Maney Publishing

Maney Publishing
A QUESTION OF STRENGTH: MILITARY CAPABILITY AND PRINCESS-BESTOWAL IN IMPERIAL

CHINA'S FOREIGN RELATIONS (HAN TO CH'ING)
Author(s): Jennifer Holmgren
Source: Monumenta Serica, Vol. 39 (1990-1991), pp. 31-85
Published by: Maney Publishing
Stable URL: http://www.jstor.org/stable/40726901
Accessed: 27-10-2015 10:01 UTC
REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/40726901?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
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].
Maney Publishing is collaborating with JSTOR to digitize, preserve and extend access to Monumenta Serica.
http://www.jstor.org
This content downloaded from 129.78.139.28 on Tue, 27 Oct 2015 10:01:26 UTC
All use subject to JSTOR Terms and Conditions
Mon. Ser. 39 (1990-91)
A QUESTION OF STRENGTH: MILITARY CAPABILITY

AND PRINCESS-BESTOWAL IN IMPERIAL CHINA'S
FOREIGN RELATIONS (HAN TO CH'ING)
Jennifer Holmgren
AustralianNational University
Contents
I. Introduction 31
II. Scope of Research 35
III. The Chinese Case
Frequency of Bestowal 36
Recipients 37
Principles and Conditions 38
Forms of the Marriage Contract 54
IV. The Non-Han Case
Steppe Traditions 58
The Conquest Dynasties 65
V. Social Background 73
VI. Conclusion 82
I. INTRODUCTION
The giftof a royalprincessto the rulerof a foreigntribeor stateis a
phenomenon observedbymanyscholarsworkingon thehistoryof impe-
rial China. In China itself,cases such as the marriageof WangChao-
chün iEBgg to a Hsiung-nu^j$ chieftainin 33 B.C. and thatof the
T'ang princessWen-ch'eng~%fàto the rulerof Tibet in A.D. 641 have
passed,via themediumof poetryand drama,intofolklore.1 Most schol-
1) See, forexample,Chao Chun: Beautyin Exile' a novelby Shu Chiung(Hong

Kong: Kellyand Walsh,1934). Demiévillehas made a studyof a late T'ang song-story
31
32 JENNIFER HOLMGREN
ars, however,fail to comment on the phenomenon other than in pass-

ing.2
It is generallyacknowledged that at the time such marriages were

seen by the Chinese as a source of sorrow and shame, carried out only
under threatfromoutside powers - a sign of weakness and thus some-
thing to be resistedat all costs.3 The phenomenon is also seen to have
flourished irregularly:as H. Serruys has remarked, there is no Ming
counterpartto Wang Chao-chün or to the many princessessent by Tang
to the Turks.4Cases which readily springto mind do in fact refereither
to the Han or T'ang dynasties,and since T'ang poems and edicts on the
subject invariablycite precedentsset during Han, it would appear that
the phenomenon was largely confined to those periods.
In the sixteenthcentury,the statesman Chang Chü-cheng jjgJgjE

ascribed absence of princess-bestowalon foreignleaders in Ming times
to the strengthof that dynasty:Chang placed foreignmarriage alliances
based on the themeof WangChao-chün'smarriage:P. Demiéville,"Quelques traitsde

mœursbarbaresdans une chantefable chinoisedes Tang," Acta Orient.Hung. 15 (1962):
71-85.
2) Onlyone detailedstudyis knownto me.Thisworkdeals withbestowalsin T'ang
times:KuangP'ing-changU^W, "T'ang tai kung-chu ho-ch'ink'ao" J§ft&Íín£B#,
Shih-hsüehnien-piao £Ç^$g 2,2 (1935): 23-68. T'ang bestowalsupon the Uighur
Khans, particularly the last in A.D. 821, are discussedin C. Mackerras,The Uighur
EmpireAccordingto the T'ang DynasticHistories:A Studyin Sino-UighurRelations,
744-840 (Canberra:AustralianNationalUniversity Press, 1972),pp. 43-47. Loewe,in
his "Introduction" to A. F. P. Hulsewé'sannotatedtranslation of chapters61 and 96 of
the Han-shu(China in CentralAsia: The Early Stage: 125 B.C.-A.D. 23 [Leiden:Brill,
1979],pp. 60-62), makessomegeneralcomments on matrimonial alliancesin Han times.
Briefgeneralcomments mayalso be foundin JagchidSechin,"Patterns of Tradeand Con-
flictbetweenChina and the Nomads of Mongolia,"Zentralasiatische Studien11 (1977):
183-84 and 190-91.
3) See, forexample,thespeechattributed to T'ang T'ai-tsung(r. A.D. 627-49) on
thisissue,reproducedin Kuang,op. cit., p. 62; also commentsby JagchidSechin,op.
cit., 183-84, and Hsiao Ch'i-ch'ingj|l?)S » "Yuan-Likuan-hsichungte wang-shih hun-
yin yü ch'iang-ch'üancheng-shih"KBMffi(i>fö££mHlittflUK¥ , in Yüan-tai
shih hsin fan 7Gftjfc#rfiÊ (Taipei: Hsin-wenfeng,1983), p. 231.
4) HenrySerruys, Sino-MongolRelationsduringtheMing,II: The TributeSystem
and DiplomatieMissions(1400-1600). MélangesChinoiset Bouddhiques14 (1967): 18.
MILITARY CAPABILITY AND PRINCESS-BESTOWAL 33
contractedduring Han and T'ang on a par with gifts(cash and goods)

made by Sung to northernfrontierstates during the tenth,eleventhand
twelfthcenturies.5In our own time,Yü Ying-shihhas similarlycontrast-
ed the ho-ch'in fp$g policy of the Han dynastywith its more successful
counterpartof later times - the tributarysystem.6Yet Han and T'ang
are often described as two very strong,successfullyexpansionist dynas-
ties. The apparent contradictionbetween this perception of Han/T'ang
historyand the idea that foreignmarriage alliances in China were con-
tractedonly under pressureis one of the subjects of the present study.
Official Chinese scholarship of the late imperial era places foreign

marriagecontractsunder the rubricof "requests" and "gifts." For exam-
ple, one of the underlyingassumptions in the Ts'e-fu yüan-kuei fñM
7ÉH , compiled at the beginningof the eleventhcentury,is that daugh-
ters and sistersof Chinese emperorswere eagerly sought after as brides
by foreignrulersbecause Chinese culturalsuperiorityenhanced the status
of the recipientnot only in relation to his own people but also before
other foreignleaders.7 Recognizing that the great majority of successful
"requests" for an imperial bride were actually "demands" backed up by
threatof militaryaction, most independentscholars dismiss this termi-
nology as simplypart of the rhetoricof the tributarysystem.Neverthe-
less, it is undeniable that the peoples around China's borders, whether
weak or strong,evinced a general willingnessto contract marriage alli-
ances with the Chinese state - a willingnesswhich was not reciprocated
by the Chinese.
5) Lien-shengYang,"HistoricalNotes on the ChineseWorldOrder,"in JohnK.

Fairbank,ed., The Chinese WorldOrder(Cambridge,Mass.: HarvardUniversity Press,
1968),p. 30.
6) Ying-shihYü, Tradeand Expansionin Han China: A Studyin the Structure
of Sino-Barbarian EconomicRelations(Berkeley/Los of California
Angeles:University
Press,1967),pp. 42-44.
7) Ts'e-fuyüan-kuei(hereafterTFYK) (Taiwan: Chung-huashu-chü,1967), eh.
978-79, pp. 11486-11507.See, forexample,theterminology used to describeWeiHsiao-
ching'sSÍ#S (r. A.D. 534-550) dealingswiththe Juan-juan:kung Jt (tribute)and
ch'iu >fc (beg) fortheJuan-juan;hsü ff (permit)fortheWeiemperor- therealitywas
verydifferent (see below). TFYK 978, p. 11493.
Mancaligroupsthenumerousand variedstrategies used bytheChi-

nese in theirrelationswithforeignneighboursunderfourmain catego-
ries:military power;dynasticweakness;Confucianism; and Chinesecul-
turalsuperiority. He places marriagealliance withinthe firstcategory;
giftsof cash and goods in thesecond.Accordingly,themarriagecontract
is seen as a vehicleforforeigngroupsto ally themselveswitha stronger
military power.8It is possiblethatMancalPsviewhas been undulyinflu-
encedbythefocusof his study,theCh'ingdynasty. SincetheCh'ingwas
a non-Han regime,it may have had a somewhatdifferent orientation
fromthatof nativeChinesestates.On this subject,Farquharhas sug-
gestedthatthewell-attested exchangeof womenbetweentheManchuand
Mongolnobilityin Ch'ingtimeswas partof an InnerAsian ratherthan
Chinesetraditionof diplomacy.9 His viewis supportedby the workof
Hsiao Ch'i-ch'ingwho arguesthatthe Yuan pursuedmarriagealliances
withforeign statesmuchmoreactivelythandid nativeChineseregimes.10
LikeCh'ing,Yuan was a strongexpansionist power,butwas also of non-
Han origin.In contrastto these views,Wang Chen-mainclaims that
ManchumarriagetieswithMongolleadersweremodelledon traditional
Chinesemethodsof dealingwithtroublesome borderpeople.11His view
seemsirreconcilable withthatof Farquhar,but,as thispaper will show,
botharetenableto somedegreewhentheCh'ingdata is placedin histori-
cal perspective.
The presentstudyarguesthat irrespective of fluctuationsin the

numberof marriagealliancesconsideredor contracted at any one time,
a consistent
approach to thebestowal
of women upon foreign leaderscan
be discernedacrossthehistoryof nativeregimesin China. In discussing
the circumstancesunderwhichforeignmarriagediplomacywas consid-
ered by Chinese states,the study revealsthe mannerin which the
8) M. Mancali,Chinaat theCentre:300 Yearsof ForeignPolicy(London:Collier

Macmillan,1984),pp. 28-33.
9) D. M. Farquhar,"The Originsof the Manchus' MongolianPolicy,"in Fair-
bank, ed., The Chinese WorldOrder,pp. 198-99.
10) Hsiao Ch'i-ch'ing,"Yüan-Li kuan-hsichung te wang-shihhun-yin"(n. 3
above), p. 231.
11)WangChen-main,"The Ch'ingDynastyand Its Influenceand Effecton Mon-
golia," ChineseCulture26,2 (1985): 78.
phenomenon of princess-bestowal seenas an effective

can be legitimately
barometerof changein militarycapability.The paper also arguesthat
whendistinctions are drawnbetweenvariousformsof themarriagecon-
tract,a sharpcontrastemergesbetweenthe Chineseapproachto foreign
marriagediplomacyand thatof her northern and thatthis
neighbours,
difference is also evidentin the conductof non-Han regimesbased in
China. Finally,the paper discussesthe social or culturalbasis of these
divergent attitudes.
II. SCOPE OF RESEARCH

As indicatedabove,the paper deals withChina's foreignrelationsbut
notexclusivelywithChineseculture:it focussesnotonlyon Chineseatti-
tudesand practicesbut also on howthosepracticeswereadopted,reject-
ed or modifiedby China-basedconquestdynastiessuch as Wei (A.D.
398-534), Liao (A.D. 916-1125), Yuan (A.D. 1271-1368) and Ch'ing
(A.D. 1644-1911).In themain,thestudydeals withperiodswhenChina
was governedby a singlestatesuchas duringHan (206 B.C.-A.D. 220),
T'ang (A.D. 618-907), Northern Sung (A.D. 960-1126) and Ming(A.D.
1368-1644). Reference is also made to occasions whenthe peoples of
InnerAsia faceda unitednorthern frontieras happenedduringtheThree
Kingdomsera (A.D. 220-280). In general,however,marriagealliances
contractedduringperiodsof division(such as that betweenSouthern
Han and thekingdomof Nan-chaoi^fg in A.D. 925) are not discussed.
Foreignmarriagecontractsmade by pretenders forthe Chinesethrone
duringperiods of civilwar (such as thatof P'eng Ch'ung ^H withthe
Hsiung-nuin A.D. 27) are also excludedfromthe discussion.
The authoracknowledgesthatit is oftendifficultto drawa clear
linebetween"Chinese"and "foreign"or to determine thatwhichis exter-
nal to and beyondthearmof statecontrol.This is particularly so when
usinggeographicor culturalfactorsto definespheresof influence.How-
ever,wheremarriageallianceis concerned,seriousdefinitional problems
as to whatis both"foreign"and "external"ariseonlyin relationto some
of the conquestdynasties,but as SectionIV will show,evenhere,such
are moreapparentthan real.
difficulties
Giftsof servingwomenand femaleentertainers such as thosebe-

stowedupon the Turksby T'ang Kao-tsuin A.D. 618, or those sentto
China byKorea,are mentionedonlyin passingwithreference to thesub-

-
ject at hand the formalmarriageof a royalprincessto the rulerof
an independent foreignstate.Here,the word"state"is used to referto
agrarian-basedregimesand nomadic tribalconfederations, i.e., it is a
termof convenienceused withoutregardto politicalstructure. Chinese
definitionsof the word "princess"are discussedbelow.
III. THE CHINESE CASE
Frequency of Bestowal
Whileit is generallycorrectto see the Han and Tang dynastiesas high

pointsin the historyof foreignmarriagediplomacyin Chinesehistory,
it shouldbe pointedout thatin all, no morethaneightimperialprincess-
es weregivenaway to foreignleadersduringthe whole of Formerand
LaterHan - a periodof some fourhundredyears.All bestowals,more-
over,occurredduringFormerHan (206 B.C.-A.D. 7), and all but one
of thoseduringthe firsthalfof thatperiod(201-105 B.C.). In contrast,
the Tang dynasty,coveringa period of 290 years,witnessedat least
twenty bestowalsspreadout over180years.In absoluteterms,theTang
certainlygave away morewomento foreignleadersthan did any other
dynasty. Yet,in termsof averagenumberof bestowalsperyear,theTang
is outclassed by its predecessor,the short-livedSui dynasty(A.D.
581-618): withina periodof less than fortyyearsSui bestowedat least
six princessesupon foreignleaders,makingan averageof one alliance
everysix yearscomparedwiththeTang averageof one every14.5 years
and the FormerHan averageof one every28.3 years.
Sui alliancescontinueda trendbegunfiftyyearsearlierat thetime

of thedivisionof northern China intotwocompetingstates:betweenthe
collapseof Northern Weiand Sui reunification
of theempire(A.D. 530s
to 590), at least eightimperialprincesseswerebestowedupon foreign
leaders.However,alliancescontractedat thistimewerehighlyunusual
in that theyofteninvolvedexchangeof womenratherthan a one-way
"gift"passingfromChinato theforeign state.Forthisand otherreasons,
it is a mootpointwhethertheseregimesshouldbe regardedas "native"
or "non-Han." The Period of Division is discussedin greaterdetail
below.Hereit sufficesto pointout thatthelatterpartof thisera through
to late Sui (534-614) was in manyrespectsmoreintensein regardto for-

eign marriagealliancesthan was the T'ang.
At theheightof itstroubleswithLiao in the 1040s,Sungconsidered
offering a womanin marriageto the heir-apparent of thatstate,but in
theendtheofferwas nevermadeor allowedto lapse (see below).In Ming
times,it was rumouredthattheOyiradinvasionin 1449was due to Chi-
neserefusalto senda womanin marriageto Esen-tayiäi (d. 1454). How-
ever,as H. Serruys has noted,thereis no firmevidencethat sucha mar-
riage was ever discussedby the Ming In
court.12 short, bestowalof
the
an imperialprincessupon theleaderof a foreigntribeor stateby native
Chineseregimesin post-Tangtimeswas as rareas fortheperiodbetween
the rise of Later Han and the fall of WesternChin (A.D. 27-316).
Recipients
Betweenthe riseof Han and the fall of T'ang, Chineseprincesseswere
bestowedupon leaders of some twelvedifferent political units: the
Hsiung-nu, Wu-sun Jg{£, Juan-juanj^Ü , Turks,T'u-yü-huntíi^S ,
Kao-ch'angÜü, Tibet,Hsi Ü, Ch'i-tan |g£h Turgas,Ferghanaand
Uighurs.Only the Hsiung-nu,Tbrks,and Uighursreceivedmore than
fourwomenduringthecourseof theircontactwithChina: mostgroups
receivedno morethantwo or threeprincesses.The low numberof be-
thepragmaticapproachtakenby the Chinese
stowalsper unithighlights
towardssuchalliances.Traditionor customplayedlittlerolein thesemar-
riages.
Apartfromtwowomenbestowedupon rulersof the Wu-suncom-

munitywhichoccupiedthe Hi Valleyto the southof Lake Balkash,all
duringFormerHan timeswereHsiung-nu.In the latterpart
recipients
of thePeriodof Divisionthrough to theend of Sui, bestowalsbeganwith
theJuan-juanand T'u-yü-hunbut weresoon concentrated almostsolely
upon the Tbrks.In all, Turkishleadersreceivedsevenbrides(A.D. 551
to 614).
Contraryto the popular perception,marriagealliances withthe

