A Dendrochronological Framework

8/12/2019 A Dendrochronological Framework

1/12

TcBA-AR VI I 2004

A DENDROCHRONOLOGICAL FRAMEWORKFOR THE A SSYRIA N COLONYPERIOD IN ASIA MINOR

.. ..

Robert J. Bra idwood an d L indaS Bra idwood in memoriam

Maryanne W. NEWTON**Peter an KUNMOLM

Keywords Dendrochronology, radiocarbon, Middle Bronze Age, Assyrian Colony Period, wiggl e-mat chingAnahtar sbzcukler: Agas halka tarihlemesi, r adyok arbon , Orta Tuns Gaol, Asur Koloni Gaol, fark eor i lerinin ortu5 mesi

Anadolu icin M . 0 . 2657-649 W 1 0 1 yillari boyunca uzanan iki parca halinde 2009 yillik bir EskiTune-Demir Cagi agac-halka kronolojisi sunuyoruz. Bu kronoloji hem Kultepenin Karum 11 veKarum Ib tabakalarini icermekte, hem de bu tabakalari Acemhoyijk ve Karahoyiik-Konya dakicagda? tabakalara baglamaktadir. Ilk defa her uc kazi alaninin Karum II tabakasina ait yapilarinakesin tarih verebiliyoruz. Ornegin, Karum Ibde, Kultepenin WarSama Sarayi M.0 1832) veAcemhoyii@n Sarikaya Sarayi ve Hatipler Tepesi (ikisi de M .0 1774te yapilmigtir) gibi me8huryapilarini tarihlendirebiliyoruz. Ayrica, agac halkalari Wargama Sarayi ninM ,0 1771 dekiyikmindanonce en az 61 sene, ve Sarikaya SarayininM . 0 . 1766daki yikimindan once en az 8 sene var olduk-larini gosteriyor.

The Assyrian Colony P e r i dToward the end of the third millennium BCAssyrian merchants began a remarkable, multi-generation-long commercial relationship, princi-pally a trade in metals, with the kings of centralAnatolia. Their typical archaeological imprint,seen best at Kultepe, anc ient Kane:, is a settle-ment o r karum of m erchants houses (T. Ozguq1986) clustered around a large mound wherethe indigenous Anatolian ruler lived, usually ina substantial palace T.Ozguc; 1999; 2003), anddocumented by the archives of thousands ofcuneiform tablets that recorded the merchants

daily business and personal transactions. Inaddition, seals and sealings record the names ofa number of rulers or magistrates both fromAnatolia and the Near East.This so-called Assyrian Colony Period ipAnatolia is conventionally divided into fourphase s, nam ed after the karum levels at Kultepe.Thus from bottom (early) to top (late) the phas-ing is: Karum IV, Karum 111, Karum 11, andKarum Ib and Ia. Not much is know n about thelower two levels because of the minimal exca-

* - **Malcolman d Carolyn Wiener Laboratory for Aegean an d Near E astern Dendrochronology, Cornell University, Ithaca, N.Y14853, U.S.A.


2/12

vation carried out to these depths, and the latterhalf of level I (Ia) encompasses everything fromthe Middle Bronze Age to the present.Therefore, this paper deals only with thechronology of Karum levels I1 (=M ound level 8 )and Ib (=M ound level 7) (T. Ozguc 1999, 77).Assigning a length to th e period of Karum LevelI1 has been aided in the past several years by theidentification of four cuneiform texts excavatedfrom the karum at Kultepe, the so-called limu ,or epon ym , lists. These are lists of the adm inis-trative officials who served annually as the limuor magistrate at the Assyrian capital Azur, andafter each one of whom the year of his admin-istration was nam ed. The most recent Kultepeeponym list, as modified by the publication of along list of these names on a single tablet(Veenhof 20031, now includes 129 names ofofficials who held the post of limu during theperiod of Karum Level 11, more than half a cen-tury longer than had been posited on the basisof the num ber of previously-known limu names(Balkan 1955). Professor Veenhof has proposedan additional 9 eponyms to fill out the KarumLevel I1 phase, for a total of 138 years.One of the difficulties of assigning absolutedates to these years has lain in the absence ofany correlation between the names of the offi-cials and the material remains of the karum.Instead, archaeologists and Assyriologists havebeen struggling to date the years of the limGs'reigns by correlating them with the dates of thereigns of Assyrian kings, Babylonian kings, andlocal kings of the sites in Anatolia in which theprincipal Assyrian merchant colonies were locat-ed, including Kultepe, Aligar, and Bogazkoy.On e presumes that karums were also located atAcemhoyiik and possibly Karahoyiik-Konya,though none have been found in 30 years ofexcavation. The vast majority of the effortexp end ed at the latter two sites has been on themo unds themselves. Professor Veenhof nowdates the Karum I1 period betw een ca. 1974 and1836 BC (absolute dates based on theM esopotam ian Middle Chronology). The e nd of

