The revise the paper of the Eurasian nomad history

ETHNOGENESIS, C OEVO LUTIO N AND POLITICAL MORPHOLO GY OF THE EARLIEST STEPPE EMPIRE:

THE XIONGNU QUESTION REVISITED

N icoJ a D i C osmo

It is genera lly ass umed tha t the X iongn u were the firs t st eppe empire, b u t how d id su ch an em ­

p ir e come about '? Th is is , fo r I nn er A sian histo ry , a "found atio nal " q ues t ion, w h ic h , if an­

sw ered , w o uld cons idera bly advan ce the field of X ion gnu stu dies , an d more genera]]y tha t o f

In ner As ian hi sto ry. M y pu rp ose in this essa y is not to pro vid e a co m p lete su m m ary o f o ur pres ­

en t state of knowledge rega rdi ng th e Xiongnu, w h ich, as wil l beco m e ap pa rent , is sim ply im­

p ossib le to d o at th e p res ent time, bu t to h ighl igh t th e limitations of su ch kn owl edge and d isc uss

th e us eful ness o f certa in assum p tio ns an d theories .

In t he first part o f th is essa y I sha ll illu strate so me of th e issues that va rio us approaches to

Xiongnu stu d ies ent ail , focusing on h isto ry, archaeo lo gy, gen et ic stud ies, and anthropol o gy. In

th e second part 1 wi ll focus o n so me characteristic s o f the" m orp holo gy " o f th e Xiongnu em pire,

which have led me to qu estion th e usefu lness of the " coevolu tion" th esis, o r, in fac t, of any th esis

based o n the assu m p tion tha t front ie r d y namics between C hina and the X iongnu could be re ­

garded as th e ch ief "l ocu s" of the form ati ve stages of th e Xi o ng nu p o li tical culture and of its

im pe r ial evolu tion .

HI STO RY

Mos t o f w hat w e kno w ab out th e Xi ong n u as a h isto ri ca l p heno m eno n is based on C h ine se

sou rces , in pa rti cul ar t he "S hiji" , th e "Ha ns h u " and th e "Hou Ha nshu ", wr itt en between the

second cen tu ry BC an d the 5th century A D2. T he d esc ri ptio n o f th e Xio ngn u th at w e rece ive

fro m these writte n re cords essentially p rov id es no acco u nt o f the his to ry o f the X iongnu p rio r

to th e esta b lish ment of th eir em pire or the narratives of their re lat ionship wi th th e Q in and the

Former an d La te r H an d yna s t ies . Mentio ns o f the X io ngn u in other sources, such as the

"Zhanguo ce " (Stra tegies o f th e Warrin g State s) , are also ext remel y few.

C im me r ians and Scyt.hians w ere also po werful s teppe nomadic ar is rocracies amo unted to a poliricallv unified peo p les, but o n the bas is of t he extant sources we can ­ em pir e. A l the same t im e , it can by no means be ex­ not sa y that these fo rmed " empires" , if, by empi re , w e e lude d th at the po litica l cu lt u re o f th e Sc yt h ian s w as mea n a po liti cal ent ity t ha t g ro ws by in tegra ting its shared broad I)' aCrOSS the steppe reg ion . con sti tuenc ies int o a un ified and cenrralized st ructu re. 2 For a qu ick refe ren ce see E. I'. Wi lkinson, C h inese H is ­ Amo ng th em, the R oy al Scyrh ians menti o ned by to r)' : A M an ual. Ha rvard -Yen ch ing Instit ut e mo no­ H erodotus co ns tit u ted p robably an elite aristocrat ic graph ser ies 52 (Ca m b ridge, Mass ., Lo nd o n 200 0) s tratu m, but there is no evid ence th at the pr esence o f 78 1- 788.

36 NI CO L A DI C O SMO

The qu esti on o f their "o r igin " is represented in a hi stori cal nar rati ve that co nfla tes a t lea st

three types o f " p eo p les" . First a re th e peoples th at w ere [h e an cestors o f th e Xi ongnu as the

"h isto rica l ene m ies " of China (X u nyu, Xianyun, R on g, Di, etc .) , lo cated in th e north , wh o had

be en at w ar with the Zhou and var iou s states in th e remote p ast. Sec ondl y, we have [h e Xi on gnu

as a n eth nograp h ic " typ e", that is, the typ ical step p e nom ad s w ho lived in tents wi th no fixed

ab ode, [ended the ir anima ls, roamed [h e land in searc h of gr ass and water, and did not pract ice

agr iculture. The people who ac tua lly ca lled th emselves Xi ongn u fo rm the third type. Of them,

we know also th e na me s of th re e, and perhaps four clan s o f [he ro yal a r istoc racy, the person al

nam es o f the ir su p rem e chief, "shan yu " o r "chanyu ", and the titles of th e top po sit ion s in their

go ve rn me m . M oreover , th e Chi nes e w rit ten so u rces in clud e a co llec ti o n of "Xion gnu " words

th at have long been mined for hypoth eses on [he lin guis[i c and possibl y ethnic id en ti ty o f [h e

Xion gnu , rang ing from Sibe rian (Ke r) to Iranian and Tu rkic.

On th e methodological level, no aaemp[ has be en made thu s far [0 un ravel the me an ings of th e vari o us terms of id enti fication [h at we find in [h e Chi ne se sourc es in referen ce to [he

Xiongnu or related p eoples, such as Hu, Lin Hu, Dong Hu, Loufan, et c. , an d fo r [he rime being

we can say remarkabl y little about [h em. For ins ta nc e, wa s [he name Xiongn u an e th n ic term

tha t lat er be came a pol it ical term [0 refe r [0 all o f those "nomads" , and poss ibl y n o n -n o m ad s,

w ho were me m bers of [h e X ion gnu empire? What was the rel at ionship be tween "Xio ngnu" and

Hu , two name s th at at t ime s seem [0 be inte rchan geable? And of co u rse we have questi on s about

the lin gui sti c identity o f the Xi ongnu o r reg arding [he rel ationship (ethni c, cu lru ra] o r polit ical )

bet w een X ion gnu an d E uro pea n Huns, w hich have be en de bated for many decades and re centl y

revi ved re specti ve ly b y A . Vovin (2000) and t . d e la Vaissi er e (2005)J. After seve ra l d ecad es of deb at es , q u es t io ns re lat ed [0 [he ethni c o r lin gui sti c idemity o f the

X ion gnu are still unan sw ered. Likewise, [he possib ility o f finding in Ch ine se so u rce s clu es ab out

[he " ro o ts" of [he Xion gnu phenome non is fai n t at be st. Over th e pa st 30 y ear s or so arc h aeo­

logi cal , am h ro p o lo gi ca l, and, more rec en tly , gene[i c exp lorati ons, ha ve add ed new d imen sions

[0 o ur understandin g o f Xi ongnu cu ltu re, ide ntitv , an d so c io - po litica l forma ti on . These stu die s

have inc re m en ta lly ad d ed to o u r kno wled ge of the "Xion gnu phenome non " and o p ened new

avenues and hypoth eses of stud y, At the same ti m e, the general p ic ture of [he emergence a nd d ura ti on of the X io ng nu em p ire in it s hi st or ical and cu ltura l dimension s ha s become far more

var ied and co mplex.

AR CHA EOLOGY

The id ent ifica ti o n o f a dist in ct " Xi o ng n u cu ltu re " in archaeo logy d epends on what is mean t by

" X io ngn u ". If by " X io ng nu" o ne means a step p e cul ture located in a given regi on and wi thin a certa in hi storical period that coin cides with the assumed Xi ongnu su p rem ac y in [he same area ,

the n, as lo ng as cert a in charac te ri stics of [he m at erial cu ltu re of it s site s a re b ro ad ly consis ten t,

it is p erh aps poss ibl e [0 spea k o f a "Xion gnu culture " . On the other hand, no historical event

o r historica l representation of the X ion gnu o ther than in ver y rare cas es can be firmly co n nec ted

For a di ffere nt view , w h ich arg ues against th e ide ntifi ­ differe nt transc ri pti o n of the ter m Xiongn u in Mi d d le cation o f H un s and X rongnu .1S we ll as in su p po rt of a Chi ne se, see Bec kwith 200 9, 72 and 404--4 0 5.

