Why Polyandry Fails

Why Polyandry Fails: Sources of Instability in Polyandrous Marriages
Author(s): Nancy E. Levine and Joan B. Silk

Source: Current Anthropology, Vol. 38, No. 3 (June 1997), pp. 375-398
Published by: The University of Chicago Press on behalf of Wenner-Gren Foundation for Anthropological
Research
Stable URL: http://www.jstor.org/stable/10.1086/204624 .
Accessed: 20/07/2013 01:01
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
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 [email protected].
The University of Chicago Press and Wenner-Gren Foundation for Anthropological Research are collaborating
with JSTOR to digitize, preserve and extend access to Current Anthropology.
http://www.jstor.org
This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

All use subject to JSTOR Terms and Conditions
Current Anthropology Volume 38, Number 3, June 1997
 1997 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved 0011-3204/97/3903-0003$3.00
Fraternal polyandry is the normative form of marriage

among the Nyinba, an ethnically Tibetan population in
Why Polyandry Fails northwestern Nepal. Despite this, polyandrous Nyinba
marriages sometimes fail, and one or more of the co-
husbands leave the common marriage to form a new
household. The goal of this paper is to identify factors
Sources of Instability in at the level of household and individual that influence
the stability of these marriages.1 The findings contrib-
Polyandrous Marriages ute to an ongoing debate about the causes of variability
in marital systems derived from two theoretical para-
digms: sociocultural anthropology and evolutionary bi-
by Nancy E. Levine and ology.
We begin with a brief review of the literature on poly-
Joan B. Silk andry, focusing on theories about why polyandry occurs
in human societies. Then we consider the factors that
sociocultural and evolutionary theories have identified
as likely to affect interpersonal and sexual relationships
within polyandrous marriages and the stability of poly-
Polyandry has long been viewed as an anomalous form of mar-
riage that raises fundamental questions about variability in hu- androus households. Next, we provide an ethnographic
man kinship systems. This paper integrates and evaluates a set account of polyandry among the Nyinba. Demographic
of hypotheses derived from sociocultural anthropological and evo- and economic data collected by Levine are used to eval-
lutionary biological theories of polyandry against data collected uate putative sources of stress in polyandrous house-
on the Nyinba, a well-studied ethnically Tibetan population liv-
ing in northwestern Nepal. In this population, polyandry is frater- holds. Finally, we consider factors that may contribute
nal; it is the normative form of marriage and highly valued cul- to the cross-cultural rarity of these sorts of marital ar-
turally. Nonetheless, certain polyandrous marriages fail—men rangements.
occasionally leave their natal households and abandon their joint
marriages. In exploring the reasons for these marital breakdowns
and the characteristics of men who instigate them, this paper of-
fers a new perspective on the presumed contradictions of polyan- Causes of Polyandry
dry and a more fruitful approach to understanding how polyan-
drous practice comes to be perpetuated from one generation to Until recently, sociocultural accounts of polyandry fo-
the next. It also contributes to discussions about how sociocul- cused largely on explaining why specific societies are
tural and evolutionary perspectives may provide complementary
viewpoints for ethnographic data analysis.
polyandrous. That is to say, scholars were concerned to
show why such an exotic arrangement—to which hu-
nancy e. levine is Associate Professor at the University of Cal- mans seemed poorly suited—occurred as a group-level
ifornia, Los Angeles (Los Angeles, Calif. 90095-1553, U.S.A.). phenomenon. The 19th-century evolutionists saw the
Born in 1948, she was educated at Reed College (B.A., 1970) and dispersed modern occurrences of polyandry as ‘‘surviv-
at the University of Rochester (M.A., 1972; Ph.D., 1978). Her re- als’’ from a more primitive epoch in human history (see
search is focused on the impact of economic and legal reforms McLennan 1876 [1865]:137, 139). Westermarck, who
on herd management and household systems among Tibetan no-
madic pastoralists. Her publications include The Dynamics of was the first to turn to synchronic explanations, identi-
Polyandry (Chicago: University of Chicago Press, 1988), ‘‘Differ- fied a set of ‘‘co-existing conditions,’’ or predictors for
ential Child Care in Three Tibetan Communities: Beyond Son polyandry, including high sex ratios at birth, resource
Preference’’ (Population and Development Review 13:281–304), limitations, geographical circumscription, and pro-
and ‘‘Fathers and Sons: Kinship Value and Validation in Tibetan
Polyandry’’ (Man 22:267–86). longed absences of husbands from home (1926:258–60,
264). These factors still figure in many contemporary
joan b. silk is Professor and Chair of the Department of An-
thropology, University of California, Los Angeles. She was born discussions of the origin and maintenance of poly-
in 1953 and educated at Pitzer College (B.A., 1975) and the Uni- andry.2
versity of California, Davis (M.A., 1978; Ph.D., 1981). Her re- British social anthropologists, who carried out field
search interests are in primate social behavior and reproductive research in polyandrous societies in the mid-20th cen-
strategies, behavioral ecology, and adoption and fosterage in hu- tury, focused on endogenous causes and holistic explan-
man societies. She has published ‘‘Patterns of Intervention in Ag-
onistic Contests among Male Bonnet Macaques,’’ in Coalitions atory frameworks. For Leach, Sri Lankan polyandry was
and Alliances in Humans and Other Animals, edited by S. Har- a means of resolving social structural contradictions
court and F. de Waal (Oxford: Oxford University Press, 1992), created when men and women both inherited substan-
(with D. L. Cheney and R. M. Seyfarth) ‘‘The Form and Function
of Post-Conflict Interactions among Female Baboons’’ (Animal
Behaviour 52:259–68), and (with R. Boyd) How Humans Evolved
(New York: Norton Press, in press). 1. Thanks are due to R. Boyd, A. T. Carter, W. Durham, S. B. Hrdy,
E. Smith, T. Fricke, M. Borgerhoff Mulder, and K. Haddix for their
The present paper was submitted 16 ix 96 and accepted 23 ix 96. comments on earlier drafts of the manuscript.
2. Cassidy and Lee have recently updated societal-level explana-
tions of polyandry and suggested that its causes lie in ‘‘the combi-
nation of the harsh environment with a limited productive role for
women’’ (1989:9).
375

376 c u r r e n t an t hr o p o l o g y Volume 38, Number 3, June 1997
tial property and brothers maintained joint ownership ber of surviving offspring produced. Males who share ac-
of land. Under such circumstances, brothers found their cess to a single female are likely to leave fewer descen-
interests divided by their wives’ distinct property dants than males who monopolize access to their
shares—a problem that could be avoided by polyandry mates. By the same token, a man who marries polyan-
(1966 [1955]). To this discussion Tambiah (1966:285, drously can expect to sire only a fraction of one
298, 303–5, 316–17) added a concern with individual woman’s children. Thus, the existence of polyandry in
choice and decision making. He described Sri Lankan human societies seems to contradict the general predic-
polyandry as one of a number of culturally sanctioned tion that evolution will favor the development of be-
strategies for dealing with growing land fragmentation. haviors that increase the ability of individuals to sur-
Polyandry benefited individuals and families by preserv- vive and reproduce.
ing large parcels of land, supplementing inadequate male Along with many sociocultural anthropologists, evo-
labor, and increasing the likelihood that one adult male lutionary biologists generally contend that polyandry is
would be at home while others were absent (citing Gough practiced when economic, ecosystemic, or demographic
1952, 1959, 1961 for parallels with the Nayar case). circumstances limit individual men’s ability to support
Subsequent studies have continued to stress the bene- women and their children adequately. Alexander (1974:
fits of polyandry for both individuals and their society 371) suggested that polyandry is ‘‘related to the low and
and have suggested that these advantages are particu- reliable productivity of farms, with the result that addi-
larly critical for economic success in resource-scarce tional labor without additional children (that is, more
environments. In the Indian Himalayas, where polyan- than a single male per family) has come to be the best
dry is commonly combined with polygyny, the benefi- route to long-term maximization of reproduction be-
cial consequences identified include less land fragmen- cause of the necessity of maintaining the minimal ac-
tation, diversification of domestic economic activities, ceptable plot of land.’’ Crook and Crook interpreted
and lower rates of population growth (Chandra 1987: polyandry in Zanskar, northwestern India, as a ‘‘re-
148; Majumdar 1962:75; Parmar 1975:127–50; Saksena sponse to an ecology where the carrying capacity of the
1962:24–25). Scholars have also argued that this sys- land is not only restricted but subject to severe seasonal
tem, commonly termed ‘‘polygynandry,’’ facilitates ad- constraints’’ (Crook and Crook 1988:99; see also van
justing the household workforce to the needs of depen- den Berghe and Barash 1977:811 and Durham 1991:100
dents and to the available resource base.3 Goldstein on this subject).5
(1976, 1978) similarly described polyandry in an ethni- A corollary of the view that polyandry is a response
cally Tibetan community in northwestern Nepal as an to specific environmental constraints is the idea that
adaptation to a circumscribed environment. Here, he men who marry polyandrously actually may have
asserted, individuals married polyandrously in order to greater reproductive fitness than those who marry mo-
avoid dividing family estates and to solidify wealth and nogamously or polygynously.6 However, the two quan-
class advantages when external economic opportunities titative studies of the reproductive consequences of
were limited (1978a:326, 329). These conscious goals polyandry have reached very different conclusions
were separate from the postulated latent functions of about this issue. While Beall and Goldstein (1981) found
the system: restraining population growth and homeo- that polyandrous men in northwestern Nepal repro-
static adjustment of group size to resources (Goldstein duced less successfully than monogamous men, Crook
1976:231; 1978:330, 335).4 and Crook (1988) reported that in Zanskar polyandrous
For those interested in the evolution of behavior, the men produced slightly more children than monoga-
existence of polyandry is problematic because it appears mous men. The conclusions of both studies must be
to limit male reproductive success, defined as the num- viewed with some caution because no information
about the paternity of children born in polyandrous
households was available.
3. Berreman (1978:343) stated that the range of marital choice ‘‘af-
fords economically advantageous flexibility to a household. Where 5. Crook and Crook also stress how polyandry enhances adaptation
an expensive brideprice is required, or where land is scarce, frater- to a high-altitude arid environment which is characterized by lim-
nal polyandry may be a good solution to the need for a wife. . . . ited carrying capacity, severe seasonal constraints, and exactions
Where land is ample and labor is short, monogamy may be the an- of state and landlords. As in Goldstein’s model, polyandry is seen
swer . . . or polygyny may be preferred’’ (see also Majumdar 1962: as curbing population and maximizing the labor force per house-
76). Both Berreman (1962:65; 1978:344) and Majumdar (1962:76) hold (Crook and Crook 1988:99, 102; 1994:760–61).
nevertheless caution that not all Himalayan hill communities 6. Although we do not specifically consider women’s perspective
practice polygynandry despite seemingly identical environmental on polyandry, it is worth noting that polyandry is likely to influ-
and economic circumstances. ence female as well as male reproductive success. If having several
4. In this model, polyandrous marriage systems serve as societal- husbands increases a woman’s access to resources or influences her
level mechanisms which adjust population to resources. They do children’s welfare, women who marry polyandrously may actually
so by permitting monogamy when economic opportunities expand. be better off than women who marry monogamously. In Ladakh,
Monogamy, however, produces population growth and increased monogamously married women produced on average 3.1 children
pressure on resources, conditions which prompt a return to frater- while their polyandrous counterparts produced 5.2 (Crook and
nal polyandry. This is postulated as a feedback relationship whose Crook 1988:106). However, polyandry is also expected to increase
adaptive advantages remain unrecognized by social actors the variance in female reproductive success, since in polyandrous
(Goldstein 1976:232–33; 1978:33; see also Crook and Crook 1988: societies with balanced sex ratios many women are likely to re-
102). main unmarried (Crook and Crook 1988:106–7).

l e v i ne a nd s i l k Why Polyandry Fails 377
Thus the functional explanations of human polyan- inba, Levine (1988:152) has suggested that co-husbands
dry proposed by sociocultural anthropologists are simi- who have different parents are less tolerant of inequities
lar to those proposed by evolutionary biologists, and than co-husbands who have the same parents.
both kinds of explanations are subject to the same criti- In evolutionary analyses, men’s contentment with
cism—namely, that it is difficult to establish a consis- their marriages is expected to be linked to the reproduc-
tent connection between specific economic, ecological, tive consequences of polyandry. Thus, disparity in age
or demographic variables and the presence or absence of among husbands and wives is expected to be an impor-
polyandry. Among Tibetan agricultural populations in tant source of dissatisfaction because men remain fer-
northwestern Nepal and western Tibet, for example, no tile far longer than women. A man who is much
sociocultural, political, economic, or demographic fac- younger than his wife may be concerned about her abil-
tor reliably predicts the incidence of polyandry. Berre- ity to produce children in the future. This may be a par-
man found no correlation between the presence of poly- ticular concern for men of high birth order, since mar-
andry and any identifiable factor among Indian riages are likely to be contracted by or for first-born
Himalayan groups and attributed the distribution of brothers.9 Cross-cultural data suggest that evolution has
polyandry to culture history (1978:344). Finally, there shaped human psychology and created strong prefer-
are no commonalities among polyandrous societies that ences among men for wives who are younger than
distinguish them unambiguously from nonpolyandrous themselves (Buss 1989). A man’s success in fathering
societies around the world.7 children also may influence his decision to maintain or
dissolve a polyandrous marriage. If reproductive success
varies among co-husbands, then the men who have fa-
Sources of Stress in Polyandrous Marriages thered few children may be prone to initiate partition.
Evolutionary theory also predicts that polyandry is
Sociocultural and evolutionary biological explanations more likely to occur among related than among unre-
of the societal-level sources of polyandry overlap at sev- lated males. The degree of relatedness among males
eral points. Analyses deriving from these two bodies of who share access to mates is important, because the
theory concur in the assumption that polyandry is a dif- theory of kin selection predicts that unreciprocated al-
ficult form of marriage that is perpetuated because of its truism will be restricted to kin. Altruism is defined by
adaptive consequences for the society and its economic biologists as an act that reduces the genetic fitness of
advantages for individual participants. A corollary of the donor and increases the genetic fitness of the recipi-
this assumption is that men will leave polyandrous ent. Kinship provides a mechanism for the evolution of
households whenever they can afford to do so. While so- altruistic interactions because relatives are descended
ciocultural models focus on the proximate sources of from a common ancestor and are therefore likely to
conflict and stresses that may disrupt polyandrous mar- share some fraction of their genetic material (Hamilton
riages, evolutionary analyses emphasize the ultimate 1964). Sharing access to a wife is a particularly striking
selective factors that underlie such conflict. example of altruism because men who permit other
Ethnographic studies indicate that polyandrous mar- men sexual access to their wives directly reduce their
riages generate certain types of problems and tensions own reproductive success (Beall and Goldstein 1981:6).
among co-husbands. Men may resent perceived inequi- A number of researchers have noted that the reproduc-
ties among co-husbands, particularly the younger broth- tive costs of polyandry may be offset if co-husbands are
ers, who are dominated by their elder siblings. Men may close kin (Hiatt 1980:587; van den Berghe and Barash
have unsatisfactory interpersonal and sexual relation- 1977:812).10 While virtually all co-husbands in Tibetan
ships with their wives, particularly those who have
older wives (Ekvall 1968:27; Goldstein 1971:73; 1976:
232; 1978:328), as youngest brothers often do. Among our only information on these practices; unfortunately, they ne-
the Nyinba disappointed expectations of fathering chil- glect the details of co-husband relationships, how these relation-
ships were affected by kinship proximity and status differentials,
dren, particularly in large sibling groups, reportedly how kinship and status affected sexual access to the common wife,
contribute to dissatisfaction with polyandrous mar- and how paternity was allocated to the different men (see Otterbein
riages. Tensions within polyandrous households may be 1968 [1963]).
exacerbated when the co-husbands do not have the 9. In the community studied here, men typically select wives a few
years younger than themselves. When parents arrange marriages,
same parentage. In Sri Lanka, where polyandry is not they try to select a girl intermediate in age between their older sons
always fraternal, co-husbands who are brothers have to ensure compatibility. Many also wish to select a girl mature
more stable marriages than unrelated men (Tambiah enough for the prompt production of heirs.
1966:286–87, 298; see also Hiatt 1980).8 Among the Ny- 10. The theory of kin selection does not imply that reproductive
equity among co-husbands is favored. Nor does it mean that the
reproductive interests of parents and their children are congruent
7. Two generalizations can be made. First, the majority of known (Alexander 1974:372). Since men are more closely related to their
polyandrous societies are agriculturalist or horticulturalist. Sec- own children than to their brothers’ children, co-husbands may
ond, fraternal polyandry is associated with patrilineal descent and compete with each other for reproductive opportunities, and repro-
virilocal residence. ductive success among brothers may vary (Crook and Crook 1988:
8. The Marquesan Islands offer the only known instance of nonfra- 110). For parents, it does not matter which of their sons fathers
ternal polyandry (often coupled with polygyny) in which the hus- their grandchildren—grandparents are equally closely related to all
bands coresided. Nineteenth-century travelers’ accounts provide of their grandchildren. If the fitness of grandparents is mainly in-