Karum Level I1 has been attributed to Assyrianking Naram-Sin based on the latest attested bul-lae found in the karum at Kultepe (Ozkan 1993).What is clear is a distinct shift in the archaeo-logical imprint on the karum after a realignm entfrom the Level I1 plan. After a maxim um intervalof perhaps a generation, Karum Level Ib isestablished ca. 1800 and runs to 1730 BC (T.Ozgug 2003, 28). However, not enough limGnames from Level Ib are known to give anychronological dimension to the period from thistype of evidence alone.An alternative set of dates continues to emergefrom another, indepe nde nt, source. This is thedendrochronological dating of a variety of mon-umental buildings from the Assyrian ColonyPeriod in central Anatolia. Although these d atesare not yet absolute (these are floating chronolo-gies, and they will remain floating until theAegean Dendrochronology Project [henceforthADPI can connect them with the long tree-ringsequences from later periods), they are securelyconnected with one another. Lacking a den-drochronological bridge to the present, our dat-ing the tree-ring sequences in absolute time hasrequired the use of a proxy method, namelyradiocarbon wiggle-matching. In the late 1980sthe ADP began a collaboration with Dr. B.Kromer at the Institut fur Umweltphysik at theUniversity of Heidelberg to wiggle-match ourlong dendroch ronological sequen ces in an effortto come up with precise radiocarbon dates-within the limits of the me thod- for all w oo dthat could be connected to two of our longesttree-ring sequences.The Karum I1 Period at Kiikepe,Acemhoytk, and Karahoyiik-KonyaKultepe Karum Level I1 is represented by a 521-year tree-ring chronology, spanning the years2544-2024 BC2, built from the juniper doo r-threshold timbers of rooms in the Eski Saray (T.Ozguc 1999, 106-110 and Plates 45-49; T. Ozguc2003, 133-137) next to a corduroy road of oaklogs from which we have built a 251-year


3/12

A DENDROCHRONOLOGICAL FRAMEWORK IN ASIA MIANOR 167

chronology (not yet date d). Acem hoyiik Earlyis represented by a 508-year chronology builtfrom burned, re-used timbers (Figure 1 andFigure 2) in the foundations of unburned wallsof kitchen structures in the Northwest Trench.Although the kitchen area s period of use wasthe 18th century BC on the basis of seals, seal-ings, small finds, and pottery (A. Oztan, pers.c o r n . ; 1992 and her figures 1-3; 1993 and herfigures 1-2), the 508-year ring-sequence datesfrom 2657 to 2150 BC. Finally, Karahoyiik isrepresented by a 198-year timber taken from thescarp of a de ep sondage in Trench C, Levels 6/7made a generation ago by Professor Sedat Alp.There was no indication from the excavatorwhether or not this sample was part of a wall ofa larger building that can be attributed to theEarly Bronze Age. But the cro ssdating againstboth the Acemhoyiik Early junipers and KultepeEski Saray junipers is exce llent. The rings spa nthe years 2359-2162 BC.

Since these sites are widely separated, there isnothing to suggest a common cause of the bur-ning. The end-dates of the rings span almost acentury and a half, and an accidental conflagra-tion once every 50 years som ewhere in Anatoliais easily conceivable. If a military campaign bysom e agg ressor is a serious possibility, w e nee dto look at the Assyriological record for candi-dates. Since only the Kultepe samples are in aprimary construction context, the latter sitedeserves the most comm ent. The threshold tim-bers of the Eski Saray were cut aro und 2024 BC,probably some while later, and then after anunknown lifespan the Eski Saray was destroyedin a conflagration. Professor Tahsin O z g u ~ ,heexcavator of Kultepe, thinks that the Eski Sarayis contemporary with Karum Level 11. Whetherthe burning up on the mound and down in thekarum is the same burning is anybodys guess.Recently Professor OzguC commented that theincineration of palace and ka rum w as d ue to thesame fire, possibly the attack of Uhna, king ofZalpa (T. Ozguc 2003, 131). At any rate, to