37 ETHN OG ENESI S, C OEVOLUTION A N D P O llTI C A L M ORPHOLO GY

w ith an actua l X iongn u archaeol ogical cu ltu re o r population . T h e et hn ic ide nt ificat io n of h is­

to rica l peoples in archaeol ogy is an o ld problem th at requ ire s theo retical aw aren ess and sophis­

tic ation" and th e archaeology of the "Xion gnu ," extended as it is over vast and d iverse reg io ns,

has to pay spec ial a ttention to th e im pl icat io ns of su ch iden tificat ions .

If in Cen tral A sian an d ea stern European archaeology t he ter m " Scythia n" is w ide ly u sed to de scri be a specific type of st eppe mounted nomadic cu lture that extend ed far beyond any iden­

tification with th e his to r ical Scyt hia ns described by H erod o tu s, in Chinese or M o ngo lian ar­

ch aeo lo g y , then, a sim ilar " all -e nco m p assin g " cu ltu ra l te r m does n o t exis t. Som e R ussi an

archaeo logists rej ect th e blank et term "Scyth ian" becau se it er ases important differences be tween

local cu ltures". Yet, for the purpose of a broad typologica l and cul tural description , with no pre­

tence of re lationsh ip wi th the w ritten records, the ter m " Scy th ian " p rese rves a ce rtain u sefu ln ess .

O n th e other ha nd, the p roli feration o f cultu res identified exclu sively by the names of typesi tes

makes it d ifficult to recognize broad cultura l complexes an d phenomena at any sca le that exceeds

th e " mic roreg ional " and loca l leve ls''. One coul d talk fo r ins tance of t he " Scyth ian " cultures

(plural) o f the U rals , o f the A ltai, o r o f t he Black Sea region ju st to indicate a familiar ity of fea ­

tu res tha t does , by no means, exclude lo cal charac teris tics and even specific local names . The re­

fore, it may be poss ible to id ent ify a " Scyt h ian" ele men t in the P azy ryk culture even w hen local features m ak e such a cultu re un ique and d iffe rent fro m o ther "Scythia n " sites .

I n co ntras t, fo r the p re -Xiongnu cu ltures of northern China coe va l wi th th e nomadic,

"Scy th ia n" cu ltu res of cent ral Asia from th e Baikal to th e B lack Sea, we d o not have an umbrella

ter m th at wou ld allow the reco gn ition o f broad ly simila r cul tura l fea ture s ass ociated w ith the

emergence of m oun ted nomadism. The ter m northern zo n e, " be ifang d iqu", often used in C hi­

ne se archaeo logy to indica te the emergence of early no mad ic cultures, is purely geograp hical,

and o ther terms drawn from th e trad it ional w ritten so urc es (O i, Rong, H u, et c.) ca n n ot be u sed

as overarchin g cu ltu ral te rms becau se of t heir spe c ific historical associa t ions . The abs ence of a

term descr iptive of a cu ltu ra l p henomenon crea tes the di storti ng impressio n tha t the emergence

of mounted steppe nomadis m, amp ly documen ted in th e archaeo logica l record, and certa inly

connected on th e ba sis o f its ma teria l cu lture to a broad er E u ras ian p heno m eno n, can be reduced

to a series of loc al expressions. Sites defined by erh non yrn s su ch as Rong, Oi, Shan Rong, Q iang

a nd others, o r s im p ly id en tified by their types ite, as in the case of the Yan glang cu lture, a re

theref o re d isconne ct ed fro m th e ge nera l ap p earance and aff irm at io n of pasto ra l nomads as new

h isto rical subjects . On the other hand, when the term Xiongnu is used to id en ti fy arc haeological

sites typi cally da ting from the 3 rd century B C onwa rds, the impression generated is that such

sites - often di sp layi ng a bro ad ra n ge of cu ltura l exp ressio ns - w ere no t just cultu rally rela ted

but b elonged to a n ew politica l configuration: the Xiongnu empire. T he artificia l b reak between

p re-X io ngnu and Xiongn u cu ltures in no rth ern Ch ina implied by the unwarranted attribution

of a variety of et hnic ter m s remai ns o ne of the most in sidious theoretica l obstacl es to und er­

standin g the evolu tion of the step pe nomads of an cient C hina . The assumption , arc haeologica lly

undemonstra ted , of a late 4th or 3rd ce n tury BC break between an inaccu ra tely identified ser ies

of pre-Xiongnu cu ltures an d a unifi ed Xi o ngn u cu lture p revents a cle ar unders ta nd ing o f act ual

break s and co nti n uiti es . In fluen ces over lon g and sh o rt d istances, and affin it ies w ith othe r cu l­

f a r a rev iew a f sa me a f the rheorer ica l w o rk see E m­ 6 See , fa r in stance, the stud y by Ill. Khud iako v ( J99 6), berling 1997. in wh ich a va riety of cu ltu res and s ites a f the A lt ai, See, far ins tanc e, t he c rit iqu e o f the use a f the term Tuva , and C entral Asia are gro u ped unde r the "Scyt h ian " in Yab lansk y 1995. "S cyth ian" and " H un n ic " gene ral term .

38 NIC O LA DI C OSMO

tures over a lo ng period o f time, th at is, the fir st seven o r eigh t centuri es during whi ch northern Chinese cultures of th e steppe no ma d ic type seem to ha ve evol ved.

As a gene ral t erm o f reference, the term Xiongnu co u ld be accepted in the same way that

"S cythian " or "Hunnic " are accepted in othe r part s of Centra] Asian an d Siberian archaeology.

But in this case to what period should it appl y ? Why sho u ld it coi nci de nece ssarily wi[h its ap ­

pe arance in the C h inese written record s ? If, o n th e othe r h and, the late and mo re ev olved no­ madic phenomenon represented h istorically by [he Xiongnu can be taken as diffe rem from [h e

p re-Xiongnu phase , th an an other term sho u ld be appl ied to th e earli er nomadic cu ltu res. Th ese

pre-Xiongnu cu lt ures of northern China share essentia l features and coll ectively co ns tit u te an

arc ha eo logi cal phenomenon certa inl y not idemical but most likely co m parab le, in terms of [h e

evolu tion o f pasto ral no mad ism, with the emergence o f Scythian cultures in w estern Eurasi a,

[h e Altai, and Sou th Siberia .

In M ongoli an ar ch aeology , we see tod ay ye[ ano ther co nfigu rat io n . Here th e " Xiong nu

(Khunnii ) period" is d at ed roug h ly f rom the 2n d cemury BC to the 2n d century AD, fo llo w ed by a non-d escript ph ase, a b lan k spot th at co n t inues u ntil the "Turk" period, dated from [he 6[h

cenwry. First, i[ is nece ssar y to b ear in m ind that the asso ciation between the historical X ion gnu

empire an d [he nomadic cu lture th at emerg ed in M ongoli a aro und the 1st cem u ry BC, whi ch is

charact erized by lar ge necropolises and elite burial s, has yet ro be demonstrated. To the north of

[he Gob i desert w e se nse a development rh at enfo lds the enti re nomadic w orld and that w as pos­

sibl y acc elera ted by th e Xiongnu expansion and " im periali zatiori" o f vast areas o f [he s[epp e re­

gion but that may nor be necessaril y full y co inc ide n tal with or causa lly rel at ed to it. In va rio us regions of Mongolia , "Xiongnu" necropolises, se t tlem ents and o ther sites have been

doc umen ted . Some o f [he burials, s im ilar to [he tombs excavated by Russian arch ae ologist s in

Noyon U ul in the 1920s, are m onumental in s ize and belonged wan " elite " th at had access to vast

local re sou rces as well as im po rt ed Chinese goods (Brosseder 2009). H owever, [hi s co ns ide rati on

d oe s n ot m ean t hat we d o know in w ha t form [he Xi ongnu establi sh ed themselves in M ongolia.

