshave learnedeither technique,

they would transmitthem throughinstructionto new learners.Instructional
learningis likely to build up a preferencefor the HarikantratechniqueMH
based on the ease of its learningand employment.Thus, a Harikantrapar-
ent who has adoptedMA is likely to teach his "culturaloffspring"the Am-
biga technique,but nonethelesswould show a bias (B) toward the easier
Harikantratechnique.If the weight of the ith parent teachingMA is wi, so
thatΣEwi = 1, his bias towardMH can be measuredas the product wij B
Following the rules of linear transmissionof culturaltraits illustratedin
Boyd and Richerson(1985, pp. 145-146),the frequencyof learningMH by
the culturaloffspringafter the biased transmissionwill be

Table I. DifferenceBetweenthe Numberof TrialsRequiredfor UntrainedIndividualsto

SuccessfullyLearnthe Ambiga(MA) and Harikantra(MH) Waysof Throwingbisu-bale
Net throwing Mean No.
technique N of trials SD t p
Learning MA 12 130.0 41.80
the 1st 6.39 <0.001
technique MH 9 34.0 18.59

Learning MA 12 57.4 18.58

the 2nd 5.28 <0.001
technique MH 9 22.0 12.23
Cast Net and Caste Memes 117

p'=p + B (1 -p)

wherep is the frequencyof the meme MHin the population,and B is the

bias towardthat meme. Therefore,the frequencyof adoptingthe preferred
meme will increase in each (cultural)generationas long as B > 0, and n
> 1. The results also shows that the frequencywill increase regardlessof
whether the respectiveweights (wi) of the parents are equal or not (see
AppendixA). With an initial frequencyp as low as 0.01 and B = 0.1, the
frequencyof learningMHwill exceed the 50% mark in the populationby
only seven generations.The entire populationis expectedto have adopted
the meme in just nine generationsif either B or p is a little higher (see
Fig. 4a). This indicatesthat even an insignificantamountof bias can lead
to rapid changes in the frequencyof differentvariantsin the population.

THE MEMEAS A MARKEROF GROUPIDENTITY

Manyobjectsof functionalutilitytend to be transformedthroughtra-

dition into objectsof socio-religiousconventionality,often to the extent of
prejudices(cf. Rappaport1971). Conventionalitymay also take the form
of a group marker,such that the item of behaviorin convention distin-
guishes the communitypracticingit, and the behaviorassumesa semantic
value for the group(s)in question.Examplesof suchmarkersof groupiden-
tity are found in all societies: the sacred thread of Brahmins,the sandal-
wood mark on the Vaishnava'sforehead, the crown of the monarch,the
white robe of clergymen,the horse and the hat of Western cowboys,the

a 1.0 b 1.0

p/ p'/
0.2 0.2-

10 20 30 40 50 10 20 30 40 50
Generation Generation
Fig. 4. (a) Biasedtransmissionof the memeMH fromthree parentswith differentamplitudes
of the preferenceB, (b) Conformisttransmissionof the meme from three and four parents
with frequencydependentbias F = 0.1.
118 Deb

judge'swig, the police and militaryuniforms,and so on all serve as markers

to identifythe specialtyof the respectivecommunities.
Differentiationby group markersis likely to occur between different
castes occupyingdifferentoccupationalniches. When a caste-specificniche
is invadedby other castes, the caste originallyoccupyingthe niche is likely
to distinguishits specialty from others by certain markersrelated to its
resourceuse behavioror norm.The more difficulta behavioris, the more
likely is it to be preferredas a markerfor the original occupant of the
niche, because that behaviorwould be the least likely to be usurped by
outsiders.Typically,the overhead net technique is extremelydifficult for
any new learner (Table I). Hence, this particulartechniqueseems to have
been chosen as a markerof specializationby the oldest fishing castes in
at least five States of India (Table II), where the traditionalfishermenin-
variablyuse the overheadtechnique,while newcomersto the occupation
(including immigrantpeople who took to fishing relativelyrecently) are
found to have adopted only the waist-leveltechnique.
The data in Table II also indicatethat the bias for the waist-levelnet
techniquehas not been sufficientfor the discontinuationof the overhead
techniqueamongspecialistfisherfolk.One reasoncouldbe thatthe advantage
of the overheadtechniquein deeper watersoverridesthe trade-offof diffi-
culty.However,since the advantagelies only in the facilityof fishingwhile
standingin deeper water, and not in terms of catchingmore fish (Bhat-
tachaiyya,1993;personalobservation),it seems more reasonableto surmise
that the drivefor groupidentityhas been strongenoughfor the "overhead"
meme to survivethe invasionby the "waist-level"meme. In this case, the