Tbrkswereproposedin T'ang times,but fewagreements everreached
12) Serruys,"Sino-MongolRelations,"p. 18.
completion.During Tang, bestowalsbegan with the T'u-yü-hunand

Tibet (A.D. 640 to 710), then turnedto the Hsi and Ch'i-tanon the
northeasternfrontier(A.D. 717 to 745) and to Ferghanaand the Türgas
- a branchof Turks(723 and 744 respectively). The Ch'i-tanreceived
four princesses,the T'u-yü-hunthree.Finally,the Uighurleadership
becamethe object of T'ang bestowals(A.D. 758 to 821). In all the Ui-
ghursreceivedfiveprincesses.13
Thereare obviousshortcomings in attemptingto drawconclusions
fromsetsof figuresrelatingto frequency of bestowal.For example,the
earlydeathof a brideor thepremature deathof herhusbandsometimes
initiateda secondbestowal,artificially
increasingthe numberof women
sentto a particulargroup.14On the otherhand, a long-livedcouple,or
a long-livedprincesswho followedthenon-Hancustomof marrying her
husband'ssuccessor(levirate),mightresultin extensionof the alliance
withoutfurtherexpenditure of women on China's part.15Again, the
highlypersonalnatureof thesealliancesmeantthat manyagreements
suffered considerabledelay beforecompletionor werenevercompleted
at all due to the untimelydeath of one or otherof therulersinvolved
in the numberof womengivenaway as
(see below). Thus, fluctuations
bridesto foreignleadersare difficultto interpretand highlyunreliable
as a guideto changingpolicy.A saferwayof approachingthelatterprob-
lemis to examinetheprinciplesbehindsuchagreements, comparingthese
withconditionsunderwhichallianceswerenevercontemplated or never
came to fruition.The followingsectionsummarizesthe resultsof such
a surveyas theypertainto the historyof nativeregimes.
Principlesand Conditions
The existence
of a stableleadershipwithwhichtheChinesecould negoti-
13) ForT'ang bestowals,see thetablein Kuang,"Tang tai kung-chuho-ch'ink'ao"

(n. 2 above), pp. 65-67.
14) The Sui bestowalupon a Turkishleaderin A.D. 599 was, forexample,simply
replacement fora womangiventwo yearsearlier.See n. 21 below.
15) For example,the womangivento the Uighurleadershipin A.D. 787 married
threeof her husband'ssuccessors.Since she died in A.D. 808 the gap betweenthe last
two bestowalson Uighurleaders(A.D. 787 and 821) was not 30 yearsbut a littleover
a decade. For similarcases in Sui times,see nn. 20-21 below.
ate was a fundamental conditionunderlying all alliances:therewas little

pointin sendingwomento leaderswho had at bestonlytenuouscontrol
overtheirown peoples.16In this and otherrespects,conditionsduring
Ming(A.D. 1368-1644)and thosefacedbythestatesof Weiand Western
Chin duringthe thirdcentury(A.D. 220-316) are directlycomparable.
In bothcases, theChinesewereconfronted by disunitedbut nevertheless
dangerousremnants -
of an earlierempire in the case of Wei/Western
Chin, by remnantsof the Hsien-pi|^^ confederation whichhad been
led by T'an-shih-huaiWBVi (d. c. A.D. 180) at the end of LaterHan;
in Mingtimes,byremnants of Northern Yuan (A.D. 1368-1389),succes-
sor stateto the Mongol regimewhichhad been drivenout of China in
1368. These forcesrepresented a significantthreatto China's integrity,
butsincetherewas no leaderof outstanding staturewithwhomto negoti-
ate, marriagealliance,as a wayof reducingthatthreat,was unfeasible.
theMongolsin Mingtimeswereled bya GreatKhaghan,
In theory,
but, as the Chineserealized,such men weremerepawns in the hands
of powerfulnobles and succeededeach otheralmostas rapidlyas did
powersbehindthe throne.The successionof leaderswas so bewildering
thatthe Chineseweresometimesunable to determine whichleaderwas
withthosewho vied
in facttheGreatKhaghan,letalone his relationship
to controlhim.Similarly,duringWei/Western Chin,theHsien-picontin-
ued to threatenChina's integrity as
just they had underT'an-shih-huai,
but thechaoticnatureof the leadershiprenderedmarriageallianceas a
formof peace negotiationsinoperable.
Timeand technological differencesalone invalidateanydirectcom-

parisonbetweenthemilitary capabilitiesof Mingand thoseof Wei/West-
ernChin,buta comparisonmaybe drawnbetweentheirrelativestrengths
vis-a-visthehostileforceswhichopposedthem.For muchof theirhisto-
ry,bothregimeswereequallymatchedby externalpowers,theirintegrity
beingmaintained onlythrougha superiordegreeof unity,i.e.,it was only
T'an-shih-huai's
deathand thesubsequentcollapseof unityamongstthe
Hsien-piwhichsavedChina froman invasionof disastrousproportions
16) See, forexample,thebackground to cancellationof a contractwiththeWu-sun

c. 60 B.C. M. Loewe,Crisisand Conflictin Han China: 104 BC to AD 9 (London:Allen
and Unwin,1974),pp. 223-24.
as LaterHan fellintodecline,and it was onlythefailureof Esen-tayiSi's

successorsto maintainMongolunitywhichsavedChina fromsimilarde-
structionduringthe fifteenthcentury.17
Ming and WesternChin also facedthe problemof managingnu-

merousunassimilated non-Hancommunities longsincesettledwithinthe
heartlandof the state.18More relevantto this studyis theircommon
approachto the constanttrickleof defectors and refugeegroupsfleeing
politicalchaos in the steppe.In the many non-hostile contactsof the
time, there no
were, doubt, leaders of breakawaygroupswho requested
Chineseprincessesas a symbolof protection. For theirpart,theChinese
invariablyrefusedsuchrequestswhentheunitsinvolvedrepresented little
or no threatto thestateor weretoo smallto serveas a publicdemonstra-
tionof China'sattractivenessto foreignpowers.Theseleadersweregiven
in
titlesand resettled frontier areas wheretheirpeoplesmightbe incor-
poratedintothe military systemand act as a bufferzone forthe heart-
land of the state.This phenomenoncan be observednot only during
WesternChin and Ming but also in the absence of marriagealliances
withdefectors fromtheTurkishempireduringearlySui (see below),and
in the phasingout of bestowalsupon the T'u-yü-hunleadershipafterit
was forcedfromitshomelandbyTibetduringtheearlypartof theT'ang
dynasty.19
17) On Tan-shih-huai'sempire,see K. H. J.Gardinerand R. R. C. de Crespigny,

"T'an-shih-huai and theHsien-piTribesof theSecondCenturyA.D.," Paperson Far East-
ern History15 (1977): 1-44. On Ming relationswiththe Mongols,see Serruys,"Sino-
MongolRelations,"esp. pp. 7-8. Mongolpowersuffered seriousset-backsin earlyMing,
butbythemiddleof thatperiodhad recovered On earlyMingrelationswith
substantially.
the equallyfragmented Jurchen see
tribes, HenrySerruys, Sino-JüraedRelationsduring
the Yung-loPeriod (1403-1424) (Wiesbaden:Harrassowitz,1955).
18) Ying-shihYü, Tradeand Expansion,pp. 203-204, estimatesthatat the end
of thethirdcentury one halfof thepopulationof Shensiwas "barbarian."In T'ai-yüan,
Shansi,severalthousandsof Hsiung-nuwereemployedas tenant-farmers and agricultural
slaves.On Mongolsresidentin China duringMing,see HenrySerruys,"The Mongolsof
KansuduringtheMing,"MélangesChinoiset Bouddhiques10 (1952-55): 215-346, and
HenrySerruys,TheMongolsin China duringtheHung-wuPeriod(1368-1398).Ibid. 11
(1956-59): 1-328.
19) See G. Mole, The Tu-yü-hunfromtheNorthernWeito the Timeof theFive
Dynasties(Roma: IstitutoItalianoperil Medio ed EstremoOriente,1970),pp. 58-59 and
notespp. 175-77.
So stringentwere the conditions under which native Chinese re-

gimes grantedprincessesto breakawaygroups that examples of such alli-
ances are difficultto find even in Han, Sui and T'ang times: it would
seem that two preconditionswere necessary for this type of alliance: a)
the division of a united and hostile force of overwhelmingstrengthinto
two camps of near equal militarycapability; and b) a corresponding
decline in Chinese power to the point where use of force in attempting
to deal witheitherpartywas unwise. Where such alliances did take place,
or were at least proposed as happened in T'ang relations with Turkish
factions, the Chinese intentionwas to maintain suspicion and discord
between the two camps; to succour the weaker side in order to prevent
reconquest by the strongerparty; and to gain a potential ally in case of
attack. For his part, the proposed recipientmightwelcome Chinese aid
in maintaininghis independence,but the power differentialbetween the
three groups was always slight.
Given the above, we can see whyoverturesmade by Turkishleaders

for Chinese princessesin the period immediatelyafter fragmentationof
theirempirein A.D. 583 were rejected: first,Chinese militarycapabilities
in this period had been considerablyenhanced by unificationof the north
in A.D. 577 and were to be furtherenhanced by reconquest of the south
in 589; secondly,over-fragmentation had led to considerable uncertainty
and instabilitywithinthe warringTurkishcamps.20 However,by 597, Sui
had become aware of the regroupingof Turkishforces and thus agreed
to a marriage alliance with one of the more compliant leaders in order
to keep alive antagonism betweenthe various groups. Due to the longev-
ity of the princess and the institutionof levirate,this alliance remained
in place throughoutthe period of Sui's decline.21It did not, however,
20) At thistime,a Chineseprincessgivento the Turksby NorthernChou (A.D.

580) was stillalive and currently
marriedto one of her firsthusband'ssuccessors.She
and herhusbandmovedsouthto be underSui protection, butlater,becauseof herconnec-
tionwiththe defuncthouse of Chou, the womanbecamean embarrassment to Sui and
a plotwas hatchedto havehermurdered. See Pei-shihik$l 99 (Peking:Chung-huashu-
chü, 1974),pp. 3293-96 fordetails.
21) Thiswomanwas actuallya secondprincess,sentin 599, herpredecessor having
died in the previousyear.The recipientand his tribesmenwereallowedto settlein Sui-
yüan,and theprincessmarried,in succession,threeof herfirsthusband'ssons. She sur-
vivedinto the T'ang era. See Pei-shih99, pp. 3296-97 fordetails.
preventTurkishcoalescenceinto two powerful,mutuallyhostilecamps,

laterknownas the Easternand WesternEmpires.
In manyrespects,conditionsduringearlyTang wereripe for a

seriesof allianceswithTurkishleaders.Yetforvariousreasons,no over-
ture,whetherinitiatedby T'ang or the Turks,everreachedsatisfactory
conclusion.22 Failurein negotiationtellsus much about the principles,
conditionsand processesbehindthe more successfulcontracts.These
cases reveal,forexample,the intenselypersonalnatureof the alliance:
since the contractwas essentiallybetweentwo individualrulersrather
thanbetweengovernments or states,thedeathof eitherrulerbeforecom-
pletionof the marriageritesannullednot onlythe agreementbut also
the considerablegroundworkwhich had gone into its making.Even
thoughthe side whichhad mostto gain mightcontinueto pressforan
alliance - oftenon the basis of previouspromisesmade (see the case
of Nan-chaobelow),negotiations had to beginall overagain. Bothsides
unquestioningly acceptedthis fact.23
T'ang relationswithTurkishleadersalso show thatmarriagealli-

ance proposedby the Chinesewitha breakawaygroupmightnot even-
tuatebecausepreliminary negotiations had theeffectof drivingthewarr-
ing factionstogether.Again, one side, perhaps intentionally,might
demandfartoo muchin returnforcompletionof theagreement, causing
to lapse. T'ang also brokeoffproposedallianceswhenthe
negotiations
intendedrecipientactivelysupportedor contractedallianceswithother
groupshostileto China.24As we shall see, the Chineseview of proper
22) See Kuang,"T'ang tai kung-chuho ch'in k'ao" (n. 2 above), p. 60.
23) The same principleexistedin tributary relationshipsin late imperialtimes:in
thewordsof TruongBuu Lam ("Intervention versusTVibutein Sino-Vietnamese Relations,
1788-1790," in Fairbank,ed., The Chinese WorldOrder,p. 179), "Tributary statuswas
granted... not to a countrybut to a ruler."In T'ang, thisconditionprovedto be the
undoingof at least one proposedagreement formarriagealliancewiththe lurks in the
periodbeforefragmentation of theEasternand Western Empiresin A.D. 627. See Kuang,
op. cit., p. 60. It also underpinnedfailureof a NorthernChou alliancewiththe Turks
in A.D. 555 (Pei-shih97, p. 3287),and was responsible
forthelapse of a proposedalliance
withTibet in A.D. 703 (Kuang,op. cit., pp. 33-34).
24) For examplessee Kuang,op. cit., pp. 39 and 60-61: cases of A.D. 625, 646
and 710.
formin foreignmarriagediplomacy did not extendto multiple,simulta-

neous alliances.
As well as unityand stabilityin leadership,therehad to be a clearly

identifiablepermanenthead of state before a bestowal mighttake place:
suitable recipientscould not be found in systemsof governmentbased
on rotatingleadership or short-termelected heads of state. During the
eighth century,the Ch'i-tan solved this problem by creating a special
"rulingfamily"to act as go-betweenin negotiationswiththe Tang court:
successive heads of the familyreceivedthe title of vassal king and when
necessaryaccepted Chinese women as brides in acknowledgementof Chi-
nese suzerainty.This ploy left the real leadership freeto negotiate with
otherpowers - an unsatisfactoryarrangementas far as the Chinese were
concerned for Tang alliances were chieflythird-partycontractsdesigned
to preventweaker groups along the frontierfromformingalliances with
stronger,more clearly hostile powers.25
Conditions under which most marriagealliances wereconsideredor

not considered by Chinese states are typifiedin the Tang relationship
with Nan-chao, a state in present-dayYunnan province.The firstpart of
this story(given below) deals withthe yearsA.D. 780-793. It shows how
marriage alliance in Tang times was often the product of a search for
allies against a foreignadversarywhose power was comparable with or
superiorto that of the Chinese state. The threads of the storyare taken
up in 784 when the Chinese court was preoccupied with mopping up
internalrebellion:26
Tibetan forces are supporting T'ang against the rebels and their Uighur
supporters.Tibet's rewardis to be T'ang withdrawalfromcertain western
areas includingthe protectorateof Pei-t'ing. Hoping to regain Uighur sup-
port to keep the Tibetans out of Pei-t'ing,T'ang renegeson the agreement.
25) See J. Holmgren, "Yeh-lii,Yao-lien and Ta-ho: Views of the HereditaryPreroga-

tive in Early Khitan Leadership," Papers on Far Eastern History 34 (1986): 52-55; also
Kuang, op. cit., p. 65 on the failureof these bestowals to preventthe Khitan fromjoining
the Turks in an alliance against China.
26) For a detailed study of Nan-chao's historyat this time, see C. Backus, "The
Nan-chao Kingdom and FrontierPolicy in SouthwestChina duringthe Sui and T'ang Peri-
ods" (Ph.D. diss., Princeton University,1978), pp. 140-44.
44 JENNIFERHOLMGREN
At thistime,theChinesecan barelycontainthespreadof Tibetanpower.

Their greatestfearis of a joint attackfromthe west by Tibet and its
satellitestateNan-chao.A plan is hatchedto cause dissentionbetweenthe
twostatesand havetheUighursattackTibetfromthenorth.The Uighurs
demanda Chineseprincessin returnfortheirco-operation.EmperorTe-
tsungíItík is eventually
persuadedto giveawayone of his own daughters
(A.D. 787).
Thus thebestowalof a princessupon theUighurleadershipin A.D.
787 was in parta desperateployto swingthebalanceof powerawayfrom
a strongadversary- in thiscase Tibet. The same conditionlay behind
China's firstmarriagealliancewiththe Wu-sun(againstthe Hsiung-nu)
in 105 B.C.,27the firstalliance withthe Turks(againstthe Juan-juan)
in A.D. 551,28and the Sui alliancewithKao-ch'ang(againstthe Turks)
in A.D. 609.29In T'ang times,it also underpinned bestowalsupon the
T'u-yü-hunin the640s and 650s,theHsi and Ch'i-tanbetweenA.D. 717
and 745, and thatupon Ferghanain the 740s. DuringT'ang, some of
these "allies," includingthe Uighurs,representeda threatto China in
theirownright- a threatwhichtheChinesecould havecontainedmili-
tarilybutat considerable cost.The realdangerlayin an alliancebetween
one of thesegroupsand a stronger powerforthiswouldhavetippedthe
balancedecidedlyagainstChina. Conversely, a Chinesealliancewithone
of these lessergroupsworkedin China's favour:this was a powerful
incentiveto turnaway fromthe use of force.
The secondpart of the accountof relationswithNan-chao(A.D.
793- 876) revealshowmarriagealliancewas neverconsideredas a control
mechanismwhenthe stateconcernedpresentedno militarythreatand
27) For details,see Loewe,"Introduction" in Hulsewé,China in CentralAsia, pp.