Exc avated sites tha t, bas ed o n th e d ive rsif ica t ion o f burial structures, indi cate [h e existen ce of an

I ron Age aristoc racy in whose hands pol itical and rnilitary p ower wa s co ncen trated, do nor auto­ mat ica lly call for an imp eria l st ru cture. Several aristo crati c groups th at exhibited a co m mo n idiom

o f taste and fo r ms of representation o f elite st atus mi gh t have est ablished [heir independem

authority over certain territories without bei ng bound to a centralize d pol itical st ruc tu re.

F or inst ance, let us assume, following th e gen eral w isd om: the X iongnu empi re w as a co n­

fede rat io n of ethni cally and lingu ist ically diverse peop les first esta blished in w ha t is tod ay Inner

Mongolia , to the south of the G obi de sert . A s the Xi ongnu expand ed terriro r ially, th eir domin­ ion w oul d al so co me to include pe oples in the north . Empowered by th e X ion gnu, a new ar is­

toe ra cy, rem arkabl y homogen eou s in terms o f its own soc ial and pol iti cal id ent ity, w ould

eme rg e, there by establishin g power centers across N orthern Mongoli a. However, [h eir politi cal

unity, as we kn ow fr o m the h isto r ical reco rds, w as fra gil e, and in practi ce may have van ish ed

with [he secessio n of some chieftai ns to C hi na already in th e 'l st century Be. A s th e Xiongnu

be cam e further fr agmented during th e l st cem u ry AD, these elites ma y h av e developed in to

sepa ra te politi cal en titi es in localized power base s. Pol itic all y speaking, they returned to what

they were at [h e beginnin g of [h e Xiongnu expan sion, but re defined thems elv es in terms of the

im p erial ide ology imported b y [he Xi on gnu . Their " grand eur" of authority, pow er and im per ial

sy m bo lis m is visible in [h e size o f [h e buri als and by [he loca l an d C h inese luxury goods that

acco m pan ied [he de cea sed . It is also nor to be excluded a grow ing X ianbei pr esence in M on gol ia, altho ug h th ei r arc ha eo logical presenc e, [Q m y kn owled ge , has ye [ [Q be ide ntified.

39 E T H NOGENESI S, C O EVO L U T I O N AND P O LI T I C A L M ORP H OL OGY

The fac t tha t su ch" ro yal" burials are m issin g in Inn er M ongol ia, how ever, still co nsti tutes a

p ro b lem when w e w ish to es tab lis h a link be tween the X io ngnu o f th e O rdos and surrounding

areas an d th e M o n goli an "Khu nnu " . Perhaps the em ergence of s tro ng local ce nters o f powe r

was a Mongo lian de ve lop me nt th at foll o w ed th e dem ise of th e Xiongn u so uth of th e G o bi . A t

any rat e, if we accept th at the Xi o n gnu w ere at some point "de- imperialize d " an d ceased to ex ist

as a poli tica lly u n ified en ti ty, it is hi ghl y p ro bable t hat the "X io ngn u " archaeologica l s ites and

nec ropolises in M o ngoli a do not re p resent an "em p ire" at all, but in stead reflect th e presenc e o f

aristocrat ic power centers th at ove rt ime mayor may no t have reached a high level of p o lit ical

coh es io n. If in th eir imper ial co nd itio n they m ay have retain ed the co llecti ve na me of X io n gnu , we m igh t assu me tha t in th eir no n -imperi al state th ey revert ed to sub- o r pre- imperi al ethnic

identi t ies, whose hi sto rica l reality is un known to us, as n eithe r h is torical no r ar chaeo lo gical

records reveal to wha t "poli tica l conf iguration" a ce rta in excavated site in M on go lia m ight have

be longed .

A n additional pro b lem in X iongn u archaeology cons ists of th e difference between two co m ­

p lete ly separa te sph eres. One, in to d ay's nor thern C h ina, ex te nds ac ross the so- called N o rt he rn

Z one of Shanxi , I nner M on goli a, a nd Liaon ing. The o t he r emb races M on go lia and su rrounding

re gions, Tr ansb aik ali a, Buria tiia, Tuva and ea stern Ka zakhs tan. T he se tw o arch eol o gical

m ac ro reg io ns have been created by two q ui te separa te tr adit ions, na mel y , the So viet - M o ngol

and the C hi nese archaeo logica l sc hools. Each of them def ines its " X io ngn u " cu ltures in enti rely

d iffere nt ways , both m eth od ol o gicall y an d typol o gicall y. In C hina , fo r ins tance, Xiongnu si tes

are ide nti fied lar gely un der the infl uence of th e his to rical so urces . If, fo r in stan ce, a no rthe rn non -Ch in ese site is locat ed in an a re a an d dated to a p eri od co nsisten t w ith th e p resence of

X io ngn u peop le, it w ould be clas sified as "Xiongn u " . In the M ongo lian region, o n th e o ther

hand, the identifica tion is m o stl y based o n element s suc h as th e typol o gy of th e si te and exca ­

va ted art ifac ts . Another import ant d iffere nce is in chro no logy.

T he re ar e also profo u nd d iffe rences in term s of the arc ha eologica l rema in s a nd the ge ne ral

context. O ne o f t he m o st puzzlin g, whi ch I have a lr eady m ention ed , is th e ab sence, in Ch ina,

of monumental buria ls co mparab le w it h th o se fo u nd in M o n goli a. Mo reove r, Xiongn u set ­

tlem ents in Mo n golia an d Transb aikalia , such as Iv ol ga an d the m o re rec ently ex cav ated B o roo

Gol (R am seye r e t a l. 2009) sh o w a level of soc ial co mp lex ity tha t has not bee n full y doc u ­

me nted in C hina . Our c u rrent k nowled ge o f the m at eria l cu lture lab eled "Xio ngn u " in C h ina

is b as ed ex clusive ly on bu r ial excava tions a t O rdos sites suc h as M aoq in ggou , Tao ho ngba la,

and Xi go upa n, bu t th e class Str uctu re as well as th e eco no m ic bas es of the X io ngnu ar e po orly

reflec ted in the arc hitec tura l struc tures o f th e to mbs and un evenl y in th e ir fune rary inven ­ to r tes.

If we take th e northern M o n goli an s ites at Egiin Gol, w hich have been attrib u ted to the Xi o n gnu, th is little resembl es ei th er a "Xio ngn u " o r an y " n omad ic" site fou nd in C h in a. T he

p eo ple of Egi in Go l were agropas tora lis ts livin g in se t tle men ts . Wha t is r em ar ka bl e about

changes in the "Xiongn u " pe rio d is th e spatia l r econf igu ra t io n of th e d ist r ib u t ion of si tes, in d i­

cating, possibly, a more cohes ive polity th at unifi ed the en tire va lley's pop ula tio n (\Xfr ig ht et al.