Table II. Cast Net ThrowingTechniquesUsed by FishingCommunitiesin Different States

Cast net throwingtechnique
State Overhead Waist level
Karnataka Ambiga Bhoi, Harikantra,a
Jaladheer,a
Moghaveeraa
Assam Jaloi-keot Kurmi, Oraon, SadgopP
Orissa Keuto Gokha,Kartia,Bhuiya,a
Mahishyab
AndhraPradesh Vadde,Jalari Bestha,aMala,a
Vadapalija,Pallae
West Bengal Kaibarta Kaibarta,Malo, Pod,a
Bagdia
aOriginallynon-fishingcommunitywhich has now invadedthe fishingniche.
bNon-fishingcommunitymigratedfrom neighboringstates, practicingfishingas a secondary
occupation.
Cast Net and Caste Memes 119

retention,or survivalof the Ambigatechniqueis betterexplainedby cultural

mechanismsthan by standardeconomicor fitnessmodels.
In Kamataka,wherecasteexogamyis relativelystrong,examplesof such
"castemarkers"(Bhattacharyya, 1993)are numerous.Differencesbetweenthe
Ambiga and Harikantrafisherfolkin the repertoireof caste-specificuse of
craftsand nets have been describedin Fig. 2. Anotherexampleis provided
by a communityof Muslinfishermen,Jaladheer(locallyknown as Jaldi in
Kumta)who live in severalmaritimevillagesof Kumtaand other parts of
Karnataka.These fishermendisplayyet anothervariantof cast-netthrowing
technique,whichis the mirrorimageof the memefor Harikantra technique-
throwingat the waist level from the right-handside insteadof the left-hand
side. Jaladheerswere earlierrenownedin Maharashtra for theirskillin naval
warfare,but in the eighteenthcenturythey lost theirgloryunderBritishrule
andmigratedto Karnatakato takeup fishingandtradefor livelihood.Clearly,
they adoptedthe Harikantra techniqueof throwingbisu-balM, yet distinguished
themselvesfromthe Hindufisherfolkby reversingthe directionof the throw.
The bias (B) towardthe waist-leveltechniquediscussedabove is likely
to get transformedinto a caste marker,after it has been spread over a
considerablesection of the population.A similarbias (say B') towardthe
overhead technique (against the waist-level technique) would operate
among the Ambiga for the purely culturalreason of distinguishingthem-
selves from the other groups(the Harikantraand Jaladheer).The bias may
now take many forms, and its spreadwould be determinedby the propor-
tion of the populationadoptingit. Thus, when the meme is adopted by
the majority-whetherfor its ease of use or its symbolicvalue-it is likely
to become a markerof group identity.Boyd and Richerson (1985) illus-
trated the rules for such "conformisttransmission"of a meme with a fre-
quency-dependentbias F, which will operate only when the number of
culturalparentstransmittingit i > n/2, and is governedby the conditions
F < | I| O < i < n/2
for for n >2
F> | I| n > i > n/2
and F = 0 otherwise.
Assumingthat each set of parentstransmittingthe meme is a random
sample of the population,and that all parents have equal weights (wi =
1/n), the frequencyof the culturaloffspringadoptingthe meme after trans-
mission (see AppendixB) will be:
n .
* Σ
E nC pi 1-_
t n-t X(F !i/n)
120 Deb