polyandry come from a single household, their relation- are, however, unambiguously Tibetan in language and
ships to each other vary, depending on the marital ar- culture, displaying strong similarities to agriculturalist
rangements of the previous generation. If the reproduc- groups of western Tibet. There are four Nyinba villages,
tive costs of polyandry are offset by the reproductive situated between 9,500 and 11,000 ft. on gently sloping
gains accrued by close kin, then it might be predicted hillsides above major rivers. Cultivated lands range ap-
that most stable and cohesive polyandrous households proximately 1,000 ft. below and above these villages,
will be composed of men who are closely related to one with alpine valleys above the fields devoted to summer
another. pasturage. Like other ethnic Tibetans in Nepal, Nyinba
Not all those who believe that evolution has shaped support themselves through a combination of agricul-
human behavior would endorse this last prediction. ture, herding, and long-distance trade. Recent years
Some researchers, who call themselves evolutionary have seen declining profits and growing risks in various
psychologists or Darwinian psychologists, believe that sectors of the economy. Agriculture has become less
evolution has shaped the psychological mechanisms profitable as population growth has created more pres-
that underlie human behavior (Cosmides and Tooby sure on land, pastoralism has contracted as grasslands
1992, Symons 1992). These mechanisms evolved during have been converted to agricultural use, and trade has
the tens of thousands of years in which humans were become less reliably profitable because of the destabili-
foragers and lived in small, egalitarian groups of closely zation of commerce with Tibet. Despite these adverse
related individuals. If polyandry plays no role in this conditions, Nyinba are adapting to change and continue
scenario, then human males may not be equipped with to maintain a respectable standard of living. The major-
psychological mechanisms that temper sexual jealousy ity of households produce adequate food even in years
or concerns about paternity when rivals are related. of poor harvest or trading failures. Their houses are well
A final source of insight about the factors that influ- built and comfortable, people are adequately clothed,
ence the stability of polyandrous marriages comes from and villages manage to maintain a rich ceremonial life.
polyandrous peoples themselves. Emic evaluations of Perhaps the most striking feature of the Nyinba socio-
this kind may influence individuals’ decisions about cultural system is the extremely high incidence of fra-
their marriages and thus merit special attention. When ternal polyandry, in rare circumstances accompanied by
asked what undermines polyandrous marriages and polygyny.
leads to their dissolution, the Nyinba mention the size Levine conducted 22 months of ethnographic field-
of the sibling group, the closeness of kinship among co- work in Nyinba villages in 1973–75 and 1982–83. This
husbands, the extent of landholdings or the ease of ac- research yielded various sources of data used in this pa-
cess to reclaimable land, the success of the relationship per, including a census and household survey, retro-
with the common wife, and the presence or absence of spective fertility and marital histories, interviews about
‘‘own children’’ within the marriage (see also Levine the precipitating causes of and property settlements fol-
1988:257). lowing household partitions, attributions of paternity
These theoretical and ethnographic sources provide a in polyandrous marriages, and genealogical materials.
rich set of hypotheses about the factors that influence Details on data collection strategies and limitations in
the stability of polyandrous marriages. We have tested the data sets are provided in appendix A. Different sam-
these hypotheses about the causes of marital dissolu- pling strategies were followed in different circum-
tion among the Nyinba in two ways. First, we examined stances. In particular, the census and household survey
whether each of the hypothesized factors was associ- involved a random cross-section of the population,
ated with the probability that a man would leave his while information was collected for every partition
polyandrous marriage. Second, we examined whether event that had occurred over the past 25 years. Conse-
the separation from the old and entry into a new mar- quently, partitions are overrepresented in the data set.
riage produced improved marital circumstances in ways Oversampling partitioners does offer one benefit: a
that the hypotheses would lead us to expect. Did men, broader vantage point for assessing factors implicated in
for example, leave older women for younger ones? Did such events.
men selectively align themselves with their closest kin
after partition?
Nyinba Polyandry
Setting and Sources of Data
In 1983 the Nyinba numbered 1,332—716 men and 616
women. This population was subdivided into two
The Nyinba are situated at the interface between
largely endogamous social strata: the numerically dom-
Nepali-speaking caste Hindu and Tibetan worlds. They
inant landholders and the descendants of slaves who
were emancipated in 1926. This paper will focus on the
fluenced by the amount of care, nourishment, protection, and so landholding stratum, which comprises the vast major-
on, that their grandchildren receive and a single man is unable to
provide for all the needs of one woman’s children, then parents
ity of the population (1,152), primarily because there are
should encourage their sons to marry polyandrously and contribute different traditions in slave kinship and marriage that
to the joint household economy (Alexander 1974:372). still affect the practices of their descendants today and

special cultural and socioeconomic constraints on those portunities for having their ‘‘own children.’’ Second, it
marriages.11 creates an ever-present risk of partition, although a
In Nyinba landholding families, brothers jointly wed number of such complex unions have persisted
a single woman when one or more of the brothers has throughout the partners’ lifetimes.13 Third, it produces
reached adolescence. The same rule obtains regardless children with different fathers and mothers. Most of
of the number of brothers, and, in consequence, approx- these mothers are unrelated women, although one-third
imately 70% of 150 surveyed Nyinba marriages had be- are sisters or half-sisters, themselves born of a polyan-
gun with two or more brothers. Over time, however, the drous marriage.
proportion of polyandrous marriages and the number of In contrast with Tibetan polyandry as reported else-
brothers within existing polyandrous marriages inevita- where, Nyinba place great emphasis on the paternity of
bly declines because of the death of co-husbands, out- children; one brother is identified as the ‘‘real’’ (ngo-
marriage, and, occasionally, partition. Thus, a cross- thog) father, that is, the man believed to be responsible
sectional analysis carried out in 1983 showed that only for the child’s conception.14 The mother generally takes
74 (49.3%) of the sampled marriages were still polyan- the initiative in deciding who the father might be in ac-
drous. One-third of the newly monogamous marriages cordance with local theories of conception, which hold
can be traced to household partitions. Once they are es- that women are likeliest to become pregnant in the sec-
tablished, these monogamous marriages, like all others, ond week of their menstrual cycles. A woman’s cer-
will persist throughout the partners’ lifetimes. tainty about the paternity of her children is likely to be
Bifraternal polyandry is the commonest situation. enhanced by the fact that husbands often are away from
Among marriages with one wife or one fertile wife, 55% home for lengthy periods of time.
involved two brothers, 28% involved three brothers, The consequences of paternity designation are sev-
and only 17% involved four or more brothers. Trifrater- eral. First, it influences interpersonal relationships
nal polyandry, nonetheless, has a high value culturally within the household. Children are said to develop es-
and is considered desirable economically, since each pecially close ties with their ‘‘real’’ fathers, and siblings
brother can specialize in one sphere of the tripartite Ny- with the same real father and the same mother are ex-
inba economy: farming, herding, or trading. Marriages pected to have more amicable relationships. This is par-
with more men are said to be prone to discord and diffi- ticularly important for brothers who are expected to
cult to sustain (Carrasco 1959:36; Goldstein 1971:68). live their lives together. Second, it determines inheri-
For this reason, young girls express reluctance to enter tance, which is calculated on a per stirpes basis. We can
into such marriages, and their concerns are echoed by illustrate this system with the example of four polyan-
their parents, who are responsible for arranging mar- drous partitioners, three of whom are the sons of one
riages. Various mechanisms were employed to limit the man while the fourth is the son of his brother. With per
number of marrying brothers in Western and Central stirpes inheritance the first three brothers ‘‘take their
Tibet as well, including sending sons off to join a mon- father’s share’’ (which means that each receives one-
astery, to find their fortunes elsewhere, or to marry sixth of the property). The fourth brother receives a full
heiresses. half for himself. However, partitioners may negotiate
As in other regions of Tibetan language and culture, about where to begin the per stirpes accounting, and
Nyinba polyandry coexists with occasional polygyny. men with grown sons may use their own generation. If
This inevitably occurs in instances of infertility and so, the partition will end up in an effectively per capita
may also occur when one or more co-husbands become division of property.15
deeply dissatisfied with the common wife and add a sec-
ond wife to their marriage. The latter circumstance, 13. By 1984, 16 out of 26 (62%) of the conjoint marriages that had
polyandry and polygyny with two fertile women, form- been contracted during recent decades—and during living individu-
ing a ‘‘conjoint marriage,’’ has important conse- als’ lifetimes—had undergone partition. Five of those marriages
had ended in widowhood without partition (see Levine 1988:174).
quences.12 It partly severs the formerly collective mar- The recent fate of the other five conjoint households (see n. 12) is
riage, since sexually exclusive partnerships tend to unknown.
form. Conjoint marriage has three further conse- 14. Accounts of Indian Himalayan polygynandry note that upon
quences. First, it offers polyandrous men additional op- partition women similarly may designate fathers of children born
into polygynandrous marriages. This is the only one of a number
of reported methods, including lot, order of birth (the first child
being attributed to the eldest brother, the second to the next-eldest
11. Slave marriages were traditionally monogamous and uxorilocal. brother, and so on), physical resemblance, and affective relation-
Only in the past generation have the wealthier members of this ships between husbands and wives (see Berreman 1975:128–29;
population begun marrying virilocally and polyandrously—compli- Parmar 1975:83). Any discussion of paternity designation in poly-
cating comparisons with traditional landholders. In the past gener- andry calls for reference to the classic cases of the Nayar and the
ation as well, the descendants of slaves have begun marrying mem- Toda, on which interested readers may consult Gough (1959:26–
bers of traditional landholding households. (On the marriage and 27) and Rivers (1906:319–23).
domestic systems of the different strata, see Levine 1988:72–74, 15. Studies of traditional Tibet suggest that inheritance followed
81–84.) per capita reckoning and Nyinba reliance on per stirpes calcula-
12. Among the 150 existing marriages surveyed in 1983, 9 (6%) in- tions may be due to the influence of Nepali law. Whatever the
volved simple polygyny (men married to two or more women be- source, this practice is congruent with their emphasis on individ-
cause of infertility), 2 (1.3%) involved polyandry combined with ual paternity. (On Nyinba property division, see Levine 1988:178–
polygyny because of infertility, and 5 (3.3%) were conjoint. 84.)

table 1
Wealth of Households by Type of Household
Total Landholdingsa Per Capita Landholdingsa

(plow days) (plow days)
Type of Number of
Household Households Mean S.E. Mean S.E.
Intact 60 14.61 0.97 5.42 0.41

Conjoint 8 22.58 1.59 6.97 0.78
Partition 18 18.29 2.00 4.84 0.63
a
Land per co-husband prior to partition.
Causes of Dissolution of Polyandrous to matter once a decision to partition is reached and

Marriages brothers take their personal shares to their new homes.
While individual expectations would seem to be more
estate size important than joint holdings, brothers contemplating
partition often do not know exactly how much land
We evaluated the prediction that resource availability they will receive after the household is divided. Their
influences the stability of polyandrous marriages by life-cycle stage and the outcome of negotiations that
comparing the estate size of polyandrous Nyinba house- take place after partition is initiated will determine
holds which experienced partition with the estate size how the per stirpes division is calculated, and these ne-
of those which did not. For these analyses, we focused gotiations can have a major impact on share sizes. The
on land rather than other economic resources, such as outcome may also be affected by the brothers’ relation-
opportunities in herding or trading, for several reasons. ships with the common wife and the number of chil-
First, landholdings are relatively inelastic because good dren each brother has produced in the existing marriage.
arable land is a scarce commodity in these communi- Another possible point of uncertainty is how the broth-
ties. Second, Nyinba see themselves as primarily agri- ers will align themselves after partition. For these rea-
culturalists, with trade as highly volatile and herding sons, the size of the new estates may not be known un-
too small a supplement to their income. Third, it would til the partition is well under way.
have been difficult to assess external economic opportu- Therefore, in evaluating how estate size may affect
nities. Finally, land inelasticity has figured prominently decisions to partition, we decided to examine both land-
in theoretical models of Tibetan polyandry. holdings of intact households and brothers’ potential
Estate size was calculated according to Nyinba re- per capita shares to see if either proved a reliable pre-
ports of their holdings, which were expressed in terms dictor of partition.
of ‘‘plow days.’’ A plow day is the amount of land that In our sample, estate size varies widely. The poorest
can be plowed by a yak-ox crossbreed in the course of a household’s landholdings amounted to only 2 plow
single day. Measurements of Nyinba fields in 1983 days, while the wealthiest household’s landholdings
show a plow day to be equivalent to 0.3 acres. Nyinba were equivalent to 34 plow days. The average house-
further categorize their fields as highland or lowland, hold’s landholdings were equivalent to 16.1 plow days.
with lowland fields being more productive and accord- There was significant variation in wealth among the in-
ingly more desirable. Levine (1988:248) has calculated tact, partitioning, and conjoint households (one-way
that the average lowland field yields 40% more food analysis of variance: F2,83 5 4.91, p 5 .0096; table 1),16
than a highland one. Our value for landholdings was de- but the pattern of this variation was not consistent with
rived by multiplying lowland plow days by 1.4 and add- predictions. The landholdings of households which be-
ing that value to the number of highland plow days for came conjoint were significantly larger than the land-
each household. holdings of households which remained intact (Scheffe
It is less clear, however, how men assess landholdings test, p , 0.05).17 Although households which remained
when contemplating partition and how best to model intact had somewhat less land than households which
their evaluations. Is the dominant factor in individual
men’s reckoning the extent of the undivided house-
hold’s estate, or is it the size of the shares they antici- 16. One-way analysis of variance evaluates the homogeneity of val-
pate receiving after partition? The literature on Tibet ues within designated samples. A significant result indicates that
the variance between groups is greater than the variance within
speaks mostly of the importance of the family corpora- groups.
tion and of maintaining household wealth intact from 17. The Scheffe test evaluates the significance of the difference be-
one generation to the next. But household wealth ceases tween any two groups compared in an analysis of variance.