2009). If that is th e case, an d if th e X iongn u p eriod is character iz ed by basic co ntinu ity in ter m s of ma teria l culture and genetic heritage, th en the Xiongnu impac t o n the area is probably to be

in vesti gated at th e level of a n ew so cio -po lit ica l art iculat ion. Th e ch an ges in the ab ility of elites

to co ntro / larger swaths of land and to re- rerrito ria lize their po p ulat io n accord ing to th eir n eed s,

w h ich m ay incl ude econom ic productio n, m o bilization fo r co mmunal lab or o r d efense, o r co n ­

t rol of trade routes, may be regarded as a more critica l tran sfo rmat io n th an de mog raph ic re­

40 NICOL A DI COSMO

placement s, changes in the mater ial cu lture, or the acquisition of luxury ob jects from far-flung region s. Above all, the rema ins at Egiin Gal and other settlements attributed to the Xiongnu per iod , show the inadequacy of anthropologically-d erived theo ries (to be d iscussed below) based on the notion that the X iongnu were a people whose econ om y was based exclusively on animal husbandry.

The advances made by Mon golian archaeology in the past ten year s or so, on the other hand , appear to widen rather than redu ce the gap with the Chinese side of Xiongnu archaeology, as it seems to me increa singly difficult to app ly mode ls of social o r pol itical organization deve loped on the basis of the necropolises, large bur ials, and sett leme nts excavated in Mongo lia to the Xiongnu sites excavated in the Ordos and other areas of northern China. On ly further explo­ rations and closer collabora tion between researchers working on both sides of the Gobi may clarify the nature of the connections - whi ch su rely exist but may be more anthropological th an cu ltura l or polit ical - between the two "X iongnu" archaeolog ies. For the time being, th e Xiongnu phenomenon in arch aeology appears fragmented , and methods of analysis as well as models of socio-pol itical evo lution are far from being able to embrace all of its aspe cts .

GENETICS

The youngest o f the disc iplines that have co ntr ibuted to Xiongnu stu d ies is genetics . The cold climate and often fro zen buri al cond itio ns of Mongo lia and Ka zakhstan have preserved the genet ic material in a go od state of co nservation and made it po ssibl e to extract DNA ma ­ te rial. Aside fro m considerations regarding the methods of analy sis , whi ch remain COntro­ versia l due to the intr insic d ifficulties of th e scienc e involved, and are in any case beyond my competence , a problem often encountered is the overlap between a biolog ical noti on o f population , and ethnic, political o r cultu ral co ncepts of "people ". This prob lem raises th e issue of the val idity o f h istorically sen siti ve conclu sions based exclusively or prima rily on genet ic data .

A recent DNA study found that an ind ividu al buried in a 2,000-year-old elite " Xiongnu " cemetery at Duurlig Nars, northeast Mongolia to be "Western Eurasian" , that is, associated with a European haplogroup (Kim et al. 2010). T his study elicits severa l possible que stions . Fo r in­ Stance, wh at is th e relatio nship between "Western Euras ian" and the "Xiongnu" attribution of the site? Should we take the "Xiongnu " label att ributed to the provenance site as a cultura l marker or as an indic ation of a specific pop ulation type? Should one see the presence of non ­ East Asian individuals in a site th at cultur ally and chronologically is und istingui shable from other "Xiongnu" sites as evidenc e o f a cultu ral d iffusion that exten ded acros s racial lines? Is "Xiongnu " a socio -politi callabel for people who, regardless of their racial identity, d isplayed certa in characteristics of elite Status and aristocra tic priv ilege wid espread across the steppes dur­ ing the early Iron Age per iod? T he authors' conclusion tha t the p rese nce of th is individual in­ dica ted that th ere was "racial tolerat ion " in the Xiongnu empire, how ever, assumes the existen ce of an imperial structure tha t cert ainly goes beyo nd the evidence provided either by the genetic samples or by the archaeological context.

On the other hand , careful genetic analys is can be very helpful in marking certain broad pat­ tern s of biological diffusion that can be historically significa nt, such as a study (Benriet t/Kaestle

E n-lN O G EN ESI S, C O EVOL U TI ON AN D P O LI T I C A L M O R PH OL O G Y 41

2006) of an cient DNA from human remain s from Lin z i (Sha nd o n g) an d E giin Go l (north ern

M on golia )? Lin zi (L ia ngc h u n site) has been d a ted to ab o ut 500 BC and Egiin Gol r ange s

chro nologically fro m ap prox im ately the m iddle of the 1s t cen tury BC to the 2 nd ce ntury AD. \X!hile the ge ne ticists assu me tha t a certain n u mber o f events ha ve tak en place in th e two reg io ns ,

and th at vari o us grou ps m ay have been p resent at d iff erent tim es , th ey are ca refu l to sta te th at "the exact degree to wh ich th ese va r ious cultural and lin guis ti c grou ps represented bi ol ogical

pop u lation s is d ebatabl e, and it is uncl ear w hether the afo remen tio ned chan ges w ere accompa ­

n ied by the movemen ts o f su ch biol o gical pop u lat io n s" (B en net t/ Ka es tle 2006, 414 ). The con­

clu sions presen ted in this s tu dy, w h ic h is o ne o f th e m o st at tenti ve in sepa ra ting OUt bi ol o gic al

and cu ltura l or h istori cal phenomen a, is th at th e p o p u la tio n o f Egii n Gol was m ore simi lar to

m odern E ast A sian peop les, w h ile that of Lin zi, ba sed howeve r on a sma ller sam p le, gives

entirely different results, and ra ther sho wed an affi n it y with th e west and esp ecially w it h peo p le

suc h as Iranian s, Turk s, Iraqi s, Hungarians a nd Slovakia ns . The p opula ti o n o f E giin Gol, o n th e

oth er hand, fi rmly r elates to northern Ea st A s ian peop les, such as Mon go lia ns, no rt hern

C h inese, ]apanese, Ko reans and Xin jian g H an peop le. In su m , the a rt icle est ab lishes that " w hat ­

ever the ex act interp retation of th e gen eti c affi nities o f the E giin Go l an d L in zi p o pulations m a y

b e, it is cl ear that th ey di ffer sign ifican tly " (B en ri ett / Ka es t le 2006, 43 0). Thi s type o f stu d y pro ­

vide s ex cellent ev id ence o f the d yn a m ism that exis te d in northern C hi na enco mpass ing both th e

Xi o ngnu and t he pre-Xion gn u peri ods. Th e presence o f a p opulation in 6th -century BC Shan ­

dong th at is cl o ser to a Central Asia n ty p e and that is di stinctl y di ssim ilar fr om the northern

C hinese m ay p oint to an ea rl y penetration in n orthern C h ina o f wh at is o therwi se desc ribed as

a Scyth e-Siber ian p opulation . Thi s fact m ay have so me implicati ons for the develop ment o f no ­

m adic cultures in north Ch in a to which the Xi ongnu may have been th e su ccessors. T hat the

p o pul ati on of E giin Gol d oes not show d is tin ct differences wi th resp ect to no rth ern Chi na is

in teresti n g, bu t fair ly co nsis te n t w ith other popu lat ion data of M o n goli a of the sa m e period.