tree at Acem hoyiik. The wiggle-m atch (illustra-ted in Figures 4-6) testing the proposed den-drochronological date with the EBA tree-ringchronology starting in year 2657 is based on 13sets of decadal tree-ring samples fromAcemhoyiik and Karahoyiik, each dated atHeidelberg by Dr. B. Kromer with subsequentanalysis by Dr. S. W. Manning.The end-dates, all t rminipost quos, of the last-preserved rings are therefore as follows:Karahoytik-Konya Early: 2162 BC (no bark,unknown number of rings missing at end, noburning visible in the scarp today); AcemhoyiikEarly: 2150 BC (no bark, unknown number ofrings missing, all partially burned) ; Kultepe Early:2024 BC (no bark, trimmed, unknown numberof rings missing at end, all badly burned).

fully) was almost total, and the combustion leftus little but ash . After years of trying we haveyet to derive a single tree-ring date for anybuilding in the k arum at Kaneg.The Karum Ib Period at Kultepe,Acemhowk, and Karahoyuk-KonyaThis period is much better represented den-drochronologically than Karum Level 11. Allthree sites have one or more major burned mo-numents with long tree-ring sequences, allpinned to our Bronze Age/Iron Age tree-ringchronology which is accurate to within a fewyears (k4/7 years in Manning et al. 2001 at 2 0 ;and less than *16/7 years at 3 ange in Man-ning et al. 2003). Moreover, repair timbers existin two monuments that allow us to make an


4/12

68 M . W. ,YEWTON- P. I . KCA7HOLM

estimate on dendrochronological grounds aloneabout the life-span of each building before itwas destroyed.At Kultepe a large number of timbers in theWarzama Sarayi (T. Ozguq 2003, 120-1251, allpreserving the bark, were cut in 1832 BC. A sec-ond building program took place in the north-west corner of the building in 1810/1808 (barkpreserved). Additional timbers wh ich m7e inter-pret as late repairs were cut as late as 1779 orpossibly later (no bark preserved), indicating aminimum of 61 years for the lifetime of thebuilding before its violent destruc tion som e timeafter 1779. At Acemhoyiik, two major buildings,the Sarikaya Palace and the Hatipler Tepesibuilding, both violently burned (T. Ozguq 2003,126-128, and our Figure 31, were cons tructed inthe same year: 1774 BC (bark preserved in bothbuildings).Two repair timbers in the Sarikaya Palace werecut in 1767 and 1766 or later (no bark pre-served), indicating that it had a lifespan of atleast 8 years. The bulk of the reported 1600 bul-lae in the Sarikaya Palace should have beendeposited there after 1774 and before itsdestruc tion som e time after 1766. Foreign roy-alty whose bullae are found in the SarikayaPalace include King gamgi-Adad of AEur, thePrincess Dugedu , daughter of King Iakhdun-Limof M ari, and King Aplakhanda of Carchemish (T.Ozguq 2003). W hen the sealings from this build-ing are fully published (Nimet Ozguq, in prepa-ration) we should k now m ore abo ut their distri-bution, and possibly how many should beassigned to which years of the buildings life-time, and the Anatolian tree-ring work will havea new set of foreign connotations. At Karahoyiik- Konya, the last-preserved rings (no bark) of yetanother burned building (majority of timbersfrom Room 4 ) in Trench X (Alp 1992; 1993) datefrom after 1768 BC.Again we need to look at the Assyriologicalrecord. Was there a military campaign in the1760s to blame for all this, or are we dealing

with three unrelated destructions of these majormounds? Professor Tahsin Ozguq has recentlysuggested attacks by competing regional kings(T. Ozguc 2003, 132). As a cautionary point, wenote that the wo oden city of Novgorod in Russiawas destroyed by fire o n average o nce every 24years over a six-century period (Kolchin 1963,8 9 , yet there is no evidence whatever in theRussian chronicles for any foreign attack, civilunrest, or the like as a causal factor.Nonetheless, given the historical informationconcerning military activities in the period , thereis more of a case to be made here in the KarumIb period than there was for the Karum I1 peri-od for an event such as a single military cam-paign that might have caused all these destruc-tions at nearly the same time.Comment on our publisheddendrochronological dates for the MBANew articles and commentary by other scholarson the Assyrian Colony Period are appearingpractically bi-monthly, most recently ProfessorKlaas Veenhofs The Old Assyrian List of YearEponyms from Karum Kanish and itsChronological Implications (Ankara, 2003), andProfessor Tahsin Ozguqs Kiilrepe Kani,%/Nek(Tokyo, 20031, and still others are in advancedstages of preparation, such as Professor CahitGunbattis limu text referred to above andProfessor Nimet Ozguqs final reports on boththe seals and sealings from Acemhoyiik as wellas the architecture volume . We therefore feel itnecessary to set the dendrochronolog ical recordstraight so that our colleagues will not inadver-tently cite on e of ou r earlier reports with thepossibly confusing dating systems noted below.We now think that our tree-ring dates, especial-ly for the Middle Bronze Age, are accurate towithin a very few years. Confirmation of allthese dates, of course, will come when theabsolute dendrochronological sequence for theAegean and eastern Mediterranean is extendedfrom the present to the second millennium BC.But if it turns out that we have to move a dateup one year, then everything moves up one