F ig. 2. Number of co-husbands by type of household.

Black, polyandrous; white, partition; striped, conjoint.
table 2
Average Number of Co-husbands per Household by
Type of Household
F i g. 1. Number of co-husbands per household. The
average polyandrous household initially contained
3.5 co-husbands; the Nyinba consider 3 the optimal Number of Co-Husbands per Household
number.
Type of Number of
Household Households Mean S.E.
partitioned, these differences are not statistically sig-
nificant.
The magnitude and significance of differences in ag- Intact 60 2.92 0.15
Conjoint 8 3.50 0.42
ricultural wealth among intact, conjoint, and parti- Partition 25 3.88 0.27
tioning households contract, however, when per capita
landholdings are considered. Partitioning households
own less land per co-husband than intact or conjoint
households. Nonetheless, there was no significant rela- The largest sibling groups were the least stable (fig.
tionship between per capita landholdings and house- 2). Among marriages with two co-husbands, 10% were
hold stability (one-way analysis of variance: F2,83 5 1.38, dissolved through partition. In contrast, 58% of mar-
p 5 0.2567; table 1). riages with four co-husbands ended in partition. There
One reason for the lack of association between estate was a significant difference in the number of co-hus-
size and partition may be economic options outside of bands among households that remained intact, became
agriculture. At the same time, land is so important to conjoint, and partitioned (one-way analysis of variance:
the household’s economic and social standing that one F2,90 5 5.84, p 5 0.0041). Households that partitioned in-
would expect to find some effect, and there is none. In cluded, on average, 3.9 brothers, whereas those that be-
any event, the consequences of partition are unambigu- came conjoint included 3.5 brothers and those that re-
ous. Households created by partition had, on average, mained intact included 2.9 brothers (table 2).18
half the land they would have held had they remained
intact (Levine 1988:253). These men, accordingly, could
disparities in age among husbands and wives
pass on to their sons only half the patrimony that would
have been their fortune had they inherited an undivided Both sociocultural and evolutionary biological models
polyandrous household. of polyandry predict that disparities in age between hus-
size of the sibling group 18. Thus we see that the households that become conjoint are
likely to include more land and more sibling co-husbands than the
In the Nyinba community, all brothers jointly marry a average polyandrous household. This may be because wealthy
single woman, so the number of co-husbands initially households have more sons to begin with and because the house-
depends upon the size of the fraternal sibling group. The holds with more sons are likelier to become conjoint. Given the
small number of cases at hand, however, it becomes impossible to
typical polyandrous household in our sample began distinguish the factors that contribute to a decision for partition
with 3.5 co-husbands (n 5 233 marriages), although from those that contribute to a decision for a continuing conjoint
some marriages initially included as many as 7 (fig. 1). marriage.

F i g. 3. Means and standard errors of age differences

between husbands of different birth positions and F ig. 4. Means and standard errors of age differences
their wives. Positive values indicate that husbands between husbands and wives for men in stable
are older than their wives, while negative values polyandrous marriages, passive partitioners, and
indicate that husbands are younger than their wives. active partitioners. Active partitioners are most
In most marriages, the oldest brother was older junior to their wives.
than the common wife, while men in higher birth
positions were generally younger than their wives.
Disparities in age among husbands and wives are
older than their common wife did so. Active partition-
most pronounced for the most junior co-husbands.
ers were nearly six years younger than their wives on
average, while their co-husbands were about one year
younger than their wives. Men who remained in stable
bands and wives will influence marital stability. To polyandrous marriages were approximately the same
evaluate these predictions we will draw comparisons age as their wives.
among three groups of men. ‘‘Polyandrous men’’ are It should be noted that the wives of men who con-
those who marry polyandrously and whose marriages tracted conjoint marriages or partitioned were not el-
remain intact over time. ‘‘Active partitioners’’ are those derly, even though they were usually older than their
who actively instigate conjoint marriages or partitions. husbands. The sample included 23 wives of known age
Men who initiate conjoint marriages are combined with whose marriages were terminated by conjoint marriages
those who initiate partitions because both events dis- or partitions. These women’s ages ranged from 19 to 69
rupt the polyandrous marriage. Men whose marriages years, but the average age was 34.9 years. Ten of the 23
are altered by conjoint marriages or severed by parti- women (43%) were no more than 30 years old at the
tions initiated by their siblings are called ‘‘passive parti- time of partition.
tioners.’’ Since birth order and disparities in husbands’ and
In the typical polyandrous marriage, the first-born wives’ ages are associated, it is not surprising that birth
brother was 3.8 years older than the common wife, order is also a good predictor of the role that men play
while all of the other brothers were younger than the in partition events. First-born brothers initiated only
common wife (fig. 3). This means that birth order is sig- 7% of all conjoint marriages and partitions, while sec-
nificantly related to the disparity in age between a man ond-born brothers initiated 22% and men born later in
and his wife (Pearson correlation coefficient: r 5 the birth order initiated 37–39%. There was significant
20.5919, p , 0.001, n 5 199). For men born late in the variation in the birth positions of polyandrous men,
birth order, the disparity in age can be substantial. The passive partitioners, and active partitioners (one-way
two sixth-born brothers were 14 and 15 years younger analysis of variance: F2,230 5 3.95, p 5 0.0206; table 3).
than their wives. When men contract new marriages, they nearly al-
The disparity in age between men and their wives is ways marry women who are younger than their first
significantly related to the role that men play in their wife (35/39 5 90%; sign test: z 5 4.80, p , 0.001).
marriages (one-way analysis of variance: F2,199 5 7.31, p Among the men who initiated conjoint marriages or
5 0.0009; fig. 4). Active partitioners are significantly partitions, 82% were younger than their first wife,
younger vis-à-vis their wives than their co-husbands while only 33% were younger than their new wife.
(Scheffe test, p , 0.05). In our sample, 27% of the men These active partitioners left first wives who were, on
who were younger than the common wife became ac- average, 5.7 years older and married second wives who
tive partitioners, while only 9% of the men who were were 3.6 years younger than themselves.

t ab l e 3 t ab l e 4
Birth Position and Marital Status Birth Positions of Co-husbands and Paternity
Birth Position Birth Position of Co-husband
Number 1 2 .2
Marital Status of Men Mean S.E.
Birth
Position Number of Number of Number of
Intact 135 2.05 0.13 of Childa Children % Children % Children %
Passive partitioner 50 2.02 0.17
Active partitioner 48 2.69 0.18
1 43 67 16 25 5 8
2 22 39 29 51 6 11
3 20 39 21 41 10 20
reproductive success 4 19 45 18 43 5 12
5 12 39 13 42 6 20
To examine the effect of reproductive success upon 6 11 50 5 23 6 27
7 6 38 5 31 5 31
men’s decisions to remain in polyandrous unions, we 8 3 25 6 50 3 25
have drawn upon cross-sectional census data collected Total 135 46 113 38 46 16
by Levine in 1982–83. These data provide information
about men’s ages, marital status, and reproductive his- a
tories. The techniques used to obtain and verify infor- Children surviving at the time of the survey.
mation about children’s paternity rely on local attribu-
tions and are described in appendix A. To account for
the fact that some men were married longer than others the roles that men played in their marriages. While men
and the fact that some men matured after they were who remained in polyandrous marriages produced, on
married, some of the analyses of reproductive success average, 1.8 children over the course of their married
that follow are based upon the number of children pro- lives, passive partitioners produced 1.1 children and ac-
duced per year of marriage for men aged 18 and over. tive partitioners produced 0.5 children during their orig-
It is almost inevitable that at the outset of the mar- inal marriages (one-way analysis of variance: F2,134 5
riage the eldest brother will have a certain advantage. 10.8964, p , 0.0001; table 5). These figures do not take
In some cases it is he who selects the wife and brings into account the fact that men who remained in polyan-
her home; in arranged marriages he is apt to be the one drous marriages had longer marriages—and thus more
who first initiates a sexual relationship with her. In opportunities to produce children—than men whose
fact, the eldest brother is most likely to be considered marriages were terminated. Significant differences in
the genitor of the first child born in the marriage: 67% individual reproductive success persist when this factor
of first-born children surviving at the time of the survey is taken into account (one-way analysis of variance:
were considered the offspring of the eldest (table 4). F2,134 5 4.3328, p 5 0.0150; fig. 5). While polyandrous
This advantage declines to some extent with subse- men and passive partitioners had produced, on average,
quent births, and the second birth is most likely to be 0.10–0.11 children per year of marriage, the active parti-
attributed to the second-eldest brother in the marriage. tioners had produced only 0.04 children per year of mar-
However, disparities in individual reproductive success riage.
among co-husbands are not completely eliminated over The low reproductive success of active partitioners is
the course of men’s lives. Birth order is negatively re-
lated to the number of children produced during their
t ab l e 5
marriages for men who remain in stable polyandrous
Reproductive Success of Co-husbands in Original
marriages (r 5 20.2740, p 5 0.028, n 5 67) and to the
Marriages
number of children produced per year of marriage (r 5
2 0.2355, p 5 0.028, n 5 67).
In households that became conjoint or partitioned, Total Number Number of
the most senior co-husbands had produced more chil- of Children Children Sired
dren than their younger co-husbands before the mar- Sired per Year
riage was terminated. In these households, birth posi- Number
tion is negatively related to the number of children Marital Status of Men Mean S.E. Mean S.E.
sired during the original polyandrous marriage (r 5
20.3810, p 5 0.001, n 5 70) and to the number of chil-
Intact 67 1.78 .20 .11 .01
dren sired per year of marriage (r 5 20.3668, p 5 0.001, Passive partitioner 39 1.08 .17 .10 .02
n 5 70). Active partitioner 31 .45 .14 .04 .01
Reproductive success is, moreover, associated with

table 6
Average Degree of Relatedness within Households
Average Degree of Relatedness
Number of
Type of Household Households Mean S.E.
Intact 29 0.37 0.02

Conjoint 7 0.39 0.02
Partition 21 0.37 0.04
0.50. The average value was 0.36, halfway between the

values for full siblings and half-siblings. There was no
consistent relationship between the number of co-
husbands in households and their average degree of re-
F i g. 5. Means and standard errors of number
latedness to one another (Pearson correlation coeffi-
of children produced per year of marriage by
cient: r 5 0.0297, p 5 0.413, n 5 57).
polyandrous men, passive partitioners, and active
The average degree of relatedness among co-husbands
partitioners. Active partitioners produced
did not vary significantly between households that re-
substantially fewer offspring per year of marriage
mained polyandrous, became conjoint, or partitioned
during their initial marriages than other men.
(one-way analysis of variance: F2,54 5 0.0918, p 5 0.9124;
table 6). Thus the degree of relatedness among co-
husbands did not seem to influence the stability of their
apparently not the result of their own infertility. If that marriages.19
were the case, we would expect little change in their It is possible that men might have taken advantage of
reproductive rates after active partitioners remarried. In partition or conjoint marriage to increase the relat-
fact, these rates significantly increased after remarriage edness between themselves and their co-husbands. To
(t 5 2.88, d.f. 5 29, p 5 0.007). If active partitioners had determine whether realignment consistently altered
been less fertile than their co-husbands, we also would the average degree of relatedness among co-husbands,
expect the same disparities in reproductive success be- we compared the composition of 27 sibling groups be-
tween active and passive partitioners before and after fore and after partition. In 12 cases, no changes in the
they remarried. However, men who were active parti- average degree of relatedness among co-husbands were
tioners displayed reproductive rates in their new mar- possible because there were only two co-husbands in
riages three times those of men who were passive par- the marriage (3 cases) or all co-husbands had the same
titioners (one-way analysis of variance: F1,67 5 5.9420, degree of relatedness to one another (9 cases). In the re-
p 5 0.0174). maining 15 households, men could have altered the de-
gree of relatedness to their co-husbands when they
kinship among co-husbands realigned themselves in new marriages. In 5 of these
households, men formed 2 new polyandrous marriages.
To test the prediction that the degree of relatedness In the remaining 10 households, one co-husband mar-
among co-husbands will influence the stability of poly- ried monogamously while the others married polyan-
androus marriages, we compared the average degree of drously. Thus, 20 new polyandrous marriages were
relatedness among co-husbands in households that re- formed. In 11 of these new marriages, the average degree
mained intact, became conjoint, and partitioned. Our of relatedness among co-husbands was higher than the
sample included the men of 57 households. The proce- average degree of relatedness among co-husbands in the
dure for estimating the relatedness among co-husbands original marriage; in 8 cases the pattern was reversed,
is detailed in appendix B. To obtain the average degree and in 1 case there was no change. Thus, there is no
of relatedness within each group of polyandrously mar- consistent tendency for men to increase or decrease
ried men, we summed the coefficients of relatedness for
each pair of co-husbands and divided the total by the
number of pairs of co-husbands in the marriage. As we 19. Nyinba state that having different parentage has a negative im-
noted earlier, data on relatedness derive from local attri- pact on fraternal relationships, and the offspring of conjoint mar-
riages do seem more likely to contract conjoint marriages them-
butions (see appendix A). selves (see Levine 1988:154). Some such unions, however, are
The estimates of the average degree of relatedness sustained for the life of the partners, that is, they do not necessarily
among co-husbands in households ranged from 0.09 to result in partition (see n. 13).