Vari ou s studies em ph asi ze th e co n nectio n be tween " K u rga n p eople" w ho w ere present in

C ent ral A sia and Sou th Sib e r ia, in pa rtic ula r th e regio n of K rasn o iarsk, in t he Bron ze and Earl y

I ron A ges and "Euro p ean" hapJ o groups, thus esta b lishi n g a genetic co ntin u u m ac ro ss C ent ra l

E urasia and hypothesizi ng a lon g- range diffusio n o f a pa rti cular ty pe o f pop u la t io n (Keyser­

T racqu i et al. 2008, R icau t er al. 20 04 ). The present sta ge of research seems to show the existenc e

of a " fau lt-line " be tw een \X!estern E u rasia n and Eas t A sian pop u la tio n gro u ps th at runs fr o m

th e terr itory ar ound the Baikal lake throu gh th e Altai re gio n in W estern Mo ngolia , dow n to t he

Gan su re gion in th e n orthwest, with po ssibl e " in filt ra t io ns " o f p eopl e ge n et ica lly relat ed to

"European-typ e" Sou th Siberian and C entral A sian po p u lation as fa r east as Sh andon g. Given th at th e d ata availabl e is s till ex t re mely sca rc e and th a t re search o n ancie nt DNA is intrinsic a lly

unstabl e, as the result s are su bject to d an gers of co ntamin at ion and o th er te ch n ical pitfa lls , it

wi ll take so me time before one can ge net ically map th e regions b etween Sou th Siberia, M ongol ia,

an d no rth Ch in a be tween the Srh and th e 1st century Be. N o neth eless, we d o know th at du r in g

th is time p er iod o f time it appears th a t someth in g of crit ica l importance h appened in th e devel ­

op m ent of m o unted cu ltures o f the "S cy thi an " type in Ea st A si a, co nn ec te d poss ib ly to demo ­

grap hic, po litica l, and cul tural ch anges. In order to Jearn m ore about su ch developm ents new

resea rch h ypotheses are re qui red that tak e into cons ide ra tion th a t th is region not onl y re ac ted

to chan ges occurrin g elsew he re (C hina o r the Iranian world ) but also produced transfo rmati on s

7 Fo r related stud ies see also Keyser-Tracgu i et al. 2003; 2006; 2008.

42 NIC O L A DI COSM O

within its o w n cultural sp here. It wou ld then be po ssible to in vestigate whether such societ ies wou ld hav e advanc ed not just in term s of technic al progr ess and arti st ic produ cti on, as is am ply

demonstrated, but ev en m o re sign ifican tly throu gh the emergen ce of new, intern ally generated

form s of social and pol iti cal o rg ani za ti o n .

Wh at o fte n co mp licates th e p o tent ia l u sefuln ess of DNA d ata is the tendenc y of scientists

to co nfir m rath er tha n proble mat ize and confu te exi st ing ar cheological o r hi st oric al con jec ­

tu re s. For ins tance, DNA dat a is often tak en as eviden ce of the fact that a giv en population

assum ed to ha ve been " no mad ic " was mo re likel y to foll ow a ce rtai n pattern of "ge ne flo w "

by conquest , wa r, migration o r other event , whe n neith er th e nature of "n omadism, " nor the

nomadi c ch aract er o f a popul at ion , n or the exi sten ce of suc h hi stori cal events ca n be demon­

strated. F or insta nce, an influential stu dy o n DNA material fro m a "Scyrho-Siberian " ku rgan

in the Altai Republic (ca. 500 BC ), ass u mes that the population in th at pla ce bel o nged to " no­ madi c trib es" full y distinct from agr icultu ral populati ons (Ri caut et al. 2004 ). Su ch an assu m p­ tion , ho w ever, shou ld be veri fied from d at a that ca n no t be purely biologica l. Thi s stu dy 's most

intri guin g co n trib u t io n is th at it d emonstrates, by ex aminin g biol o gica l ev id ence th at a co n­

ne ction exis ted between a g ive n a rea at a spec if ic p o int in time a nd p o pulations o f other

regi ons. T he authors a re al so on the wh o le cau ti o us a bo u t mak in g hi st ori cal assum p tions .

Still, on e h as to gu ard aga in st sim p li fy ing the socio- economic context t o fit bi ologi cal dat a.

In stead, DNA data is most he lpfu l wh en it leads us to re -th ink o ld theo ries and to mod ify

th em accordi ng to a comprehens ive an a lysis . Future genet ic re se arch, co m p lem en te d with

ar chaeol o gical and an thropol o gical data , sho u ld indeed co ntri bu te to formulate more accurat e

his tor ical hy po the ses regard in g population movements and localized dem ographic shifts, thus

allowing a cr it ical re -asses sm ent o f exis ti ng the ories.

THEORETJCAL ApPR O A CH ES T O EMPIRE F ORM ATI ON

While hi storical so urces , material cu lture , and informat ion gat hered fro m the ex cava tion o f se t­

tlements and burials, as w ell as bi ol ogica l data, are essential to the definiti on o f the Xi ongnu as

a histori cal, cultural and demogra phic phenom enon, th e m er e accumul at ion o f data can no t an­

swer th e questi on of th e formati on of the Xi ongnu empire . His torical ant h rop o logis ts have

given us their most important contributions, by propo sing a series o f models th at mi ght explain

th e ri se of the X iongnu emp ir e in terms of state fo rma tion an d th e creatio n of larger so ciet ies in

th e step pes. I n the present con tex t I will lim it myse lf to theories o f state building and to co n ­

ceptual q uesti on s th at hav e co llec tively co n tribu ted to a general reflection of so cial change and

im peria l fo rmati o n am o n g the step pe nomad s and the X io ng n u in particular.

The fo u nda t ional w ork o f O. L attim ore (1940) as sumed the exi stence o f a sh arp divide - eco­ lo gical, econ o m ic, a nd p ol iti cal - between pastoral nom ad s and agri culturali sts in northern

China. The rising specializ ation between the two " cam ps" increasingly created a s ituati o n o f

"d epend enc y " o f the nomads upon the ag r icu lturalis ts of C h ina, leadin g to dy namics so me ti m e

d escribed as "t rade o r raid " (jagch id/Syrnons 1989), that is, o f fo rceful ap p ro p riat ion by the no­ mad s of cert ai n goods lacking in their o w n econom y a nd soci ety. In a nu tshe ll, th e rise of no­

madic empires alo n g the northern front ier of C hina w as seen as a dram at iza t ion of that bi nary

relati onship provoked b y th e unifi cati on o f C hina.

43 E THNOGENESlS, C O EVO LU TI O N AN D P O LIT IC A L M ORPHOLOGY

Acco rd ing to a we ll-know n theory, the steppe no mad s, and the Xiongnu in particular, form ed

"s hadow empires " when Ch ina w as uni fied , strong, and able to provide great quantities of "tri b­

ute" to a m ighty no mad ic con fede racy (Barfiel d 200 b ). The extract ive power of the nom ad s

w as directl y p roporti onal to t he military pressure wi th whic h they would challenge China, an d

th e need for inc reas ing th eir pressu re w hen co nfro n ted wi th a stronger China led to th e exp an­

sio n o f ste p pe polit ies and to the formati on , even tu ally, of the Xiongn u empire. Im po rt an t el­

eme nts o f th is interpretati on, d eri ved fr om Latt im ore 's th eories, ar e t he no tion s of a sha rp d emar cation between step pe and sow n, o f th e inh erent need for no mad s to suppl ement their

imbalanced economic basis by imposin g a t ribute o n C h ina th at wo u ld provi de them with the

produ cts o f ag r icul ture, suc h as co tton and cer ea ls, as w ell as with lu xury product s, suc h as go ld and silk, and o f the po litical relevancy o f th is d epend en cy. As Barfield put it : " E mp ires es ta b ­

lished by nom ad s in Mongo lia were di sti nct ive " sh ad ow em pires" that ar ose as seconda ry ph e­

nomena in response to imp erial expa nsion by th e Chi nese. Their sta bi lity depended on ex to rt ing

vast amounts o f we alt h fro m Ch ina through pi llage , t rib ute payments, bo rd er t rade, and int er ­

national re-export of lu xury goods - not by taxin g ste ppe no ma ds . When C hi na was cen t ra lize d

and powerful, so we re nomadic em pires ; when C h ina co llap sed into polit ica l anarchy and eco ­

nomic depression, so did the un ified step pe polities that had prospered by its exto rt io n " (Barfield

2001a, 10). In other word s, the nomadic worl d united politically when it req u ired an equal

amount of "force" to be ab le to extract goods from a neighbor that had be com e not just r icher

but also more powe rful.