5/12

A DENDROCHRONOLOGICAL FRAMEWORK IN ASIA MLilOR 69

Publication m .Fublilicxtion Date (1989)

year; if we have to move a date down one ortwo Years, then everything rnOveS down one Or

it is the column on the right in bold charactersthat should be cited from now o n. Columns 1-5

Syria W Nahug Skimce Amtipi- -1992) (1993) (1396) 2001) ( 2 0 0 3 )

years, and so forth*We change Onedate without changing all the others. The inter-vals between the dates listed below remain con-

are provided for readers who have seen somebut not all of our earlier publications. Super-

Last preserved ring nla nla 1785+37RC 1746 BC 17&+4/--TRC 17@1771BC

Hittite city wall, 1439 1439 16212378Cinner pos tern (MBARD) (MBARD)constructionHittite city wall, nla 1470 1590+373Couter postem (MBARD)construction

1562 BC 1604+4/--Tf3C 1 )41 yf3C

2551 R C 1573+4 / -mC l575 576BC

Table 11.Our first announ cem ent was in 1989 in A 677Year Tree-Ring Chronology for the MiddleBronze Ag e, in Anatolia and the Ancient NearEast: Studies in Honor of Tahsin Ozgiic (Ankara:Turk Tarih Kurumu). Here all dates wereexpressed in terms of a Middle Bronze AgeRelative Dating system, which we built upon adate assigned for the first measured sample fromKultepe. The ADP relative dating procedures

we re adop ted from the L aboratory for Tree-RingResearch, University of Arizona. The systemarbitrarily assigns the first measured ring of thefirst sample to a year 1001. This allows flexibili-ty for crossdating any tree-ring sample againstthe first on e either to any year up to 1000 yearsbefore, or 1000 years after, the year 1001 Theseyears have always bee n relative, and are in partthe legacy of the days of punch-cards and an


6/12

170 M . W. NEWTON - P. . KUNIHOLM

old computer system wh en the machines couldnot handle negative (BC) numbers. This nolonger applies, but we still maintain a relativedating system for all BC dates that is linked tothe first measured sample from Gordion(Kuniholm 1977).Since the Middle Bronze Age chronologies hadnot (in 1989) been linked with the Gordionchronologies, the tree-ring chronologies fromthe Warxama Sarayi at Kultepe, the SarikayaPalace and Hatipler Tepesi Building at Acem-hoyiik, and the postern gate at Porsuk werelinked to the same MBA relative dating system.Though an absolute date could not then beassigned, the relative dates reported in 1989remain the same as those reported today, withthe Warxama Sarayi at Kultepes being built 58years3 before the two palatial buildings atAcem hoyiik. In addition we reported , in pre-liminary fashion, a 321-year tree-ring chronolo-gy under development for the postern gate atPorsuk (Kuniholm, et.al. 1992). This sequenceextended the Middle Bronze Age tree-ringchronology from Kultepe and Acemhoyiik by113 years o n the recent/lower end , providing atotal of 677 years for what we called the MiddleBronze Age chronology , spanning MBA-RD 763-1439.2. A major tree-ring anomaly at Porsuk -anupw ard spike- the m ost singular anomaly in thelast 9000 years and an apparent reaction to aseries of cool, wet summers (of which morebelow ), was noted in a Preliminary Report onDendrochronological Investigations at Porsuk /Ulukigla, Turkey 1987-1989, in Syria LXIX(1992), 379-389, but at the time of that publica-tion the event was described only in terms ofsignificance based on the number of treesrecording it. Thirty-one trees (then; the total isnow 61) recorded the growth as deviating fromnormal as a positive anomaly of between 167%and 207% of normal in the years, according tothe Middle Bronze Age Relative Chronology,occurring in MBA-RD 1356-1357.