40-45; and Yü, Tradeand Expansion,p. 139.
28) The Turkishleader of thistimeat firstsoughta daughterof the Juan-juan
leaderas rewardfordefeating theT'ieh-le g|$Ö • Whenturneddown,he soughtan alli-
ance withtheChineseand subsequently attackedand defeatedtheJuan-juan.See Pei-shih
99, pp. 3286-87.
29) Bestowalupon the rulerof Kao-ch'angwas in partrewardforhelp in a Sui
attackon Koreaand in parta desireto haveKao-ch'angas an allyagainstthelurks: for
his part,therulerof Kao-ch'angresentedthewaytheTurkstriedto forcehimto marry
his stepmother, a Turkishwoman.See Pei-shih97, pp. 3212-15; and TFYK 978, p. 11495.
whentherewas no dangerof, or threatinvolvedin, an alliancebetween

that stateand a strongerpowerhostileto China:30
Whilethe UighursattackTibet fromthe north,the Chinesecommander
in Szechuansurreptitiouslyweans Nan-chaoaway fromTibet. An open
breakbetweenthetwostatesoccurswhenTibetperemptorily demandssol-
diersfromNan-chaoto help itsdefenceagainst Uighurs:an alliance
in the
withT'ang is formally in 793 AD, and
establishedthroughgift-exchange
forthenext50 yearsNan-chaosendsnumerousmissionsto China. How-
ever,thebilateraltreatybetweenChina and Tibetin 821 AD and thecol-
lapse of Tibetanunityin the840s greatlyreducesNan-chao'simportance.
The finalphase of relationswithNan-chao(A.D. 876-883) high-

lightsTang reluctanceto engagein any formof marriagediplomacy.It
showshow subtlechangesin therelativestrengthof thetwopartiespro-
duced immediatedelay or negationof an agreement:31
By the860s Nan-chaois strongenoughto invadeVietnamand Szechuan.
At the same time,T'ang hold overthe southernreachesof its empireis
becomingincreasingly precarious.The Chinesecommanderin Szechuanis
authorizedto negotiatewithNan-chao.Amongstotherthings,he promises
a princess.However,therulerof Nan-chaois moreinterested in theword-
ingof his relationshipwiththeT'ang courtand holdsout forrecognition
of equalityas had happenedin ChinesetreatieswithTibet.
In the following year,the rulerof Nan-chaodies and his stateis thrown
into disarray.Althoughweakerunderhis successor,it managesto make
directcontactwiththe courtin 878 to pressits claims. But thisis also
a yearof T'ang victoryagainstrebelsin the south,and withthe threat
of militaryretaliation
byNan-chaoconsiderably reduced,theChinesestall
on theagreement. In 880, whenChina is once morebesetby internaldif-
ficulties,the emperoragreesto the marriage.When Nan-chao sends its
mostable statesmen to collectthebride,theChinesehavethemmurdered.
Nan-chaocontinuesto declineand continuesto pressitsclaim;T'ang con-
tinuesto stall.Finally,withthe quellingof the Huang Ch'ao SÄ rebel-
lion and the recaptureof Ch'ang-an g$ in 883, T'ang is confident
enoughto refuseoutrightany further demands.
30) See Backus,"Nan-chaoKingdomand FrontierPolicy,"pp. 144-53 and 171

fordetails.
31) Ibid., pp. 242-59.
The factthatT'ang historycontainsas manypostponedand can-

celledagreements as it does real bestowalsalso suggeststhatT'ang atti-
tudesdid not differsubstantially fromthosewhenmarriagealliancewas
rareor non-existent:32it wouldseemthatT'ang bestowalsweretheprod-
uct of specificconditionsratherthanof a uniquemind-set.Thesecondi-
tionsmaybe summarizedas follows:a) theexistenceof a unitedhostile
powerof comparablemilitarycapability;and b) the contemporaneous
presenceof a numberof other,lesser,but independentand internally
unitedgroups,each representing a seriousnuisancefactoralongthefron-
tier.The likelihoodof one or severalmarriagealliancesbeingcontracted
in successionwas highwhenthenumberof statesinvolvedwas relatively
greatand thepowerdifferential betweenthemand China relatively small.
As we have seen,theseconditionsappliedduringT'ang but not during
Ming.Nor weretheyin effectfromtheend of LaterHan throughto the
collapse of NorthernWei in the middleof the sixthcentury.
T'ang reluctance to engagein anyformof marriagediplomacycan

also be seen in its refusalto give away real daughtersof the emperor:
womenenfeoffed as princessesand sentout as bridesto otherstateswere
mostlydistantrelativesof thethrone.Some weredaughtersof trueprin-
cesses,in whichcase theywerenotof theroyalor imperialsurname.Oth-
ers wereniecesor cousinsof the emperorthroughthe male line. A few
wereadopteddaughters. The T'ang evenmadeuse of womenof non-Han
origin,i.e., the daughteror sisterof a surrendered leader,whosepeople
had settledwithinChina proper,was enfeoffed and sent out to marry
a foreignruleron China's behalf.33The temporary"banishment"of
PrincessT'ai-p'ing ^c^Äi to a Taoistmonastery duringthecourseof
negotiations with Tibet in the670s illustrates
thelengthsto whicha ruler
would go to avoid sendinga daughter.34
Refusalto give away real princessesin T'ang timesmirrorsthe

historyof Han and Sui bestowals:WangChao-chünis said to havebeen
32) See n. 36 belowforan earliercase of a cancelledalliance - thistimeinvolving

the Syr-tardush.
33) For an exampleof the latter,see bestowalsupon theTurgasand Sha-t'o $>pb
in the 720s. Kuang,"T'ang tai kung-chuho-ch'ink'ao," p. 46.
34) Hsin T'ang-shufflîtt 83 (Peking:Chung-huashu-chii,1975),p. 3650.
chosento marrya Hsiung-nuchieftainin Han timesbecause she was

presumedto be theugliestof theemperor'smanypalace women;during
Sui, PrinceHuai-yangrii§ was degradedin rank for revealingthat a
bridesentto the T'u-yü-hunwas not the emperor'sreal daughter.35 In
all, onlythreecases are known where true daughtersor sisters
were given
away.36These wentto Uighurleadersduringthelatterpartof theT'ang
dynasty. The firstwas marriedin A.D. 758 whenChina desperately need-
ed Uighursupportto helpquell an internalrebellion.Ratherthanbeing
forcedto marryher husband'ssuccessoras was usuallythe case, this
womanwas broughtback to China afterherhusband'sdeath.The sec-
ond case occurredin A.D. 787 when T'ang requiredUighursupport
againstTibet (see above); and the third,in 821,was aimed at stallinga
threatened Uighurinvasionat a timewhenthestatewas once morepreoc-
cupied withTibet. Thus, exceptunderextremecircumstances whenthe
statetotteredon theedge of collapse,normalChinesedefinitions of the
termkung-chu&± (princess)did notapplyin foreign marriagerelations
evenin T'ang times.37
Apart fromthird-party alliances occasioned by the conditions

describedabove,nativeChineseregimescommonlyproposedor agreed
to proposalsforallianceswithhostilestatesof apparentor demonstrably
superiormilitarycapability.FormerHan marriagecontractswiththe
Hsiung-nubetween200 and 100 B.C. fall into thiscategory,as do alli-
ances withthe Juan- juan in the 540s and thosewiththe Turksbetween
560 and 580. In T'ang times,we findthistypeof bestowalupon Tibet
35) For storiesassociatedwithWangChao-chün'smarriage, see n. 1 above.On the

T'u-yü-huncase, see Mole, The Tu-yü-hun,pp. 171-72, n. 467.
36) It seemsthatT'ang T'ai-tsunghad been readyto givea daughterto the Syr-
tardushin A.D. 642 but the alliancewas brokenoff by China at the last momentdue
to Syr-tardush's to presentthepromisedhead of cattle.It was not failureto keep
inability
a promisewhichworriedT'ang, but ratherthatdifficulties in the Syr-tardush economy
indicateda suddenweaknessin theirpositionwhichwouldmakeseriousreprisalsagainst
China extremely See, Kuang,op. cit., p. 62; and Hu San-hsing's$jH^î com"
unlikely.
mentaryto Tzu-chiht'ung-chienJlreMSí (hereafter TCTC) (Hong Kong: Chung-hua
shu-chü,1971),p. 6180.
37) Kuang,op. cit., p. 29. On the Uighursin particular, see Kuang,op. cit., pp.
49-56; and Mackerras,The UighurEmpire,pp. 43-47.
(A.D. 641 and 710), the Turgas(A.D. 723) and the Uighurs(especially
A.D. 769).
Two otherconditionsunderwhichnativeregimesgave womenin
marriageto foreign leadersmustbe mentioned: a) whenthestaterequired
in
help combating internal and
rebellion; b) when thestatewas interested
in oustingtherulerof a hostilegroupand replacinghimwithone more
favourableto China. Both thesestrategies wereused sparinglyeven in
Han and Sui/T'angtimes.In all cases, the recipientgrouprepresented
a threatto China in its own right,or had in the recentpast constituted
a threatof significantproportion.An exampleof thelattercase is Wang
Chao-chün'smarriageto a claimantforleadershipof the Hsiung-nuin
33 B.C. This was the firstand only bestowalupon a Hsiung-nuleader
afterthe divisionof theirconfederation (see above).38A variantsitua-
tion occurredin A.D. 610 whenSui promiseda womanin marriageto
an uncle of the leader of the WesternTurksif he would eliminatehis
obstreperous nephew.The nephewwas forcedout of officeand fledto
Sui wherehe himselfreceiveda princessfor help in two campaigns
againstKorea.39Bestowalsupon Uighurleadersin A.D. 758 and 760 fall
into the categoryof allies againstinternalenemies,but were,in major
part,insuranceagainstUighuraid forT'ang rebels.Similarly, T'ang Kao-
tsu's attemptsto negotiatean alliance withthe Turkswere aimed at
breakingtheirsupportforforceswhichopposedtheestablishment of the
T'ang dynasty.
When China was demonstrably strongerthan the otherparty,or
was at least confidentof her superiority in the field,she was unlikely
eitherto proposeor agreeto any proposalsformarriagealliance.One
earlyexception,involvingthe Wu-sunin 110B.C., had provedmostun-
as Loewe has pointedout, the Chinesewereof the opinion
satisfactory:
thatgranting a princessin marriageto the rulerof a lesserstatelocked
themintoa relationship of politicaland military
supportforat leastthe
durationof the woman'slifetime.This could well provedetrimental to
thenationalinterest.40Thus Hsiung-nurequestsmade duringtheperiod
38) See Loewe, Crisis,ch. 7; and Hulsewé,China in CentralAsia, p. 126 for

backgrounddetails.
39) Pei-shih99, pp. 3301-302.
40) See Loewe,Crisis,pp. 216-19; and Hulsewé,op. cit., p. 127.
of theirdecline at the end of Later Han were invariablyturned down as

were those made by Turkishleaders at the height of T'ang influencein
the Far East: at these times, the Chinese were able to dictate the terms
of the relationship,and for various reasons their vision of foreignrela-
tions did not embrace the marriage contract (see Section V).
Marriage alliance, then, did not occur when Chinese forces were
demonstrablystrongin relationto the peoples ranged around the frontier,
or when frontierforces opposing the state sufferedfrom lack of unity
and marked instabilityof leadership. Neither did it occur when Chinese
forceswereequally matched and in stalemate,i.e., in times of comparable
and stable militarycapability: at the least, there had to be a perceived
potential for change in the balance of power - change which would be
significantlydetrimentalto China. This lattercondition is typifiedin the
historyof foreignrelations during Northern Sung (A.D. 960-1126).
Not unnaturally,contemporariesof the Ming dynasty(1368-1644)

tended to blame Northern Sung for the Mongol threat which existed
along the frontierin Ming times (see above). The Ming perception of
Sung as weak regimeaddicted to compromisepersistedwell into the nine-
teenthand early part of the twentiethcenturieswhere it served as a con-
venient focus of discussion for two contemporary(but not necessarily
related) issues - Ch'ing weakness and ineptitudein dealing with alien
westernpowers; and the Ch'ing's own alienness. However,as Labadie has
pointed out, the very question of whetherSung was weaker than Han
or T'ang tends to obscure the fact that each regimefaced a verydifferent
set of circumstancesalong the northernfrontier.41 The followingdiscus-
sion attemptsto dispel furtherthe mythof a weak pacifist regime and
to show why foreignmarriage diplomacy was rarelyconsidered in Sung
times.
Ming scholars were certainlycorrectin their assessment of North-

ern Sung historyas lacking a period of dramaticconquest and expansion
of the kind seen under Han Wu-ti, T'ang Kao-tsung and the firstthree
41) See J. Labadie,"Rulersand Soldiers:Perceptionand Managementof theMili-

taryin Northern SungChina (960-ca. 1060)" (Ph.D diss.,Washington 1981),
University,
pp. 225-29.
rulersof Ming - conquestswhichextendedChinesepowerintoMongo-

lia, CentralAsia and Korea. At the same time,NorthernSung is also
unique in thatit saw no significantcontractionof its bordersas hap-
pened duringHan, T'ang and Ming times.Moreover,unlikemanyre-
gimesin Chinesehistory, Sung was able to maintainits positionwithout
numbersof soldiersof non-Hanorigin.It should
recourseto significant
also be recognizedthatat the timeof its foundationthe statewas con-
fronted notbya singleconfederationof nomadictribesas happenedwith
FormerHan, nor by a plethoraof groupsas in T'ang times,but by two
empires(Liao and Hsi-hsia)bothof which
stable,in partagrarian-based,
withinthe GreatWall. The factthat Sung was
had long held territory
able to conquerand reunitethe restof China, and thento maintainits
positionagainstthesestatesforsome 160 years,is evidencenot onlyof
fortuitous but also of considerablestrength.
circumstance,
Both McGrathand Labadie have shownthat neitherthe rhetoric

northeactivitiesof Northern Sungsuggestpassivity. Twoperiodsin par-
ticularshowa highdegreeof activeaggressionand expansionistintent.
Together,these phases representsome 90 yearsof the historyof the
regime.The first(A.D. 960-1004) saw belligerence towardsboth Liao
and Hsia. It ended in a treatyof equalitywithLiao in A.D. 1005 (see
below).The secondphase (A.D. 1070-1119)saw aggressiondirectedpri-
marilytowardsHsia duringwhichtimethatstatewas forcedto call on
Liao forsupport:as McGrathhas remarked, thereis littlereasonto deni-
gratethe veryconsiderablemilitaryachievements of NorthernSung.42
The treatybetweenLiao and Sung in 1005 concludeda periodof

activehostilityand successfulrepulsionof a Liao invasion.As a signof
theirequality,the rulersof Liao and Sung agreedto addresseach other
as brothers.In the past, thishad been a conditionof treatiesconcluded
42) M. McGrath,"Militaryand RegionalAdministration in NorthernSung China

1982),pp. 295 and 297. See also Labadie,
(960-1126)" (Ph.D. diss.,PrincetonUniversity,
op. cit., pp. 47-50, 73 and esp. 230. L. Kwanten,"Chio-ssü-lo(997-1065), a Tibetan
Ally of NorthernSung," RocznikOrientalistyczny 39,2 (1977): 104, comes to the same
conclusion.On Hsia pleas forLiao aid fromthe 1090sthroughto the early 1100s,see
Liao-shih Jg£ (Peking:Chung-huashu-chü,1974),chs. 25-27, pp. 300-324.
withthe Hsiung-nu,Turksand Tibetans.43However,the similarity be-

tweenthoseperiods of weaknessand Sung's conditionis moreapparent
thanreal.Certainly,Sungregretted itscondition,butthisregretstemmed
fromthwarted aims ratherthaninabilityto protectcurrent
expansionist
borders:in A.D. 1005,China's capacityto oppose forwardexpansionby
hostileforeignpowerswas fargreaterthanin anyof thoseearliertreaty-
periods.This can be seen not only in the militaryachievementsof the
statebut also in the absenceof proposalsfora marriagealliance with
Liao or any otherstatealong the frontieruntilthe 1040s(see below).44
The treatyof 1005was in parta resultof stalematebetweenLiao

and Sung, and in part the resultof Sung's fearof a rearguardattack
by Hsia whileengagedin hostilities withLiao. Therewas also thelikeli-
hood of invasionby Liao duringa campaignagainstHsia.45 However,
if Sung hoped thatthe treatywould allow her to pursuea relationship
witheitherstateindependent of influencefromtheother,shewas mistak-
en: from989 throughto 1038,Hsia rulerswereallied throughmarriage
to theLiao leadership(see SectionIV). This was not,however, a simple
client/patronrelationshipsinceLiao fearedHsia forthesame reasonsas
did Sung. Hsia independence,then, was maintainedby the wariness
whichexistedbetweenthetwostronger powers.In thisway,generalstale-
mate ensuedalong the frontier. Minor skirmishesaside, this condition
lasteduntilthe collapse of Liao underpressurefromthe JurchenChin
in the 1120s.
The Hsi-hsiaapproachto relationswithLiao was typicallynon-

in marriagefromthatstatebothwhen
Han in thatit requestedprincesses
Liao neededits supportand whenit, Hsia, was in need of Liao support
43) See Lien-shengYang,"HistoricalNotes"in Fairbank,ed., The ChineseWorld