their relatedness to their co-husbands when they re- capital of Ladakh and the availability of new job oppor-
align themselves in new polyandrous marriages. tunities were followed by a similarly dramatic decline
in polyandrous marriages (Crook and Crook 1988:105),
although the government-instituted proscription on
Discussion polyandrous marriage may have played a role in this de-
cline as well.
We have reviewed a range of models of polyandrous sys- The abandonment of polyandry under such circum-
tems deriving from the paradigms of sociocultural stances may reflect processes of adjustment between
anthropology and evolutionary biology and tested hy- systems for organizing labor and forms of marriage, as
potheses about factors influencing the stability of poly- studies of polygyny suggest (Boserup 1970, Goody
androus marriages. These findings have substantiated 1976). Goody has argued that traditional Tibetan poly-
some of the common presumptions about polyandrous andry is most appropriately viewed as a special form of
systems and called others into question. Although it is household organization that enhances viability under
hazardous to generalize from correlations to the causes conditions of high demand for male labor while pre-
of behavior, our results do allow us to make several cau- venting the diminution of property and status (1990:
tious observations about factors that contribute to the 139, 153). This view fails to explain why a system rely-
stability of polyandrous marriages—for Nyinba land- ing so heavily on male labor developed in the first place.
holders at least. It also leaves unresolved long-standing questions about
First, the Nyinba data do not support the common ar- the role of economic and environmental factors in the
gument that polyandry is maintained by economic and development of polyandrous institutions.
ecological factors that constrain men’s ability to sup- Four factors were consistently associated with the
port their families. The total and per capita landhold- stability of Nyinba polyandrous households: (1) the size
ings of intact households are approximately the same as of the sibling group, (2) the magnitude of disparities in
the landholdings of households that partitioned, contra- age among husbands and wives, (3) birth position, and
dicting the notion that men leave polyandrous mar- (4) the number of children that men fathered. All these
riages whenever they can afford to do so. Men with the factors are closely correlated. It would be ideal to subdi-
largest estates in our sample contracted conjoint mar- vide the sample and examine the effects of each variable
riages but did not partition. Moreover, poor men did not separately. However, the numbers of households and in-
necessarily remain in polyandrous marriages.20 dividuals in our sample are too small and the factors are
This finding, however, does not refute hypotheses too closely related to make this a practical option.
that polyandry originated under conditions of severe re- Nonetheless, we can make some reasonable deductions
source constraints in the unrecorded past, nor does the about the relative importance of these factors in men’s
Nyinba case contribute to our understanding of how decisions to partition.
modernization and changes in subsistence strategies There are several reasons that the number of co-hus-
may undermine commitments to polyandry. Reports on bands might influence the stability of polyandrous mar-
traditional Tibet have suggested that polyandry pre- riages. First, it is possible that large sibling groups are
dominated among landed agriculturalists and was less unstable simply because of their size. The Nyinba be-
common among pastoral nomads and traders (Aziz lieve that trifraternal polyandry is highly desirable and
1978:157–58). Recent studies of indigenous ethnic Ti- that larger marital unions are problematic because they
betan communities in India and Nepal have described are prone to discord. If each pair of co-husbands must
a sharp decline in polyandry following economic establish and maintain a cooperative, cordial, or at least
changes and fuller incorporation into the modern world tolerant relationship in order to sustain the marriage,
economy. In Limi, for example, men began leaving their then it is easy to see why the number of co-husbands
polyandrous marriages after the community shifted might influence marital stability. Second, following
from dependence on agriculture to major involvement evolutionary biological predictions, larger sibling
in long-distance trade and an expansion of herding groups might be undesirable because they impose
(Goldstein 1978:331–32). Urban development near the greater limits upon men’s reproductive opportunities
(Beall and Goldstein 1981:9; Crook and Crook 1988:
108). Third, following sociocultural predictions, sexual
20. While these analyses show no consistent relationship between jealousy might intensify as more men share sexual ac-
landholdings per se and the rate of partitions, there are a number cess to a single wife.
of related circumstances whose effects we have not been able to However, if the size of the sibling group were the only
test. One concerns the rare occasions in which landed households
acquire second estates from families without heirs. Such house- factor influencing men’s decisions to dissolve their
holds do seem likelier to partition, but the reasons seem more com- marriages, then all co-husbands in large sibling groups
plex than ownership of large parcels of land (see Levine 1988:247, should be equally likely to initiate partitions. Instead,
251). Another involves households in new hamlets, which have the men who take the initiative in contracting conjoint
ready access to new lands and are likelier to undergo partition.
However, the majority of such households are descended from marriages and partitions are consistently those who are
slaves, who, as we have seen (n. 11), were customarily monoga- most junior to their wives and co-husbands and least
mous (see Levine 1988:276). successful in producing children of their own.

The dissatisfaction of the most junior brothers with successful in fathering children. The average active par-
their marriages may be related to their subordinate sta- titioner is 29 years old and is married to a 34-year-old
tus within the household. In Tibetan households, the woman. He has produced half as many children per year
eldest brother has greater authority and may take ad- of marriage as his co-husbands. These men are capable
vantage of his position to delegate unpleasant work to of fathering children and have successful reproductive
his junior siblings or sexually monopolize the common careers when they remarry. Thus, active partitioners are
wife. It is plausible that strong-willed younger siblings old enough to have produced several children already,
resent this. While we lack the necessary data to exam- they are physically capable of producing children, and
ine the subjective experiences of polyandrous men, the their wives were generally young enough to have borne
high stability of bifraternal polyandrous marriages sug- them children. What accounts for their poor reproduc-
gests that such feelings are not prominent in small sib- tive performance?
ling groups. It is possible, however, that a man’s stand- One possibility is that these men were denied equal
ing in his family declines a notch with each move access to the common wife by their brothers. From both
higher in the birth order and that any disadvantages as- an evolutionary and a sociocultural perspective, such
sociated with holding junior status would increase cor- competition might be expected. Crook and Crook
respondingly. (1988:110) point out that the ‘‘reproductive advantage
Birth order is also correlated with the magnitude of of the older brother can be increased by the elimination
age differences among husbands and wives. When from co-husbandry of the brother nearest to him in
brothers marry, their common wife is typically some- age.’’ They suggest that this may be the reason that sec-
what younger than the oldest brother but older than the ond-born brothers predominate among monks in two
other brothers. The largest disparities in age among hus- Zanskari monasteries (p. 110, citing Crook and Shakya
bands and wives are found in the largest sibling groups, 1988).22 Among Nyinba, sexual exclusivity is strongly
making it difficult to assess the relative importance of disapproved (see Levine 1988:164–65). If so, contraven-
these two factors in men’s dissatisfaction with their tions of the norm may be particularly galling to an ex-
marriages. There are several reasons, however, that we cluded brother and serve to justify partition.
might suspect that disparities in age among husbands A second possibility is that the low reproductive suc-
and wives influence decisions to partition. First, as cess of active partitioners is a consequence of a decision
sociocultural anthropologists and evolutionary psychol- to partition rather than its cause. That is, active parti-
ogists have suggested, young men married to older tioners may limit their own reproductive activities as a
women may find them less satisfactory companions, deliberate strategy to facilitate their later separation
less attractive physically, and less satisfying sexually. from the family. Nyinba say that men whose marriages
Moreover, men’s reproductive prospects are enhanced have broken down and who are planning partition pur-
by marriage to a younger woman. There is some reason posely avoid sexual relations with the common wife.
to believe that Nyinba men do prefer younger women Any partitioner who has had such a relationship is pe-
as wives. When men remarry, disparities in age among nalized; he is said to have spurned the common wife
husbands and wives are generally reduced or elimi- and is obliged to compensate her with part of his share
nated. Virtually all men married women who were of the household’s heirloom jewelry. More problematic,
younger than their first wives, and two-thirds of the a man who fathers one or more sons by the common
men married women who were younger than them- wife and fails to take them with him is obliged to leave
selves. Although it is plausible that men initiate con- a part of his property share behind for them. A third pos-
joint marriages and partitions because they are dissatis- sibility is that the common wife is trying to precipitate
fied with much older wives, we have no direct evidence partition by avoiding sexual relations with a husband
that men are dissatisfied with their wives or have not she does not like. However, Nyinba women do their
established satisfactory sexual and personal relation- best to retain all their husbands in order to maintain
ships with them; moreover, the fact that partitioners their households’ wealth and have it passed on intact to
marry younger women may simply reflect the fact that their children.
more young women are available for marriage.21 While such strategies may contribute to the markedly
In addition to being junior to their co-husbands and low reproductive success of active partitioners, other
much younger than their wives, men who initiate con- factors clearly are at work. Even in stable polyandrous
joint marriages and partitions have generally been un- households, younger brothers fail to reproduce as suc-
cessfully as their older siblings. Birth position is nega-
tively related to the total number of children produced
21. Nyinba men seek young unmarried women for a number of rea-
sons. Disentangling a childless woman from her marriage involves during the marriage and to the number of children pro-
heavy compensation payments and often embroils the pair in long- duced per year of marriage. Of course it is possible that
lived disputes with the former husband and his allies. Married
women with children almost never leave their husbands, and wid-
owed women with children always stay with those children in
their natal home. Unmarried women in their twenties and above 22. In Indian polygynandry, such problems are resolved by incorpo-
are known by the unflattering term ‘‘left over’’ and tend to be in rating additional wives into the household, which mitigates disad-
this situation because of perceived personal flaws. vantages experienced by younger brothers.

this negative correlation is created by men who are this subject in the published literature on other polyan-
planning to terminate their marriages but have not yet drous societies. The cross-cultural literature also re-
done so. However, the third- and fourth-born brothers, veals marked variations in the way genealogical kinship
who suffered the greatest reproductive decrements in is understood and how its importance is assessed (see
polyandrous households, had been married 19 years on Schneider 1984). Many societies play down the genea-
average. logical aspects of paternity, the most notable examples
The degree of relatedness did not have any consistent being in Oceania.24 This is one example of the great
influence upon the stability of polyandrous men’s mar- cross-cultural diversity in behavioral obligations among
riages, and when men formed new polyandrous mar- kin.
riages after partition they did not consistently increase As our study has shown, sociocultural and evolution-
their degree of relatedness to their co-husbands. These ary theories may complement one another in testing
data are somewhat surprising in view of the fact that hypotheses about human behavior. The two bodies of
the Nyinba suggest that the most stable marriages in- theory are concerned with different levels of explana-
volve groups of men who are closely related to one tion, sociocultural theory with the proximate and evo-
another and the fact that kinship is associated with lutionary theory with the ultimate factors that shape
polyandry in other animal species. Evolutionary psy- human behavior. For most sociocultural anthropolo-
chologists might argue that this result is due to the fact gists, human behavior is culturally conditioned and em-
that polyandry is a recent innovation in human societ- phasis is placed on the plasticity of culture. Beyond this,
ies and our evolved psychology is not designed for such the field subdivides. Some sociocultural anthropolo-
situations. gists see culture as autonomous and not necessarily
In summary, we conjecture that Nyinba men’s deci- adapted to the material conditions of existence. Others
sions to leave their marriages are mainly influenced by treat culture as constrained by features of the political
their disadvantages within the sibling group and their and economic environment or other features of the
relationships with much older wives, which reduce sociocultural system.
their chances of reproducing successfully within the Recently, however, moves have been made to bridge
polyandrous marriage. The importance of these factors the sociocultural and evolutionary divide. Many biolog-
is underscored by the fact that when men remarry, they ical anthropologists have become sensitive to the ways
remedy these circumstances by marrying women in which social and cultural systems shape mating and
younger than their first wives and by siring more chil- parenting behavior, while certain sociocultural anthro-
dren than they had in their first marriages. pologists have begun to test evolutionary predictions
The associations we have established contravene cer- (see, e.g., Betzig, Borgerhoff Mulder, and Turke 1988,
tain predictions drawn from evolutionary and sociocul- Hewlett 1992). It may be that a meaningful synthesis—
tural theories of human behavior while supporting oth- which encompasses presumptions about the power of
ers. Our findings suggest that the most important factor culture, the forces of chance in cultural history, and pre-
prompting marital dissolutions is the reproductive dis- sumptions about how natural selection has favored be-
advantages experienced by younger men, often in the haviors that enhance reproductive success—will prove
larger sibling groups. While the argument that re- elusive. Nonetheless, a fruitful discourse between the
stricted reproductive opportunities contribute to poly- two paradigms has begun; it undoubtedly will continue
androus marital breakdowns is consistent with the pre- and may enhance our understanding of human be-
dictions of evolutionary theory, it is not prominent in havior.
sociocultural theory.23 Nyinba culture and ideas about
kinship, however, stress such expectations, and many
sociocultural anthropologists would agree that emic Appendix A: Sources of Data
evaluative criteria of these kinds can influence men’s
assessments of the satisfactions of their marriages and The data cited in this paper derive from different strate-
their willingness to continue them. Such decisions, gies of data collection carried out at different times,
moreover, are not made in a vacuum. Each man may which can be subdivided as follows:
have to decide individually about his marriage, but oth- 1. Marriage, landholdings, and fertility. These data
ers may influence this decision. Men’s wives, siblings, derive from a structured questionnaire and census ad-
and parents, as well as friends, may be deeply concerned ministered in 1982–83 to every household in two Ny-
about and try to play a role in their marital decisions. inba villages and a random sample of households in two
Sociocultural theory suggests that Nyinba views other villages (which were difficult of access). Members
about the importance of having ‘‘own’’ children may be of 126 out of 184 Nyinba households completed the
culturally idiosyncratic. Certainly we read little about questionnaire; another 4 interviews were incomplete.
23. In his classic article on polyandry and the definition of mar-

riage, Leach speaks of the rights of establishing legal parenthood 24. By this we mean the classic cases of the Trobriands and aborigi-
over a spouse’s children and monopoly over a spouse’s sexuality nal Australians, whose members reportedly did not acknowledge
but not rights for reproductive purposes per se (1966 [1955]:107). men’s biological roles in reproduction (see Weiner 1992:73–75).