When this chronology was co nnected to the onefrom Gordion via the discovery of exceptionallylong-lived juniper boards used in the construc-tion of the Phrygian tumulus at Kizlarkaya in1991, the relative years for the spike becameGordion MMT-RD 854-855 (see #3 below ). Thisis the positive growth anomaly we would laterpublish in Nature in 1996 (see 4 below) asoccurring in these years, and in that publicationwe correlated these years with the then growingconsensus for a major tree-ring growth anom alyin the northern hemisphere in 1628-1627 BC(also suggested in work published 1984-1995 tobe perhaps correlated with the Thera eruption).This was not the best match for the AD 1996wiggle-match (which was c.1641 BC +/-), butwas chosen for the sim ple reason that it seemedlikely that the Porsuk extraordinary growthanom aly correlated with the other recorded tree-ring growth anomalies around the northernhemisphere and because we thought that whatwe had come to refer to as the PorsukAnomaly was exactly the kind of responseexpected from trees growing in the easternMediterranean after the eruption of Thera. Thefact that Porsuk is situated within the arc of therecorded ash fallout only increased our confi-dence in this connection. In AD 1996 thishypothesis was possible within the then estab-lished dating error on the radiocarbon wiggle-match. But this situation has subsequentlychanged: see 5 below.3. In 1993 w e reported, in a third paper, ADate-List for Bronze Age and Iron AgeMonuments Based on Combined Dendrochro-nological and Radiocarbon Evidence.Aspects of rt and Iconography: Anatolia and ItsNeighbors-Studies in Honor of Nimet Ozguc(Ankara, Turk Tarih Kurumu, 1993), the con-nection of the Middle Bronze Age dendrochro-nology (Middle Bronze Age Relative DatingYears 763-1439) with the Late Bronze Age-IronAge dendrochronology developed from woodfrom Gordion, including the K alarkaya Tumulus(spanning Gordion Relative Dating Years 739-


7/12

A DEIDROCHROA\lOLOGICAL FRAMEWORK IiY ASlA MINOR 171

1647). We converted all of our dendrochrono-logical dates for MBA wood to the GordionRelative Dating system based on the so-calledMidas Mound Tumulus dendrochronology(MMTRD).The 677-year Middle Bronze Age tree-ring series(mostly juniper) now spanned the yearsMMTRD 257-933. It had an overlap with woodof the same species from the KizlarkayaTumulus of 216 years based on the then-current1986 radiocarbon calibration curve which gaveus cutting dates of 1849 BC +/-37 years for theWarzama Sarayi and 1785 BC +/-37 years for theAcemhoyiik buildings. The Porsuk growthanomaly was now back in the 1660s (+/-37), afact immediately noted by Professor F. H.Schweingruber who commented that it was apity that the center of the anomaly did not lineup with one of the proposed eruption dates(1628/1627) for Therahantorini.4. By 1996 not only was the 1993 radiocarboncalibration curve available, but we had enoughradiocarbon dates in hand to report inAnatolian tree rings and the absolute chrono-logy of the eastern M editerranean 2220-718 BC,Nature 381 that we thought the Porsuk anomalyshould be placed at 1641 BC *76/22. Since thiswindow included 1628, we opted for the latter,even though it was 13-odd years lower than thecenter of the chi-squared fit function (in retros-pect an ill-advised move, although it seemedthoroughly reasonable at the time). The effecton the dating of the big MBA buildings was aconstruction date for the WaGama Palace of1810 BC, construction dates for AcemhoyLik of1752 BC, and a date for the Porsuk outerpostern at 1551.5 By 2001 we had not only the 1998 radiocar-bon ca libration curve but also nearly three timesthe number of radiocarbon determinations thathad been available for the Nature article. Themorphology of the fit with the radiocarboncurve showed that the earlier downward place-ment of the Gordion tree-ring chronology was