Order,p. 20; and Chusei Suzuki, "China's RelationswithInnerAsia: The Hsiung-nu,
Tibet," ibid., p. 190.
44) Sung-shih5fc£ 490 (Peking:Chung-huashu-chii,1977),pp. 14114-17makes
the pointthatduringthe Five Dynastiesand NorthernSung,the Chinesecontinuedto
referto Uighurleadersas wai-sheng^fg (nephewsthroughthe sister'sline) whilethe
UighursaddressedChineserulersas chiù jj (unclesthrough themother'sline).The con-
dition,however, stemmedfromlate T'ang bestowalsupon the Uighurleadership.
45) See Labadie, "Rulersand Soldiers,"pp. 40 and 63-65.
againstSung (see below). Sung,however, remainedaloof fromthisform

of diplomacyevenduring attemptto woo theTibetansas alliesagainst
its
Hsia:46 the treatywith Liao, the only state withcomparablemilitary
strength,appearedto obviatethe need for such alliances.
In the 1040s,however,conditionschanged.At thattime,Sungmili-
tarycapabilitiesvis-a-visHsia wereat an all-timelow and Liao began
usingthisto pressfora radicalrevisionof the treaty,demandingSung
withdrawalfromsubstantialareas along the frontier betweenthe two
states.As part of its appeasementplan, the courtconsideredincreased
annual giftsor a marriageallianceinvolvinga Sung "princess"and the
heir-apparent to the Liao throne.The proposedalliance failedto even-
tuate,firstbecauseFu Pi %Wi (A.D. 1004-83), who was sentto negoti-
ate thetermsof therevisedtreaty, disapprovedof theidea on thegrounds
thatsuch contractswereof no necessarylastingvalue and did not save
the state anythingin termsof monetaryexpenditure;47 and secondly
becauseLiao, in contrastto Hsia, did notencouragetheentryof foreign
womenintoitsimperialharem.The natureof themarriagesystemadopt-
ed by the foundersof Liao also meant that the presenceof a Sung
womanin the imperialharemwould have had no effecton Liao policy
(see SectionsIV and V). Fu Pi, who knewtheLiao courtwell,was proba-
bly aware of this fact.
Apartfromtheabove factors,Sung's positionwas not as desperate
as firstthought.In fact,it seemsthattheChinesecourtpanickedunnec-
essarilyoverLiao demands.AlthoughSung forcesappearedto be weak
at thistime,considerabletensionhad arisenbetweenHsia and Liao over
thedeathof one of theLiao princessessentto thatstate(thusconfirming
Fu Pi's view of the risksinvolvedin such contracts).48Hsia was also
veryconcernedby the presenceof a united,hostileTibetanforceon its
southeastern flank.That forcewas being activelywooed by Sung and
mightwell, as futureeventswereto prove,also be wooed by Liao.49
46) On relationsbetweenSung, Hsia and the Tibetan leader Chio-ssú-lo,see

Kwanten,"Chio-ssú-lo(997-1065)."
47) See Sung-shihchi-shihpen-mo %-&.fâ^$S 21 (Peking:Chung-huashu-
chü, 1977),pp. 150-53.
48) Liao-shih18, p. 220.
49) See Kwanten,op. cit., esp. pp. 100-102; and Sung-shih492, p. 14162.
In short,therewas littlereal danger of a united attack by Liao and Hsia

at that time, a fact which can be seen in Hsia failure to support Liao
during an attemptedinvasion of Sung in 1044 (see below).
During negotiationsover the revisedtreaty,Liao had suggesteduse

of the term hsien jfô(sacrificial offering)in regard to annual payments
and giftsmade by Sung. Fu Pi argued that this was inappropriatesince
the Sung emperor was "brother" to the Liao ruler,i.e., his equal. The
fact that Liao accepted this indicates Sung's high standingin the tri-par-
tite relationship.When the word na $j (present) was suggested, Fu Pi
repliedthat such words had been used only once under veryunusual cir-
cumstances - when T'ang Kao-tsu had tried to borrow Turkishtroops
to help stabilize his rule within China. The reply suggests Sung confi-
dence in its position - at least in regard to conditions faced by earlier
regimes.Nevertheless,the Chinese eventuallyagreed to the termna. Two
years later,when Liao invaded Sung, Hsia gave its support not to Liao
but to Sung, thus ensuring Liao withdrawaland a returnto the condi-
tions described above.50
It could be argued that Sung survivalrestedlargelyon the fact that

for much of the time Hsia and Liao had little interestin extendingthe
southern or China-bounded frontiersof their states, i.e., that border
problems at this time stemmed largely from the Sung obsession with
regainingterritorylost duringlate T'ang and the Five Dynasties.51How-
ever, it could be equally argued that Sung strengthstemmed from the
absence of internaldisruptionof the kind seen duringlate T'ang and the
Five Dynasties. As McGrath has pointed out, Sung success in this field
was a major achievement.52It was also an importantfactorin obviating
50) See Sung-shihchi-shihpen-mo21, p. 153.

51) Many instancescan be foundin Liao historyof "failure"to take concerted
actionagainsttheSungat an opportunemoment.Occasionscan also be found,as in A.D.
1008,whenLiao ceased borderviolationsaftera verbalreprimand fromSung. Labadie,
"Rulersand Soldiers,"pp. 96-103, citesa contemporary Sung view thatLiao was not
interestedin warwithSung.Similarly, Kwanten,"The roleof theTangutin Chinese-Inner
Asianrelations," Acta Oñentalia39 (1978): 191-98, has shownthatexceptforone particu-
lar era, Hsi-hsiawas mainlyinterested in extending its westernfrontiers
at the expense
of the Uighurs.
52) McGrath,"Militaryand RegionalAdministration," p. 310.
by T'ang withthe
the need formarriageallianceof the typecontracted
Uighurleadershipin the period of its decline.
In sum,we can say thatwhatever thecause,NorthernSung history

in Chineseforeign
providesan unparalleledexampleof stability relations,
and thatmarriagealliancewas not a significantfeatureof foreignpolicy
at thistimebecause the statewas able to maintainits militarycapacity
to deal withanythreatposed byhostileexternalforces.Thiswas no isola-
tionistperiod,however,for á regularstreamof diplomatic-and gift-
exchangemissionskept each state informedon eventswhich might
indicatea changein thebalanceof power.That changecame in theunex-
pectedformof a challengeby the JurchenChin in the second decade
of thetwelfth century- a challengeto Liao whichbothSung and Hsia
soughtto exploit.For Sung the consequencesweredisastrous.Hsia sur-
viveduntilthe rise of the Mongols underChinggisKhan one century
later.
We have seen thatwhateverthe frequency of bestowalor intended

bestowalin anyone period,nativeChineseregimesweregenerallyreluc-
tantto indulgein marriageallianceas a formof foreignrelationsdiplo-
macy.Thus bestowaldid not occurwhenChina was in a positionto dic-
tatethetermsof itsrelationshipwithforeignpowers,i.e., in timesof ter-
ritorial
expansionor equilibriumwhenthestatewasdemonstrably stronger
thanor at least confidentof its abilityto containany perceivedthreat.
The Sung case also demonstrates the complexity be-
of the relationship
tweenprincess-bestowal and dynasticweakness:proposalsformarriage
alliance with a foreignpower could be a functionof detrimental or
perceiveddetrimental changein the balance of powerwithinan overall
contextof relativestrength.
Forms of the Marriage Contract
It willbe apparentfromthediscussionabovethatChinesestatesdid not

normallyindulgein multiplealliances,i.e., contractswereusuallywith
one groupat a time,and thatthiswas in contrastto non-Hanmoresand
attitudes.In T'ang times,forexample,theTibetanswerehighlyincensed
thattheChineseshouldbestowwomenupon theT'u-yii-hunor,in later
times,upon the Uighurswhile denyingthat courtesyto Tibet:53non-

Han leadersviewedpreferential or exclusivemaritalties betweenother
powersas an act of hostility;
theChinesesaw multiplealliancesas secre-
tiveand underhand.In A.D. 727, whenthe Turgasleader supporteda
Tibetaninvasionof An-hsi$ffi, the Chinesesaw his simultaneousac-
ceptanceof womenfromChina,Tibetand thenorthern Turksfouryears
earlieras evidenceof maliceaforethought. In theirview,acceptanceof
a Chinesewomanin marriagehad committed him to an exclusiverela-
tionshipwithT'ang forthe durationof the woman'slifetime.54
It is also apparentthatwhatever theirmilitary
positionand status
vis-a-visChina, non-Han leaders readilyacceptedChinese women as
brides.Since fewinsistedon or evensuggestedthatthe Chineseaccept
one of theirprincesses allianceswerealmostalwaysof thetype
in return,
wherewomenpassedfromChina to thestatein question.However,non-
Han willingness to acceptChinesebridesonlypartlyexplainstheabsence
of marriage-exchange withChina. One incidentconcerningthe Turkish
leaderMo-cho jgftlgg (d. A.D. 716)in the690s suggeststhatevenin T'ang
timestheChineseadamantlyrejectedtheidea of acceptingforeignwom-
en into the imperialharem:55
AfterhavinghelpedT'ang crusha Ch'i-tanrevoltin the northwest, Mo-
cho insiststhathis daughtermarrya T'ang prince.The courtsendshim
a memberof theempress'sfamily(a Wu S£ ratherthana Li $ ), hoping
thatif thisdoes not deceivehim,he mightat least be mollified.Mo-cho
is notpleased:he wantshisdaughterto marrytheheirto theT'ang throne
and knowsverywellhe has beengivena memberof theempress'sfamily.
The resultis thatHopei and Shansi provincessuffercontinuousraiding
53) In the firstcase, TibetattackedtheT'u-yü-hun,thenpresentedan ultimatum

to China. The outcomewas the firstof two marriageallianceswithT'ang (A.D. 641).
Afterthis,Tibet refrained fromattackingChina directlyfor some 30 years.See Chiù
Tang-shu Ü Jgit 196A (Peking:Chung-huashu-chii,1975),pp. 5220-23. On China's
relationswithTibetand theUighursin A.D. 821,see Mackerras,The UighurEmpire,pp.
45-47.
54) See TCTC, pp. 6754, 6775-76 and 6833; Chiù Tang-shu 194B,pp. 5191-92;
and Hsin Tang-shu215B,pp. 6067-68.
55) For accountsof Mo-cho's relationswith T'ang as given below, see Chiù
Tang-shu 194A,pp. 5168-73; and Hsin Tang-shu215A, pp. 6045-49.
56 JENNIFERHOLMGREN
formorethanfiveyears.In A.D. 703, Mo-chomakesanotherattemptto

havehis daughtermarrytheheirto theTang throne.This time,theChi-
nese appear to consent:betrothalgiftsare exchanged,but whenChung-
tsung cf^ ascendsthe throne(A.D. 707), the courtcancelsthe agree-
ment.Furtherraidingensues.Mo-cho eventually abandonshis ambition
to becomegrandfather to a Tang emperor:whenJui-tsung§^ ascends
the throne(A.D. 710), he asks fora princess.
It is clearthenthatmarriagecontractswhichsaw the introduction

of foreignwomeninto the imperialharemin China wererarenot only
because of willingnessby non-Han regimesto accept Chinese women
withoutreciprocity butalso becausetheChineseregardedsuchproposals
as insupportable: the statewouldhave had to be in a parlouscondition
indeedforthecourtto accepta foreign bridefortheemperoror heir-ap-
parent(see Section V). In timesof nationalunity,thatconditionnever
arose. We can, however,see it in periodsof extremeweaknesssuch as
in theNorthern and SouthernDynastiesafterthecollapseof theNorth-
ern Wei empirein the 530s.
Analysisof alliancescontractedduringEasternand WesternWei

betweenA.D. 533 and 546 is somewhatcomplicatedbythestrandof non-
Han ideas runningthroughnorthernChinesecultureat thistime:it is
possible that the presenceof non-Han ideas facilitatedacceptanceof
marriage-exchange (see SectionIV). Analysisis also complicatedby the
factthatherewe are dealingwithsetsof de factoand puppetrulers.Yet,
it is clearthatthemainconditionunderlying acceptanceof foreign wom-
en as senioror principalwivesby bothde factoand puppetleaderswas
military weakness,a weaknesswhichderivednot onlyfromthe strength
of theJuan-juanbutalso frompoliticaluncertainty withinbothsegments
of the empireat thistime.The accountbelow showshow fluctuations
in military positionbetweenA.D. 530 and 550 werematchedby changes
in both occurrenceand formof the marriagecontract:56
In A.D. 533, thepuppetemperorof EasternWeisendsa womanin mar-
riageto theJuan-juan.The womanis thedaughterof a Weiprince.Shortly
afterwards,theemperordefectsto WesternWei.Thus theallianceis trans-
56) On eventsoutlinedbelow,see Pei-shih13-14, pp. 507 and 516-18; and ibid.

98, pp. 3263-64.
ferredto that state. When the emperor dies in A.D. 535, the Juan-juah
demand that his successor accept a Juan-juan woman in marriage. They
back up the demand with border incursions,and in 538 at the age of 13,
a daughterof the Juan-juan leader becomes empressof WesternWei. West-
ern Wei is now freenot only fromattack by the Juan-juan but also from
a Juan-juan alliance with Eastern Wei.
In A.D. 540, the empressof WesternWei dies. Angryand suspicious over
the death of his daughter,the Juan-juan leader invades Western Wei. His
forces are beaten back and he turns to Eastern Wei for an alliance. He
is sent the daughter of a Wei prince. However, the alliance must be more
firmlycemented, and in 542 the de facto ruler of Eastern Wei agrees to
take a granddaughterof the Juan-juan leader as wife for one of his own
sons. In 546 he accepts a daughter of the Juan-juan chieftainas his own
principal wife.
The de facto ruler of Eastern Wei dies in 547 in mid campaign against
WesternWei. The campaign has been a disaster,and his eldest son takes
the Juan-juan woman as his own principal wife (levirate) in an effortto
preventJuan-juan defection to WesternWei.
These unique relations with the Juan-juan were broughtto an end

by the rise of the Turks and internalconsolidation of both the Eastern
and WesternEmpires in the 550s. Return to the more normal form of
alliance in the subsequent era (A.D. 551-580) suggests that even in the
NorthernDynasties there was considerable reluctanceto accept foreign
women as wives for the imperial harem. This point is also brought out
in the historyof foreignrelationsduring NorthernWei (Sections IV and
V) and in relationsbetweenEastern Wei and the T'u-yü-hun in the 540s.
By the 540s, it was clear that the puppet rulersof Eastern and West-
ern Wei would soon be replaced by theirmasters - the Kao Sb and Yü-
wen ^^t . Thus in 542 and 546 the Juan-juan made sure that their
women were placed not with the Wei emperor of the time but with a
senior member of the Kao family (see above). However, a T'u-yü-hun
woman sent to Eastern Wei in 545 was relegatedto the harem of the pup-
pet emperor:to have the T'u-yü-hun as an ally against WesternWei was
valuable, but theydid not representa threatto the viabilityof the state
in the same way as the Juan-juan. In consequence, they were in no
positionto dictatethe formor courseof the alliance.57