2. Partition. These data were collected in the course tions, a severe constraint. The second would be to com-
of unstructured participant observation and semistruc- pute two parallel estimates, one based upon a minimum
tured interviews during 1973–75 and 1982–83. This in- estimate of relatedness between men and the other
cluded interviewing at least one participant from every based upon the maximum estimate. The true figure
partition known to have occurred during the preceding would be assumed to lie somewhere in between. The
25 years about the causes of the breakup, how the fam- problem with this method is that the range between
ily was subdivided, and the property share each man re- these values is broad, and it does not incorporate what
ceived. we know about mating patterns within the population.
3. Paternity. Attributions of paternity were collected The third approach, which is the one taken here, is to
in 1982–83 for all children in three of the four Nyinba generate an estimate of the average degree of relat-
villages. Because paternity can be a sensitive matter, edness within the population and use this unbiased esti-
some of these questions had to be referred to close rela- mate to compute relatedness when information is in-
tives outside the household. These data were cross- complete.
checked against genealogies and partition histories in We attempted to estimate the average degree of relat-
which paternity affected property divisions. These attri- edness within the population as follows: We tabulated
butions may not be empirically accurate but do reflect the number of pairs of children in the same household
local ideas. who were known to have the same father, the same
4. Genealogical materials. These were sought from mother, different fathers, and different mothers. Infor-
one member of every Nyinba household, gradually, over mation about 386 pairs of children in polyandrous
the course of the two field studies. Because people typi- households was available. There were 192 pairs of sib-
cally specify the parentage of children born of complex lings with the same mother and same father, 38 pairs
polyandrous-polygynous unions in genealogies, they with the same father and different mothers, 129 pairs
were used to validate the data on paternity. with the same mother and different fathers, and 27
pairs with different mothers and different fathers. These
values were used to compute the proportion of pairs of
children in the same cohort that were full siblings, pa-
Appendix B: Procedure for Computing ternal half-siblings, maternal half-siblings, or unrelated.
Relatedness among Co-husbands Each of these proportions was multiplied by the appro-
priate degree of relatedness to obtain the expected de-
The degree of relatedness (r) between two individuals is gree of relatedness between any given pair of children
defined as the probability that two individuals carry in the same household. Thus, (192/386) 3 0.5 1 (38/
copies of the same gene through descent from a com- 386) 3 0.25 1 (129/386) 3 0.25 5 0.3569. This figure is
mon ancestor. This is, in turn, a function of the number an estimate of the expected degree of relatedness be-
of common ancestors two individuals have and the tween any pair of children raised in the same house-
number of generations that they are displaced from hold. The estimate underestimates relatedness to some
their common ancestor(s). The relatedness among extent because it does not allow for the fact that the fa-
parents and offspring is 0.5. Siblings are related by thers of children with different fathers may have been
0.25 through their mothers and/or 0.25 through their related. We are unable to rectify this omission because
fathers. When siblings have the same mother and this is the value that we are trying to estimate. Any
the same father, these components are summed to method that relies upon incomplete genealogies must
equal 0.5. make some assumption about relatedness in the most
Computations of degrees of relatedness among the distal generation for which information is available.
Nyinba are complicated by the complexity of polyan-
drous marital arrangements. In general, related men are
married jointly to a single woman. There are also cases,
however, in which groups of men marry more than Comments
once, unrelated men marry jointly, children are fathered
illegitimately, and so on. This means that the degree of
relatedness among children in the same household may m . b o r g e r h o f f m ul d e r a n d k. a. h a dd i x
theoretically vary from 0 to 0.5. Although these compu- Department of Anthropology, University of California
tations are tedious, they do not present significant tech- at Davis, Davis, Calif. 95616, U.S.A. 13 x 96
nical obstacles when genealogies are complete. How-
ever, a major practical difficulty arises for children who Levine and Silk conclude with a brief discussion of the
are sired by related men whose genetic relationship to compatibility between evolutionary (by which they
each other is not known precisely. mean behavioral ecological) and sociocultural perspec-
There are several different ways to resolve this prob- tives within anthropology, and their paper demon-
lem. One would be to exclude all pairs of siblings for strates the fruits of such collaboration. Using insights
which genealogical data were incomplete. Unfortu- from both perspectives, they make some simple predic-
nately, this would restrict our data to households for tions of when polyandrous households might split,
which we had complete information over many genera- shedding direct light on the instabilities within polyan-

drous marriage among the Nyinba. What can we, as evo- tistical analyses of these data because of the small sam-
lutionary anthropologists, say more generally about ple size. We also appreciate the difficulty in collecting a
polyandry on the basis of this study? data set of this quality. In addition, we strongly support
The results of these analyses suggest that the evolu- collaborative work between evolutionary biologists and
tionary biologists’ model for polyandry in humans has sociocultural anthropologists. We therefore raise these
not been well specified. The hypothesis, first proposed issues to point out that researchers in the future will
by Alexander and elaborated on by Goldstein and need larger samples or stratified sampling techniques in
Crook, is that a group of brothers will share paternity order to make more precise tests of the current behav-
whenever economic and demographic circumstances ioral ecological model for the distribution of polyandry
limit the ability of each to support a monogamous fam- in human societies. In our opinion Levine and Silk’s
ily of his own. If this hypothesis were correct we would, study, along with Durham’s (1991) analysis, stands as
amongst other things, expect (a) polyandrous house- the best test to date.
holds to have smaller estates than nonpolyandrous
households and (b) partitioning polyandrous households
to have larger estates than nonpartitioning polyandrous r am e sh c ha n d r a
households. Neither of these hypotheses is supported. Anthropological Survey of India, Udaipur 313001,
Levine and Silk do not seem to appreciate the signifi- India. 13 xii 96
cance of the refutation of this widely cited model for
polyandry. Instead, they focus on the evidence for sex- Levine and Silk are to be congratulated for opening a
ual jealousy among the Nyinba and note its congruence new vista in research on polyandry, which has previ-
with evolutionary predictions. While we agree that this ously been studied in terms of what it is, where it is,
finding is very interesting, wasn’t the original hypothe- how it functions, and what factors accentuate it. Postu-
sis that men would be willing to share paternity when lating ‘‘sources of instability in polyandrous mar-
resources were stressed? The refutation of this long- riages,’’ prefixed by ‘‘why polyandry fails,’’ seems a half-
standing hypothesis warrants more thorough analysis baked approach to polyandry as having socio-ritual and
and discussion. We need a multivariate model that physical components (sexuality and the siring of chil-
looks at the risk of a marriage’s partitioning (or, indeed, dren) that undermine the process of evolution of human
becoming polyandrous) with respect to the interaction marriage (entailing both components). Given that poly-
between the availability of resources and a set of vari- andry has a long history in a number of different socie-
ables indicative of potential sexual jealousy (birth order, ties around the world—the Paiute and Shoshone in
sibset size, and spousal age differences). Logistic regres- North America and others in Australia, Africa, and Asia
sion, ideally with time-varying covariates, is the appro- (Sri Lanka, Tibet, India)—the notion that it ‘‘fails’’
priate statistical procedure for testing a model devel- needs rethinking. As is evident from Indian Himalayan
oped along these lines, since it is designed to analyze a societies, among others, and in the light of the explana-
dichotomous outcome such as ‘‘partition’’ versus tions of the evolution of human marriage of McLennan,
‘‘don’t partition’’ or ‘‘join your brothers’’ versus ‘‘don’t Spencer, and Westermarck (see Chandra 1992:154–59),
join your brothers.’’ which stress the dynamic nature of human societies
In addition, estate size is only one way of measuring and humans’ capacity for manipulations to suit their re-
resource availability. Though Levine and Silk recognize quirements, it seems inappropriate to describe social re-
this point, they do not pursue it. Himalayan households ality as ‘‘failing’’ rather than ‘‘changing.’’ There are dif-
produce wealth not only through agriculture but ferences, after all, between decay and transformation.
through the herding of domestic stock and investments Levine and Silk’s hypothesis that age disparity among
in long-distance trade. Depending on the extent to co-husbands results in disadvantages in sharing the
which a household’s economic activities are diversified, wife’s sexuality and in turn diminishes chances of sir-
labour, particularly male labour, may be limiting. While ing children needs to be substantiated with a set of dia-
one set of brothers may inherit a large herd of pack ani- chronic data from different societies reported as poly-
mals suitable for large-scale involvement in long-dis- androus rather than merely a few synchronic data from
tance trade, allowing that household to exploit trade a single society—even one in which polyandry is
and agricultural opportunities, another set of brothers normative such as the Nyinba—especially when the
may inherit few animals and instead concentrate its patterns emerging from the study are somewhat similar
economic activity upon agriculture. Thus, though all to those produced by earlier studies. In explaining poly-
households theoretically have equal access to each of andry the authors themselves seem to waver, for inter-
the economic spheres in this tripartite economy, some nal contradictions appear with regard to the role of soci-
households in fact require more males of working age etal profile and economic and ecological factors and the
than others. Therefore we suggest that two further mea- limited generalizability of results based on so small a
sures be introduced into the multivariate analysis—the sample.
number of men of working age and some ranking of I wish that Levine and Silk had recognized the au-
household economic diversification. thenticity of the sacredness of the institution of mar-
We are very sympathetic to Levine and Silk’s com- riage, of which, at least in the Eastern context, polyan-
ments about being unable to conduct more complex sta- dry is only a derivative. The socio-ritual aspect

predominates over the sexual aspect of marriage. Even proaches. The bipartisan approach adopted here goes far
today, Indian Himalayan polyandrous societies depend beyond the disputations between adherents of these ap-
on the kinship bond in polyandrous families to keep the proaches that have been all too prominent in recent
economy intact at the sibling level. Whatever malad- years and moves us on to a promising discourse be-
justments there may be over the sharing of the common tween the two. The value of this approach is clearly re-
wife’s sexuality or her wifehood are minimized in the vealed by the fact that the two perspectives can now be
interest of the success of the polyandrous union neces- seen to support one another in the examination of the
sary for survival under harsh environmental conditions cultural and the demographic aspects of village life and
(fraternal polyandrous societies are mostly found at alti- enable us to test overlapping hypotheses in a highly cre-
tudes of 2,000 m and above), despite the social change ative manner.
associated with development efforts and a shift in the Levine and Silk point out that it is difficult to estab-
economy from kind to cash, which is reorienting minds lish a consistent connection between specific eco-
from collectivism to individualism and introducing the nomic, ecological, or demographic variables and the oc-
notion of a joint venture at the sibling level and rela- currence of contrasting marital forms in the Himalayas.
tions between spouses devoid of compulsory sharing This should not surprise us. The marital outcomes are
(Prasanna et al. 1996). Changing scenarios of human ac- clearly due to the systemic interaction of several such
tion resulting in dents here and there in social arrange- factors and, in addition, cultural constraints such as the
ments may too quickly be identified as ‘‘weakness’’ and presence or absence of large monasteries and ideological
‘‘failure’’ of the system. Even with monogamy, often contrasts.
considered the ultimate marriage form, concepts such To determine both the proximate sources of stressors
as living together without marriage and unwed mothers determining marital practice and an estimation of be-
have crept in to stay. Does this mean that monogamy havioural effects on reproductive fitness, large and ef-
is failing? It is, however, reasonable to suggest that as fective data bases are required. So far only a limited
changing conditions make life easier people may aban- number of these are available, but a comparative survey
don polyandry over time in the course of the evolution across the trans-Himalayan region suggests that model-
of human marriage, as there are still societies with ling of local variation using common parameters may
some distinct ethos about this. One needs to examine soon become possible, elucidating the systemic pro-
the data and their interconnections within their context cesses generally at work (Crook 1995). Future work will
rather than in isolation. require that the materials be collected in similar ways.
Levine and Silk have analyzed the decline in number Each research location needs to be subjected to ques-
of co-husbands from many to two, but in all such cases tioning in common, and for this to be productive the
polyandry strictly speaking prevails. The three sources data must be in a comparable format. Research collabo-
of instability they identify stress sexuality and procre- ration will be the order of the day. Contrasts between
ation at the expense of the sociocultural legitimation Ladakh, Humla, Dolpo, Mustang, and Manang are
and ritual attached to acquiring the status of husband/ likely to be especially valuable because of the ecological
wife and married person and the ecological/environ- and cultural differences between these places.
mental conditions that operate as causes. Paradoxically, The factors determining the occurrence of polyandry
the sexual aspect is emphasized primarily when the au- become particularly visible when cultural and demo-
thors seem uncomfortable about polyandry as such. The graphic parameters are on the move. A recent review
Nyinba’s disapproval of exclusive rights over the com- has therefore attempted a preliminary analysis treating
mon wife’s sexuality repudiates these conjectures. The not only comparisons between locations but also com-
idea of compensation to the common wife on deserting parisons through time (Crook 1996). Data collected at
the polyandrous union after fathering a child clearly fa- one time should be clearly comparable with data col-
vors the maintenance of polyandry rather than the re- lected after a period of years. The common ethno-
verse. Perhaps for want of clarity on a man’s moving graphic practice of disguising names is called into ques-
from one polyandrous union to another or to a monoga- tion here, because accurate statements of place and
mous one, no conclusions are drawn singling out household location will be essential. Data collection in
sources of instability. Therefore the results at best pro- this research project needs to be as exactingly numeri-
vide suggestions for further research employing the so- cal as is the rule in analyses of social life in animal pop-
ciocultural and evolutionary approach for which the au- ulations and conducted in such a way as to be repeat-
thors undoubtedly deserve kudos. able by others in later years. Levine and Silk have set
an excellent example.
j oh n h. c r o o k
Winterhead Hill Farm, Shipham, North Somerset m il d re d di c ke m an n
BS25 1RS, U.K. 29 xi 96 2901 Humphrey Ave., Richmond, Calif. 94804, U.S.A.
10 xii 96
I greatly welcome this contribution to the analysis of
polyandrous behaviour in the Himalayas and Tibet. At This article is an interesting attempt to address the
last we have a friendly dialogue between theoretical po- ‘‘puzzle’’ of polyandry, entering, so to speak, by the
sitions arising from sociocultural and evolutionary ap- back door because the front door has up to now been