This theory ha s bee n su bjec t to cr itica l assess ments, based on both archaeological and histo r ­

ical evidences. M. D rompp, for ins tance, afte r a careful review of the emergence o f t he fir st and

second Turk and U ighu r emp ires, co ncludes that "the picture of state formation and d isin te ­

grat io n that em erges ... is more com plex t han th e model presented by Barfield suggest s. We ca n

see the importa nce of a mu lti polar sys tem of dipl omacy and trade, rather than the bipolar mod el upon which Barfield re lies " (D ro m pp 2005, 109).

P. Turch in, an evol ut ionary b iologist and an thropol o gist , has critiqued the th eo ry of a sudden emerge nc e of th e X io ng nu em pire and ta ken issue in partic ul ar with the notion of a " u nid irec­

tion al " relat ionship be tw een s teppe and sown whereby nomads typ ically u n ify as a reac tion to

th e un ification of C h ina to app ro p ria te th e riches provided by a w ealthi er agricultural wor ld

(Turchi n 2009). After clos e exa mi natio n o f Bar field 's "s ha dow em pi res " theory , Tu rchin firs t

po int s to th e absen ce o f any ac tua l "tim e lag" betw een the un ificat ion of C hina and the rise of

steppe emp ires . Secondl y, he und erscores th e fact that in several cases s teppe sta tes came into

existence whe n Ch ina was wea k and d ivid ed . Tu rchin also su bsc ribes to th e noti o n o f the "i rn ­ perio genet ic " (Turch in 's word) ro le of fro ntier zones, but ta kes a mu ch longer tim e fram e, w hic h

leads to a theory of " co evo lu tio nary " d evelo p me nt of nomadi c and ag r icultural societ ies.

Hi s model, w hich he appli es in parti cu lar to th e Xi o ngnu, w oul d lead us to co ns ider a pe rio d

o f time o f severa l ce nturi es d uring which on both sid es alon g the front ier between C hina and th e

nomad s in creasin gly mo re co mplex po litie s were fo rm ed . As th e smaller Chi nese sta tes were

go bbl ed up by larger an d more pow erfu l ones, so did the no mad s grow into more powerful

" scaled-up " soci eties . According to this model we should not assume that the unificat ion of Chi na

" gen erated " a step pe em pire th at would "exploit" and extract go ods from Chi na, but t hat the

Xi on gnu em pire w as the product of a long-term coevolutionary pro cess on both sides of the

Am on g the mos t recent crit ics , see Bec kw it h 2009, 329 and pp.

44 NICOLA DI C OSMO

fro n tier that drove the fo rma tion o f at least tWO im peri al p olities. U n fortunatel y, w hile we k now

w hat happened in C hi na, th e existe nce of a para llel deve lo p m ent in th e step pe is pure ly hypo­

thet ical and to my m ind im possib le to de mo nstra te o n ei th er text ual or arc h aeo logical grou nds .

A related area of debate concerns th e kind of "polity" (if w e can use th at term) that was crea ted

by th e X io ng nu; na mely, w as it a " su pratri b al " state or w as it a more co mplex version of a tri b al

chi efdom? The state -vs. -c hiefdom d eb at e has been em b raced mo stl y by an thropologists. T he

iss u e revo lves aro u nd the de ve lop m ent o f stat e inst itutio ns. In parti cular, sch ola rs su ch as

N . K rad in co nte nd th at an ea rly sta te shou ld hav e the folJo w ing ch arac te ristics: (1) acc ess to mana gerial position s by some k ind of me rit- bas ed or, at any rate, extra-c lan and non-kin- based

sel ection; (2) reg u lar taxa t io n to p ay th e off ici als' wages; (3) the crea t ion o f a sp ecialized judicial

power se parate from t he p ol it ical one; and (4) a class o f sta te fun cti ona ries en gaged in run n in g

a state m achi nery co nsisting o f serv ices for the adm inistrat ion of the w ho le pol it ical co mmun ity.

Th e X io ng nu "s ta te", acco rd ing to Krad in , d id not sa t isfy th ese re q uire ments ot her than at an

em b ryo n ic lev el, and therefore sh ou ld b e co nsidered not a state bu t a su p erior k ind of "chief­

do m" , o r, as it is d ef ined here , a " su per-c o mple x ch iefdom" (Krad in 200 2b) .

O n t he ot her ha nd , it is also p ossibl e to see the "early st ate" of th e X io ngnu as art icu lated

in to distinct military and civili an ap paratu ses sep arate from kin -based hierarch ies. Th e top co m ­

mand ers and func tionaries received the ir wa ges, rewa rd s o r co m p ensa tio n (in w ha tever form)

from a po lit ica l center head ed by the "cha rismatic" lead er. This lea d er, the cha nyu, w as in ch arge

o f ritu als an d ce remo nies that we re meant to includ e th e wh ole pol it ica l co m mu nity . Added to

the afo rement ioned se parat io n o f the ju d ic ial pow er from th e stric tly politica l an d mi litary o ne,

it may also be po ssible to regard the X iongn u po lity as clo ser to a "stat e" th an to a ch iefdo m.

Lo o k ing at t h is prob lem from ano ther pe rs pec tive, it is q ui te clear that th e su rv iva l o f the

X io ngnu as a u n ified po lity dep ended u pon the acq u isit io n o f extern al resources th at were re­

di st ributed by the cha ny u acro ss t he to p ec helo ns o f th e ar istoc racy and p resu m ably " trickled

down " to the low er rungs of so ciet y according to posit ion held by va riou s people wi thin th e

sta te m achin er y (Di C osmo 1999 b). It is t h erefo re a reaso nab le as sumptio n th at th e "stat e" ap para tu s that had co me in to existence was mainta ined financia IJy o n the basis of a redi stributive

p rinciple cen ter ed on bo th aristocratic p r ivil ege and so cial fun ct ion. It cannot b e dou bted, given the com plexi ty o f the Xiong nu em pire, th at it incl ud ed a lar ge number of mid -r ankin g positions ,

such as hea d s of sm aller m ilitary units, loc a l ad minist rators, and su p erintendents in ch arge of

po st al se rv ice, sup pl ies an d va rio us de partm ents that serviced the cou rt an d the ari sto cratic es­

tabli shmen ts. These peop le mu st ha ve been su pported acc ording to p rinc ip les that cou ld not

have been exc lus ively b ased on ari stocra tic lineage. We do no t know exactly wh o oc cu pied th ese

positio ns in th e X iongnu emp ire, as we do no t know wh o w er e the people in c ha rge of a nu m ber

of th e 24 " G reat Dign itary " posit ions, or how these w ere assigned. T hese pos it ions w ere mo st

lik ely su bjec t to some kind o f aristoc ra tic m o no pol y, bu t thi s do es not mean th at al l im porta n t

positi ons in the re alm w ere fil led by m embers of the th ree ro y al clans ment io ned in th e "S hi ji",

which ap pare n tly co ns titu ted t he sol e a ris toc racy o f the em pire . G iven the exte nt o f the em pire,

an d th e nu m ber of pe o p les that converged into it, it is hig h ly un likely th at all mili tar y an d c iv il

p osi tion s in t he govern ment, both centrally and locall y , were the pr ivate d omain of su ch an ex­

t rem ely rest r icted elite. The po int is d est in ed to rem ain mo ot fo r lack of suff icient in fo rma tion ,

but it canno t be sim ply assumed th at access to th e political, military, and ad m inistrat ive p o sit ion s

was exc lusively based o n k in -based selec t io n.

W heth er we accept th e definit ion o f the Xiongn u as a "sta te" or as a "super-com p lex c hi ef­

do m" th ere sh o ul d be no d ou b t tha t it was an emp ire , th at is, a po litica l form ation th at extended

45 E T H N O G ENES IS, COE VO L UT ION AND POLIT IC A L MORPHOLOGY

fa r be yond its o rig ina l terri torial o r ethnic confine s and em braced , by direc t conqu est o r by the

im pos itio n o f its polit ica l a u thor ity , a variety o f peop les and land s tha t m ay have had diffe re nt

types o f re lations wi th the imperia l cente r, co ns ti tu ted by the imperial cla n and b y its cha ris ma tic

leader. Ev en if th e Xiongnu as a polit y do no t sa tisfy the requ irem ents fo r be ing regarded as an

"early state", th ey still had th e poli tical ability and military power to crea te a "step p e empi re " .