It would seem,then,thatnon-Hanrulerswerenot only prepared
to enterintomaritalcontracts withstatesof equal or lessermilitary
capa-
bility,but thatthey were also preparedto receive as well as givewomen
in marriageand to indulgein simultaneousor multiplealliancesspread
across a varietyof friendly and mutuallyhostilepowers.Whythiswas
so is discussedin SectionV below.For the moment,the paper attempts
further
to illustrate thedifferent viewsheldby Chineseand stepperulers
about the place and formof the marriagecontractin foreignrelations.
IV. THE NON-HAN CASE
Steppe Traditions
LikeChina,non-Hanstatesengagedin marriagediplomacyin an attempt
to staveoffmilitary threatby stronger parties.Alliancesmightbe pro-
posed witha third but
party, in generaltheywerecontracted with
directly
thestrongerpower.The responseof thestronger partyto sucha proposal
varied accordingto intention:rejectionimplied aggressiveintent -
unwillingnessto abandonor moderateexpansionist aims. Proposalswere
also made by stronger,potentiallyhostile to
powers weakerpartiesand
to partiesof comparablestrength. These werenot merely"demands";
theyweretokensof goodwill- signsof willingness, howevertemporary,
to relinquishaggressivedesignson the statein question.
Since refusalto engagein marriagediplomacywas seen as tanta-
mountto declaredwar,the Chineseresponseinvariably provokedhostil-
ity.Sometimes,angeroverthe issue eruptedinto violence - imprison-
mentor executionof envoysand/orimmediateresumption of war. On
otheroccasions,hostilitywas limitedto verbalsparring.Illustrationof
the lattercase is to be foundin the historyof T'ang relationswiththe
Turksduringthe timeof Mo-chi-lienSft$fö£ ¡n the 720s. The primary
issue at pointwas preferentialtreatment by China for otherpowers.58
57) TFYK 978, p. 11493;Pei-shih96, p. 3187; and Mole, The Tu-yü-hun,p. 19.
58) For an accountof the affairwithMo-chi-lien,see Hsin T'ang-shu215B, p.
6053. For a moreaggressiveresponse,see ibid. 216A, pp. 6071-89; and the discussion
above relatedto n. 53 on earlyT'ang relationswithTibet.
Mo-chi-lien'sofficialsarguedthat the Tibetan,Hsi and Ch'i-tan

leadershad all receivedT'ang womenin marriageand thattherewas no
reasonto excludetheTurksfromsuchan arrangement. Theirpersistence
led theChineseenvoyto seek refugein theexcusethatMo-chi-lien, who
had been grantedthe statusof son of the T'ang emperor,could hardly
marrytheemperor'sdaughter.Mo-chi-lienpointedout theinconsistency
in this argument:otherleaderswho had been giventhe T'ang family
name (Li) had all been grantedwomenin marriage.He concludedwith
the retortthat anywayeveryoneknew that the womenwere not true
daughtersof the emperorand thatT'ang was attempting to make him
the laughingstockof nations.
The abovecase is interestingin thatit revealsthehostilityprovoked

by Chinese refusalto engage marriagediplomacy; way in which
in the
a Chineseenvoymightattemptto placatenon-Hanangeroverthisissue;
non-Hanperceptiveness about the natureof excusesput forwardby the
court;and how steppeleaderswerereadyto acquiescein the deception
overdaughters but refusedor failedto understand Chinesereluctanceto
in
engage any kind of marriagediplomacy at all. As indicatedabove,
anger over the of
question apparentpreferential treatment oftenerupted
intoa militaryoffensive againstChina or againstthoseperceivedto have
a special exclusiverelationshipwiththe Chinesestate.
While some stepperegimesattemptedto conductrelationswith

China as if dealingwithanothersteppepower,othersmodifiedtheir
approachto suit needs and objectivesvis-a-visa prosperousagrarian
regime,i.e., proposalsforan alliancemightbe aimedat obtainingthird-
partysupportor gainingtemporary militaryrespiteas occurredin rela-
tionsbetweensteppepowers(see below),butin othercases marriagealli-
ance was part of a systematic programforexploitationof the Chinese
economy.Twotypesof exploitative The
stepperegimescan be identified.
firstreliedon periodiclow levelmilitaryactivitycoupled withindirect
economicexploitation. Here,calculatedlevelsof violenceweregearedto
obtainingmaterialwealththroughtradeconcessions,subsidiesand gifts.
Wherehighlysuccessful,as withthe Hsiung-nu,59 regularbestowalof
Organizationand For-
59) See T. Barfield,"The Hsiung-nuImperialConfederacy:
eign Policy,"Journal of Asian Studies41,1 (1981): 45-85.
womenbytheChinesecourtwas an important sourceof revenue(see Sec-

empire- as seenin thevarious
tionV). The secondkindof exploitative
-
conquestdynastiesdiscussedbelow requireda higher,moreconsistent
militaryprofilesince these regimesrelied on more directavenues of
achievedthroughpermanent
exploitation acquisition.In accor-
territorial
dancewith no
steppetradition, proposal or demand formarriagealliance
was forthcoming fromtheseregimes.
The principlesgoverning foreignmarriagediplomacybetweensteppe

powers are most clearlyseen in the behaviourof T'o-pa Í5í8c, Mongol
and Manchuleadersin predynastic timesbeforethe foundationof their
respective China-basedempires.The historiesof thesepeoples not only
illustrateexactlywheresteppetraditionsdiffered fromthe Chinese,but
also how the place and relativeimportanceof foreignmarriagealliance
variedwithtime.These variationsmusthaveadded considerably to Chi-
nese difficultiesin understanding past and currentrelationswith the
peoples of this region.The followingdiscussionindicatessome of the
similarities betweentraditional
and differences T'o-pa, Mongoland Man-
chu approachesto foreignmarriagealliance.
In predynastic times,marriagewas a normalpartof foreignrela-

tions for T'o-pa leaders.Allianceswerecontractedundera varietyof
conditions:a) whenthe T'o-pa werestrongerthan the otherparty,as
occurredwiththe Yü-wenand Hsiung-nuin A.D. 293 and 318; b) when
in need of third-partysupport,as withtheYü-wenallianceof A.D. 299;
and c) in mutualnon-aggression pacts betweenequals, as withthe Mu-
jung ^^ allianceof A.D. 344. These involvedbothgivingand
contracts
receivingwomen as well as simultaneous exchange.Womengivenaway
as brideswereusuallydaughtersof the rulerconcerned.Simultaneous
exchangetendedto occur in pacts betweenequals.60
Apartfromtheabove,therewerealso "alliances"withtriballeaders
suchas theHo-lan MBB.However,sincetheseinvolvedpeopleswho were
they can be classifiedas
part of the establishedpolitical structure,
60) J.Holmgren,Annalsof Tai,EarlyT'o-pa HistoryAccordingto theFirstChap-

terof the Wei-shu(Canberra:AustralianNationalUniversity Pressin associationwiththe
Faculty of Asian Studies,1982),pp. 30-41, 57, 59, 64, 68-69, and 71.
internalcontractsof the kindseen betweenthe imperialfamilyand the

civilianeliteduringT'ang or thosebetweenthe imperialfamilyand the
militaryeliteduringNorthernSung.61
In general,the distinction betweeninternaland externalalliances

is less clear fortribalconfederations than it is forcentralizedbureau-
craticstatesas foundin China. For T'o-pa history, however,it is relatively
straightforward because, unlikethe Mongols, the T'o-pa used foreign
marriage allianceas a defencemechanism rather thana tool forconquest
and rapid territorial expansion.In the thirteenthcentury,the Mongols
exploitedthe full potentialof the marriagecontractas an adjunct to
military conquestand rapidterritorial expansion.Theirverysuccessin
thismatter- the factthatprincess-bestowal came to playa crucialrole
in maintaining the rateof expansionand the politicalviabilityof their
empire(see below),createsa definitional problemas to whichmarriage
groupshould be seen as foreignand properlyindependentof Mongol
control.Similarquestionsarisewhendealingwiththerelationship between
the Ch'i-tanrulersof Liao and theirUighurconsorts,or Manchu ties
withthe Mongol eliteduringthe Ch'ing dynasty.
UndertheMongolsin thethirteenth peopleswhoseleaders

century,
readilysubmitted to marriagealliancewithChinggisKhan wereregarded
as alliesand theirgovernments and traditional wereleftintact.
leaderships
In a fewcases, a regularsystemof marriage-exchange ensued between
the leadersof thesepeoplesand the rulingMongol clan. In othercases
- mostlythose involvingnon-Mongolsand peoples who had agreed
somewhattardilyto marriagealliance, a systemof one-waybestowal
ensuedin whichdaughtersand sistersof the Khan weregivenin mar-
riage, but women fromthese groups were only irregularly taken as
61) On T'o-pa intermarriagewiththe Ho-lan, see ibid., pp. 34-35, n. 76, 66, 75
and 110-12. Data on imperialmarriagesduringT'ang maybe foundin the chapteron
distaffkinin Chiù T'ang-shu183,pp. 4721-51; in workssuchas TCTC; and in thestudy
by KarenTang, "EmpressWei, ConsortShang-kuanand the PoliticalConflictsin the
Reignof Chung-tsung" of BritishColumbia,1976).On Northern
(M.A. thesis,University
Sung,see Sung-shih 248, pp. 8771-90; and PriscillaChingChung,Palace Womenin the
NorthernSung (960-1126) (Leiden: Brill, 1981).
wives.62Even thoughthiskindof alliancewas meantto be symbolicof

a pact betweentwo equal, independentpowers,it was actuallya mecha-
nismof control:sons of Chinggisidprincessessucceededto the leader-
shipof theseother"states";and thewomenthemselves governedinternal
affairsbothas regentsforyoungor absentrulers,and as widowsin the
periodbetweenthedeathof one rulerand theaccessionof another.The
wayin whichbestowaland exchangeworkedto thedetriment of theother
powerhas been discussedin detailelsewhere.63
Here it sufficesto indi-
cate the relationship
betweenmarriageallianceand the processof con-
quest and expansion undertakenbytheChinggisids in thethirteenthcen-
tury.
In the systemdevelopedby ChinggisKhan, marriagewas an inte-

gralpartof a feudalcontractinvolving
militaryservicein returnforlegit-
imationand protectionof thetraditional i.e.,it becamea tool
leadership,
forproviding themilitary for
supportnecessary conquestof trulyhostile
groups.In thisrespect, forcibletakingof wivesand daughtersfrom
the
leadersof stateswhichattemptedto resistthe Mongol advance has no
bearingon thecontractsdiscussedabove sincethetraditionalleadership
was destroyed.Marriagewas thusa once-onlyphenomenonfromwhich
no regularsystemof bestowalor exchangecould ensue.64
In sum,marriage-exchangeand princess-bestowal
in Mongolsociety
duringthethirteenthcenturyprovidedthemilitarywherewithal fora suc-
cessfulprogramof rapidexpansion.The resultwas a relatively
decentral-
ized empireheld togetherby an extensive
networkof enforcedmarriage
62) ExchangedevelopedmainlywiththeOnggiratand Ikirasand to a lesserextent

withthe Baya'ut and Oirat.One-waybestowalsdevelopedwiththe Uighurs,Önggütand
Arulatand perhapswiththe Olqunu'ut.
63) See J.Holmgren,"Observations on Marriageand Inheritance Practicesin Early
Mongol and Yuan Society,withParticularReference to the Levirate,"Journalof Asian
History20,2 (1986): 127-192 (part 1.3: "Simultaneousexchangeand preferential mar-
riage";part1.11:"Levirateas a politicalstrategy";
and partII. 5: "Inheritance
and succes-
sion").
64) Examplesbeingthe forcibletakingof wivesand daughtersof Kereyitleaders
(A.D. 1200),and of theTartar(A.D. 1202),Naimanand Uduyut-merkit (A.D. 1204),and
Tangut(A.D. 1227).
contracts.The whole systemwas geared to continualrapid conquest.

However,sincemaintenance of themarriagesystemhad to be backedup
bythreatof forcedintervention in theaffairsof each alliedgroup,crude
militaryforcewas essentialto the survivalof the empirein a way not
seen in the T'o-pa historyeitherin predynastic or latertimes.Nor did
this problemgreatlyaffectthe empiresof the Ch'i-tan and Manchu
peoples.
We have almostno information about Ch'i-tanmarriagearrange-

mentsin predynastic times,but can say withsome certainty thatthesys-
temof leadershipin the periodimmediately priorto the foundationof
Liao wouldhave negatedany military or politicalbenefitsderivedfrom
marriage alliance contractedat confederation level:as indicatedabove,
therewas littlepointto marriagealliancein systemsof leadershipwhich
operatedon the principleof electionand regularrotationof the head
of state.It is also doubtfulwhethermarriagescontractedat the tribal
levelregularly involvedallianceswithpowersoutsidethe confederation:
in thesole case wherewe knowthisoccurred,it led to rapiddestruction
of thetraditional systemof government - to theinstitutionof dynastic
ruleundera permanent head of statesupportedbytheforeign power(see
below).65
LiketheMongolempirein thethirteenth thatof theMan-
century,
chus blossomedthroughjudicious use of forceand marriagealliance.
However,in the periodimmediately priorto the foundationof Ch'ing,
the formof alliance mostlycontractedwithnon-Manchupowerswas
mainlyof theChinesekind(althoughmoreprolific)wherewomenother
than daughtersand sistersof the rulerwerebestowedon the partyin
question.66Only Mongol leaders, whose supportwas crucial to the
65) The marriagecontractof the 680s, discussedin Holmgren,"Yeh-lü,Yao-hen

and Ta-ho,"pp. 51- 52, was almostcertainly an intertribalor sub-confederationalliance
betweenKhitanleaders.
66) For example,in about A.D. 1600,Nurhaciordereddaughters of his "officials"
to be givenin marriageto variouschieftainsof the Hurkapeoples.This typeof alliance
was stillcurrentin the Ch'ien-lungera in the latterpart of the eighteenth
century, but
mayhave died out afterthattime.See RobertH. G. Lee, The ManchurianFrontierin
Ch'ing History(Cambridge,Mass.: HarvardUniversity Press, 1970),pp. 43-44.
of conqueringChina, wereregularly
enterprise givensistersand daugh-
in
ters marriage.67
In theory,Manchu-Mongolcontractsof earlytimesconformedto
the T'o-pa ratherthan the Mongol model of alliancein thatmarriage-
exchangewas meantto be symbolicof a mutualnon-aggression pact
betweentwoindependent unitsof roughlycomparablemilitary
capability
(see above). Yet these contractsalso involvedmilitarysupportby the
MongolsfortheManchucause. In thatrespect,theyfollowedthepattern
of alliancedevelopedby theMongolsin thethirteenth It is also
century.
clearthatin realitythesecontractsquicklyevolvedtowardstheChinggis-
id model of control(see below).
In sum,the fourcases discussedabove reveala similarbut by no
meansidenticalapproachto foreignmarriagealliance.Whileconditions
withinCh'i-tan societymade marriagecontractslargelyirrelevantto
foreignpolicy,the T'o-pa, Mongols and Manchus utilizedmarriage-
exchange,bestowaland acceptanceof womenas an integraland normal
partof foreign All threepowerssaw themarriagecon-
relationsstrategy.
tractas a tokenof goodwill,but each developedits own particularout-
look and approach. The Mongols emphasizedmarriage-exchange and
princess-bestowalratherthan acceptanceof foreignwomenas brides:
underthem,exchangeand bestowalbecame strategies crucialto a pro-
of
gram conquest and expansion.The T'o-pa used foreignmarriagealli-
ance in a morelimitedway - primarily as a defencemechanism:while
theirleadersindulgedin all threestrategiesand wereequally prepared
to giveor receivewomen,foreignmarriagecontractsweregearedto sur-
vivalratherthanto theprogramof expansion.Bothpowerswillingly sent
daughters and sisters to
away foreign states.In contrast,the Manchus
adopteda mixedattitudeand approach:theInnerAsianmodeofexchange
was used in relationswiththe Mongols; Chinese-style bestowalmostly
prevailedin othercases.
It would seem thatvariationsin the steppeapproachwerein part
the resultof cumulativeexperience,i.e., new techniquesand attitudes
67) See Ch'ing shih kao jf £jfi| 166 (Peking:Chung-huashu-chü,1977), pp.