incorrect. In two articles in Science (Kromer etal., and Manning et al., December 2001, and asupplementary comment by Reimer in the samevolume), w e reported a modification of our pre-vious position, thereby m oving the constructiondates of the Warxama Sarayi up 22 years to circa1832 BC and the Acemhoyiik buildings to circa1774 BC. This time the error margins were rela-tively negligible, plus 4 or minus 7 years at2 o (95.4%) confidence, and the Porsuk anomalymoved u p to a round 1650, n o longer having anyconnection to any 1628BC northern hemispheretree-ring growth anomaly.6. Most recently in a paper in the March, 2003Ant iquiy , Confirmation of near-absolute datingof east Mediterranean Bronze-Iron Dendro-chronology, (available online at http://antiqui-ty.ac.uk/Proj Gal l /manni ng/Manni ng.htm)e report-ed that the likely best-fit margins varied by per-haps 0-3 years within a 30 (99.7%) confidencerange of less than *16/7 calendar years, proba-bly even narrower (see Manning et al. 2001,2535 n.17). While noting that an error rangeapplies (given above and below at 30 confi-dence - see also n.1 above), w e cite in Table 1the specific best-fit (0-3 years variation therein)as the approximate dates that should be used atpresent for the ADP Bronze Age-Iron Agechronology (as of AD 2003). These dates are thecolumn of figures in bold type o n the right. Asnoted, these dates are shown without the errormargins (see next paragraph) that should beused and remembered in any discussion. Thesecurrent best-fit dates are robust, but could movevery slightly if new samples have the rings wecurrently lack, or if minor modifications aremade to the radiocarbon calibration curve itself.7. On the basis of the published exercises inwiggle-matching in Science (2001) and Antiquity(2003) we believe a fit for the last preservedrings, across various scenarios and options, lieswithin the four year span shown in the last co-lumn and that an overall 3 error range of lessthan ?16/7 calendar years, and likely *9/5 ca-lendar years, exists around this 4 year fit zone.


8/12

172 M . W. NEWTON - P. I. KUNIHOLM

This seems to agree well with new, as yetunpublished data and analysis for the late endwhich has been reported to us by our collabo-raters Drs. Kromer and Manning in recentmonths from the work on the East MediterraneanRadiocarbon Intercomparison Project.

Connecting the Two Chronologies

ter if we could find missing rings, thereby push-ing the Eski Saray earlier by a few years.Figures 5 and 6 illustrate the quality of the fit forthe wiggle-matches according to two scenarios.The first is with all existing 13 radiocarbon data(Figure 5), and clearly shows two competingdates for the whole. The second is with the mostsignificant outlier removed (Figure 6), and clear-ly shows a preference for an earlier date for the

The ADP now reports a significant addition onthe early end to the long Bronze Age-Iron Agetree-ring chronology that began with the collec-tion of timbers in the Midas Mound Tumulus atGordion. The 1599-year Bronze-Iron tree-ringchronology as published in Science 2001spanned the years 2247-649 +4/7 BC. We havebeen aware of the long overlap (now 223 years)with the Early Bronze Age master tree-ringchronology for some time, but had been unableto connect the two convincingly because almostall of the overlap depended largely on a 440-year-old, highly erratic juniper timber (KUL-23)from the Wargama Sarayi. We think we havefinally worked out the problems, and the ADPBronze Age-Iron Age tree-ring chronology nowspans the years 2657-649 BC (with the caveatthat there could be surprises in the form of miss-ing rings in the area of the overlap). The den-drochronological linkages are shown in Figure7, with the associated t-scores as a measure ofthe quality of fit. Clearly, the retrieval of addi-tional timbers from the 21st century BC (avai-lable in the Eski Saray at Kultepe) would helpconfirm this placement. Indeed, the discovery ofadditional missing rings in the earliest 150 yearsof KUL-23 or the latest 100 years of KUL-85 andKUL-88 (Eski Saray juniper door thresholds)would improve the quality of both the visualand statistical match by helping align a numberof the tree-ring signatures.However, we do not want to invent ings wecannot actually see in the material currentlyavailable. We note that the radiocarbon wiggle-match in Figure 4, while supporting the pro-posed dendrochronological date, would fit bet-

EBA chronologies. The effect on the datesreported here for the EBA sequences wouldthen require our shifting them up by perhaps asmuch as a decade, increasing the span of timefor the Kultepe Karum Level I1 dendrochrono-logical sequence to 202 years. Work on resolv-ing this discrepancy (both by radiocarbon datingof more samples that are linked dendrochrono-logically in the Early Bronze Age master, and bya sampling strategy to retrieve additional den-drochronological samples from the Eski Saraydoor thresholds and additional fragments ofKUL-23) is ongoing. However, the dendrochro-nological fit between the EBA sequence and theMBA sequence reported here, supported by thetwo sets of wiggle-matches, is the best we canachieve as of December 2003. It should bethought of as tentative, subject to verification ormodification as samples become available in thefuture.ConclusionsThese observations still do not tell us whetherthe karum buildings or the palaces came first.Was the prosperity obvious in the palatial struc-tures on the mound above a by-product of thecommercial activities in the karum below? Ordid the merchants come to an already prosper-ous center? Clearly, if we had datable buildingsin the karum, that would be a big help. A build-ing date of 1832 BC for the Wargama Sarayi islater than the last-preserved ring (2024 BC) ofthe Eski Saray by 192 years. If, say, a half-cen-tury of rings is missing from the latter, does theresulting difference of 142 years have anythingto do with the long limu lists of circa 138 yearsbeing published for the Karum I1 period? We