In o ther w o rd s, the re would be no co nt rad ictio n betw een being at the sam e time a "ch iefd o m "

- in terms of th e level of evo lu tion of th e ad mini st rati ve machi n er y, th e sp ecific qu alit y of the

sovereign's power, and th e forms in w hich different d ut ies (m ilitary, fiscal, or o th er) w ere allo ­

cated to the popu lat io n - and an em p ire .

CHAR ACTER OF THE XI ONGNU EMPI RE

Current theori es encou n ter diffic u lt ies not only w hen try in g to d et ermine th e n at u re o f the

Xiongnu em pire , but also in explainin g the qu antum leap fro m a rela tively sparse society - w here

agrop asto ra l com m u n ities lived in close proxi m ity with more specializ ed pastoral nomads , nei­

th er especially powerful nor especially wealthy - to the imperial expa ns io n that took p lace under

the Xiong nu . Let us remind ourselves that the Xiongnu empire appa rently ex te nd ed fro m the

Ordos to th e Ba ika l and fro m M an churia to eas te rn Kazakhst an , inc lud ing und er its domi nion

Mongo lia, N o rthern Ch ina, the " o asis- st at es" of the "Wes tern Regio ns" (Xinj iang), th e A lta i,

Tuva, and Trans baikalia. Wh at re mains un expl ained is how the Xiongnu cou ld forge the necessary

tools (po litica l, administrative, military and ideo lo gical) extensively documented in Ch in ese

records, without which, arguably , their imper ial rise and expan sion would not hav e ma te rialized .

Th e o rga niza tio n o f th e ar m y, the imperial rituals, th e go vernment structure and politica lly cen­

tral ized fu nc tio ns o f trad e and d ipl omacy all bespeak o f a polit ical mac hi ne ry and imperial, o r at

an y rate "su pratr ibal ", ideo logy that so far re mains unexpl ain ed as to th eir origin and fo rma tio n .

It may be argued th at, if w e have in the step pe gra d ually more complex and larger soci eti es,

the not io n of a "supratrib al " leader must h av e em erged so m e ti me befo re. Th e inner struc tu re

of the Xiongnu empire, wh eth er we see it as a weak and d evelo ping state or as a particularly

s t ro ng and successfu l chiefdom, can not have been cre ated by M o d u n as he proceed ed to assem ­

bl e an empire, a nd in deed exist ed alrea dy, as a no n -impe rial poli tical ent ity, under th e rule o f

h is fa ther Turn en , whom Modu n d efeat ed an d rep lace d . If th e imper ia l political an d ad m inis­ trati ve " too l- box" of the Xiongnu was the p roduct of social evolution, was th is pro du ct interna l

to th e steppe world's own deve lop m ent - ma kin g allowances for so m e extern al conta cts , bu t

onl y of an incidental an d non -s t ructu ral na t ure - or was it rather a frontier ph enomenon, that

is, the un ique resu lt of a long interac tio n w ith C h in a m ade of mutual stimu li and "feed back

lo ops" (to use Turch in 's terminol ogy )?

In o rde r to beg in to answer this qu esti on o ne has to look in to the inst itut ions th at mos t cle arly

ch ar act erized the X io ngnu as an emp ire, and see whether it is possib le to trace th em to a potenti al

source. I ca n see only th ree p o ssibilities : a p reex ist ing, bu t as yet h istori cally invisib le, steppe

t ra dition of large su pratri bal p o lities, Chinese state institutions, o r an imperi al mod el imp o rt ed

fro m the West. Sin ce the Xiongn u ph en o m eno n h as bee n assumed to be the p ro d u ct of the in­

teract ion between C h ina and th e nomad s, an d beca us e of time -honored notion s of Ch ine se cul­

tural influence ra dia ti ng into th e sur rou ndi ng areas and on to non -Ch ine se peop les, Chi na has

46 NI C O LA DI COS MO

long been co nsidered as the only "originaror" of an impe rial ideolo gy and practice. Hence, in this final port ion of this essay 1 shall consid er briefly a few elements that have been presented as evidence th at the roots of the Xiongnu imperial formation should be sou ght in China.

It is pos sible that the appell ation "Son of Heaven ", if this is what the Xiongnu words " Chengli gu tu " mean, as reponed in the "Hanshu " (Ban G u 1962,3751 ), w as added ro the title "chanyu" (meaning " mag nificence") fo r the Xion gnu ruler at a later, "imperial " time ro provide a mor e universal meanin g ro a native title, and make it "sacral" in the sense of charismatic and supratribal (Pulley blank 200 0). Ca n then such a dev elopm ent be regarded as a direct loan from the C hinese idea of the "Son of Heaven" ("T ianzi")? To phrase it d ifferently, can the idea of a " supratribal " head be derived from an existing Chinese noti on of sacral kin gsh ip?

In my view the we ight of the evidence speaks against a C hinese loan for two reasons. First of all the Xiong nu word "chengli" has been lon g ago recon structed, rath er uncontrover sially, as the Tur co-Mon gol divin ity " tengri". It seems ro me imp ossible that a borrowing from China could also pr odu ce a native word different from the Chinese one . More over, the w ord "gutu" , wh ich the Chinese sources render as "s on", has not been satisfacrorily resolved ety mol ogically. If the non-Chinese word "tengri" existed already in a Xion gnu context, there is no reason ro believe that the conc ept ro which it is closely associated, that is, a sacral, heaven-in vested concept of sovereignty, did not also exist w ith in a non-Ch inese politi cal cont ext pr ior ro th e emergence of the Xion gnu em pire". Th e term "chan yu ", o r "shanyu ", has been interpreted in various ways and linguisticaJly associated by E. Pulleyblank (1962, 91) with a later Altaic word "darva ", which again points ro a bod y of po litical terms total ly extraneous ro the Ch inese trad ition.

Possibly the most compelling argument of C hinese influence has been made by Pulleyblank (2000) and ant icipated by \V SamoJin (195 7/58). Thi s is based on the Xion gnu govern ment 's char­ acteri stic to have officials "paired" inro a "lef t" and "right" position for each title, wh ereb y the left was the higher rank. From the early Zh ou dyn asty down ro th e Warring States and finally to the Qin and Han dynasties, several State titles in Ch ina were divid ed in to two separate posts with a left and a right denomination, whereb y the left indeed took preceden ce and repres ented a higher position than the right. Acco rdin g ro Pull eyblank, th is is because the ruler in China position ed himself in the north facing south, and the officials ro his left (i.e., the east), too k pre cedence over those ro his right (west), as this was the dire ction of the rising sun. H owever this arr angement is especially relevant fo r Chinese ritual consideration s (Graham 1986,79-80; 82-84) and is not really consi stent with what we know abo ut Xiongnu rituals . We can see fro m a "Shiji" referen ce ro a ritual per formed by th e chanyu that off icials did no t sit on both sides of the ruler: "When the chanyu in the morning leaves the camp, he pra ys ro the incip ient binh of the sun, in the evening, he prays to the moon. When they sit down, their senio r officials sit on the left and face north" (Sima Qian 1959,2892). Wh at this means is th at the chan yu faced south and the senior officials, all of them, not just those "of the left", Sat on the side of the rising sun (east), left of the ruler, and faced north, that is, thei r gaze was dir ected to the ruler. Hence, the den omin ation of the officials as "l eft" and "r ight " do es not apply, as far as we can see, ro ritual considerations.