where nC1 is the binomialcoefficientfor the combinationof i parents, and

p is the initial frequencyof the meme within the population.(The prime
of the resultantfrequencyis differentfrom that in the previousequation,
merely to show the differencein their derivations.)Since this model pre-
supposes an equal weight of all parents,only the numberi > n/2 is impor-
tant, regardless of which parent can teach the meme. The pattern of
transmissionwith three parentsis elaboratedin AppendixB.
Obviously,once the frequencyof culturalparentstransmittingthe fa-
vored meme even slightlyexceeds the value p = 0.5 as described above,
p" will progressivelyincrease over generations.Figure 4b illustratesthat
when n = 3 and F = 0.1, the whole populationwill adopt the favored
meme in 40 generationswhich, in the context of continuous and active
social learning,especiallywith horizontalcomponentof transmission,would
span a fairlybrief period of time.
Thus, with increasingvalue of p", the rule of conformitywould ensure
that no Harikantraindividualis likely to employ MA and conversely,no
Ambiga individualwould employMH. The outgroupmeme would now be
consideredwithineither groupas an unwantedvariantof the conventional,
ingroup meme. While the Harikantrawould find no reason for adopting
the difficult meme MA, the Ambiga are likely to retain the meme as a
markerof their distinctionas the specialistcaste. In either case, the caste
markerwould reinforceconformityto the group'snorm.

CONCLUSION

The specialist/generalistdichotomybetween the Ambigaand Harikan-

tra which gives rise to differencesin economic and social achievementsis
also likely to strengthenthe caste distinctions.The economicallybetter-off
generalist caste Harikantratoday considersthe Ambiga inferior in social
status.Thus, even the moderntechnologiesof purseseine and trawlfishing,
which requiregroup enterprises,are owned by separatecaste cooperatives
in the district.The caste memes will possiblycontinue to remain distinct
as long as there is no pressureon the castes to coordinatefishingactivities.
The fact that the Bhoi fishermen,who originallywere palanquinbear-
ers (Campbell, 1883), like the Bestha of Andhra Pradesh (Thurstonand
Rangachari,1909), or other originallynon-fishingcastes who have invaded
the fishing niche (Table II), also throw the cast net from the waist level
lends supportto the hypothesisof memeticdifferentiationof groupidentity.
It is probable that the meme MH was not invented by Harikantras,but
learned instead from some other group of fisherfolk,e.g., Bhoi, or Kharvi,
who employ the same technique.However, this alternativescenario does
Cast Net and Caste Memes 121

not preclude the conjecture that this meme originated through efforts (of
whichever group) to imitate the overhead technique of a specialist fishing
caste, nor that the two memes evolved into caste-specific characteristics
reinforcing the distinction between the castes.

ACKNOWLEDGMENTS

This work was accomplished during my visit in 1990 to the Centres

for Ecological Sciences and Theoretical Studies, Indian Institute of Science,
Bangalore. I am grateful to M. Gadgil, N. V. Joshi, and Mrinal Ghosh for
their helpful suggestions, to R. B. Harikant for his assistance, and to two
unknown referees for their most perceptive criticisms.

APPENDIX A

Biased Transmission

For any number of cultural parents transmitting the favored meme,

the amount of bias toward the meme shown by any parent is wr.B, where
Σ
wi is the weight of the ith meme such that ΣEwi = 1, and B is the bias
toward the favored meme. Even when none of the parents are using the
favored meme, they would nonetheless show a bias toward it while teach-
ing the offspring (Boyd and Richerson, 1985, pp. 145-146). Thus, when
MH is favored, the transmission of the memes will follow the pattern
shown in Table Al for n = 2. Regardless of the weights of the parents,
the frequency of the favored meme in the next generation will be (in my
notation)

Table Al. The Patternof BiasedTransmissionof the Two Memes from Two Cultural
Parents,Each Showinga Bias TowardMH
No. of parents Probabilitythat a naive
transmitting individualacquires
MA MH MA MH
2 0 1-B B
1 1 w1(1-B)+w2(1-B) w1(l+B)+w2(1+B)
0 2 0 1
122 Deb

p' =p2 +p (1 -p) (1 +B) (w1+ W2)+B (1 _p) 2

i.e., p' =p +B (1 -p)

This result obtains for any number n > 1 of cultural parents.