9/12

A DEATDROCHRO~ \'OLOGICALRAME WORK IAYASIA MIXOR 73

have the impression, simply from the den-drochronological results, that the AssyrianColony Period, at least the last two phases of it,was a longer, more stretched-out affair thansome scholars have been prepared to admit.Acknowledgments:The Malcolm and Carolyn Wiener Laboratory forAegean and Near Eastern Dendrochronology atCornell University is supported by the NationalScience Foundation, the Malcolm H. WienerFoundation, and individual Patrons of theNOTES1 For convenience and clarity we cite absolute dates according to

the exercise published by M anning et.al. in Science in 2001. For amore conservative calculation of the error see Manning et.al.(2003). The error margins stated in the main text ap ply to all theabsolute dates reporied in this paper. especially those in Column

ALP . 1992Konya-K arahoyek 1991 Kazisi , XIV. Kazi Sonuflari Toplantisi.Ankara. T. C. Kultur Bakanligi. 301-306.

Konya-K araho)ak 1992 Kazisi m. Kazi Sonu g h Top lant is i,Ankara. T. C. Kultur Bakanligi. 267-270.Obsemations on the Chronological Problems of the KarumKanic Ankara. Turk Tarih Kurumu..Radiocarbon calibration and ana lysis of stratigraphy the OxCal

program , Radiocarbon 37. 425-430.in preparation. Second long limu text from Karum Level 11Novi Metodi v Archeologii 111: Trudi ,VorgorodskoiArcheologicheskoi Ekspeditsii. Moscow , Akadem ii Xauk.

ALP, . 1993

BALKAN, K. 1955

BRONK RAMSEY, C. 1995

GUNBATIT, c.KOLCHIN, B. A., 1 3

KROMER, B., S. W. MANNING, P. I KUNIHOLM,M. W. NEWTON,M. SPURK, I. LEVI N, 2001

Regional 14CO2 Offsets in the Trop osph ere: Magnitude,Mecha nisms. and Consequenc es . Science 294. 2529-2532.Dendrochronologyat Gordion and on the Anatolian Plateau.Philadelphia unpublished Ph.D. dissertation).A 677 Year Tree-Ring Chronology for the M iddle Bronze Age.

Anatolia and the .\ear East: Studies in H ono r of Tahsin Ozgiic,Ankara. Turk Tarih Kurumu. 371-373.KUNIHOLM, P. I., S L. TARTER,M. W. NEWTON,C. B. GRIGGS, 1992

KUNIHOLM,P. I., 1977

KUNIHOLM, P. I., M. W. NEWTON, 1989

Preliminary Report on Dendrochronological Investigations atPorsukiUlukigla, Turkey 1987-1989 ,Syria LXIX. 379-389.A Date-List for Bronze Age an d Iron Age Mon uments Base d onCombined Dendrochronological and Radiocarbon Evidence .

Aspecrs of An and Iconography: Studies in Honor of X m e tOzguc, Ankara, Turk Tarih Kurumu. 279-293.

KUNIHOLM, P. I., 1993

KUNMOLM, P. I., B. KROMR, S W. MANNING, M. W. NEWTON,C. E. LATIM, M. J. BRUCE, 1996

Anatolian tree rings and the absolute chronology of theeaster n Me diterrane an. 2220-'18 BC . .\ature 381. 730-783.

Aegean Dendrochronology Project. For funda-mental research permissions we thank theappropriate governmental and religious authori-ties in all the countries in which we work, aswell as the many excavators, especially thosecited in this report, who not only take time outto explain the intricacies of their sites but whomake us welcome at their excavation housesyear after year. We owe special thanks to Dr.Bernd Kromer and Dr. Sahra Talamo for radio-carbon dating and Dr. Sturt Manning for statisti-cal commentary and graphical help.

6 in the table below, that are linked dendrochronologically2 See figure and following text for an explanation for these and the

following dates)3 [Sid We reported the difference as both 58 (p.279) and 68 (page293) years We thank Professor Klaas Veenhof for noting the dis-crepancy and communicating it to us.