5265-80; also Farquhar,"Originsof Manchus*MongolianPolicy,"and WangChen-main,
"The Ch'ing Dynasty."
weredevelopedfromtheexperienceof earlierregimesand fromcumula-

tivecontactwithChina. The followingdiscussionshowshow variations
in formand approachalso appear in the historiesof conquestdynasties
withinor alongtheChinesefrontier,
established and howthesetoo added
to thestoreof cumulatedtechniquesand experiences to be used by later
regimes.
The ConquestDynasties
Duringthetimeof thefirstemperorof Wei(r. A.D 398-409), therewas

a markedchangein T'o-pa attitudesto foreignmarriagealliance:impe-
rial daughterswereno longersentawayas bridesto foreignstates,and
foreignwomenwereno longerfreelyadmittedto the imperialharem.
However,thesystemwhichdevelopedwas not mereimitationof Chinese
practice, and newlyacquiredideas.
buta uniqueadaptationof traditional
Ratherthanbeinggivento leadersof independent hostile
and potentially
foreignpowers,daughtersand sistersof the emperormarriedinto sur-
renderedrefugeegroupsfleeingthose states.Once the recipientleader
had been assimilatedintotheWeipoliticalsystem,i.e.,
and his followers
whentheyno longerrepresented bestowals
a threatto internalstability,
ceased. In short,as withnativeChineseregimes,the marriageof Wei
imperialdaughtersconcernedmattersof internalratherthan external
emphasison foreign
import,yetthetraditional relationsremainedintact.68
In general,marriagecontractswith independentforeignpowers
ceased altogetherafterthe establishment of the Wei dynasty,i.e., no
womenwhetherdaughters,sisters,or more distantrelatives,weresent
awayas bridesforforeignrulers.This reflectsboththemilitarystrength
of Weivis-a-visothercontemporary powersand its aggressive
expansion-
thechangein termsof behaviourevinced
istoutlook.We could interpret
by native Chinese but it should be remembered
regimes(sinification),
that steppepowersdid not normallypropose,or accept proposalsfor,
68) Each defectorgroup received no more than one or at the most two princesses
because it representedonly a minimal threat to security - either because of its size or
because its defectionwas a sign of the imminentdisintegrationor conquest of the state
in question. See J. Holmgren, "Wei-shu Records on the Bestowal of Imperial Princesses
during the Northern Wei Dynasty," Papers on Far Eastern History 27 (1983): 21-97.
marriagealliancewhentheintention was immediate conquestand annihi-

lation.In otherwords,Weipolicyaccordedwellwiththetraditional ethos
which saw marriagealliance primarilyas a defencemechanism(see
above). In thisrespect,it is interesting
to see thattwo exceptionsto the
ruleof absencein Weitimesreflecta stepperatherthanChineseoutlook
and thatthisoutlookpresagedthe veryaggressiveapproachto foreign
marriagediplomacyadopted by Mongol rulersof the WorldEmpire.
The two exceptionsinvolvedalliances with a steppe power,the

Juan-juan,and withNorthernLiang, a statein Kansu province.Both
occurredin the 430s at the heightof an aggressiveexpansionistphase
whenWeistillwas ringedbya multitude of hostileand potentially
hostile
foreignpowers.Some groups,then, needed to be placatedbeforethecon-
questof otherscouldgo ahead. In spirit,therefore,thesecontracts
resem-
ble thetypeof allianceused so successfully by the Mongolsin the thir-
teenthcentury;but in formtheyadheredto T'o-pa traditionin thatthey
did notinvolvemilitarysupportbytheotherpartybutweresimplymutu-
al non-aggressionpactssealedby an exchangeof daughtersand sisters.69
The circumstances surrounding Wei alliances withthe Juan-juan

(A.D. 434) and Northern Liang (A.D. 430-37), and themannerin which
thesepacts werebroken,70 revealthatwhilenativeChineseregimessaw
thelengthof an alliancebeingdetermined by the life-spanof the bride,
thiswas clearlynotso in thesteppe:theremilitaryaspectsof thecontract
could be dissolvedwithoutreference to the marriage.In short,alliances
betweensteppepowerspersistedjust as long as the political,economic
or militaryconditionswhichled to theirinstitutionremainedintact.How
maritalaspectsof thecontractcould continuebeyondbreakdownof the
69) Ibid., pp. 48-51.

70) In A.D. 436, theJuan-juanbrokethealliancewhenWeiinvadedNorthern Yen
in Liaoningand was simultaneouslyconfrontedwithuprisingsin newlyacquiredterritories
in thewest.In A.D. 439, twoyearsafterhavingquashedtroublein thewest,drivenback
the Juan-juanand completedthe pacificationof Liaoning,Wei brokeits alliance with
Liang. WhileoccupiedwithLiang,it suffered anotherJuan-juaninvasion.Althoughthe
excuseforinvadingLiang was couchedin somewhatChineseterms(see SectionV), it is
clearthatin each case,themilitary
pactwas severedonlyafterchangingmilitaryand polit-
ical circumstances appearedto favourone of the partiesinvolved.
militarypact is discussedin SectionV. Here we merelynotethedifficul-

ties facedby any regimein attempting to forgean empirebased upon
a permanent systemof marriage-exchange betweenallies of independent
or semi-independent status.
Paradoxically,it was not the Mongols,but ratherthe Ch'i-tanand

to a lesserextenttheManchuswho weremostsuccessfulin findinga way
to use foreignmarriagecontracts as a controlmechanismin combination
witha centralized system of rulein dynastictimes.In returnformilitary
and politicalsupportagainst otherKhitanpowers(triballeaders and
paternalkin),Liao rulerspermitted thedevelopment of an exclusivesys-
tem of marriage-exchange betweenthe throneand the Uighurrelatives
of the firstempress,i.e., whatbeganas a foreignmarriagealliancecon-
ducted at tribal level in the period prior to the foundationof Liao
became a tool fortransformation of the traditionalpoliticalstructure
intoa centralized bureaucratic regimebased on theprincipleof a heredi-
tarymonarchy.The marriagesystemwhich so developedformedthe
backboneof the newstructure: suprememilitary and politicalauthority
was to be sharedin perpetuity betweenthe throneand seniormembers
of theconsortclan.71These marriagecontracts, then,wereused not for
conquestand expansionbut forinternalconsolidationof powerand the
emergence of a newChinese-style imperialgovernment fromtheold trib-
al confederation.
In the periodafterthe consolidationof imperialpower,marriages

involving members of theconsortclan clearlybelongto internalworkings
of the Liao empire,i.e., theycan no longerbe regardedas externalcon-
tracts.Whereallianceswithexternalpowerswereconcerned,the Liao
formwas identicalwiththat of the Chinese: daughtersand sistersof
rulerswerenevergivenawayin marriage(theywerereservedas wivesfor
highrankingofficialsfromtheconsortclan); and therewas no place for
foreignwomenin the imperialharemas principalwivesand empresses.
Thus marriagealliancewithotherstatesinvariably involvedbestowalof
distantrelativesratherthan exchangeor acceptanceof women.
71) The processof thisdevelopment

has been discussedin fullin J. Holmgren,
"Marriage,Kinshipand Successionunder the Ch'i-tan Rulersof the Liao Dynasty,"
Toung Pao 72 (1986): 44-91.
UnlikeLiao, the Hsi-hsiaformof foreignmarriagediplomacywas

close to the traditionalsteppemodel in thatTangutrulerswerewilling
to accept foreignwomenas brides.For this reason,and because Hsia
occupiedsuch a strategicpositionin relationsbetweenLiao and Sung
(see above),marriagecontractsnegotiatedbytheLiao courtmainlycon-
cernedHsia. The firstalliance(A.D. 989) was contracted underpressure
at the timewhenLiao was occupiedwitha southerncampaignagainst
Sung.72The ensuingtreaty, characterizedby a seriesof one-waybestow-
als fromLiao to Hsia, lasted for 50 years.It ended in bitterness over
the suddenand unexplaineddeath of one of the womenin A.D. 1038.
Twentyyearslater,Liao bestoweda woman upon the Tibetansin an
attemptto maintainthe statusquo duringa period of active Tangut
aggression.73 This was one of the fewcontractsmade withgroupsother
thantheTangutHsia.74Marriagealliancewiththelatterbeganagain in
A.D. 1105.Here,bestowalwas also aimed at maintaining the statusquo
- thistimebysupporting Hsia duringa periodof activeSungaggression
(see above).
From the above, it would seem that in spiritLiao allianceswere

similarto those conductedby the T'o-pa in pre-dynastic timesin that
they weregeared to defenceratherthanto conquestand expansion.Over-
all, then,Liao comes close to Chinesepatternsof behaviour. However,
whilethe formand in some respectsthe spiritof thesealliancescon-
formedto Chinesepractice,it can be arguedthatreluctance to giveaway
daughterswas less a matterof "sinification"than a functionof the
unique marriagesystemdevelopedby the foundersof the state.More
importantly, the relationshipwithHsia showsthatLiao rulerswerewell
72) See K. A. Wittfogel and Feng Chia-sheng,Historyof ChineseSociety,Liao

(907-1125) (Philadelphia,1949),p. 166, n. 24 and 583; Liao-shih 11,pp. 126-27; and
Liao-shih 12, p. 134.
73) Liao-shih18, p. 220; Kwanten,"Chio-ssü-lo(997-1065); and Sung-shih492,
p. 14162.
74) Liao refusedseveralUighurrequestsformarriagealliance.See Wittfogel and
Feng,p. 579 and Liao-shih 13, p. 148. In 996, a womanwas sentto Korea (Liao-shih
13, p. 147). In 1021,an alliancewas concludedwiththe Ta-shih~fc& Moslemsto the
westof theUighursin EasternTurkestan. Wittfogeland Feng,pp. 51,and 357; and Liao-
shih 16, p. 189.
aware of the potential of the marriage contract as a political/military

weapon, and that judicious use of that weapon helped to maintain the
status quo along the frontierfor more than a hundredyears. Liao under-
standing of the role of marriage alliance in steppe politics was also re-
sponsible for laws such as that of 1094 which attemptedto control inter-
marriagebetweenborder peoples and those under the rule of neighbour-
ing states.75
In sum, we can say that forboth Wei and Liao, the feudal contract,
underpinnedby princess-bestowalor marriage-exchange,played little or
no role in advancing the programof expansion or in holding the empire
together.Thus problemswithfragmentedcontrol,enforcedmarriagecon-
tracts,and the need fora uniformand permanentlyhigh militaryprofile,
were not significantfeaturesof T'o-pa or Ch'i-tan historyin dynastic
times. Yuan history,however,differsconsiderably on these points.76
Mongol rulers made spasmodic but concerted attemptsto replace

the traditional,fragmentedsystemof governmentwith a set of central-
ized bureaucratic controls. However, even though they were somewhat
successfulin Yuan times (in areas such as taxation, law and appointment
of personnel),thefeudalcontract,cementedbymarriagealliance, remained
an integralpart of the political and militarysystemwell into the four-
teenthcentury.77Thus princess-bestowaland marriage-exchangeinvolv-
ing the traditionalcircle of allies gradually begin to relate more to inter-
nal structureand politics than to foreignpolicy.The transformationfrom
foreign,externalalliances to internalcontractscan be said to have been
more or less complete with consolidation of the geographic boundaries
of the state by the end of the thirteenthcentury.
75) Liao-shih 25, p. 303.

76) For a recentreviewof the degree of centralizationof governmentin Yuan times,
see D. Farquhar, "Structure and Function in the Yuan Imperial Government," in J. D.
Langlois, ed., China under Mongol Rule (Princeton UniversityPress, 1981), pp. 25-55.
The problem of centralizationand bureaucratizationin Wei, Liao and Yuan and conflict-
ing scholarly views on this subject in regard to Yuan have been discussed at length in J.
Holmgren, "Political Organization of Non-Han States in China: the Role of Imperial
Princes in Wei, Liao and Yuan," Journal of Oriental Studies 25 (1987): 1-48.
77) Idem.
One of themoreclear-cutcases of marriagealliancewitha foreign

powerin Yuan timesconcernsrelationswiththe Kingdomof Korea, a
lateadditionto themarriagecircle.Yüan-Koreanmarriagerela-
relatively
tionshave been studiedin fullby Hsiao Ch'i-ch'ing.78 Here,it suffices
to pointout thateventhoughKorea regularly sentgroupsof womento
Yuan as entertainers, servantsand concubinesforthe imperialharem,
thesewomenwereby law not permitted to become principalwivesor
empressesof Yuan monarchs,79 and that the formof the relationship
followedcloselythe WorldEmpiremodel of one-waybestowalwhere
daughtersand sisterswereregularly givenin marriageto the rulerof an
alliedforeignpower as a control In short,Chinggisidprincesses
strategy.
sent to Korea duringthe Yuan acted as empresses,mothersof heirs-
apparentand regentssupervising internalaffairson behalfof the Yuan
court.As withsome of theinternaland moretraditional alliancesbegun
during the WorldEmpire,marriage withKorea was maintained bythreat
of militaryintervention in the internalaffairsof the state.
Marriagepracticesin Yuan timeswerethusmodeledon traditional

predynastichabitsdevelopedduringtheWorldEmpire.The lattersystem,
however, derivedfroma varietyof sources.In externalrelations,the ap-
proachapproximated to thattakenby Wei in the 430s, but the problem
of controllingexternalinterference in Mongol affairswas solved by
adoptingtheLiao idea of a singlepreferred consortclan in combination
withthe WorldEmpireidea of successionto powerof the mostcompe-
tentadult son. Persistenceof the traditionalmode of decentralizedcon-
trol,in particularthe role of in-lawsas fief-holders
in Yuan timesalso
helpedto limitexternalor distaffinterference in theaffairsof theimpe-
rial domain.
Fletcherand FarquharhaveobservedthatMongolprincesin Ch'ing

timesweremilitaryallies who occupieda special place overand above
theregularCh'ingbureaucracy and thatthispositionderivedfromInner
78) Hsiao Ch'i-ch'ing,"Yüan-Li kuan-hsichung te wang-shihhun-yin"(n. 3

above), p. 231.
79) The law was brokenwithsome difficulty by the last emperor.See JohnW.
Dardess,Conquerors and Confucians:Aspectsof PoliticalChangein Late Yuan China
(New York/London:Columbia University Press),pp. 92-93.
Asian ratherthan Chinesemodes of diplomacy(see above). This is so,

but it shouldbe emphasizedthatrulersof the Ch'ing,likethoseof Wei
bureaucratized
and to a lesserextentLiao, presidedovera relatively and
centralizedsystemof government, and that Ch'ing relationswiththe
Mongolswerein manyrespectsexceptional(see above). Moreover,rela-
tions with the Mongols did not remainstatic,but graduallyevolved
towardsthemoretraditionally commonLiao/Chineseformof bestowal.
Threephasescan be discernedin marriagetiesbetweentheimperial
Ch'ing house and Mongol princesfromthe seventeenth throughto the
nineteenth As
century. indicatedabove, theseallianceswereinitiallysym-
bolic of a pact formilitarysupport- a pact betweentwo equal and
independent powers.However,in the period followingthe conquestof
Chinatheybecamea mechanismforManchusubjugationof theMongol
population.80Finally,as the militaryrationaleforcontinuationof the
allianceswanedwiththegradualintegration of InnerMongoliaintothe
centralized state,thepatternof alliancesseemschangedyet
bureaucratic
again.
in theCh'ingshihkao suggeststhatafterconsolida-
Data presented
tion of the China-based empiretherewas a gradualdeclinein the fre-
quencyof allianceswiththe Mongol princes;an increasein the number
of cases whichinvolvedadopted ratherthan naturaldaughtersof the
Ch'ing monarch;and a decline in the numberof Mongol women
appointedas principalwivesand empressesforthe Ch'ing court.81In
otherwords,the decliningthreatposed by the Mongolsto the stability
of the empireseemsto havebeen matchedby a changein boththe fre-
quencyand formof thecontract:therewas a swingawayfromexchange
commontypeof
of daughtersand sistersback to the moretraditionally
bestowal.
Detailedstudyof all imperialmarriagesin Ch'ingtimesis needed
beforemorecan be said on thislastphase sincetrendsin theCh'ingshih
80) On thevariousmeansof controlling the Mongol populationin Ch'ingtimes,

see WangChen-main,"The Ch'ingDynasty,"pp. 73-81; and J. Fletcher,"Ch'ing Inner
Asia c. 1800,"in J. K. Fairbank,ed., The CambridgeHistoryof China,vol. 10, part 1
(London/NewYork/Melbourne, CambridgeUniversity Press, 1978),pp. 48-58.
81) Ch'ingshih kao 166, pp. 5265-301; 214, pp. 8859-933.
kao forthelatterpartof thedynastyare not veryclear.It is clear,how-

ever,thatthe highpointof marriageties withMongol princeswas the
periodimmediately priorto conquestand consolidationof the China-
based empire.It is also clearthatchangein frequency and formof these
alliancescan be equallyinterpreted in termsof traditionalManchu be-
haviouror theChineseapproachto theproblemof dealingwithtrouble-
some borderpeoples.As we have seen,similardifficulties in interpreta-
tion also occur in analysisof the behaviourof Wei and Liao rulersat
timeswhenprevailing conditionsin thoseempireswould have provoked
identicalor similarresponsesfromnon-Han and nativeregimesalike.
We haveseenthata cleardistinctioncan be drawnbetweenChinese

and non-Hanapproachesto foreignmarriagediplomacy;thata number
of different or variantformsand attitudesexistedwithinthe steppe
model of behaviour;and thatconsiderablevariationalso existedin the
behaviourof conquestdynastiesin China. We have also seen that the
behaviourof conquestdynastiescannotbe describedin purelyChinese
terms.Nor was it necessarilyrelatedto traditionalbehaviour.Wei and
Liao, forexample,adopteda radicallynewapproachto foreignmarriage
alliance.The Weiapproachhad its originsin steppe,althoughnot neces-
sarilyT'o-pa, tradition;thatof Liao, whichapproximated quite closely
to Chineseprocedure, derivedfroma judiciousmixtureof Chinese,non-
Han and newlydevelopedideas all of which lay outside the native
Ch'i-tantradition.In contrast,
theYuan systemseemsto havebeensolid-
ly based on traditionalMongol practices,but in factsome of thosepre-
dynastictraditionsclearlyhad theiroriginsin theconductof earliercon-
quest regimes such as Wei and Liao. Manchu/Ch'ingpracticesseem to
havebeensimilarly builtupon a cumulativebutselectivetradition
derived
fromthe experienceof other,earlierregimes.
The finalsectionof thispaperprovidesan explanationforChinese