APPENDIX B

Frequency-Dependent Transmission

The minimum number of cultural parents required for a frequency

dependent transmission of a meme is n = 3, in which case the pattern of
transmission of the memes under consideration is shown in Table BI fol-
lowing Boyd and Richerson (1985).
When all parents have equal weights, i.e., wi = 1/3, the frequency of
the meme H in the subsequent generations will be

p"=p+3Fp(1-p)(2p-1)

where p is the initial frequency of individuals adopting MH, and F is the

frequency dependent bias factor (see text) which assumes a positive value
as soon as the number of parents transmitting the biased meme exceeds
n/2, or p > 0.5.
With larger n, the frequency will increase, following the formula

Table BL.The Patternof Frequency-Dependent

Transmissionof the
Meme MH from Three CulturalParentsof EqualWeights
No. of parents Probabilitythat a naive
transmitting individualacquires
MA MH MA MH
3 0 1 0
2 1 3(2/3 + F) 3(1/3 - F)
1 2 3(1/3 -F) 3(2/3 + F)
0 3 0 1
Cast Net and Caste Memes 123

Σ
p" =EnC,pi (1 -p) ni. F+ i/n
Σ

where nCiis the binomialcoefficientfor the combinationof i parentsfrom

a set of n parents, and the sign of F depends on i, as discussed above.
Thusp" increasesfaster for n = 4 than with n = 3, for the same initial
frequency p = 0.51, and F = 0.1 (Fig. 4b).

REFERENCES
Bandura,A. (1977). SocialLeamingTheory.Prentice-Hall,EngelwoodCliffs,NJ.
Beteilles,A. (1965). Caste,Class,andPower.Universityof CaliforniaPress,Berkeleyand Los
Angeles.
Bhattacharyya,S. (1993). Ecological Organization of Indian Rural Populations. PhD
dissertation,IndianInstituteof Science,Bangalore.
Boyd, R., and Richerson,P. J. (1985). Cultureand EvolutionaryProcess.ChicagoUniversity
Press,Chicago.
Campbell,J. M. (1883). Gazettierof BombayPresidency(Vol. XV, Pt. I). KanareGovernment
CentralPress,Bombay,pp. 300-307.
Datta, B. N. (1944). Studiesin IndianSocialPolity.PurabiPublishers,Calcutta.
Dawkins,R. (1990). TheSelfishGene (2nd Ed.). OxfordUniversityPress,London.
Gadgil,M., and Guha, R. (1992). ThisFissuredLand:An EcologicalHistoryof India. Oxford
UniversityPress,Delhi.
Gadgil,M., and Iyer, P. (1988). On the diversificationof commonpropertyresourceuse by
Indian society. In Berkes, F. (ed.), Common PropertyResources: Ecology and
Community-Based SustainableDevelopment.Belhaven,London,pp. 240-255.
Gadgil,M., and Malhotra,K. C. (1983).Adaptivesignificanceof the Indiancaste system:An
ecologicalperspective.HumanEcology10:465-478.
ISST (1986).Postscriptto Fisherwomen of Tadri:A Socio-EconomicSurvey.Instituteof Social
StudiesTrust,Banagalore,pp. 6-7.
Klass, M. (1980). Caste:The Emergenceof the SouthAsian Social System.Institutefor the
Studyof HumanIssues,Philadelphia.
Owens,R. L., and Nandy,A. (1977). TheNew Vaishyas.Allied Publishers,New Delhi.
Rappaport,R. A. (1971).The sacredin humanevolution.AnnualReviewof EcologicalSystems
2:23-44.
Thurston, E., and Rangachari,K. (1909). Castes and Tribesof SouthernIndia (Vol. 1).
GovernmentPress,Madras.