MANNING, S. W., B. KROMER, P. I. KUNMOLM,M. W. NEWTON, 2001ha to li an Tree Rings and a N ew Chronology for the EastMediterranean B ronze-Iron Ages . Science 294. 2532-2535.MANNING, S. W., B. KROMER, P. I. KUNIHOLM,M.W. NEWTON, 2003Confirmation of near-absolute dating of east Mediterranean

Bronze-Iron Dendrochronology , Antiquity 77. No 295 ;hlarch.http:i /antiqui~,ac.uMProjGalL'Mannin~manning,tmlThe Seal Impressions of Tw o Old Assyrian Kings , Aspects of

An and Iconography: Studies in Honor of Ximet Ozguc,Ankara, Turk Tarih Kurumu, 501-502.

dZKAN, S., 1993

&TAN, A.1992 1991 Yili Acemhoyiik Kazilari . XIV. Kazi Sonuglan Toplantisi.1993 1992 Yili Acemho)iik KazilarC. XI'. Kazi Sonu & Toplantisi.bZGUC N.

Ankara, T. C. Kultur Bakanligi. 281-300.Ankara. T. C. Kultur Bakanligi, 245-254.in preparation. Final reports on the bullae and architecture ofAcemhoyiik.Kultepe-Kanis II:Kew Researches a t rhe Trading Center of theAncient .Year East, Ankara. Turk Tarih KurumuThe Palaces and Temples ofKiiltepe-KaniF'&e&, Ank ara. TurkTarih Kururnu.Kiilrepe KaniTLVe&: The earliest international trade cente r an dthe oldest capital city of he Hittites. Tokyo , MECCJ.A Nevi Twist in the Radiocarbon Ta le , Science 294, 2494-249

OZGUC, T., 19%

OZGUC, T., 19

OZGUC, T., 2003

REUIER, P. J., 2001STUIVER, M., PJ. FEIMER, E. BARD, J.W. BECK, G.S. BURR,K.A. HUGHEN, B. KROMER, G . MCCORMAC,J. van der PLICHT,M. SPURK, 1998INTCAL98 radiocarbon age calibration. 24,000-0 cal BP. .

Radiocarbon, 40, 1041 1083,Th e Old Assyrian List of Year Eponym s from Karum Kanish an dits Chronological Implications, Ankara, Turk Tarih Kurumu.

VEENHOF, K., 2003


10/12

174 M . W NEWTON - P. I. KUNIHOLM

Fig. 1: Partially burned EBA juniper timber at Acemhoyiik.Photograph courtesy A. Oztan.

Fig. : Partially burned EBA juniper timbers atAcemhoyiik. Photograph courtesy A. Oztan

INTCAL98KBKl7MACM 90

Relative Years

Cal BC Yea rs

I

Fig. 3: Wiggle-matchfor Acemhoyiik Early and Karahoyiik Early.


11/12

A DENDROCHRONOLOGICM FRAMEWORK IN ASLA MINOR

Fig. 4:Burned juniper foundation timbers in one room of the Hatipler Tepesi buildingat Acemhoyiik. All were cut in the same year, 1774 BC.

1 o0.80.6

First First68.2% probability2669.7BC (36.3 ) 2661.7E3C2657.T;rBC(31.9 ) 2651BC2705.2BC 1.9 )2701.3BC2477.7BC (93.5 ) 2448.6BC

95.4 probability

9.7 probability2708.8BC (4.6% ) 2688.8BC2685BC (95.1 )2646.7BC0.40.20.0

2750BC 2650BCCalendar date

6UUBC

75

Fig. 5: The statistical f it for the EBA Wiggle-Match using all 13 data.


12/12

176

t=7.6 t=5.4

M . W. NEWTON - P. I. KUNIHOLM

t=6.8

0.80.60.49.29.0

L +Kulteoe Wargamabaray (without KUL-23)

First irst68.2% probability

99.7 probability2705.4BC (99.7 ) 2646.9BC

I I I I I I I I I I I I I I I I I l l

2750BC 2700BC 2659BC 26QOBCCalendar date

Fig. 6: The statistical fi t for the EBA Wiggle Match using only 12 data, with the biggest outlier removed.

t=4.1KulteDe Eski SaravKarahtivuk-FonvaV Earlv t=6.8

Acemhtiyuk Ear ly (NW T re n c I~ )) = ~ ~ t=3.4

t=2.6EBA Jun iper Chrono logy

-2700 -2500 -2300 -2100Calendar Years BC -19 -1700

Fig. 7: Aegean Dendrochronology Project EBA and MBA sample spread and depth, for Anatolian juniper only, 27th-17thcenturies BC

Documents

A Dendrochronological Framework