objectionsto marriagediplomacy,in particularChineserefusalto accept
womenfromotherstatesas bridesforthe emperoror his offspring. It
showswhythis was generallynot an issue in relationsbetweensteppe
powers,and explainswhyearlyemperors of Weiand Liao weremoreada-
mantthan those of Yuan and Ch'ing in rejectingthe entryof foreign
womeninto the imperialharemin dynastictimes.
V. SOCIAL BACKGROUND
Twoconditionsgovernedtheapproachto foreignmarriagediplomacyin
Chineseand steppesociety:marriagepracticesin generaland thesystem
of successionto leadership(in theChinesecase,successionto thethrone).
Ideas associatedwithmarriagegovernedwillingness to sendwomenaway
to foreign the
states; system of successionin combinationwithcertain
-
elementsin the marriagesystem determinedwillingnessto accept
foreignwomenas brides.
Two elementsin particularwerecommonto nearlyall non-Han

societieson China's northern
frontier:polygamousmarriage,and a sys-
tem of successionwhich excludedrule by minorsand incompetents.
Together, thesetwo elementsfacilitated
the operationof a complexsys-
tem of foreignmarriagediplomacybased on multiplesimultaneous
exchangeof daughters.Theyprovidethe basic patternof contrastwith
conditionsin China.
We have seen how the Ch'i-tansystemof leadershipat confedera-

tionlevel - based on a mixtureof regularrotationof office,fixed-term
appointments and election - did not permiteasy developmentof a
coherentforeignrelationsstrategybased on marriagediplomacy.For
othersteppepowers,such as the Hsiung-nu,Turks,T'o-pa, Manchus,
and Mongols(aftertheestablishment of theWorldEmpire),thiswas not
a problem.For them,leadershipwas based on the principleof lifetime
rule ratherthan ante-mortem succession.Withinthis largergroup of
peoples,a variety of succession systems butin essenceeach was
prevailed,
geared to some extent to the principle electingthemostcapable male
of
withina well-defined kingroup.Differences lay chieflyin determination
of the kinshipgroup.Since each rulercame to poweras an experienced
and competentadult,therewas no place in most of thesesocietiesfor
regenciesdominatedbymothersor wives,maternalunclesor in-laws,and
thusno reasonto fearmarriagealliancewithoutsidepowersas a poten-
tial threatto independence(see below).
As indicatedabove,theothergeneralfactorwhichfacilitated
accep-
tance of foreignwomenas bridesby steppeleaderswas polygamy, or
ratherthe absenceof truemonogamy.Polygamyhad severaladvantages
overmonogamy.First,the foreignbridewas easilyassimilatedinto the
familywithoutissue overherstatus,i.e., polygamy offeredno barrierto

multiplealliancesbut rathermaximizedthe potentialfora wide range
of contactswitha varietyof otherpowers.Secondly,the foreignbride
could noteasilyriseto a positionabovethatof otherwives.In part,this
explainsacquiescencein thedeceptionoverthestatusof Chinesebrides:
for steppepowers,the natal statusof the woman was probablyof no
more importthan the questionof her place in the husband's family.
Finally,equalitywithotherwomenin the husband'sfamilymeantthat
sons of foreignwomenwererarelyif evergivenpreference in succession
procedure.When they did come to power, it was usuallylong afterthe
death of seniormaternalkin (see above).
T'o-pa historyprovidesa classic illustration of the operationof

theseprinciples:in theperiodimmediately priorto the foundationof Wei
(c. A.D. 277-377), theT'o-pa followed
the "Turkic" mode of succession
theoldestmostcapablemalein thecollateralline.Alongwith
in selecting
an absenceof monogamy, the systemof fraternal successioneffectively
blockeddistaffand in-lawhopes forsustainedpoliticalinfluenceon the
leadership.82A radicalchangein successionprocedurein the periodaf-
terthe foundationof Wei was to alterall this (see below).
Mongol and Manchuhistoryprovidetwo interesting variationson

the above theme.Underthe systemof leadershipand successionestab-
lishedby ChinggisKhan in the thirteenth century, therewas a limited
-
role in the Mongol empireforwomenas regents not foryoungor
immaturerulersas in China, but duringthe interregnum betweenthe
deathof one Khan and theaccessionof another.In theory, femaleregen-
cies wereto last no morethana fewmonths;in practice,because of dif-
ficultiesin recallingdispersedtroops,theymightlast foryears.At this
time,therulingelitealso beganto adopt a systemof monogamousmar-
riage:onlysonsof theprincipalwifewereeligibleforelectionto theposi-
tionof Khan. In thisrespect,it is interesting
to see thatemphasisduring
theWorldEmpirelay withthebestowalof womenratherthanwithmar-
riage-exchange or acceptanceof foreignbrides.Simultaneousadoption
82) Holmgren,Annals of Tai, p. 34; and J. Holmgren, "Women and Political Power
in the Traditional T'o-pa Elite: A PreliminaryStudy of the Biographies of Empresses in
the Wei-shu,"MonumentaSerica 35 (1981-83): 33-74.
of the Liao idea of a single,preferredconsortclan of Mongol origins

also restricted forexternalinterference
possibilities in leadershipmatters.
The mostimportant defence,however, lay in retention of thetraditional
idea of electingthemostsuitableor experienced leader:althoughthecir-
cle of heirswas now restrictedto sons of principalwives,the emphasis
was stillon maturity and capabilityratherthansuccessionof the eldest
son as occurredin China.83
LiketheMongolsand Chinese,theManchusemphasizedsuccession
in thedirectline.Therewas, however, no attemptin pre-Ch'ingtimesto
distinguishbetweensons accordingto the statusof the mother.Rather,
the emphasiswas on competenceand leadershippotential.Nurhaci(d.
A.D. 1626),however, finallyoptednot fora singleheir,but forjoint or
collectiverule:all sons wereto participatein government as membersof
a supremecouncilof rulers,withthe councilbeingchairedby a trusted
outsiderelectedbytheprincesthemselves. The conceptof collectiverule
remainedtheideal but was in facttemperedby therealityof differences
in capabilityand personality, i.e., collectiverulealternatedwithand was
to some extentmodifiedby the "Chinese"idea of a singleheir-apparent
appointedduringtheruler'slifetime. In thelattercase, theemphasison
leadershippotentialremainedintact.More importantly, whereregents
governedfor minors,theywereinvariablyprincesof the rulinghouse
ratherthan wives,mothersor maternalkin.84
As studentsof Chinesecultureare aware,the Chinesesystemof

successionwas verydifferent both in formand spirit:there,the throne
to
passed directly the eldestson of the principalwife.The emphasison
monogamy(withconcubinage)and primogeniture a general
necessitated
acceptanceof regencygovernment, and it was the empressor principal
wifewho, in conjunctionwith her own kin, played a crucial role in
government affairsduringthe reignof a young,sick or incompetent
83) For furtherdiscussionon thesepoints,see J.Fletcher,

"The Mongols:Ecologi-
cal and Social Perspectives,"
HarvardJournalof AsiaticStudies46,1 (1986):24-28; also
Holmgren,"Observations on Marriageand Inheritance Practices,"parts1.3; 1.8 and 1.11.
84) This conditionlastedwellintothe Ch'ing dynasty.See R. B. Oxnam,Ruling
fromHorseback:Manchu Politicsin the Oboi Regency,1661-1669 (Chicago/London:
Universityof Chicago Press, 1975); also n. 87 below.
monarch.85 In thisway imperialmarriagesin China had an important

of power.Thus,giventhehighrateof infant
bearingon thedistribution
mortalityand otherunpredictablefactorssuch as humaninfertility,
the
Chinesehad considerablereasonto rejectout of hand any demandby
externalpowersthatthe emperoror his heiraccepta foreignwomanin
marriage.
The foundersof Weiand thoseof Liao made a deliberatedecision

to replacethetraditional systemof successionwithprimogeniture. While
reasonsdiffered, the effectwas the same: each statewas suddenlyfaced
withthe "Chinese" problemof how to cope withmaternalor distaff
influencein government in theeventof successionby a minor.The Liao
responsewas to developthe conceptof monogamyand regencygovern-
mentto a pointwheredistaffinfluencewas restricted to one favoured
consortclan. In contrast,Weiattempted to retainthetraditionalempha-
sis on equal statusforall wives(polygamy) and to developa set of rules
and tacticswhichwouldcompletely excludedistaffinfluenceon thethrone.
In each case, thechiefconcernwas withdistaffratherthanfemaleinflu-
ence,i.e., withthe powerof the mother'snatal relatives.86 It was under
these circumstances that both states to
attempted denyforeignpowers
accessto theimperialharemin dynastictimes:underthemoreaggressive
and successfully expansionist Weiregime,therewas a markedabsenceof
foreignmarriagealliance; underLiao, denial of access to the imperial
haremwas achievedby emphasizing one-waybestowalsfromLiao to the
statein question.
85) On the legalityof femaleregenciesin China, see Yang Lien-sheng,"Female

Rulersin ImperialChina," in J. L. Bishop,ed., Studiesof Governmental Institutions
in
ChineseHistory(Cambridge,Mass.: HarvardUniversity Press,1968),pp. 55-169. For a
specificcase see
study, Chung, Palace Women.
86) On Wei,see J.Holmgren,"The Haremin Northern WeiPolitics,398-498 AD:
A Studyof T'o-pa Attitudes Towardsthe Institution
of Empress,Empress-Dowager and
Regency Governments in theChinese DynasticSystemduringEarly Northern Wei,MJour-
nal of theEconomicand Social Historyof theOrient26,1(1983):71-96; and Holmgren,
uWei'ShuRecordson the Bestowalof ImperialPrincesses,"pp. 54-55; on Liao, see
Holmgren,"Marriage,Kinshipand Succession."
Unlikethe foundersof Wei and Liao, those of Yuan and Ch'ing

made no dramaticchangeto successionprocedure.87 In part this was
becauseMongolsand Manchuswerealreadyfamiliar(in theirowndiffer-
ent ways)withthe conceptof successionin the directline and the idea
of regency
government (see above). Thus thesystemof successionin dy-
nastictimesevolvedmoreslowly,almostimperceptibly, towardstheChi-
nese model. For thisand otherreasons,therewas greatercontinuity in
traditional
marriagepractices, includingacceptanceof theidea of admit-
tingforeignwomeninto the imperialharem.
WhenLiu Ching gljgfc of theFormerHan dynastyadvocatedmar-
riageallianceas a long-termstrategyforcontrollingtheHsiung-nu,88 he
was apparently unawareof how thevarioussystemsof successionin the
steppeprotected thoseregimesfrommanipulation by in-lawsand mater-
nal kin: unlessone could enforcemonogamousmarriagecoupled with
successionof a son of the principalwife(as did the Mongols in their
WorldEmpire),princess-bestowal could only be used as a short-term
strategy to gain temporarymilitaryrespite.As we have seen, non-Han
powersrealizedand acceptedthis - a matterwhichled to considerable
frustrationand angeron China's partwhenever she witnessedthe failure
of marriageallianceto providelong-term solutionsto contemporary po-
liticaland militaryneeds.
Loewehas arguedthatprincessessentto thewesternregionsin Han

timeswerein effecthostagesholdingthe courtto ransomin the same
waythatthe Chineseheld sons of Asiatickingsas hostagesin Ch'ang-
an.89To some extentthiswas truewhererelationswiththeWu-sunwere
concerned,but in generalthe hostageeffectwas chieflyan artefactof
the Chinesemind,in particularof the Chineseview of marriage.
Earlierstudieshaveshownhowmarriagepracticesin China,in par-

ticularthegrantingof dowryto thebridebyherown family,encouraged
87) On successionduringYuan, see Holmgren,"Observationson Marriageand

Inheritance Practices,"partII. 5. On successionduringearlyCh'ing,see Oxnam,Ruling
fromHorseback-, and Silas H. L. Wu, Passage to Power:K'ang-hsiand His Heir Appar-
ent, 1661-1722(Cambridge,Mass.: HarvardUniversity Press, 1979).
88) Shih-chi£!E 99 (Peking:Chung-huashu-chii,1975),p. 2719.
89) Loewe,"Introduction" to Hulsewé,China in CentralAsia, p. 60.
continued ties between a woman and her natal relativesafter marriage.

These ties found concreteexpressionin the practiceof permittingwomen
to returnto their own familiesafterdivorce in widowhood; in the habit
of fundingthe educational expenses of a sister's son; in support given
by wives to their natal kin; and in the mourning duties of a married
woman for her own parents.90The idea of continued material and psy-
chological support by parents and siblings after marriage underpinned
the belief that princess-bestowalcommittedthe state to a long-termrela-
tionship - to a pattern of political and militarysupport which could
only be broken by death (or divorce).
While the rulingelite in China werepreparedto sacrificetraditional

ideals and values about proper formin marriage (see below), repugnance
at the thoughtof a complete physical, psychological and cultural break
between the woman and her natal kin/homelandwas a major factor in
Chinese reluctance to indulge in foreignmarriage diplomacy. This can
be seen in remarksmade by T'ang T'ai-tsung upon receiptof two women
fromKorea: theirdispatch back to Korea may have involvedpolitical con-
siderations,but sympathyfor their plight was genuine.91It is this sym-
pathy which gives credence and popularityto the literarytheme of exile
for women condemned to such marriages in Han and T'ang times.
Apart from reluctance to send women away to foreignstates, the

Chinese view of the proper relationshipbetween a married woman and
her natal kin was responsibleforanotherwell-knownphenomenon - the
lavish giftsand presentsregularlybestowed upon foreignrulerswho had
been given women in marriage. Much has been said about the economic
relations between China and her neighbours. In the main, these studies
have concentratedon tributaryrelationsin Ming and Ch'ing times, with
the assumption that the relationship was controlled by China from a
position of strength.Certainly,this was so for the Ch'ing, but for earlier
regimes the situation is less clear: studies dealing with pre-Ming times
90) See PatriciaB. Ebrey,"Womenin the KinshipSystemof the SouthernSong

UpperClass," HistoricalReflections
8,3 (1981):117-25; and J.Holmgren,"The Economic
Foundationsof Virtue:Widow-Remarriage in Earlyand ModernChina," TheAustralian
Journalof ChineseAffairs13 (1985): 4-14.
91) Chiù Tang-shu 199A, p. 5326.
usuallyplace giftsand tradingfacilities,

along withbestowalof women,
underthecategoryof military weakness.The twoactivities, however,are
generallyseen as separateor distinctformsof appeasement.92 Jagchid
Sechin'sworkgoes somewaytowardsrectifying this:as he pointsout,the
magnificentdowrieswhichaccompaniedChinesebrides,theartisansand
labourerssentto maintaintheircustomarystandardof living,and the
annual giftsof silk and agricultural
producesentto recipients, werein
facta largepartof the attractionof an alliancewithChina forsteppe
powers.93
The economicattractionof marriagealliance withChina lay not
onlyin the different
economiesof China and the steppebut also in the
factthatmarriagerelationsbetweensteppepowerswere,materially, often
in
veryunprofitable: T'u-yü-hun,earlyMongol,and othersteppesociet-
ies whichoperatedon theprincipleof bride-price
(wherethegreaterpart
of marriagewealthpassed fromthe husband's familyto that of the
bride),takinga wifecould be an expensiveendeavour.94 This is why
independentpowers of similar or equal status in the steppe often
in
indulged exchangeof daughtersratherthan one-waybestowal.
Whereone-waybestowalwas associatedwithcontrol- as in par-
ticularlystrongand aggressiveempires- the wide statusgap between
recipientand donor led to a curiousanomaly:here,the notablyhigher
statusof thebride'sfamilynecessitated
institution
of a dowry-like system
in whichmarriagewealthpassed chieflyfromthe bride'sfamilyto that
of the husband.This made marriagedoublyunprofitable forthe ruling
line in thata substantialdowryhad to be suppliedforthe marriageof
a daughteror sister,whilethetraditionalbride-price
stillhad to be paid
forincomingwomenfromallied groups.The best thatcould be hoped
forwas mutualexchangewhichobviatedthe need forbride-price, but
92) Barfield,forexample(n. 59 above),passes overprincess-bestowal in his study

of economicfactorsin therelationship betweenChina and organization of theHsiung-nu
empire:marriageallianceis attributed to desirefor "prestige."
93) JagchidSechin,"Patternsof Tradeand Conflict,"pp. 183-84.
94) For a detailedstudyof the social and politicaleffectof bride-pricein early
Mongol society,see Holmgren,"Observationson Marriageand InheritancePractices,"
part I.
thehigherstatusof the rulinghouse limitedthisformto all but one or