unbreachable. However, in regard to the fundamental calculations about reproduction, or do they somehow
Darwinian proposition that human behaviors are gener- arise in parallel with but distinct from innate tenden-
ally adaptive in terms of reproductive success, the cies?
means of measurement is critical. I think it can be said The authors’ difficulty in connecting the biological
that there is, as yet, no adequate measure of the relative with the cultural stems from their adoption of the Dar-
reproductive success of polyandry. The studies included winian social science position (Symons 1992), the view
here measure ‘‘surviving offspring,’’ but numbers of sur- that behavior is an attempt to maximize fitness. Evolu-
viving births may be an inadequate representation of re- tionary psychology, in contrast, asserts that behavior is
productive gain. First, humans may achieve short-term shaped by mental modules each of which was selected
reproductive gains through large numbers of offspring for in response to specific problems that confronted an-
or long-term gains through the longevity and quality of cestral humans. The authors imply that evolutionary
fewer descendants through several generations, de- psychology is of limited applicability in the current dis-
pending on circumstances. In the latter case, first-gener- cussion, asserting that it does not predict that fraternal
ation numbers are not an adequate reflection of long- polyandry will be less fractious than polyandry among
term reproductive success. (This point is also made by unrelated men. This is based on the assumption that
Fernandez [1981] in her critique of Beall and Goldstein evolutionary psychology views an act as the product of
[1981].) Secondly, in kin societies with coercive paren- a single mental module. However, a given context may
tal control of offspring decisions, the reproductive acts elicit responses from multiple modules, and behavior
of individuals in early adulthood may accrue to the re- may therefore reflect interactions among them. Nyinba
productive benefit of parents rather than of the offspring polyandry exemplifies this.
themselves. In this case, greater generational depth is Levine and Silk echo popular and scholarly sentiment
required for a realistic reproductive-success assessment. when they state that ‘‘humans seem poorly suited’’ to
Thirdly, where such parental control exists, reproduc- polyandry. We expect participants in polyandrous mar-
tive-success measures must include gains through fe- riages to chafe in the arrangement, which is not, we pre-
male as well as male offspring. sume, congruent with their general inclinations. Our
Only Crook and Crook (1988) attempt to address this intuition is that this arrangement is especially problem-
aspect, though with reservations, employing a limited atic for men. There is a consensus that individuals par-
set of genealogies collected by Prince Peter and updated ticipate in polyandrous marriage because it is advanta-
and computing the reproductive success of grandmoth- geous to do so. Darwinian social science suggests that,
ers resulting from their children’s marriages. One result under particular conditions, actors marry polyandrously
is the ‘‘slightly higher average grandmaternal fitness’’ of so as to maximize their reproductive success. However,
women who had married polyandrously (pp. 107–8). if men are calculating gene-counters, then why should
Part of our problem in understanding fitness is our polyandry be ‘‘difficult’’? Seeking to maximize repro-
too-easy acceptance of the dominant kin ideology of ductive success might lead polyandrously married men
such patrilineal societies, namely, that descent through to try to monopolize the reproductive resources of their
males is the primary route to success. This may be falla- wives, resulting in conflict among co-husbands. How-
cious. In hypergynous dowry societies (Dickemann ever, what this explanation fails to capture is the emo-
1979a,b, 1981), families strive at great economic cost to tional tenor of these sociosexual relationships. I suggest
place daughters in marriages of equal or higher status. that co-husbands do wish to monopolize their wives,
A similar effort, less overtly acknowledged in official but they do so principally for one reason—they are jeal-
ideology than the passage of land to male heirs, may be ous. Male jealousy is part of a proprietary attitude to-
operating here. The general theoretical problem of re- wards female sexuality (Wilson and Daly 1992). It is this
productive-success measurement remains a sticking which leads observers to view polyandry as counterin-
point in human studies. However, Levine and Silk have tuitive and inherently problematic. Male attitudes are
offered a provocative step toward the resolution of the best explained as the product of a mental module which
polyandry problem. was originally selected for because its possession in-
creased reproductive success. This module continues to
influence male sociosexual behavior irrespective of re-
d an i e l m. t. f e s s l e r productive consequences—husbands do not stop caring
Department of Sociology and Anthropology, Hofstra about their wives’ sexual fidelity following tubal liga-
University, Hempstead, N.Y. 11550-1090, U.S.A. tion.
([email protected]). 13 xii 96 Does evolutionary psychology therefore exclude poly-
andry? Absolutely not. Moreover, the prediction used to
Levine and Silk’s essay is of the type needed to end the dismiss evolutionary psychology (namely, that fraternal
antiquated ‘‘nature versus nurture’’ dichotomy. How- polyandry ought to be more harmonious than nonfrater-
ever, it lacks a well-integrated explanatory model. Al- nal) is in fact generated by it. Humans universally dis-
though they state that sociocultural and evolutionary tinguish kin from nonkin and generally favor the former
theories are concerned with different levels of explana- over the latter. Evolutionary psychology explains this
tion, the authors never articulate the relationship be- behavior by reference to its original adaptive conse-
tween these analytic levels. Are cultural understand- quences. Today the module responsible continues to
ings about polyandry the product of biologically driven operate independent of fitness consequences—people

care about and for kin even when the latter are dying. both are likely to find women younger than their wives
Hence, should circumstances be such that men recog- more attractive. Younger husbands, being more dis-
nized the utility of struggling against jealousy and en- gruntled for reasons described above, are more likely to
tering into polyandrous marriages, the existence of this leave and remarry, and when they do their choices are
module would make cooperation most likely when co- influenced by an attraction to youth.
husbands were close kin. Although jealousy would not
go away, its centrifugal consequences would be miti-
gated by both the practical advantages of polyandry and b a r r y h e wl e t t
warm feelings towards kin. Department of Anthropology, Washington State
Jealousy is not all-or-nothing: A co-husband who is University, Pullman, Wash. 99164, U.S.A. 15 xii 96
able to exercise more control over his wife’s sociosexual
behavior is likely to be less unhappy with the situation This is a substantial contribution to our understanding
than a co-husband who is able to exercise less control. of polyandry because it utilizes precise field data to
Because an age-based hierarchy among Nyinba fraternal evaluate competing hypotheses regarding the nature of
co-husbands may affect access to the common wife, the polyandrous marital relations. The most interesting
youngest co-husband is likely to be the most disturbed contributions are that (1) land availability does not
by jealousy and therefore the most likely to partition. seem to be an important factor in brothers’ remaining
Age-related prestige discrepancies among Nyinba fra- married to the same wife and (2) the degree of relat-
ternal co-husbands, though subtle, are emotionally sa- edness of co-husbands does not influence marital stabil-
lient (Levine 1988:166). Evolutionary psychology as- ity. Thus, important aspects of previous ecological and
serts that men possess a module which makes prestige evolutionary hypotheses about polyandrous marriages
rewarding and its absence punishing (Daly and Wilson are rejected. The article is also important because the
1988). Independent of sexual issues, eldest brothers are authors seriously consider sociocultural hypotheses,
therefore likely to find polyandry more rewarding. Thus distinguish proximate and ultimate explanations, and
they experience a positive emotion (pride) from the incorporate Nyinba explanations for polyandrous rela-
prestige module which helps to counteract the negative tions into their evolutionary analysis. They demon-
emotion (jealousy) from the proprietary module. Young- strate sensitivity to the role of culture in shaping hu-
est brothers experience negative emotions from both man behavior and the complementarity of sociocultural
modules (shame 1 jealousy), making partition attrac- and evolutionary approaches.
tive. While the article provides many insights into polyan-
Nyinba households share pooled production (Levine drous marital relations, the methods generate some
1988:111, 125). Discrepancies in power interfere with questions about the results. What do men do when they
sharing. Egalitarianism is therefore explicitly prescribed are away for long periods of time for trade and herding?
(Levine 1988:115–16). Because of the antithetical rela- Presumably these are the younger brothers, and pre-
tionship between sharing and prestige seeking, pride is sumably they are having sexual relations while they are
considered morally reprehensible (Levine 1981:109). away (the BBC film Dragon Bride gives this impres-
Likewise, because well-being is contingent on the suc- sion). Do Levine’s birth histories of men include births
cess of fraternal polyandrous households, jealousy is outside of legal marriage? Are births outside of marriage
considered a principal moral failing, while brotherly common, and are they likely to be reported? Also, what
solidarity is a principal virtue (Levine 1988:109, 125). explanations do younger brothers give for partition (e.g.,
Via emotions, Nyinba culture thus counteracts those do they have problems with access to the wife, or are
behavioral tendencies which interfere with the Nyinba they simply attracted to a younger woman in the vil-
mode of adaptation and encourages those tendencies lage?), and do per stirpes inheritance negotiations
which reinforce it. among brothers influence marital relations? While land
Men are prone to feel proud when they succeed, even availability does not seem to be a prime mover in parti-
if they are impotent. Elderly couples still struggle with tion, there is some indication that ecology at least in
jealousy. And people love relatives who are terminally part influences polyandry. For instance, the cultural
ill. Human beings are motivated by emotions produced preference for three husbands in order to make a reason-
by mental modules once shaped by selection. Today, ev- able living suggests that this is not a bountiful natural
ery culture builds on and grapples with these elemen- environment.
tary components. The article does a wonderful job of illustrating intra-
A point of clarification: Most Nyinba men who parti- cultural diversity of polyandrous marital relations but
tion are younger than their wives and leave to marry seldom addresses the emergence or ecological adaptabil-
women younger than themselves. Levine and Silk ex- ity of this type of marriage. Nyinba men’s behaviors are
plain this with reference to a universal preference for generally consistent with evolutionary theory (i.e.,
younger wives. However, that preference results from a younger brothers in large sibling groups seek other mar-
module which gauges female sexual attractiveness on riage partners because of decreased reproductive success
the basis of cues indicative of age, hormonal status, par- in polyandrous marriage), but this all takes place in a
ity, fecundity, and health (Symons 1995). Because older cultural context in which polyandry is accepted if not
and younger co-husbands possess the same module, preferred. The authors’ results, concluding remarks, dis-

cussion of Berreman’s culture-history hypothesis, and stock from their parents (Mace 1996). Low (1991) found
statement that ‘‘there are no commonalties among that younger brothers had lower reproductive success
polyandrous societies which distinguish them unam- than elder brothers amongst monogamous 19th-century
biguously from nonpolyandrous societies’’ suggest that Swedish farmers because of the inability of younger
evolutionary biology is useful for understanding intra- sons to acquire farmland. In all these cases, the birth-
cultural variability in polyandrous cultures but that order effect is shown for males rather than females.
cultural evolutionary processes (Cavalli-Sforza and What all these groups have in common with the polyan-
Feldman 1981, Boyd and Richerson 1985) may be more drous Nyinba farmers is a culture based on the inheri-
useful for explaining the cross-cultural distribution of tance of resources over which brothers are in competi-
polyandry. This is consistent with recent cross-cultural tion. A less direct parallel might be found in the levirate
studies which indicate that the cultural diversity and system of widow inheritance by younger brothers, com-
geographic distribution of kinship and family features mon in Africa, wherein younger brothers are sometimes
of cultures may be better explained by cultural evolu- pressured into marrying older women somewhat
tionary than by ecological forces (Guglielmino et al. against their will. Sulloway (1996) argues that birth or-
1995). der influences fundamental aspects of our personality,
perhaps predisposing later-born sons (he does not pre-
sent much information on women) to be more rebel-
r ut h m a c e lious and embark on riskier and more unconventional
Department of Anthropology, University College methods of making a living. This study mainly con-
London, London WC1E 6BT, U.K. 10 i 97 cerns men of European origin. What may possibly be
one of the more intriguing consequences of low repro-
I am not convinced that polyandrous marriage is as un- ductive success for younger brothers is a phenomenon
usual as it at first appears, at least in some respects. Lev- published in recent work from Canada (Blanchard and
ine and Silk present, amongst other data, evidence that Zucker 1994, Blanchard and Bogaert 1995) indicating
boys with many elder brothers tend to have lower repro- that birth order was ‘‘the single most reliable demo-
ductive success amongst the polyandrous Nyinba. This graphic difference between homosexual and heterosex-
is partly due to their wives’ being older. Similar birth- ual men.’’ The significant effect relates strictly to num-
order effects are certainly not unique to polyandrous ber of elder brothers, not number of elder sisters, sibling
marriage. My own data from the Gabbra—a mildly po- sex ratio, or parental age.
lygynous pastoralist group from the north of Kenya— If inherited resources matter, the economic and repro-
show a very similar effect (fig. 1). In this case, the de- ductive prospects of children may be largely in the
cline in male reproductive success with birth order is hands of their parents. Sibling competition is frequently
due both to later marriage and to smaller gifts of live- resolved by parents on the basis of birth order, and
younger brothers may have few options in the face of
this discrimination. Could it be that that polyandrous
marriage is just one solution to a problem faced by par-
ents throughout the world as populations approach car-
rying capacity, of which primogeniture, the levirate,
and even homosexuality are other examples? These are
all fitness-maximizing strategies which parents use to
maximize their grandparental reproductive success
when heritable resources are scarce.
a. r. n. s ri v a s t a v a
Department of Social Anthropology, University of
Allahabad, Allahabad, India.
This paper demonstrates very clearly the ways in which

demographic variables in conjunction with economic
factors shape the patterning of the polyandrous behav-
iour of the Nyinba. Over the few past decades (more pre-
F i g. 1. Fitness as a function of birth-order in the cisely, during the 1970s), social-cultural anthropolo-
Gabbra (a polygynous, patrilineal African society). gists have demonstrated how demographic factors such
Residual fertility is number of live births relative to as distribution of siblings, their relative ages, their ages
same-age same-sex individuals. Number of elder relative to their fathers’/marriage ages, age at the birth
brothers is a significant negative correlate of fitness of first and last sons, spacing of births, distribution of
for males (white bars), but birth order does not affect household estate (land and forest), etc., determine or
fitness for females (stippled bars). Data from Mace regulate the forms of domestic organization over time.
(1996). This has been aptly called the ‘‘developmental cycle in