Going back to ancient China, inscriptions of the Zhou peri od bear evid ence of the left and right pairing of official titl es (Shau ghnessy 1999,320). In various Warring States govern ment s

9 N o te, inc id e ntall y , t har th e no t ion o f sacral kingship an ex ternal trad ition po ssib ly asso cia ted w ith C e ntr a] embodi ed in the term "t ian " and " tian z i " , w hi le it m"y Asi a. See C hen Sanping, So n o f H eaven a nd So n o f ha ve a lre ad y exi s ted as a pol irica ] co nce p t in Ch ina God : In tera ct ions am ong A nc ient Asiati c C u ltu res re­ wit h the Sh ang d ynast y, w as certainl y modified by t he ga rd ing Sacral Kin gs hip and Theop hori c N ames. Jo ur ­ intro du cti on of th e d eit y " [ian " b y [he Zho u . This w as nal of the Ro ya l As iatic Socie ty 12, 3,2002,289-325..

47 ETHNOGENESIS, COEVOLUTI ON AND P OLIT ICA L MORPH O LOG Y

thi s separation ei ther prevailed in lower-rank bu reaucrat ic posi tions, o r was rese rved to some

spec ial titles, or was a ves tigial " relic" of the Zhou state . In o ther w o rds , rig ht an d left pai rs of

o ff icia ls exi sted at differe nt ti mes an d in variou s fo rms in Chi nese h isto ry, but does it mean that

the Xiongnu mimicked th is sys tem? In fact, the X io ngn u titles do not refer to the position ing

o f offic ia ls to the left and ri gh t o f th e ru ler, bu t to a terri to rial o rganization o f the em pire th at

d oes not seem to fin d a parallel app lication in China . The sys tema tic use of officia ls of the left

and r ight reflec ts the existence of two branch es of the governmen t cha rged with the ad min is­

t ratio n of th e lef t (o r we ste rn) and the r igh t (or eastern) pa rts of the empire (Yii 1986, 385) .

Whi le we d o no t know exac tly how these tw o "branches" stood in relat ion to eac h other, and

wheth er th ere w as "mo bility " o f o fficial s between them, th ere is no qu estio n tha t it reflect s a

re gio nal or terr ito r ial structure fu nd am entall y different from the C hinese o ne . A mong th e Inn er

Asian emp ire -bu ilders that came afte r th e Xi ongn u , in part icu lar the Tu rk s, left and ri gh t re p ­

resented ea st and w est, and in deed th e polit ica l bifurcation of the empire of the Turks, divided

in to an E as tern and Wes te rn hal f, is the closest rep resentat ion of the lef t- r igh t division of titles

th a t w e find among the Xi o ng nu. T hi s is a principl e en tirel y d iffere nt fro m the left and r igh t

connotation in Chinese titles , as they were not based on a regiona l or geog raph ical pr incip le of

state administration.

A th ird element br o ug h t fo rth in sup port o f a Chinese influence consists of one sentence in

the "Shiji" in w hic h the Xiongnu appea r to foll o w the same color symbolism as the Chinese .

P u lley blan k (2000) and C h. C hen (1989) no ted a pa ssage in w h ich it is said that when Emp eror

H an Gaoz u was surrounded by the Xiongn u cava lry, th eir ho rses were w hite in the/fast, blue ". ,

in the east , black in the north, and red in the sou th . T he pa ssage in q uesti on re ads as fol lows:

"As for th e X iongnu ca valry , its w este rn side consisted completely of white horses, the eastern

side entire ly o f g rey-spo tted horses, the no rth ern side was enti rely [made of] bl ack horses, the

sout hern sid e w as o nly ba y ho rses " (Si m a Q ian 1959, 2894). T he d iffere nce in th e defi nition the

horses' natura l co loration shows that th e sym bolic associa tion between co lor and d irection is

indeed there, and cou ld be connec ted wi th the C hinese system o f the Fiv e Ele ments. B u t can

these ass ociations no t have ex isted among the Xiongn u independently of China)

K . Chang, the p reem inent C hinese archaeologist, saw a shamanistic substratu m in ea r ly Chi­

nes e civ iliz atio n of w h ich colo r- d irec tio n co rrela t ions w ere o ne as p ect (Ch ang 1989) . T he same

reasoning could be app lied, in my view, to the shama n istic world of the Xiongn u, which possibly

sha red the same su bs tra tum o f na tural associa tio ns and co lor direction s th a t w e m ay find in

early Chi nese and o t her Asi an cul tures . I t seems to me, therefore, that the argument that the

X iongn u simp ly took over Chinese symbolism could be considerably less simple, and in stead

reveal a sh ared se t o f associa tions that m ay go ba c k to a more an cient cultural stra tum , as far as

the early Bronze Age and even th e N eol ithi c.

The weight of the ev idence is d efin itely agains t any close res embla nce bet wee n th e C h inese

and the X io ng n u empire. The decimal system of the organization of th e army h as it s closest,

and to my k nowledge o nly paralle l in the A chaem en id emp ire (Dan damaev 1989, 194; B ri ant

2002, 19). Th e ritua l of mixing w ine wi th th e blood o f a sacrificed an ima l w hen swearing an

oat h and then stirring it wi th a "ji ng lu" sword ta kes us to a Scy th ian world", as would the habit

of tak ing enemies' head s in battle an d being th en re w ard ed wi th a cup o f w ine, wh ich is a known

10 A descriptio n of a ritual o ath -tak ing ceremony, in which such a blade is used , is giv en in the " Han shu" (Ban G u 1962,380 1).

48 NI COLA DI COSMO

Scythian ritu al. The very term of the sacrificial dagger, "jinglu" in Chinese, has been recon ­

structed by Pulleyblank as an Iranian word (2000,63-64). If we are look ing for the "roots" of the political culture and institutions of the Xiongnu em­

pire, they are not going to be lo cated in China, or at very least a Chinese or igin has yet to be

demonst rated . A s long as we can exclude a Chinese influence in shaping the morphology of the

Xiongnu empire, it is difficult to endorse a "co-evolutionary" hypothesis, and we must the refo re

con cl ude that events occ u rri ng on the Sin o-nomadic frontier, even if we take the long view pro­

posed by Turchin, ca nno t be th e sole catalyst for the em ergence of the Xiongnu empire. Our

discu ssio n points, rather, to the existenc e of a cultural continuum between the nomadic world

to the north of China and the "S cythian " Siberian and Central Asian cultures that d oes, in all

probab il ity, acco unt not onl y f?,r the substantia l sim ilar ities in their material cu ltu res, but also

"'"' for the possible sim ilar it ies in of for ms o f politic al organization whose paths and roots remain,

to d ate, obscure. Th is conjecture is undoubtedly diffi cult to prove, and requires historians,

archaeo lo gists and other res earchers to refocus thei r attention and investigat e mo re closely the

cultu ral and pol itic al rel ations with in the "s tepp e" world. However, w eanin g o u rse lves off the

notion that Xiongnu and C h ina formed an ex clu sive relation ship , and th at the interact ion

between them w as the birthplace of all that was complex and "evolved" in the Xiongnu e mp ire,

would be a positive step forward, that wou ld allow researchers of d ifferent dis cipl ines to con ­

sid er hypotheses that linked the genesis of the Xiongnu empire - as well as on the ethnogenesis

of the Xion gnu as a people - to the evo lu t io n of po litical cu ltures in various parts of Inn er A sia.

How to re-orient Xiongnu Stu d ies towards a greater integration of the d isciplines involved and

toward s the elaboration of theor ies of the evolu t ion of steppe nomadic polities ba sed o n internal

dynami cs, p ro gr ess, and tr ansformation is one o f the most important challenges for th e fu ture

advan cement of our kn owledge of the Xiongnu and of the history of an cient nomads.