two privilegedgroups.In otherwords,therewas rarelyany directeco-
nomicgain in marriagediplomacyforstronger powersin theirrelations
withsubordinate steppepeoples.This was not so whereChina was con-
cerned:there,steppeleadersfoundthatacceptanceof womenas brides
was a highlyprofitable As we haveseen,somepowersdeliber-
enterprise.
atelyexploitedthisfactbydeveloping policybased on low levels
a foreign
of violencecalculatedto cowerthe courtinto a marriageagreement.
In T'ang times,the Chinese made some attemptto exploitthe
steppetraditionof paymentby the recipientgroup,and it would seem
thatsubstantial giftsof cattleand horsesdid sometimespass to thestate
fromthesteppepowerin question.In general,however, Chinesedemands
for paymentwereexcessiveand wereclearlydesignedto dissuade the
steppe powersfrommarriagealliance altogether.95 The fact that this
rusewas used onlysparingly indicatesthattheChineserealizeditslimita-
tions: it workedonly when the proposedrecipientperceivedthat his
military positiondid not allow forsuccessfulharassment of the border.
In suchcases,theChineseusuallypreferred to dropanypretenceat nego-
tiation(see above).
In marriagesystemswhichoperatedon principlesassociatedwith
thetreatment
bride-price, of a wifeaftercompletionof themarriagecon-
tracttendedto be seen as an internalmatter,i.e., marriagesevereda
woman'sties withher natal kin. Full integration into the new clan -
oftenas a property-owning memberof her husband'sfamily- meant
thatshe did not returnto live withher own relativesafterwidowhood
but remainedwithherhusband'skineitheras a singlewomanor as wife
A woman'sloyalties,then,
to one of his youngerbrothers(levirate).96
laywithherhusband'sclan,and herpositionremainedlargelyunaffected
by externaldevelopments involvingnatal kin. Accordingly, eitherparty
in a steppealliancewas freeto dissolvepoliticalor militaryaspectsof
the pact as circumstancesdictated.
95) See n. 36 above and the case described in TCTC, p. 6236 (A.D. 646) where
the Chinese demanded secession of five city states subordinate to the Western Turks.
96) This is discussed in full in Holmgren, "Observations on Marriage and Inheri-
tance Practices," parts I.I; 1.9 and 1.10.
The conditionsdescribedaboveappliedbothto groupswhichprac-

tisedbride-priceand to thosein whichother,variantmarriagepractices
led to a similarlytightdegreeof culturaland physicalintegration into
the husband'sfamily.Acceptanceof the idea of a completebreakwith
natalkinmade steppepeoplesmoreamenablethantheChineseto giving
daughtersand sistersawayto independent foreigngroups.However,we
mustnot assumethatall steppesocietiesfollowedthe same patternof
development, nor thatstepperulersnecessarily
took littleor no interest
in the fateof theirdaughtersaftercompletionof themarriagecontract.
The behaviourof Mongolrulersin theperiodaftertheestablishment of
the WorldEmpireshould be warningenoughagainstsuch an assump-
tion.
Mongol behaviourduringthe WorldEmpireshowsthattradition

oftencountedforlittlein the face of pragmaticpoliticaland military
needs: on the one hand, rulersof the WorldEmpireexpectedto retain
thetraditionalloyaltyof wivesto thehusband'sfamily;on theother,they
attempted to institutea systemof controlin whichdaughtersand sisters
wouldremainloyalto theimperialinterest evenaftermarriage.As indi-
catedabove,thislatteridea was one of thecorner-stones of theEmpire.
It did, however,completely flouttraditional it was onlyin later
thinking:
times,withthe introduction of ChinesemarriagepracticesintoMongol
society thatthe idea of supportby and loyaltyto natal kin became an
integralpart of the Mongol ethos.97A similardisregardfor tradition
can be seen in Chineseattemptsto use the otherwiseabhorredpractice
of leviratein theirsearchforbettercontrolover steppepopulations.98
Whererulersof the NorthernWei wereconcerned,attitudeswere

at leastconsistent:
on theone hand,womentakenintotheimperialhar-
em wereseen as outsiderswhose loyaltiesprobablylay withnatal kin;
97) See ibid., partsII and III forYuan and post-Yuandevelopments in Mongol
society.
98) Examplesof Chinese"princesses"beingorderedby the courtto marrytheir
husband'ssuccessorare numerous.See n. 15 aboveon theT'ang; nn. 20-21 on Northern
Chou and Sui; and Loewe,"Introduction" to Hulsewé,China in CentralAsia, p. 62, on
theHan period.On Chineseabhorrence of thecustom,see Holmgren,"Observations on
Marriageand Inheritance Practices,"partsII.6 and II.7.
on the other,daughtersand sisterswereexpectedto remainloyalto the

imperialinterestaftermarriage.The excuseforbreakingthealliancewith
NorthernLiang in 439 (see above) - thattheemperor'ssisterhad been
badlytreated- was in basic accordwiththisapproach.Unfortunately,
it is uncertainwhetherthe emphasison continuingties withnatal kin
aftermarriagederivedfromthinking relatedto marriagepracticesin pre-
dynastictimes,or whetherit was a functionof changingcircumstances
associatedwiththe introduction of primogeniture.
VI. CONCLUSION
Irrespective in the numberof foreignmarriagealliances

of fluctuations
consideredor contracted by nativeChinesestates,a consistentapproach
to thisissue can be discernedacross the entireimperialera fromHan
timesthroughto the end of Ming. That approachmaybe summarized
as follows:alliance was neverconsideredwhenthe statewas confident
of beingable to cope withmilitary threat;multipleallianceswereavoid-
ed; real daughtersand sistersof the emperorwererarelygivenin mar-
riage;and thestatestrenuously resistedanyattemptto haveforeign wom-
en enterthe imperialharem.
Foreignmarriagediplomacydid not takeplace betweenChina and

her neighboursin timesof stabilityand equilibriumalong the frontier
(NorthernSung). Nor did it take place when the power differential
betweenChina and the major opposingpartywas relatively large.Here,
two phenomenamust be distinguished: non-Han regimeswhich used
militaryforceforperiodic,indirectexploitation of the Chineseeconomy
(the Hsiung-nu); and strongerpowers intent on direct exploitation
throughconquestand permanent territorialacquisition(the Mongols).
Normally,only the formerhad an interestin marriagealliance. Thus,
because of limitationsimposedby both parties,alliancesoccurredonly
whenChinawas confronted bya unitedhostileleadershipof comparable-
military
to-slightly-superior capability.The smallerthepowerdifferential
betweenthe two parties,the greaterthe chance of agreement(Former
Han and Sui). When conditionswerecomplicatedby the possibilityof
coalitionbetweenvarioussteppegroupseach of whichrepresented a seri-
ous nuisancefactorin its ownright,thelikelihoodof a Chinese-initiated
proposal was even higher:the greaterthe numberof groups,and the
betweenthem,the higherthe chance of

smallerthe powerdifferential
in
Chineseinvolvement marriagediplomacy(T'ang).
The frequency of bestowalin Han, Sui and T'ang times,then,did

not derivefroma mind-set uniqueto theearlyimperialperiod.Nor was
absenceof bestowalin latertimestheresultof a radicalchangein tactics:
any significant
developments in the Chinese approachto dealingwith
foreignpeoplesmainly occurredduringearlyFormerHan. In brief,for-
in
eignmarriagediplomacy earlytimeswas theproductof a specificset
of conditionswhich,fora varietyof reasons,neverappliedin the post-
T'ang era.
Althoughthe Chinese consistently refusedsteppe proposals for

marriagealliancewhenever the statewas confidentof its abilityto cope
withperceivedthreat,the equationbetweenbestowalof womenand dy-
nasticstrength is hardlystraightforward:absenceof bestowal,forexam-
ple, could be relatedto confidencein militarysuperiority (Northern
Chou/Suirelationswiththe TurksbetweenA.D. 575 and 595), extreme
weakness(LaterHan relationswiththeHsien-pi),or simplybe a function
of conditionswhichmade marriagealliancetechnically unfeasible(Wei/
Western Chin and Ming). For theseand otherreasons(discussedabove),
absence of marriagealliance is a highlyunreliablemeasure of the
strength of a particularregimein relationto its predecessors.
WehaveseenthatT'ang and Sungbothfaceda uniquesetof condi-

tionsalong thenorthern and thatwhenfullaccountis takenof
frontier,
all the circumstancesfacedby each regimein its dealingswithfrontier
peoples,the strength of the Ming dynastyis betterjudged againstthe
period between the fall of LaterHan and the collapse of WesternChin
thanagainsteitherHan, T'ang or Sung. In sum,then,thephenomenon
in Chinesehistory
of princess-bestowal cannotbe used to indicaterelative
strength acrossa numberof successiveregimes.It is, however, an effec-
tivebarometer of perceivedchangein thebalanceof powerwithina nar-
row,well-defined temporaland spacial context(videT'ang relationswith
Nan-chaobetweenthe780s and 880s,or Sung relationswithLiao in the
1040s): here,we findthata proposal(or agreementto a proposal) for
marriageallianceis oftenmorea functionof perceiveddeterioration in
militarycapability within an overallcontextof strength(Sung in the
1040s)than a sign of long-term, absoluteor extremeweakness.
Steppeand Chineseapproachesto foreign marriagediplomacywere

determined bymarriagepracticesin generaland bytheparticularsystem
of successionto leadership(thethrone)in each society.Marriagepractic-
es in the generalpopulationgovernedthe degreeof willingness to send
womenawayas bridesto otherstates;thesystemof successionin com-
binationwithcertainmarriagepracticessuchas polygamy or monogamy
- determinedwillingnessor reluctanceto accept foreignwomen as
brides.
Monogamyand primogeniture, along withregenciesdominatedby

maternalkin, underpinnedChineseresistanceto the idea of accepting
foreignwomenintotheimperialharem.Reluctanceto sendwomenaway
to otherstatesderivedfroma peculiarlyChineseviewof marriagewhich
led to thebeliefthatsuchalliancescommitted thestateto unconditional
supportforat least the durationof the woman's lifetime.In earlyHan
times,it was hoped thatdisadvantagesinherent in such long-term com-
mitments wouldbe offsetby major benefitsderivedfromcontrolof the
foreignleadership.The idea was premisedon social and politicalcondi-
tionswhichappliedonlyin Chineseculture.Accordingly, it failedto take
into accountthatcontrolmechanismsbased on in-law and distaffrela-
tionsneededto be supportedby a constantand uniformly highmilitary
profile(vide the Mongol World Empire). The failureof early Han alli-
ances to measureup to expectation,along with non-Han refusalto
respecttheconceptof permanency in anyrelationship,added to Chinese
warinessabout the value of foreignmarriagediplomacy.
Non-Haninterest in marriagealliancewithChina derivedfromtwo
sources:the factthatmarriagecontractsformeda normaland integral
partof foreignrelationsstrategybetweensteppepowers;and thematerial
benefitsto be gainedin acceptingChineseratherthansteppewomenas
brides.UnlikeChina,marriageand successionpracticesin steppesociety
acceptanceof foreignwomenas membersof the rulingclan
facilitated
and ensuredthatleadersweremorethanwillingto send daughtersand
sistersawayto foreignstates.Steppepowers,therefore, indulged
regularly
in bestowal,acceptance,and exchangeof women.The steppeformof
marriagealso permitted multiplealliancesinvolvingsimultaneouscon-
tractswitha varietyof groups.WhereChina was concerned,however,
relationswerelimitedto one-waybestowalwherewomenpassed from
China to the state in question: for steppepowers,this was the most
economically advantageousformof alliance;it was theonlyformaccept-

able to the Chinese.
Steppemarriagepracticesalso engendered a relativelyflexibleap-

proachto the mannerin whichalliancesmightbe dissolved:in effect,
thelengthof an alliancewas determined notbythelife-spanof thewom-
an but bychangein thepolitical,military or economicconditionswhich
had led to its institution, aspectsof a pact could be broken
i.e., military
withoutreference to themarriage-state. Untiltheriseof theMongolsand
the development of theirunique systemof marriage-control in the thir-
teenthcentury, marriagealliance,then, tendedto be seen as a short-term
non-aggression pact which did not involvepoliticalor militarysupport
forenterprises undertaken bytheotherparty.In thissense,therewas no
limitto thenumberof allianceswhichcould be contracted withdifferent
states.Thesetwoissues - multiplealliancesand thelengthof a contract
- werea majorsourceof conflictbetweenChina and thevariouspowers
along the northernfrontier in Han, Sui and Tang times.
While Chineseattitudesto foreignmarriagediplomacyremained

constant,thoseof steppepowerschangedand evolvedwithtime,i.e., in
contrastto China, steppebehaviourwas not onlyselectivebut based on
a cumulative and highlyadaptivetradition.Severalsourcesof inspiration
can be pinpointed:1) attitudesderivingfromvariationsin nativesteppe
practices;2) patternsof behaviourdevelopedby conquestdynastiesin
theirattempts to come to termswithnewsocial and politicalconditions
in China; and 3) the gradualinfiltration
of Chinesemarriagepractices
intosteppesociety- especiallyfromaboutthesixteenth century.Although
the China connectionplayedan importantpart in the development of
newideas,no regime(whethersteppeor conquest)everadopteda purely
Chinese approach to foreignmarriagediplomacy:each group was in
some wayunique and each added somethingto the reservoir of experi-
ence whichunderpinned the development of new attitudesand ideas.
Summer1987

Maney Publishing

Uploaded by

Copyright:

Available Formats

Maney Publishing

Uploaded by

Document Information

Original Description:

Original Title

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

Maney Publishing

Uploaded by

Copyright:

Available Formats

Maney Publishing

A QUESTION OF STRENGTH: MILITARY CAPABILITY AND PRINCESS-BESTOWAL IN IMPERIAL

You may need to log in to JSTOR to access the linked references.

A QUESTION OF STRENGTH: MILITARY CAPABILITY

1) See, forexample,Chao Chun: Beautyin Exile' a novelby Shu Chiung(Hong

ars, however,fail to comment on the phenomenon other than in pass-

It is generallyacknowledged that at the time such marriages were

In the sixteenthcentury,the statesman Chang Chü-cheng jjgJgjE

based on the themeof WangChao-chün'smarriage:P. Demiéville,"Quelques traitsde

contractedduring Han and T'ang on a par with gifts(cash and goods)

Official Chinese scholarship of the late imperial era places foreign

5) Lien-shengYang,"HistoricalNotes on the ChineseWorldOrder,"in JohnK.

Mancaligroupsthenumerousand variedstrategies used bytheChi-

The presentstudyarguesthat irrespective of fluctuationsin the

8) M. Mancali,Chinaat theCentre:300 Yearsof ForeignPolicy(London:Collier

phenomenon of princess-bestowal seenas an effective

II. SCOPE OF RESEARCH

Giftsof servingwomenand femaleentertainers such as thosebe-

China byKorea,are mentionedonlyin passingwithreference to thesub-

III. THE CHINESE CASE

Whileit is generallycorrectto see the Han and Tang dynastiesas high

Sui alliancescontinueda trendbegunfiftyyearsearlierat thetime

to late Sui (534-614) was in manyrespectsmoreintensein regardto for-

Apartfromtwowomenbestowedupon rulersof the Wu-suncom-

Contraryto the popular perception,marriagealliances withthe

12) Serruys,"Sino-MongolRelations,"p. 18.

completion.During Tang, bestowalsbegan with the T'u-yü-hunand

13) ForT'ang bestowals,see thetablein Kuang,"Tang tai kung-chuho-ch'ink'ao"

ate was a fundamental conditionunderlying all alliances:therewas little

Timeand technological differencesalone invalidateanydirectcom-

16) See, forexample,thebackground to cancellationof a contractwiththeWu-sun

as LaterHan fellintodecline,and it was onlythefailureof Esen-tayiSi's

Ming and WesternChin also facedthe problemof managingnu-

17) On Tan-shih-huai'sempire,see K. H. J.Gardinerand R. R. C. de Crespigny,

So stringentwere the conditions under which native Chinese re-

Given the above, we can see whyoverturesmade by Turkishleaders

20) At thistime,a Chineseprincessgivento the Turksby NorthernChou (A.D.

preventTurkishcoalescenceinto two powerful,mutuallyhostilecamps,

In manyrespects,conditionsduringearlyTang wereripe for a

T'ang relationswithTurkishleadersalso show thatmarriagealli-

formin foreignmarriagediplomacy did not extendto multiple,simulta-

As well as unityand stabilityin leadership,therehad to be a clearly

Conditions under which most marriagealliances wereconsideredor

25) See J. Holmgren, "Yeh-lii,Yao-lien and Ta-ho: Views of the HereditaryPreroga-

At thistime,theChinesecan barelycontainthespreadof Tibetanpower.

27) For details,see Loewe,"Introduction" in Hulsewé,China in CentralAsia, pp.

whentherewas no dangerof, or threatinvolvedin, an alliancebetween

The finalphase of relationswithNan-chao(A.D. 876-883) high-

30) See Backus,"Nan-chaoKingdomand FrontierPolicy,"pp. 144-53 and 171

The factthatT'ang historycontainsas manypostponedand can-

T'ang reluctance to engagein anyformof marriagediplomacycan

Refusalto give away real princessesin T'ang timesmirrorsthe

32) See n. 36 belowforan earliercase of a cancelledalliance - thistimeinvolving

chosento marrya Hsiung-nuchieftainin Han timesbecause she was

Apart fromthird-party alliances occasioned by the conditions

35) For storiesassociatedwithWangChao-chün'smarriage, see n. 1 above.On the

38) See Loewe, Crisis,ch. 7; and Hulsewé,China in CentralAsia, p. 126 for

of theirdecline at the end of Later Han were invariablyturned down as