domestic groups’’ (Goody 1958, McNetting, Wills, and they play in this scenario. Mating systems are always
Arnold 1984). In my view, Levine and Silk’s argument only temporary (and under the circumstances very frag-
is in no way very different from earlier processual analy- ile) compromises in the ‘‘battle of the sexes.’’ And yet
ses of social structure. In other words, how polyandrous polyandrous systems correspond more to genuine fe-
practice comes to be perpetuated or abandoned from male than to male interests. After all, polyandrously
one generation to the next can be grasped quite well by married women achieve greater reproductive success
studying Nyinba households from a developmental per- than women in monogamous marriages. Their advan-
spective. Elsewhere I have demonstrated the methodol- tage continues even into the grandchild generation
ogy for determining the phases/paths in domestic (Crook and Crook 1988). The more men that contribute
groups of the Korwa, a tribal community of Central In- to the family economy, the better for the lifetime repro-
dia (Srivastava 1979, 1994). Extending this analysis to ductive success of the women, which is why the latter
the Nyinba it can be established and even predicted should make every effort to avoid a partitioning of the
how the phases of households are associated with the household. Surely specific behavioral strategies on the
stages of polyandrous marriages—beginning, maturity, part of the women are necessary to convert this ‘‘struc-
and decline. It is possible that a terminated household tural bonus’’ of polyandry into a personal advantage as
may develop into a polyandrous household. great as possible.
My intention here is only to point out that predicting What could these strategies look like? For example,
social-cultural phenomena such as kinship behavior, it could be surmised that explanations concerning pa-
marriage and inheritance patterns, etc., from demo- ternity are not always made according to the mother’s
graphic variables began to appear in the anthropological best knowledge and belief but, under certain circum-
literature some two decades ago. With respect to poly- stances, also according to ‘‘strategic’’ viewpoints, in
androus marriages, however, Levine and Silk’s intellec- order to bind a potentially ‘‘active partitioner’’ or a
tual exercise is a contribution. particularly productive worker permanently to the
household. Are there any ethnographic indications of
this?
eckart voland Women may also—insofar as their sexual autonomy
Zentrum für Philosophie und Grundlagen der permits this—turn to the younger co-husbands who
Wissenschaft Universität Giessen, D-35394 Giessen, have been disadvantaged by the senior husband for sex
Germany. 11 xii 96 in a manner comparable with the situation of polyan-
drous dunnock females. If they have to fear that the beta
Males are reproductive opportunists. What Darwinian male may renounce polyandry because of the monopoli-
anthropologists and Darwinian psychologists have pre- zation attempts of the alpha male, then they will in-
dicted from their knowledge of the principles of biologi- creasingly encourage the beta male to engage in copula-
cal evolution has been empirically confirmed by Levine tion (Davies 1990). In any case, women will oppose the
and Silk. Even under the rather unusual conditions of latent instability of polyandrous marriages with efforts
cooperative polyandry, males are competing for pater- to achieve marital stability. Therefore, I should like to
nity. They are successful if they constantly keep an eye ask the authors to supply us with the second part of the
on their reproductive opportunities, weigh the costs and story: ‘‘Sources of Stability in Polyandrous Marriages:
benefits of their options, and, if need be, are willing to Female Tactics.’’
be flexible and renounce a previous way of life in favor
of a more advantageous alternative one. Levine and Silk
have shown us which factors are included in the bal-
ance. Their study is behavioral ecology par excellence. Reply
Only one result does not fit into the picture. Why
does the degree of relatedness among co-husbands not
seem to influence marital stability? Could it be that n a n c y e. l e v in e a n d j o a n b. s i l k
Levine and Silk did not choose the best of all possibili- Los Angeles, Calif., U.S.A. 7 i 97
ties for testing their thesis concerning the impact of ge-
netic relationships on the stability of marriages? Would We are gratified to find that our efforts to bridge the the-
it not make more sense and be more sensitive to com- oretical perspectives of sociocultural anthropology and
pare the average genetic relationship of the ‘‘active par- evolutionary biology and apply them to the study of hu-
titioners’’ with that of the ‘‘passive partitioners’’? The man kinship behavior have met with a receptive audi-
hypothesis would thus be: If a household divides, there ence. The authors of these comments have raised a
is a greater probability—other things being equal—that number of important issues. In our response, we will
the genetically most remote co-husband will be the one first consider general questions about the nature of our
to dissolve the cooperation. Perhaps the material would analysis; then we will address issues directly related to
allow a somewhat more differentiated reanalysis along our results and expand on a number of points meriting
these lines. further clarification. Finally, we will discuss the kinds
However, females are also reproductive opportunists. of additional data needed to answer a number of funda-
One would have been only too glad to know what role mental questions raised in these comments.

Conjoining sociocultural and evolutionary theories. economic and demographic circumstances limit the
Both Fessler and Hewlett call for an integrative theoret- ability of each to support a monogamous family’’ and
ical framework to link the perspectives of evolutionary its logical corollary that households poorer in land are
biology and sociocultural anthropology. Fessler advo- more likely to be polyandrous than their better-off
cates an approach based on evolutionary psychology, neighbors. However, we are reluctant to embrace their
while Hewlett suggests that it may be useful to focus conclusion that these data constitute a convincing ‘‘ref-
on cultural evolutionary processes. We agree that a syn- utation of this widely cited model for polyandry.’’
thetic explanatory framework would be desirable, and While men’s decisions to leave their polyandrous mar-
our results suggest that such a synthesis would be fruit- riages do not seem to be consistently influenced by the
ful. However, we do not expect this to be a simple en- size of their household’s landholdings, it is plausible
deavor; the construction of a general theory that con- that constrained ecological or economic circumstances,
nects evolutionary processes to human psychology, particularly land shortages, may have shaped the ways
social structure, and culture and that encompasses his- in which polyandry developed as a social institution in
torically particular and culturally idiosyncratic systems the Himalayas. Once established, other cultural beliefs
of meaning and the ultimate selective pressures on hu- and values may bias individual decisions about the de-
man behavior will be an enormous intellectual achieve- sirability of polyandrous unions.
ment. Mace suggests that the negative relationship between
Fessler speculates that mental modules which medi- birth order and reproductive opportunities which we
ate nepotism, sexual jealousy, and prestige jointly in- observed may extend beyond the boundaries of Nyinba
fluence the emotional tenor of marital relationships and society. She suggests that polyandry, primogeniture, the
influence men’s satisfaction with their marriages. He levirate, and homosexuality may be alternative mecha-
also asserts that we ‘‘imply that evolutionary psychol- nisms for maximizing grandparental reproductive suc-
ogy is of limited applicability in the current discussion cess in conditions of economic scarcity. If so, then there
. . . [because] it does not predict that fraternal polyandry should be a consistent relationship between such cul-
will be less fractious than polyandry among unrelated tural practices and ecological circumstances or eco-
men.’’ We did not intend to question the general utility nomic opportunities. If the hypothesis is confirmed,
of evolutionary psychology; rather, we wanted to point then we will face the intriguing question of why one al-
out that humans may not be psychologically prepared ternative is favored in some societies but not in others.
to cope with all current contingencies. On the one And if these are functionally equivalent alternatives,
hand, as Fessler notes, men’s responses to polyandry then we must ask why certain alternatives, such as pri-
might be influenced by several mental modules—the mogeniture, are common while others, such as polyan-
modules for sexual jealousy, altruistic feelings toward dry, are quite rare.
kin, and prestige. Thus the negative emotions associ- Dickemann points out that there has been no ade-
ated with sexual jealousy might be counterbalanced by quate evaluation of the Darwinian proposition that
the positive value placed on cooperating with kin. On polyandry, like other human behaviors, is adaptive in
the other hand, if mating practices akin to polyandry terms of reproductive success, since this would require
were not present in the environment of evolutionary assessing men’s and women’s genetic contributions
adaptedness, men might find coping with these compet- over several generations. Following a similar line of ar-
ing inclinations to be stressful. One way of answering gument, Voland points out that polyandry may be more
this question would be to see if men find fraternal poly- advantageous to women than to men; in Ladakh, for ex-
andry more congenial than nonfraternal polyandry. ample, polyandrously married women have more grand-
Such a resolution seems unlikely, however, because of children than monogamously married women have.
the rarity of polyandrous practice cross-culturally. These issues are important but lie beyond the scope of
Hewlett notes that we illustrate intracultural vari- the present paper. We have not attempted to evaluate
ability but do not address the broader question of inter- whether polyandry is an adaptive strategy for men or
cultural variability, that is, why polyandry arises in women; instead, we have focused on the proximate fac-
some societies and not others. He suggests that evolu- tors causing men to sever their polyandrous marriages.
tionary biology may be useful for understanding varia- In order to evaluate whether polyandry is adaptive, one
tion within a society, while cultural evolutionary pro- must compare the fitness of individuals who marry
cesses may hold the key to understanding variation polyandrously, monogamously, and polygynously for
between societies. We are not convinced, however, that fitness. Such analysis is complicated by the necessity of
there is such a clear distinction between forces shaping tracking several components of fitness. Men who marry
intercultural and intracultural variation. Boyd and polyandrously, for example, achieve individual fitness
Richerson (1985), for example, have explored the dy- benefits via their own offspring and inclusive fitness
namics of a set of cultural evolutionary forces that in- benefits via the offspring of related co-husbands. Since
fluence individual behavior within cultures, sometimes the Nyinba data set can be used to compare the fitness
in nonadaptive ways. of men and women who marry polyandrously, polygy-
The implications of the data. As Borgerhoff Mulder nously, and monogamously, we plan to conduct this
and Haddix point out, our results do not support the hy- comparison.
pothesis that ‘‘brothers will share paternity whenever Points of clarification. Speaking in cultural terms,

Fessler is on the right track when he posits that younger tiates partition. We have gone back to the data to exam-
brothers dislike their junior and therefore marginal sta- ine this question. In 9 of the 21 partitioning households
tus in the marriage and household. One major source of for which we have appropriate data, all the co-husbands
trouble in polyandrous marriage is the affront to were equally closely related. In the remaining 12 house-
younger brothers’ prestige. Junior co-husbands fre- holds the degree of relatedness among co-husbands var-
quently complain that they are dissatisfied with their ied. In 7 of these households, one co-husband initiated
marriages because they are treated poorly and do not re- the partition. The active partitioner was the least
ceive a fair share of sexual attention from the common closely related co-husband in only 3 of these 7 house-
wife. However, it cannot be said that they typically suf- holds. When partitions were jointly initiated by more
fer from jealousy. Nyinba, like other ethnic Tibetans, than one co-husband, the picture becomes more com-
have a term (mig ser) for this emotion and describe it in plicated, but the results are more or less the same. In 3
a manner similar to Euro-American commentaries on cases the active partitioners were more closely related
the subject. Although Levine did collect rare reports of to each other than to the other co-husbands, while in 2
men’s being jealous of their polyandrous co-husbands, cases the active partitioners were less closely related to
this was not the predominant theme in men’s com- one another than to the other co-husbands. Thus the de-
plaints about their marriages. gree of relatedness among co-husbands does not seem
Fessler’s note about why men may find younger to affect the role men play in partitions.
women more attractive notwithstanding, we suggest In response to Hewlett’s question, we note that Ny-
that active partitioners may be marrying younger inba traders report visiting prostitutes in Indian border
women not only because this is what they prefer to do towns, but no one has ever reported having a child out
but also because there are more young women eligible of wedlock with a non-Nyinba woman. By contrast,
for marriage (see n. 21). And, as Hewlett’s comment re- Levine collected a number of reports of illegitimacy
minds us, when men partition and marry younger within the Nyinba community. Most such children die
women they do not report their attraction to the young, but some do survive. Although we can never be
younger woman (or dissension with their brothers) as certain who fathered a given child, Nyinba keep careful
the cause for the breakup of their former marriage but track of allegations of illegitimate paternity in order to
rather mention dissatisfaction with their prior marriage avoid incestuous marriages (see Levine 1988:167n).
and relations with their former wife. Three of the men whose birth histories we surveyed
Land-short Himalayan households do have the option were reported to have had an additional child out of
of securing additional income through animal hus- wedlock. One was the eldest brother in a marriage of
bandry and long-distance trade, as Borgerhoff Mulder two brothers, one was the second brother in a marriage
and Haddix point out. We did not test the impact of of three brothers, and one was the youngest brother in
these alternative economic opportunities on partition a marriage of three brothers. In our calculations of relat-
for a number of reasons. Each Nyinba household has an edness, we treated reported illegitimate sons as fathered
equal chance of expanding its animal holdings or engag- by men unrelated to their co-husbands’ fathers.
ing in trade—this depends only on the number of Srivastava correctly points out that polyandry must
healthy brothers. Both endeavors require minimal ini- be understood in terms of domestic developmental cy-
tial capital investments. For example, a man could start cles. This is because the brothers in any given genera-
with a few pack animals and after a few years of careful tion are likely to partition during what has been termed
husbandry and, possibly, working as a herder for others the family-building stage of domestic development.
increase his herd to more than adequate size. This is However, regularized cyclical processes are quite a dif-
why Levine (1988:215) describes trading as ‘‘the flexible ferent matter from social change, as Fortes (1949:54–55)
option.’’ In addition, we note that even in households pointed out in his pioneering article on developmental
with the largest farms, only one man is called upon to processes in domestic groups. It was our contention
stay home to supervise agriculture. that understanding the sources of stress in polyandrous
We would like to correct one mistaken impression marriages would support more accurate predictions
voiced in Chandra’s comments. Our focus in this paper about which marriages would undergo fission and
was not the ‘‘failure’’ of polyandry as a marriage system which would persist until the partners’ deaths. Ideally,
but rather why particular marriages succeed or fail. understanding points of stress would also contribute to
However, we thank Chandra for drawing attention to fuller explanations of why polyandry has declined in
a commonly noted phenomenon: transitions to a cash frequency in some locales while persisting in others.
economy and increasing integration into nationwide so- The prevailing view in the literature has been that poly-
ciocultural systems are often accompanied by a decline andry everywhere entails major contradictions; while
in polyandry’s incidence. the Nyinba data support some of the commonly held
As Voland notes, our results deviate from evolution- assumptions, our findings have called others into ques-
ary biological predictions in only one way: the degree of tion. Evolutionary biology suggests that men are more
relatedness among co-husbands does not affect marital likely to be dissatisfied with marriages in which their
stability. He suggests that this may be because we did reproductive opportunities are limited, while sociocul-
not choose the most sensitive test of this hypothesis, tural anthropology and evolutionary psychology suggest
that is, whether the least closely related co-husband ini- that men are more likely to be dissatisfied with mar-

riages in which their prestige is low and they have less p a u l t u r k e. Editors. 1988. Human reproductive behavior:
control over their wives’ sexuality. Nonetheless, de- A Darwinian perspective. Cambridge: Cambridge University
Press.
spite the disadvantages for some participants, polyandry b l a n c h a r d, r. , a n d k. z u c k e r. 1994. Reanalysis of Bell,
developed and has persisted over generations in a num- Weinberg, and Hammersmith’s data on birth-order, sibling sex-
ber of locales where it became culturally valued and in- ratio, and parental age in homosexual men. American Journal
tegrated with other facets of the social system—the of Psychiatry 151:1375–76. [rm]
household system and the system of domestic labor or- b l a n c h a r d, r. , a n d a. b o g a e r t. 1995. Homosexuality in
men and number of elder brothers. American Journal of Psychi-
ganization in particular. The fact of the matter is that atry 153:27–31. [rm]
marriages in all societies create stress for participants, b o s e r u p, e s t e r. 1970. Women’s role in economic develop-
and some marriages in every society ‘‘fail,’’ that is, lead ment. New York: St. Martin’s Press.
to disaffection, separation, or divorce. For many individ- b o y d, r. , a n d p. j. r i c h e r s o n. 1985. Culture and the evo-
lutionary process. Chicago: University of Chicago Press. [bh]
uals, however, marriage ‘‘succeeds,’’ in the sense of per- b u s s, d a v i d. 1989. Sex differences in human mate prefer-
sisting throughout the participants’ lifetimes and pro- ences: Evolutionary hypotheses tested in 37 cultures. Behav-
viding them a host of perceived advantages. The same ioral and Brain Sciences 12:1–49.
can be said for Nyinba polyandry. As Levine has argued, c a r r a s c o, p e d r o. 1959. Land and polity in Tibet. Seattle:
‘‘most men and women find polyandry a personally University of Washington Press.
c a s s i d y, m a r g a r e t l. , a n d g a r y r. l e e. 1989. A study
comfortable form of marriage and one that suits cultur- of polyandry: A critique and synthesis. Journal of Comparative
ally defined practical goals’’ (1988:159). Some individu- Family Studies 20:1–11.
als do not find their polyandrous marriages satisfactory, c a v a l l i - s f o r z a, l. l. , a n d m. f e l d m a n. 1981. Culture
however, and the aim of this paper was to delineate the transmission and evolution: A quantitative approach.
Princeton: Princeton University Press. [bh]
factors which weaken such marriages and to test empir- c h a n d r a, r a m e s h. 1987. ‘‘Polyandry in the north-western Hi-
ical evidence of marital breakdowns against prevailing malayas: Some changing trends,’’ in Polyandry in India. Edited
theoretical models for understanding human kinship by Manis Kumar Raha and Palash Chandra Coomar, pp. 130–
behavior. 54. Delhi: Gian.
Additional data needed. Several commentaries em- ———. 1992. Highlanders of Northwestern Himalayas. New
Delhi: Inter-India Publications. [rc]
phasize the need for larger samples and more extensive c o s m i d e s, l e d a, a n d j o h n t o o b y. 1992. ‘‘The psycholog-
analysis. Borgerhoff Mulder and Haddix suggest that a ical foundations of culture,’’ in The adapted mind. Edited by
multivariate analysis is needed to tease apart the effects Jerome Barkow, Leda Cosmides, and John Tooby, pp. 19–136.
of various factors on decisions to partition. Chandra ar- Oxford: Oxford University Press.
c r o o k, j o h n h. 1995. ‘‘Psychological processes in cultural
gues that robust data from a number of societies are and genetic co-evolution,’’ in Survival and religion: Biological
necessary to show whether the pattern of younger evolution and cultural change. Edited by E. Jones and V. Reyn-
brothers’ siring fewer children is widespread. Crook ar- olds, pp. 45–110. London: Wiley. [jhc]
gues that larger, comparative data sets are needed to ex- ———. 1996. ‘‘Ecology and culture in the adaptive radiation of
amine the factors that influence marital systems in the Tibetan-speaking peoples in the Himalayas,’’ in Recent re-
search in Ladakh, 4 & 5. Edited by H. Osmaston and P. Den-
Himalayan region. Dickemann calls for assessment of wood. Delhi: Motilal Banarsidass. In press. [jhc]
reproductive success over several generations. The c r o o k, j o h n h. , a n d s t a m a t i j. c r o o k. 1988. ‘‘Tibetan
scope of our analysis and the power of our conclusions polyandry: Problems of adaptation and fitness,’’ in Human re-
were clearly limited by the fact that partition is a rela- productive behavior: A Darwinian perspective. Edited by L.
Betzig, M. Borgerhoff Mulder, and P. Turke, pp. 97–114. Cam-
tively uncommon event among the Nyinba and our bridge: Cambridge University Press.
sample of such events was relatively small. Thus, we ———. 1994. ‘‘Explaining Tibetan polyandry: Socio-cultural, de-
agree that there is a pressing need for more extensive mographic, and biological perspectives,’’ in Himalyan Bud-
information about the factors that influence the dynam- dhist villages. Edited by John H. Crook and Henry A. Osmas-
ics of polyandry in the Himalayas, and we hope that this ton, pp. 735–86. New Delhi: Shri Jainendra Press.
d a l y, m a r t i n, a n d m a r g o w i l s o n. 1988. Homicide.
article will help to stimulate such work. New York: Aldine de Gruyter. [dmtf]
d a v i e s, n. b. 1990. ‘‘Dunnocks: Cooperation and conflict
among males and females in a variable mating system,’’ in Co-
operative breeding in birds: Long-term studies of ecology and
References Cited behavior. Edited by P. B. Stacey and W. D. Koenig, pp. 457–85.
Cambridge: Cambridge University Press. [ev]
d i c k e m a n n, m i l d r e d. 1979a. ‘‘Female infanticide, reproduc-
a l e x a n d e r, r i c h a r d d. 1974. The evolution of social behav- tive strategies, and social stratification,’’ in Evolutionary biol-
ior. Annual Review of Ecology and Systematics 5:325–83. ogy and human social behavior: An anthropological perspec-
a z i z, b a r b a r a n. 1978. Tibetan frontier families. New tive. Edited by Napoleon Chagnon and William Irons, pp. 321–
Delhi: Vikas. 67. North Scituate, Mass.: Duxbury Press. [md]
b e a l l, c y n t h i a, a n d m e l v y n c. g o l d s t e i n. 1981. Ti- ———. 1979b. The ecology of mating systems in hypergynous
betan fraternal polyandry: A test of sociobiological theory. dowry societies. Social Science Information 18:163–95. [md]
American Anthropologist 83:5–12. ———. 1981. ‘‘Paternal confidence and dowry competition: A bio-
b e r r e m a n, g e r a l d d. 1962. Pahari polyandry: A compari- cultural analysis,’’ in Natural selection and social behavior:
son. American Anthropologist 64:60–75. Recent research and new theory. Edited by Richard Alexander
———. 1978. Ecology, demography, and domestic strategies in and Donald Tinkle, pp. 417–38. New York: Chiron. [md]
the western Himalayas. Journal of Anthropological Research d u r h a m, w i l l i a m. 1991. Coevolution. Genes, culture, and
34:326–68. human diversity. Stanford: Stanford University Press.
b e t z i g, l a u r a, m o n i q u e b o r g e r h o f f m u l d e r, a n d e k v a l l, r o b e r t b. 1968. Fields on the hoof: Nexus of Ti-

betan nomadic pastoralism. New York: Holt, Rinehart and m a c e, r. 1996. Biased parental investment and reproductive
Winston. success in Gabbra pastoralists. Behavioral Ecology and Sociobi-
f e r n a n d e z, r e n a t e. 1981. Comments on Tibetan polyandry: ology 38:75–81. [rm]
A test of sociobiological theory. American Anthropologist 83: m c l e n n a n, j o h n f. 1876 (1865). Studies in ancient history.
896–97. [md] (First series, comprising a reprint of Primitive marriage.) Lon-
f o r t e s, m e y e r. 1949. ‘‘Time and social structure: An Ashanti don: Bernard Quaritch.
case study,’’ in Social structure: Studies presented to Radcliffe- m c n e t t i n g, r. , r. r. w i l l s, a n d e. j. a r n o l d. 1984.
Brown. Edited by Meyer Fortes. Oxford: Clarendon Press. Household: Comparative and historical studies of domestic
g o l d s t e i n, m e l v y n c. 1971. Stratification, polyandry, and groups. Berkeley: University of California Press. [arns]
family structure in Tibet. Southwestern Journal of Anthropol- m a j u m d a r, d h i r e n d r a n a t h. 1962. Himalayan polyan-
ogy 27:64–74. dry. London: Asia Publishing House.
———. 1976. Fraternal polyandry and fertility in a high Himala- o t t e r b e i n, k e i t h f. 1968 (1963). ‘‘Marquesan polyandry,’’ in
yan valley. Human Ecology 4:223–33. Marriage, family, and residence. Edited by Paul Bohannan and
———. 1978. Pahari and Tibetan polyandry revisited. Ethnology John Middleton, pp. 287–96. New York: Natural History Press.
17:325–37. p a r m a r, y a s h w a n t s i n g h. 1975. Polyandry in the Himala-
g o o d y, j a c k. 1958. The developmental cycle in domestic yas. Delhi: Vikas.
groups. Cambridge: Cambridge University Press. [arns] p r a s a n n a, k. s. , e t a l. 1996. Polyandry in a Central Hima-
———. 1976. Production and reproduction. Cambridge: Cam- layan community: An eco-cultural analysis. Man in India 76:
bridge University Press. 51–65. [rc]
———. 1990. The Oriental, the ancient, and the primitive: Sys- r i v e r s, w. h. r. The Todas. London: Macmillan.
tems of marriage and the family in the pre-industrial societies s a k s e n a, r. n. 1962. Social economy of a polyandrous people.
of Eurasia. Cambridge: Cambridge University Press. London: Asia Publishing House.
g o u g h, e. k a t h l e e n. 1952. Changing kinship usages in the s c h n e i d e r, d a v i d m. 1984. A critique of the study of kin-
setting of political and economic change among the Nayars of ship. Ann Arbor: University of Michigan Press.
Malabar. Journal of the Royal Anthropological Institute 82:71– s r i v a s t a v a, a. r. n. 1979. The development dimension of
87. domestic group in India: An anthropological analysis of de-
———. 1959. The Nayars and the definition of marriage. Journal mography and economy of a tribe of Central India. New
of the Royal Anthropological Institute 89:23–34. Delhi: Books Today. [arns]
———. 1961. ‘‘Nayar: Central Kerala,’’ in Matrilineal kinship. Ed- ———. 1994. On social structure and demography. current an-
ited by David M. Schneider and E. Kathleen Gough, pp. 298– thropology 35:456–57. [arns]
384. Berkeley: University of California Press. s u l l o w a y, f. 1996. Born to rebel. New York: Pantheon. [rm]
g u g l i e m i n o, c. r. , c. v i g a n o t t i, b. h e w l e t t, a n d s y m o n s, d o n a l d. 1992. ‘‘On the use and misuse of Darwin-
l. l. c a v a l l i - s f o r z a. 1995. Cultural variation in Africa: ism in the study of human behavior,’’ in The adapted mind.
Role of mechanisms of transmission and adaptation. Proceed- Edited by Jerome Barkow, Leda Cosmides, and John Tooby, pp
ings of the National Academy of Sciences (U.S.A.) 92:7585– 137–59. Oxford: Oxford University Press.
89. [bh] ———. 1995. ‘‘Beauty is in the adaptations of the beholder: The
h a m i l t o n, w i l l i a m d. 1964. The genetical theory of social evolutionary psychology of human female sexual attrac-
behavior. Pts. 1 and 2. Journal of Theoretical Biology 7:1–52. tiveness,’’ in Sexual nature, sexual culture. Edited by P. R. Ab-
h e w l e t t, b a r r y s. Editor. 1992. Father-child relations: Cul- rams and S. D. Pinkerton, pp. 80–118. Chicago: University of
tural and biosocial contexts. New York: Aldine de Gruyter. Chicago Press. [dmtf]
h i a t t, l. r. 1980. Polyandry in Sri Lanka: A test case for pa- t a m b i a h, s t a n l e y j. 1966. ‘‘Polyandry in Ceylon—with spe-
rental investment theory. Man 15:583–602. cial reference to the Lagala region,’’ in Caste and kin in Nepal,
l e a c h, e d m u n d. 1966 (1955). ‘‘Polyandry, inheritance, and India, and Ceylon. Edited by Christoph von Fürer-Haimendorf,
the definition of marriage—with particular reference to Sin- pp. 264–358. London: Asia Publishing House.
halese customary law,’’ in Rethinking anthropology. London: v a n d e n b e r g h e, p i e r r e l. , a n d d a v i d p. b a r a s h.
Athlone Press. 1977. Inclusive fitness and human family structure. American
l e v i n e, n a n c y e. 1981. ‘‘Perspectives on love: Morality and Anthropologist 79:809–23.
affect in Nyinba interpersonal relationships,’’ in Culture and w e i n e r, a n n e t t e. 1992. Inalienable possessions. Berkeley:
morality: Essays in honour of Christoph von Fürer-Haimen- University of California Press.
dorf. Edited by A. C. Mayer, pp. 106–25. Delhi. Oxford Univer- w e s t e r m a r c k, e d w a r d. 1926. A short history of marriage.
sity Press. [dmtf] London: Macmillan.
———. 1988. The dynamics of polyandry: Kinship, domesticity, w i l s o n, m a r g o, a n d m a r t i n d a l y. 1992. ‘‘The man
and population on the Tibetan border. Chicago: University of who mistook his wife for a chattel,’’ in The adapted mind: Evo-
Chicago Press. lutionary psychology and the generation of culture. Edited by
l o w, b. 1991. Reproductive life in 19th-century Sweden: An J. H. Barkow, L. Cosmides, and J. Tooby, pp. 289–322. Oxford:
evolutionary perspective on demographic phenomena. Ethology Oxford University Press. [dmtf]
and Sociobiology 12:411–48. [rm]


Why Polyandry Fails

Uploaded by

Copyright:

Available Formats

Why Polyandry Fails

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Why Polyandry Fails

Uploaded by

Copyright:

Available Formats

Why Polyandry Fails: Sources of Instability in Polyandrous Marriages

Author(s): Nancy E. Levine and Joan B. Silk

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

Fraternal polyandry is the normative form of marriage

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

Total Landholdingsa Per Capita Landholdingsa

Intact 60 14.61 0.97 5.42 0.41

Causes of Dissolution of Polyandrous to matter once a decision to partition is reached and

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

F ig. 2. Number of co-husbands by type of household.

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

F i g. 3. Means and standard errors of age differences

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

Birth Position Birth Position of Co-husband

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

Average Degree of Relatedness

Intact 29 0.37 0.02

0.50. The average value was 0.36, halfway between the

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

23. In his classic article on polyandry and the definition of mar-

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This paper demonstrates very clearly the ways in which

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

This content downloaded from 206.212.0.156 on Sat, 20 Jul 2013 01:01:41 AM

You might also like